| author | Claus Gittinger <cg@exept.de> |
| Thu, 23 Feb 2006 19:42:30 +0100 | |
| changeset 9191 | 7f1797f8d0a3 |
| parent 9190 | 87dd824f1ebb |
| child 9192 | 2fb261f753f8 |
| permissions | -rw-r--r-- |
| 64 | 1 |
" |
2 |
COPYRIGHT (c) 1994 by Claus Gittinger |
|
| 9191 | 3 |
All Rights Reserved |
| 64 | 4 |
|
5 |
This software is furnished under a license and may be used |
|
6 |
only in accordance with the terms of that license and with the |
|
7 |
inclusion of the above copyright notice. This software may not |
|
8 |
be provided or otherwise made available to, or used by, any |
|
9 |
other person. No title to or ownership of the software is |
|
10 |
hereby transferred. |
|
11 |
" |
|
12 |
||
|
5342
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
13 |
"{ Package: 'stx:libbasic' }"
|
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
14 |
|
| 3926 | 15 |
ByteArray variableByteSubclass:#CharacterArray |
| 9191 | 16 |
instanceVariableNames:'' |
17 |
classVariableNames:'PreviousMatch DecoderTables EncoderTables DecodingFailedSignal |
|
18 |
EncodingFailedSignal' |
|
19 |
poolDictionaries:'' |
|
20 |
category:'Collections-Text' |
|
| 64 | 21 |
! |
22 |
||
|
1849
9f35f5934ec6
added msdos-latin1 en/decoder (code = #msdos850)
Claus Gittinger <cg@exept.de>
parents:
1669
diff
changeset
|
23 |
!CharacterArray class methodsFor:'documentation'! |
| 64 | 24 |
|
| 82 | 25 |
copyright |
26 |
" |
|
27 |
COPYRIGHT (c) 1994 by Claus Gittinger |
|
| 9191 | 28 |
All Rights Reserved |
| 82 | 29 |
|
30 |
This software is furnished under a license and may be used |
|
31 |
only in accordance with the terms of that license and with the |
|
32 |
inclusion of the above copyright notice. This software may not |
|
33 |
be provided or otherwise made available to, or used by, any |
|
34 |
other person. No title to or ownership of the software is |
|
35 |
hereby transferred. |
|
36 |
" |
|
37 |
! |
|
38 |
||
| 64 | 39 |
documentation |
40 |
" |
|
| 255 | 41 |
CharacterArray is a superclass for all kinds of Strings (i.e. |
| 68 | 42 |
(singleByte-)Strings, TwoByteStrings and whatever comes in the future. |
| 154 | 43 |
|
| 8874 | 44 |
This class is abstract, meaning that there are no instances of it. |
| 2537 | 45 |
All this class does is provide common protocol for concrete subclasses. |
| 1290 | 46 |
|
47 |
[author:] |
|
| 9191 | 48 |
Claus Gittinger |
| 1290 | 49 |
|
50 |
[see also:] |
|
| 9191 | 51 |
String TwoByteString |
52 |
StringCollection |
|
| 64 | 53 |
" |
54 |
! ! |
|
55 |
||
|
2728
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
56 |
!CharacterArray class methodsFor:'initialization'! |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
57 |
|
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
58 |
initialize |
| 8874 | 59 |
DecodingFailedSignal isNil ifTrue:[ |
| 9191 | 60 |
DecodingFailedSignal := DecodingError. |
61 |
DecodingFailedSignal notifierString:'error during decode'. |
|
62 |
||
63 |
EncodingFailedSignal :=EncodingError. |
|
64 |
EncodingFailedSignal notifierString:'error during encode'. |
|
| 5440 | 65 |
] |
|
2728
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
66 |
|
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
67 |
" |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
68 |
CharacterArray initialize |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
69 |
" |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
70 |
|
| 2833 | 71 |
"Modified: 3.8.1997 / 18:15:59 / cg" |
|
2728
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
72 |
! ! |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
73 |
|
|
1849
9f35f5934ec6
added msdos-latin1 en/decoder (code = #msdos850)
Claus Gittinger <cg@exept.de>
parents:
1669
diff
changeset
|
74 |
!CharacterArray class methodsFor:'instance creation'! |
| 64 | 75 |
|
76 |
basicNew |
|
77 |
"return a new empty string" |
|
78 |
||
79 |
^ self basicNew:0 |
|
80 |
! |
|
81 |
||
| 2734 | 82 |
fromBytes:aByteCollection |
83 |
"return an instance of the receiver class, |
|
84 |
taking untranslated bytes from the argument, aByteCollection. |
|
85 |
Only useful, when reading twoByteStrings from external sources." |
|
86 |
||
87 |
|mySize nBytes newString dstIdx| |
|
88 |
||
89 |
nBytes := aByteCollection size. |
|
90 |
mySize := self basicNew bitsPerCharacter. |
|
91 |
mySize == 16 ifTrue:[ |
|
| 9191 | 92 |
newString := self basicNew:(nBytes // 2). |
93 |
dstIdx := 1. |
|
94 |
aByteCollection pairWiseDo:[:hi :lo | |
|
95 |
newString at:dstIdx put:(Character value:(hi bitShift:8)+lo). |
|
96 |
dstIdx := dstIdx + 1 |
|
97 |
]. |
|
98 |
^ newString. |
|
| 2734 | 99 |
]. |
100 |
||
101 |
^ (self basicNew:nBytes) replaceFrom:1 with:aByteCollection |
|
102 |
||
103 |
" |
|
104 |
TwoByteString fromBytes:#[16r21 16r21] |
|
105 |
" |
|
106 |
||
107 |
"Modified: 30.6.1997 / 20:08:37 / cg" |
|
108 |
! |
|
109 |
||
| 64 | 110 |
fromString:aString |
111 |
"return a copy of the argument, aString" |
|
112 |
||
| 68 | 113 |
^ (self basicNew:(aString size)) replaceFrom:1 with:aString |
114 |
||
115 |
"TwoByteString fromString:'hello'" |
|
| 581 | 116 |
! |
117 |
||
118 |
fromStringCollection:aCollectionOfStrings |
|
119 |
"return new string formed by concatenating a copy of the argument, aString" |
|
120 |
||
121 |
^ self fromStringCollection:aCollectionOfStrings separatedBy:'' |
|
122 |
||
123 |
" |
|
124 |
String fromStringCollection:#('hello' 'world' 'how' 'about' 'this')
|
|
125 |
" |
|
126 |
||
127 |
"Created: 20.11.1995 / 15:26:59 / cg" |
|
128 |
! |
|
129 |
||
130 |
fromStringCollection:aCollectionOfStrings separatedBy:aSeparatorString |
|
131 |
"return new string formed by concatenating a copy of the argument, aString" |
|
132 |
||
133 |
|newString first| |
|
134 |
||
135 |
newString := ''. |
|
136 |
first := true. |
|
| 8874 | 137 |
aCollectionOfStrings do:[:s | |
| 9191 | 138 |
first ifFalse:[ |
139 |
newString := newString , aSeparatorString |
|
140 |
] ifTrue:[ |
|
141 |
first := false |
|
142 |
]. |
|
143 |
newString := newString , s |
|
| 581 | 144 |
]. |
145 |
^ newString |
|
146 |
||
147 |
" |
|
148 |
String fromStringCollection:#('hello' 'world' 'how' 'about' 'this') separatedBy:' '
|
|
149 |
" |
|
150 |
||
151 |
"Created: 20.11.1995 / 15:32:17 / cg" |
|
| 608 | 152 |
! |
153 |
||
| 5757 | 154 |
fromUTF8Bytes:aByteCollection |
| 5758 | 155 |
"return a new string which represents the characters as decoded |
156 |
from the utf8 encoded bytes, aByteCollection. |
|
157 |
Returns either a normal String, or a TwoByteString instance. |
|
| 6810 | 158 |
Only useful, when reading twoByteStrings from external sources. |
159 |
This only handles up-to 16bit characters" |
|
| 5757 | 160 |
|
| 7876 | 161 |
^ self decodeFromUTF8:aByteCollection. |
| 5757 | 162 |
|
163 |
" |
|
| 8874 | 164 |
CharacterArray fromUTF8Bytes:#[ 16r41 16r42 ] |
165 |
CharacterArray fromUTF8Bytes:#[ 16rC1 16r02 ] |
|
166 |
CharacterArray fromUTF8Bytes:#[ 16rE0 16r81 16r02 ] |
|
| 6810 | 167 |
CharacterArray fromUTF8Bytes:#[ 16rEF 16rBF 16rBF ] |
168 |
||
169 |
rfc2279 examples: |
|
| 8874 | 170 |
CharacterArray fromUTF8Bytes:#[ 16r41 16rE2 16r89 16rA2 16rCE 16r91 16r2E ] |
171 |
CharacterArray fromUTF8Bytes:#[ 16rED 16r95 16r9C 16rEA 16rB5 16rAD 16rEC 16r96 16rB4 ] |
|
172 |
CharacterArray fromUTF8Bytes:#[ 16rE6 16r97 16rA5 16rE6 16r9C 16rAC 16rE8 16rAA 16r9E ] |
|
| 6810 | 173 |
|
174 |
invalid: |
|
| 8874 | 175 |
CharacterArray fromUTF8Bytes:#[ 16rC0 16r80 ] |
176 |
CharacterArray fromUTF8Bytes:#[ 16rE0 16r80 16r80 ] |
|
| 5757 | 177 |
" |
178 |
! |
|
179 |
||
| 608 | 180 |
new |
181 |
"return a new empty string" |
|
182 |
||
183 |
^ self basicNew:0 |
|
| 64 | 184 |
! ! |
185 |
||
|
2728
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
186 |
!CharacterArray class methodsFor:'Signal constants'! |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
187 |
|
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
188 |
decodingFailedSignal |
| 2833 | 189 |
"return the signal, raised when decoding of a string is not possible |
|
2728
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
190 |
due to invalid characters contained in the source. |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
191 |
This may happen for example, if a non EUC coded 8-bit string |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
192 |
is attempted to be decoded into a JIS string." |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
193 |
|
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
194 |
^ DecodingFailedSignal |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
195 |
|
| 2833 | 196 |
"Created: 28.6.1997 / 20:09:55 / cg" |
197 |
"Modified: 3.8.1997 / 18:16:47 / cg" |
|
198 |
! |
|
199 |
||
200 |
encodingFailedSignal |
|
201 |
"return the (query-) signal, raised when encoding of a string is not possible |
|
202 |
due to invalid characters contained in the source." |
|
203 |
||
204 |
^ EncodingFailedSignal |
|
205 |
||
|
2728
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
206 |
"Modified: 28.6.1997 / 20:09:35 / cg" |
| 2833 | 207 |
"Created: 3.8.1997 / 18:16:40 / cg" |
|
2728
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
208 |
! ! |
|
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
209 |
|
|
1849
9f35f5934ec6
added msdos-latin1 en/decoder (code = #msdos850)
Claus Gittinger <cg@exept.de>
parents:
1669
diff
changeset
|
210 |
!CharacterArray class methodsFor:'cleanup'! |
|
991
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
211 |
|
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
212 |
lowSpaceCleanup |
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
213 |
"cleanup in low-memory situations" |
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
214 |
|
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
215 |
DecoderTables := EncoderTables := nil |
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
216 |
|
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
217 |
" |
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
218 |
CharacterArray lowSpaceCleanup |
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
219 |
" |
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
220 |
|
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
221 |
"Created: 22.2.1996 / 16:30:30 / cg" |
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
222 |
"Modified: 22.2.1996 / 17:58:05 / cg" |
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
223 |
! ! |
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
224 |
|
| 7421 | 225 |
!CharacterArray class methodsFor:'encoding & decoding'! |
|
1202
200f7cd0a457
moved encode/decode stuff into class methods / added big5/gb support
Claus Gittinger <cg@exept.de>
parents:
1145
diff
changeset
|
226 |
|
| 7876 | 227 |
decodeFromUTF8:aStringOrByteCollection |
228 |
"return a string which represents the characters as decoded |
|
229 |
from the utf8 encoded bytes, aByteCollection. |
|
230 |
Returns either a normal String, or a TwoByteString instance. |
|
231 |
Only useful, when reading twoByteStrings from external sources. |
|
232 |
This only handles up-to 16bit characters." |
|
233 |
||
|
8124
ca774158b317
code cleanup (removed old obsolete encoding stuff)
Claus Gittinger <cg@exept.de>
parents:
8102
diff
changeset
|
234 |
^ CharacterEncoderImplementations::ISO10646_to_UTF8 decodeString:aStringOrByteCollection |
| 7876 | 235 |
|
236 |
" |
|
| 8874 | 237 |
CharacterArray fromUTF8Bytes:#[ 16r41 16r42 ] |
238 |
CharacterArray fromUTF8Bytes:#[ 16rC1 16r02 ] |
|
239 |
CharacterArray fromUTF8Bytes:#[ 16rE0 16r81 16r02 ] |
|
| 7876 | 240 |
CharacterArray fromUTF8Bytes:#[ 16rEF 16rBF 16rBF ] |
241 |
||
242 |
rfc2279 examples: |
|
| 8874 | 243 |
CharacterArray fromUTF8Bytes:#[ 16r41 16rE2 16r89 16rA2 16rCE 16r91 16r2E ] |
244 |
CharacterArray fromUTF8Bytes:#[ 16rED 16r95 16r9C 16rEA 16rB5 16rAD 16rEC 16r96 16rB4 ] |
|
245 |
CharacterArray fromUTF8Bytes:#[ 16rE6 16r97 16rA5 16rE6 16r9C 16rAC 16rE8 16rAA 16r9E ] |
|
| 7876 | 246 |
|
247 |
invalid: |
|
| 8874 | 248 |
CharacterArray fromUTF8Bytes:#[ 16rC0 16r80 ] |
249 |
CharacterArray fromUTF8Bytes:#[ 16rE0 16r80 16r80 ] |
|
| 7876 | 250 |
" |
| 443 | 251 |
! ! |
252 |
||
|
1849
9f35f5934ec6
added msdos-latin1 en/decoder (code = #msdos850)
Claus Gittinger <cg@exept.de>
parents:
1669
diff
changeset
|
253 |
!CharacterArray class methodsFor:'pattern matching'! |
| 64 | 254 |
|
|
5342
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
255 |
matchEscapeCharacter |
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
256 |
"return the character used to escape a matchCharacter |
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
257 |
(i.e. make it a regular character in a matchPattern)" |
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
258 |
|
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
259 |
^ $\ |
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
260 |
! |
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
261 |
|
| 327 | 262 |
matchScan:matchScanArray from:matchStart to:matchStop with:aString from:start to:stop ignoreCase:ignoreCase |
263 |
"helper for match; return true if the characters from start to stop in |
|
264 |
aString are matching the scan in matchScan from matchStart to matchStop. |
|
| 328 | 265 |
The matchScan is as created by asMatchScanArray. |
| 327 | 266 |
|
| 8874 | 267 |
This algorithm is not at all the most efficient; |
268 |
for heavy duty pattern matching, an interface (primitive) to the regex |
|
| 328 | 269 |
pattern matching package should be added." |
| 327 | 270 |
|
| 8874 | 271 |
|matchEntry |
| 327 | 272 |
mStart "{ Class: SmallInteger }"
|
273 |
mStop "{ Class: SmallInteger }"
|
|
274 |
sStart "{ Class: SmallInteger }"
|
|
275 |
sStop "{ Class: SmallInteger }"
|
|
276 |
mSize "{ Class: SmallInteger }"
|
|
277 |
sSize "{ Class: SmallInteger }"
|
|
278 |
index "{ Class: SmallInteger }"
|
|
| 2561 | 279 |
quickCheck matchLast nextMatchEntry |
| 986 | 280 |
checkChar included| |
| 327 | 281 |
|
282 |
mStart := matchStart. |
|
283 |
mStop := matchStop. |
|
284 |
sStart := start. |
|
285 |
sStop := stop. |
|
286 |
||
287 |
[true] whileTrue:[ |
|
| 8874 | 288 |
"/ Transcript showCR:('match: ''' , (aString copyFrom:sStart to:sStop) ,
|
| 3537 | 289 |
"/ ''' against:' , (matchScanArray copyFrom:mStart to:mStop) printString). |
290 |
||
| 9191 | 291 |
mSize := mStop - mStart + 1. |
292 |
sSize := sStop - sStart + 1. |
|
293 |
||
294 |
"empty strings match" |
|
295 |
(mSize == 0) ifTrue:[^ (sSize == 0)]. |
|
296 |
||
297 |
matchEntry := matchScanArray at:mStart. |
|
298 |
||
299 |
"/ the most common case first: |
|
300 |
(sSize ~~ 0 |
|
301 |
and:[(checkChar := (aString at:sStart)) = matchEntry]) ifTrue:[ |
|
302 |
"advance by one and continue" |
|
303 |
mStart := mStart + 1. |
|
304 |
sStart := sStart + 1 |
|
305 |
] ifFalse:[ |
|
306 |
(matchEntry == #any) ifTrue:[ |
|
307 |
"restString empty -> no match" |
|
308 |
(sSize == 0) ifTrue:[^ false]. |
|
309 |
"# matches single character" |
|
310 |
((sSize == 1) and:[mSize == 1]) ifTrue:[^ true]. |
|
311 |
"advance by one and continue" |
|
312 |
mStart := mStart + 1. |
|
313 |
sStart := sStart + 1 |
|
314 |
] ifFalse:[ |
|
315 |
(matchEntry == #anyString) ifTrue:[ |
|
316 |
"* alone matches anything" |
|
317 |
(mSize == 1) ifTrue:[^ true]. |
|
318 |
"restString empty & matchString not empty -> no match" |
|
319 |
(sSize == 0) ifTrue:[^ false]. |
|
320 |
||
321 |
" |
|
322 |
try to avoid some of the recursion by checking last |
|
323 |
character and continue with shortened strings if possible |
|
324 |
" |
|
325 |
quickCheck := false. |
|
326 |
(mStop >= mStart) ifTrue:[ |
|
327 |
matchLast := matchScanArray at:mStop. |
|
328 |
(matchLast ~~ #anyString) ifTrue:[ |
|
329 |
(matchLast == #any) ifTrue:[ |
|
330 |
quickCheck := true |
|
331 |
] ifFalse:[ |
|
332 |
matchLast == (aString at:sStop) ifTrue:[ |
|
333 |
quickCheck := true |
|
334 |
] ifFalse:[ |
|
335 |
matchLast isString ifTrue:[ |
|
336 |
quickCheck := matchLast includes:(aString at:sStop) |
|
337 |
] |
|
338 |
] |
|
339 |
] |
|
340 |
] |
|
341 |
]. |
|
342 |
quickCheck ifTrue:[ |
|
343 |
" |
|
344 |
quickCheck ok, advance from the right |
|
345 |
" |
|
346 |
mStop := mStop - 1. |
|
347 |
sStop := sStop - 1 |
|
348 |
] ifFalse:[ |
|
349 |
"/ no quick check; |
|
350 |
"/ look for the next character(s) |
|
351 |
"/ and try matching there |
|
352 |
"/ (to avoid recursion) |
|
353 |
||
354 |
mStart < mStop ifTrue:[ |
|
355 |
nextMatchEntry := matchScanArray at:mStart+1. |
|
356 |
nextMatchEntry isCharacter ifTrue:[ |
|
357 |
sStart <= sStop ifTrue:[ |
|
358 |
[true] whileTrue:[ |
|
359 |
ignoreCase ifFalse:[ |
|
360 |
index := aString indexOf:nextMatchEntry startingAt:sStart |
|
361 |
] ifTrue:[ |
|
362 |
index := aString findFirst:[:c | c asLowercase = nextMatchEntry asLowercase] |
|
363 |
startingAt:sStart. |
|
364 |
]. |
|
365 |
(index == 0 or:[index > sStop]) ifTrue:[ |
|
366 |
^ false |
|
367 |
]. |
|
368 |
(self matchScan:matchScanArray |
|
369 |
from:(mStart + 1) |
|
370 |
to:mStop |
|
371 |
with:aString |
|
372 |
from:index |
|
373 |
to:sStop |
|
374 |
ignoreCase:ignoreCase |
|
375 |
) ifTrue:[ |
|
376 |
^ true |
|
377 |
]. |
|
378 |
sStart := index + 1. |
|
379 |
] |
|
380 |
] |
|
381 |
] |
|
382 |
]. |
|
383 |
||
384 |
" |
|
385 |
no quick check possible; |
|
386 |
loop over all possible substrings |
|
387 |
" |
|
388 |
index := sStart. |
|
389 |
[index <= sStop] whileTrue:[ |
|
390 |
(self matchScan:matchScanArray |
|
391 |
from:(mStart + 1) |
|
392 |
to:mStop |
|
393 |
with:aString |
|
394 |
from:index |
|
395 |
to:sStop |
|
396 |
ignoreCase:ignoreCase |
|
397 |
) ifTrue:[ |
|
398 |
^ true |
|
399 |
]. |
|
400 |
index := index + 1 |
|
401 |
]. |
|
402 |
^ false |
|
403 |
]. |
|
404 |
] ifFalse:[ |
|
405 |
(matchEntry isString) ifTrue:[ |
|
406 |
"testString empty -> no match" |
|
407 |
(sSize == 0) ifTrue:[^ false]. |
|
408 |
||
409 |
included := false. |
|
410 |
"/ checkChar := aString at:sStart. |
|
411 |
included := matchEntry includes:checkChar. |
|
412 |
included ifFalse:[ |
|
413 |
ignoreCase ifTrue:[ |
|
414 |
checkChar isUppercase ifTrue:[ |
|
415 |
included := matchEntry includes:checkChar asLowercase. |
|
416 |
] ifFalse:[ |
|
417 |
included := matchEntry includes:checkChar asUppercase. |
|
418 |
] |
|
419 |
]. |
|
420 |
]. |
|
421 |
mStart := mStart + 1. |
|
422 |
mSize := mSize - 1. |
|
423 |
included ifFalse:[^ false]. |
|
424 |
||
425 |
((sSize == 1) and:[mSize == 0]) ifTrue:[^ true]. |
|
426 |
"cut off 1st char and continue" |
|
427 |
sStart := sStart + 1 |
|
428 |
] ifFalse:[ |
|
429 |
"/ must be single character |
|
430 |
||
431 |
"testString empty ?" |
|
432 |
(sSize == 0) ifTrue:[^ false]. |
|
433 |
||
434 |
"first characters equal ?" |
|
435 |
"/ checkChar := aString at:sStart. |
|
436 |
ignoreCase ifFalse:[^ false]. |
|
437 |
(checkChar asUppercase ~= matchEntry asUppercase) ifTrue:[^ false]. |
|
438 |
||
439 |
"advance and continue" |
|
440 |
mStart := mStart + 1. |
|
441 |
sStart := sStart + 1 |
|
442 |
] |
|
443 |
] |
|
444 |
] |
|
445 |
] |
|
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
446 |
]. |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
447 |
|
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
448 |
" |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
449 |
|scanArray s| |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
450 |
|
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
451 |
scanArray := self matchScanArrayFrom:'*hello'. |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
452 |
s := 'foo bar hello world'. |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
453 |
CharacterArray |
| 9191 | 454 |
matchScan:scanArray |
455 |
from:1 |
|
456 |
to:scanArray size |
|
457 |
with:s |
|
458 |
from:1 |
|
459 |
to:s size |
|
460 |
ignoreCase:false |
|
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
461 |
" |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
462 |
" |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
463 |
|scanArray s| |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
464 |
|
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
465 |
scanArray := self matchScanArrayFrom:'*hello*'. |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
466 |
s := 'foo bar hello world'. |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
467 |
CharacterArray |
| 9191 | 468 |
matchScan:scanArray |
469 |
from:1 |
|
470 |
to:scanArray size |
|
471 |
with:s |
|
472 |
from:1 |
|
473 |
to:s size |
|
474 |
ignoreCase:false |
|
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
475 |
" |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
476 |
|
| 3879 | 477 |
"Modified: / 15.10.1998 / 13:39:25 / cg" |
| 608 | 478 |
! |
479 |
||
480 |
matchScanArrayFrom:aString |
|
481 |
"scan a pattern string and decompose it into a scanArray. |
|
482 |
This is processed faster (especially with character ranges), and |
|
483 |
can also be reused later. (if the same pattern is to be searched again)" |
|
484 |
||
| 8874 | 485 |
|coll |
| 608 | 486 |
idx "{ Class: SmallInteger }"
|
|
5342
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
487 |
end c1 c2 matchSet previous escape| |
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
488 |
|
| 5487 | 489 |
previous := nil. |
|
5342
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
490 |
escape := self matchEscapeCharacter. |
| 608 | 491 |
|
492 |
coll := OrderedCollection new. |
|
493 |
idx := 1. end := aString size. |
|
494 |
[idx <= end] whileTrue:[ |
|
| 9191 | 495 |
|char this| |
496 |
||
497 |
char := aString at:idx. |
|
498 |
char == $* ifTrue:[ |
|
499 |
previous ~~ #anyString ifTrue:[ |
|
500 |
this := #anyString |
|
501 |
] |
|
502 |
] ifFalse:[ |
|
503 |
char == $# ifTrue:[ |
|
504 |
previous ~~ #anyString ifTrue:[ |
|
505 |
this := #any |
|
506 |
] |
|
507 |
] ifFalse:[ |
|
508 |
char == $[ ifTrue:[ |
|
509 |
matchSet := IdentitySet new. |
|
510 |
idx := idx + 1. |
|
511 |
idx > end ifTrue:[^ nil]. |
|
512 |
char := aString at:idx. |
|
513 |
c1 := nil. |
|
514 |
[char ~~ $]] whileTrue:[ |
|
515 |
((char == $-) and:[c1 notNil]) ifTrue:[ |
|
516 |
idx := idx + 1. |
|
517 |
idx > end ifTrue:[^ nil]. |
|
518 |
c2 := aString at:idx. |
|
519 |
c1 to:c2 do:[:c | matchSet add:c]. |
|
520 |
c1 := nil. |
|
521 |
idx := idx + 1. |
|
522 |
] ifFalse:[ |
|
523 |
(char ~~ $]) ifTrue:[ |
|
524 |
matchSet add:char. |
|
525 |
c1 := char. |
|
526 |
idx := idx + 1 |
|
527 |
] |
|
528 |
]. |
|
529 |
idx > end ifTrue:[^ nil]. |
|
530 |
char := aString at:idx |
|
531 |
]. |
|
532 |
this := matchSet asString |
|
533 |
] ifFalse:[ |
|
534 |
char == escape ifTrue:[ |
|
535 |
idx := idx + 1. |
|
536 |
idx > end ifTrue:[ |
|
537 |
"/ mhmh - what should we do here ? |
|
538 |
this := char |
|
539 |
] ifFalse:[ |
|
540 |
this := aString at:idx. |
|
541 |
] |
|
542 |
] ifFalse:[ |
|
543 |
this := char |
|
544 |
] |
|
545 |
] |
|
546 |
] |
|
547 |
]. |
|
548 |
this notNil ifTrue:[coll add:this. previous := this]. |
|
549 |
idx := idx + 1 |
|
| 608 | 550 |
]. |
551 |
||
552 |
^ coll asArray |
|
553 |
||
554 |
" |
|
| 8874 | 555 |
String matchScanArrayFrom:'*ute*' |
556 |
String matchScanArrayFrom:'**ute**' |
|
557 |
String matchScanArrayFrom:'*uter' |
|
558 |
String matchScanArrayFrom:'\*uter' |
|
559 |
String matchScanArrayFrom:'[cC]#mpute[rR]' |
|
560 |
String matchScanArrayFrom:'[abcd]*' |
|
561 |
String matchScanArrayFrom:'[a-k]*' |
|
562 |
String matchScanArrayFrom:'*some*compl*ern*' |
|
563 |
String matchScanArrayFrom:'[a-' |
|
564 |
String matchScanArrayFrom:'[a-zA-Z]' |
|
565 |
String matchScanArrayFrom:'[a-z01234A-Z]' |
|
| 608 | 566 |
" |
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
567 |
|
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
568 |
"Modified: 2.4.1997 / 16:20:29 / cg" |
| 9190 | 569 |
]. |
| 6870 | 570 |
^ ds contents. |
571 |
! |
|
572 |
||
573 |
formatWith:aString |
|
574 |
"same as #bindWith: for dolphin compatibility" |
|
575 |
||
576 |
^ self bindWith:aString |
|
577 |
||
578 |
" |
|
579 |
'hello%1world' formatWith:'123' |
|
580 |
" |
|
581 |
! |
|
582 |
||
583 |
formatWith:arg1 with:arg2 |
|
584 |
"same as #bindWith: for dolphin compatibility" |
|
585 |
||
586 |
^ self bindWith:arg1 with:arg2 |
|
587 |
||
588 |
" |
|
589 |
'hello%1 %2world' formatWith:'123' with:234 |
|
590 |
" |
|
591 |
! |
|
592 |
||
593 |
formatWith:arg1 with:arg2 with:arg3 |
|
594 |
"same as #bindWith: for dolphin compatibility" |
|
595 |
||
| 9190 | 596 |
^ self bindWitcopyExpanding:expandTable |
597 |
"return a copy of myself, with translations from the expandTable sliced in. |
|
598 |
The argument is supposed to map from characters to either characters or strings." |
|
599 |
||
600 |
|ds| |
|
601 |
||
602 |
ds := WriteStream on:(self species new). |
|
603 |
self do:[:eachChar | |
|
| 9191 | 604 |
|repl| |
605 |
||
606 |
repl := expandTable at:eachChar ifAbsent:nil. |
|
607 |
repl isNil ifTrue:[ |
|
608 |
ds nextPut:eachChar |
|
609 |
] ifFalse:[ |
|
610 |
repl size == 0 ifTrue:[ |
|
611 |
ds nextPut:repl |
|
612 |
] ifFalse:[ |
|
613 |
ds nextPutAll:repl |
|
614 |
] |
|
615 |
]. |
|
| 9190 | 616 |
]. |
617 |
^ ds contents. |
|
618 |
! |
|
619 |
||
620 |
formatWith:arg1 with:arg2 with:arg3 |
|
621 |
"same as #bindWith: for dolphin compatibility" |
|
622 |
||
| 6870 | 623 |
^ self bindWith:arg1 with:arg2 with:arg3 |
624 |
||
625 |
" |
|
626 |
'hello%1 %2 %3world' formatWith:'123' with:234 with:345 |
|
627 |
" |
|
| 6391 | 628 |
! ! |
629 |
||
| 7261 | 630 |
!CharacterArray methodsFor:'Compatibility-ST/V'! |
| 608 | 631 |
|
632 |
byteAt:index put:aByte |
|
633 |
"store a byte at given index. |
|
| 9190 | 634 |
|
635 |
||
636 |
byteAt:index put:aByte |
|
637 |
"store a byte at given index. |
|
| 608 | 638 |
This is an ST/V compatibility method." |
639 |
||
|
1333
d3fdca956f0d
allow storing of 0-bytes (be careful with printing then ...)
Claus Gittinger <cg@exept.de>
parents:
1312
diff
changeset
|
640 |
"/ (aByte == 0) ifTrue:[ |
|
d3fdca956f0d
allow storing of 0-bytes (be careful with printing then ...)
Claus Gittinger <cg@exept.de>
parents:
1312
diff
changeset
|
641 |
"/ "store a space instead" |
|
d3fdca956f0d
allow storing of 0-bytes (be careful with printing then ...)
Claus Gittinger <cg@exept.de>
parents:
1312
diff
changeset
|
642 |
"/ ^ super basicAt:index put:(Character space) |
|
d3fdca956f0d
allow storing of 0-bytes (be careful with printing then ...)
Claus Gittinger <cg@exept.de>
parents:
1312
diff
changeset
|
643 |
"/ ]. |
| 9190 | 644 |
^ super at:index put:(C |
| 608 | 645 |
|
646 |
replChar:oldChar with:newChar |
|
647 |
"return a copy of the receiver, with all oldChars replaced |
|
648 |
by newChar. |
|
649 |
This is an ST/V compatibility method." |
|
650 |
||
| 3656 | 651 |
^ self copyReplaceAll:oldChar with:newChar |
| 608 | 652 |
|
653 |
" |
|
| 8874 | 654 |
'12345678901234567890' replChar:$0 with:$* |
| 608 | 655 |
" |
| 3656 | 656 |
|
657 |
"Modified: / 18.7.1998 / 22:52:57 / cg" |
|
| 608 | 658 |
! |
659 |
||
660 |
replChar:oldChar withString:newString |
|
661 |
"return a copy of the receiver, with all oldChars replaced |
|
| 9190 | 662 |
by newString (i.e. slice in the newStplChar:$0 withString:'foo' |
663 |
'a string with spaces' replChar:$ withString:' foo ' |
|
664 |
" |
|
665 |
! |
|
666 |
||
667 |
replString:subString withString:newString |
|
668 |
"return a copy of the receiver, with all sequences of subString replaced |
|
669 |
by newString (i.e. slice in the newString in place of the oldString)." |
|
670 |
||
671 |
^ self coreplChar:oldChar withString:newString |
|
672 |
"return a copy of the receiver, with all oldChars replaced |
|
| 608 | 673 |
by newString (i.e. slice in the newString in place of the oldChar). |
674 |
This is an ST/V compatibility method." |
|
675 |
||
676 |
|tmpStream| |
|
677 |
||
678 |
tmpStream := WriteStream on:(self class new). |
|
679 |
self do:[:element | |
|
| 9191 | 680 |
element = oldChar ifTrue:[ |
681 |
tmpStream nextPutAll:newString |
|
682 |
] ifFalse:[ |
|
683 |
tmpStream nextPut:element |
|
684 |
]. |
|
| 608 | 685 |
]. |
686 |
^ tmpStream contents |
|
687 |
||
688 |
" |
|
| 8874 | 689 |
'12345678901234567890' replChar:$0 withString:'foo' |
690 |
'a string with spaces' replChar:$ withString:' foo ' |
|
| 608 | 691 |
" |
| 9190 | 692 |
ithStreplString:subString withString:newString |
|
4244
5deefa3144fb
added #replString:withString:
Claus Gittinger <cg@exept.de>
parents:
4156
diff
changeset
|
693 |
"return a copy of the receiver, with all sequences of subString replaced |
|
5deefa3144fb
added #replString:withString:
Claus Gittinger <cg@exept.de>
parents:
4156
diff
changeset
|
694 |
by newString (i.e. slice in the newString in place of the oldString)." |
|
5deefa3144fb
added #replString:withString:
Claus Gittinger <cg@exept.de>
parents:
4156
diff
changeset
|
695 |
|
| 8347 | 696 |
^ self copyReplaceString:subString withString:newString |
|
4244
5deefa3144fb
added #replString:withString:
Claus Gittinger <cg@exept.de>
parents:
4156
diff
changeset
|
697 |
|
|
5deefa3144fb
added #replString:withString:
Claus Gittinger <cg@exept.de>
parents:
4156
diff
changeset
|
698 |
" |
| 8874 | 699 |
'12345678901234567890' replString:'123' withString:'OneTwoThree' |
700 |
'12345678901234567890' replString:'123' withString:'*' |
|
701 |
'12345678901234567890' replString:'234' withString:'foo' |
|
|
4244
5deefa3144fb
added #replString:withString:
Claus Gittinger <cg@exept.de>
parents:
4156
diff
changeset
|
702 |
|
|
5deefa3144fb
added #replString:withString:
Claus Gittinger <cg@exept.de>
parents:
4156
diff
changeset
|
703 |
('a string with spaces' replChar:$ withString:' foo ')
|
| 9191 | 704 |
replString:'foo' withString:'bar' |
|
4244
5deefa3144fb
added #replString:withString:
Claus Gittinger <cg@exept.de>
parents:
4156
diff
changeset
|
705 |
" |
|
5deefa3144fb
added #replString:withString:
Claus Gittinger <cg@exept.de>
parents:
4156
diff
changeset
|
706 |
|
| 8347 | 707 |
"Modified: / 12-05-2004 / 12:00:27 / cg" |
| 9190 | 708 |
|
|
4409
86e049b21f8d
added #includesSubstring:caseSensitive:
Claus Gittinger <cg@exept.de>
parents:
4389
diff
changeset
|
709 |
|
| 6148 | 710 |
asDate |
711 |
"Many allowed forms, see Date.readFrom:" |
|
712 |
||
713 |
^ Date readFrom: (ReadStream on: self) |
|
714 |
||
715 |
" |
|
| 8874 | 716 |
'30 Apr 1999' asDate dayName capitalized |
| 6148 | 717 |
" |
718 |
! |
|
719 |
||
| 8971 | 720 |
asUrl |
721 |
^ self asURL |
|
722 |
! |
|
723 |
||
| 4989 | 724 |
capitalized |
725 |
^ self asUppercaseFirst |
|
726 |
||
727 |
" |
|
728 |
'hello' capitalized |
|
729 |
" |
|
730 |
! |
|
731 |
||
| 5632 | 732 |
displayProgressAt:aPointOrNil from:start to:stop during:aBlock |
733 |
ProgressIndicator |
|
| 9191 | 734 |
displayProgress:se |
| 9190 | 735 |
|
736 |
displayProgressAt:aPointOrNil from:start to:stop during:aBlock |
|
737 |
ProgressIndicator |
|
| 9191 | 738 |
displayProgress:self |
739 |
at:aPointOrNil |
|
740 |
from:start |
|
741 |
to:stop |
|
742 |
during:aBlock. |
|
| 5632 | 743 |
|
744 |
" |
|
745 |
'dobedobedoobedoo' |
|
| 9191 | 746 |
displayProgressAt:(Screen current center) |
747 |
from:0 to:100 |
|
748 |
duri |
|
| 8971 | 749 |
"/ ]. |
750 |
"/ ^ self size + 1 |
|
751 |
! |
|
752 |
||
| 4981 | 753 |
findTokens:delimiters |
| 7270 | 754 |
"cg: I am not sure, if this is really the squeak semantics (w.r.t. empty fields)" |
755 |
||
| 4981 | 756 |
^ self asCollectionOfSubstringsSeparatedByAny:delimiters |
| 7270 | 757 |
|
758 |
" |
|
| 8874 | 759 |
'a|b#c||e' findTokens:#($# $|) |
| 7270 | 760 |
" |
| 4981 | 761 |
! |
762 |
||
| 4974 | 763 |
includesSubString:aString |
764 |
"return true, if a substring is contained in the receiver. |
|
765 |
The compare is case sensitive." |
|
766 |
||
767 |
^ self includesString:aString |
|
768 |
||
769 |
" |
|
| 8874 | 770 |
'hello world' includesSubString:'Hel' |
| 9190 | 771 |
'hello world' incfindDelimiters:delimiters startingAt:start |
772 |
"Answer the index of the character within the receiver, starting at start, that matches one of the delimiters. If the receiver does not contain any of the delimiters, answer size + 1." |
|
773 |
||
774 |
|idx| |
|
775 |
||
776 |
idx := self indexOfAny:delimiters startingAt:start. |
|
777 |
idx == 0 ifTrue:[^ self size + 1]. |
|
778 |
^ idx. |
|
779 |
"/start to: self size do: [:i | |
|
780 |
"/ |char| |
|
781 |
"/ |
|
782 |
"/ char := self at: i. |
|
783 |
"/ delimiters do: [:delim | delim = char ifTrue: [^ i]] |
|
784 |
"/ ]. |
|
785 |
"/ ^ self size + 1 |
|
786 |
||
787 |
||
788 |
includesSubString:aString |
|
789 |
"return true, if a substring is contained in the receiver. |
|
790 |
The compare is case sensitive." |
|
791 |
||
792 |
^ self includesString:aString |
|
793 |
||
794 |
" |
|
795 |
'hello world' includesSubString:'Hel' |
|
796 |
' |
|
| 4974 | 797 |
|
798 |
! |
|
799 |
||
| 9190 | 800 |
isAllDigits |
801 |
"Answer whether the receiver's characters are all digits" |
|
802 |
||
803 |
^ self conform:[:eachChar | eachChar isDigit] |
|
804 |
||
805 |
" |
|
806 |
'hello world' isAllDigits |
|
807 |
'12344' isAllDigits |
|
808 |
" |
|
| 4890 | 809 |
! |
810 |
||
| 9190 | 811 |
lastSpacePosition |
812 |
^ self lastIndexOfSeparator |
|
813 |
! |
|
814 |
||
815 |
padded:leftOrRight to:paddedSize with |
|
816 |
||
| 6979 | 817 |
isAllDigits |
818 |
"Answer whether the receiver's characters are all digits" |
|
819 |
||
820 |
^ self conform:[:eachChar | eachChar isDigit] |
|
821 |
||
822 |
" |
|
823 |
'hello world' isAllDigits |
|
824 |
'12344' isAllDigits |
|
825 |
" |
|
826 |
! |
|
827 |
||
| 4890 | 828 |
lastSpacePosition |
829 |
^ self lastIndexOfSeparator |
|
830 |
! |
|
831 |
||
| 7022 | 832 |
padded:leftOrRight to:paddedSize with:padCharacter |
833 |
leftOrRight == #left ifTrue:[ |
|
| 9191 | 834 |
^ self leftPaddedTo:paddedSize with:padCharacter |
| 7022 | 835 |
]. |
836 |
^ self paddedTo:paddedSize with:padCharacter |
|
837 |
||
838 |
" |
|
| 9190 | 839 |
'hello' padded:#righ |
840 |
||
841 |
padded:leftOrRight to:paddedSize with:padCharacter |
|
842 |
leftOrRight == #left ifTrue:[ |
|
| 9191 | 843 |
^ self leftPaddedTo:paddedSize with:padCharacter |
| 9190 | 844 |
]. |
845 |
^ self paddedTo:paddedSize with:padCharacter |
|
846 |
||
847 |
" |
|
| 8874 | 848 |
'hello' padded:#right to:10 with:$. |
| 9190 | 849 |
***7890' skipDelimiters:'*' startingAt:3 |
850 |
'123***7890' skipDelimiters:'*' startingAt:10 |
|
851 |
'123*******' skipDelimiters:'*' startingAt:10 |
|
| 8874 | 852 |
" |
853 |
! |
|
854 |
||
| 9190 | 855 |
substrings |
856 |
^ self asCollectionOfWords |
|
857 |
||
858 |
" |
|
859 |
'foo bar baz' substrings |
|
860 |
" |
|
861 |
! |
|
862 |
||
863 |
truncateTo:smallSize |
|
864 |
"return myskipDelimiters:delimiters startingAt:start |
|
| 8874 | 865 |
"Answer the index of the character within the receiver, starting at start, |
866 |
that does NOT match one of the delimiters. |
|
867 |
If the receiver does not contain any of the delimiters, answer size + 1. |
|
| 7691 | 868 |
Assumes the delimiters to be a non-empty string." |
| 8874 | 869 |
|
870 |
start to:self size do:[:i | |
|
| 9191 | 871 |
delimiters detect:[:delim | delim = (self at:i) ] ifNone:[ ^ i ] |
| 7691 | 872 |
]. |
873 |
^ self size + 1 |
|
874 |
||
875 |
" |
|
| 8874 | 876 |
'123***7890' skipDelimiters:'*' startingAt:4 |
877 |
'123***7890' skipDelimiters:'*' startingAt:3 |
|
878 |
'123***7890' skipDelimiters:'*' startingAt:10 |
|
879 |
'123*******' skipDelimiters:'*' startingAt:10 |
|
| 7691 | 880 |
" |
| 9190 | 881 |
|
| 5795 | 882 |
|
| 4890 | 883 |
withBlanksTrimmed |
884 |
"Return a copy of the receiver from which leading and trailing blanks have been trimmed." |
|
885 |
||
886 |
^ self withoutSpaces |
|
887 |
||
888 |
" |
|
| 8874 | 889 |
' hello world ' withBlanksTrimmed |
| 4890 | 890 |
" |
891 |
||
892 |
||
893 |
||
894 |
! |
|
895 |
||
896 |
withNoLineLongerThan: aNumber |
|
897 |
"Answer a string with the same content as receiver, but rewrapped so that no line has more characters than the given number" |
|
898 |
||
| 9190 | 899 |
| list " |
900 |
||
901 |
||
902 |
||
903 |
! ! |
|
904 |
||
905 |
!CharacterArray methodsFor:'Compatibility-V''Age'! |
|
906 |
||
907 |
addLineDelimiter |
|
908 |
"replace all '\'-characters by line delimiter (cr) - characters. |
|
909 |
This has been added for VisualAge compwithNoLineLongerThan: aNumber |
|
910 |
"Answer a string with the same content as receiver, but rewrapped so that no line has more characters than the given number" |
|
911 |
||
| 4890 | 912 |
| listOfLines currentLast currentStart resultString putativeLast putativeLine crPosition | |
913 |
||
914 |
aNumber isNumber not | (aNumber < 1) ifTrue: [self error: 'too narrow']. |
|
915 |
listOfLines _ OrderedCollection new. |
|
916 |
currentLast _ 0. |
|
917 |
[currentLast < self size] whileTrue: |
|
| 9191 | 918 |
[currentStart _ currentLast + 1. |
919 |
putativeLast _ (currentStart + aNumber - 1) min: self size. |
|
920 |
putativeLine _ self copyFrom: currentStart to: putativeLast. |
|
921 |
(crPosition _ putativeLine indexOf: Character cr) > 0 ifTrue: |
|
922 |
[putativeLast _ currentStart + crPosition - 1. |
|
923 |
putativeLine _ self copyFrom: currentStart to: putativeLast]. |
|
924 |
currentLast _ putativeLast == self size |
|
925 |
ifTrue: |
|
926 |
[putativeLast] |
|
927 |
ifFalse: |
|
928 |
[currentStart + putativeLine lastSpacePosition - 1]. |
|
929 |
currentLast <= currentStart ifTrue: |
|
930 |
["line has NO spaces; baleout!!" |
|
931 |
currentLast _ putativeLast]. |
|
932 |
listOfLines add: (self copyFrom: currentStart to: currentLast) withBlanksTrimmed]. |
|
| 4890 | 933 |
|
934 |
listOfLines size > 0 ifFalse: [^ '']. |
|
935 |
resultString _ listOfLines first. |
|
936 |
2 to: listOfLines size do: |
|
| 9191 | 937 |
[:i | resultString _ resultString, Character cr asString, (listOfLines at: i)]. |
| 4890 | 938 |
^ resultString |
939 |
||
940 |
" |
|
941 |
#(5 7 20) collect: |
|
| 9191 | 942 |
[:i | 'Fred the bear went down to the brook to read his book in silence' withNoLineLongerThan: i] |
| 4890 | 943 |
" |
944 |
||
945 |
||
946 |
||
|
4409
86e049b21f8d
added #includesSubstring:caseSensitive:
Claus Gittinger <cg@exept.de>
parents:
4389
diff
changeset
|
947 |
! ! |
|
86e049b21f8d
added #includesSubstring:caseSensitive:
Claus Gittinger <cg@exept.de>
parents:
4389
diff
changeset
|
948 |
|
| 9190 | 949 |
!CharacterArray |
| 608 | 950 |
|
951 |
bindWith:string1 with:string2 |
|
952 |
"return a copy of the receiver, where a '%1' escape is |
|
953 |
replaced by string1 and '%2' is replaced by string2. |
|
954 |
This has been added for VisualAge compatibility." |
|
955 |
||
956 |
^ self expandPlaceholdersWith:(Array with:string1 with:string2) |
|
957 |
||
958 |
" |
|
959 |
'do you prefer %1 or rather %2 ?' |
|
| 9191 | 960 |
bindWith:'smalltalk' with:'c++' |
| 608 | 961 |
" |
962 |
! |
|
963 |
||
964 |
bindWith:str1 with:str2 with:str3 |
|
| 9190 | 965 |
"return a copy of the receiver, where a '%1', '%2' and '%3' escape |
966 |
||
967 |
bindWith:str1 with:str2 with:str3 |
|
| 608 | 968 |
"return a copy of the receiver, where a '%1', '%2' and '%3' escapes |
969 |
are replaced by str1, str2 and str3 respectively. |
|
970 |
This has been added for VisualAge compatibility." |
|
971 |
||
972 |
^ self expandPlaceholdersWith:(Array with:str1 with:str2 with:str3) |
|
973 |
||
974 |
" |
|
| 9190 | 975 |
'do you prefer %1 or rather %2 (not talking abou |
| 608 | 976 |
|
977 |
bindWith:str1 with:str2 with:str3 with:str4 |
|
978 |
"return a copy of the receiver, where a '%1', '%2', '%3' and '%4' escapes |
|
979 |
are replaced by str1, str2, str3 and str4 respectively. |
|
980 |
This has been added for VisualAge compatibility." |
|
981 |
||
982 |
^ self expandPlaceholdersWith:(Array with:str1 with:str2 with:str3 with:str4) |
|
983 |
||
984 |
" |
|
| 9190 | 985 |
'do you prefer %1 or rather %2 (not talking about %3 or even %4) ? |
| 608 | 986 |
|
| 2344 | 987 |
bindWith:str1 with:str2 with:str3 with:str4 with:str5 |
988 |
"return a copy of the receiver, where a '%1' .. '%5' escapes |
|
989 |
are replaced by str1 .. str5 respectively. |
|
990 |
This has been added for VisualAge compatibility." |
|
991 |
||
992 |
^ self expandPlaceholdersWith:(Array with:str1 with:str2 with:str3 with:str4 with:str5) |
|
993 |
||
994 |
"Created: 31.1.1997 / 16:25:42 / cg" |
|
995 |
! |
|
996 |
||
| 5343 | 997 |
bindWith:str1 with:str2 with:str3 with:str4 with:str5 with:str6 |
| 9190 | 998 |
"return a copy of the receiv |
999 |
||
1000 |
bindWith:str1 with:str2 with:str3 with:str4 with:str5 with:str6 |
|
| 5343 | 1001 |
"return a copy of the receiver, where a '%1' .. '%6' escapes |
1002 |
are replaced by str1 .. str5 respectively. |
|
1003 |
This has been added for VisualAge compatibility." |
|
1004 |
||
| 8874 | 1005 |
^ self expandPlaceholdersWith:(Array with:str1 with:str2 |
| 9191 | 1006 |
with:str3 with:str4 |
1007 |
with:str5 with:str6) |
|
| 5343 | 1008 |
|
1009 |
! |
|
1010 |
||
| 608 | 1011 |
bindWithArguments:anArrayOfStrings |
1012 |
"return a copy of the receiver, where a '%i' escape |
|
1013 |
is replaced by the coresponding string from the argument array. |
|
1014 |
'i' may be between 1 and 9 (i.e. a maximum of 9 placeholders is allowed). |
|
1015 |
This has been added for VisualAge compatibility." |
|
1016 |
||
| 9190 | 1017 |
^ self expandPlaceholdersWith:anArrayOfStrin |
| 608 | 1018 |
|
1019 |
subStrings |
|
1020 |
"return an array consisting of all words contained in the receiver. |
|
1021 |
Words are separated by whitespace. |
|
1022 |
This has been added for VisualAge compatibility." |
|
1023 |
||
1024 |
^ self asCollectionOfWords |
|
1025 |
||
1026 |
" |
|
1027 |
'hello world, this is smalltalk' subStrings |
|
1028 |
" |
|
1029 |
! |
|
1030 |
||
1031 |
subStrings:separatorCharacter |
|
1032 |
"return an array consisting of all words contained in the receiver. |
|
1033 |
Words are separated by separatorCharacter. |
|
| 9190 | 1034 |
This has been added f |
1035 |
||
1036 |
subStrings:separatorCharacter |
|
1037 |
"return an array consisting of all words contained in the receiver. |
|
1038 |
Words are separated by separatorCharacter. |
|
| 608 | 1039 |
This has been added for VisualAge compatibility." |
1040 |
||
| 9190 | 1041 |
^ self asCollectionOfSubstringsSeparatedBy:separatorCharac |
| 5964 | 1042 |
|
| 6139 | 1043 |
asLogicalFileSpecification |
1044 |
^ self asFilename |
|
1045 |
||
1046 |
"Created: / 30.10.2001 / 17:29:53 / cg" |
|
1047 |
! |
|
1048 |
||
| 6119 | 1049 |
asQualifiedReference |
1050 |
^ BindingReference pathString:(self string) |
|
1051 |
||
1052 |
! |
|
1053 |
||
| 5964 | 1054 |
expandMacros |
1055 |
"ST80 compatibility - expand '<..>' macros with |
|
1056 |
argument strings. Similar to #bindWith:. |
|
1057 |
Read the comment in #expandMacrosWithArguments: about |
|
1058 |
limited compatibility issues." |
|
1059 |
||
1060 |
^ self expandMacrosWithArguments:#() |
|
1061 |
||
1062 |
" |
|
1063 |
'hello<n>foo' expandMacros |
|
1064 |
" |
|
1065 |
||
| 9190 | 1066 |
"Modified: / 18. |
1067 |
||
1068 |
expandMacros |
|
1069 |
"ST80 compatibility - expand '<..>' macros with |
|
1070 |
argument strings. Similar to #bindWith:. |
|
1071 |
Read the comment in #expandMacrosWithArguments: aboexpandMacros |
|
| 5964 | 1072 |
"ST80 compatibility - expand '<..>' macros with |
1073 |
argument strings. Similar to #bindWith:. |
|
1074 |
Read the comment in #expandMacrosWithArguments: about |
|
1075 |
limited compatibility issues." |
|
1076 |
||
| 9190 | 1077 |
^ self expandMacrosWithArguments:#() |
1078 |
||
1079 |
" |
|
1080 |
'hello<n>foo' expandMacros |
|
1081 |
" |
|
1082 |
||
1083 |
"Modified: / 18.6.1998 / 16:03:02 / cg" |
|
1084 |
||
| 5964 | 1085 |
|
1086 |
expandMacrosWith:arg1 with:arg2 |
|
1087 |
"ST80 compatibility - expand '<..>' macros with |
|
1088 |
argument strings. Similar to #bindWith:. |
|
1089 |
Read the comment in #expandMacrosWithArguments: about |
|
1090 |
limited compatibility issues." |
|
1091 |
||
1092 |
^ self expandMacrosWithArguments:(Array with:arg1 with:arg2) |
|
1093 |
||
1094 |
"Modified: / 6.7.1998 / 21:58:14 / cg" |
|
1095 |
! |
|
1096 |
||
| 9190 | 1097 |
expandMacrosWi |
1098 |
||
| 5964 | 1099 |
expandMacrosWith:arg1 with:arg2 with:arg3 |
1100 |
"ST80 compatibility - expand '<..>' macros with |
|
1101 |
argument strings. Similar to #bindWith:. |
|
1102 |
Read the comment in #expandMacrosWithArguments: about |
|
1103 |
limited compatibility issues." |
|
1104 |
||
1105 |
^ self expandMacrosWithArguments:(Array with:arg1 with:arg2 with:arg3) |
|
1106 |
! |
|
1107 |
||
| 9190 | 1108 |
expandMacrosW:46 / cg" |
1109 |
! |
|
1110 |
||
1111 |
isCharacters |
|
1112 |
"added for visual works compatibility" |
|
1113 |
^ true |
|
1114 |
! ! |
|
1115 |
||
1116 |
!CharacterArray methodsFor:'JavaScript support'! |
|
1117 |
||
1118 |
+ aStringOrCharacter |
|
1119 |
"alternative string-concatenation. |
|
1120 |
For JavaScript only" |
|
1121 |
||
1122 |
^ self , aStringOrCharacter |
|
1123 |
||
1124 |
" |
|
1125 |
'hello' + $1 |
|
1126 |
'hello' + '1expandMacrosWithArguments:argArray |
|
| 5964 | 1127 |
"ST80 compatibility - expand '<..>' macros with |
1128 |
argument strings. Similar to #bindWith:. |
|
1129 |
WARNING: possibly not all ST80 expansions are supported here." |
|
1130 |
||
1131 |
"/ supported expansions: |
|
1132 |
"/ |
|
1133 |
"/ <#p> # is arg Number; slice in the args printString |
|
|
8222
f2c454a9a038
replaced '' writeStream by String writeStream
Claus Gittinger <cg@exept.de>
parents:
8124
diff
changeset
|
1134 |
"/ <#s> # is arg Number; slice in the arg itself (must know asString) |
| 5964 | 1135 |
"/ <#?s1:s2> # is arg Number; slice in s1 if the arg is true, s2 otherwise |
| 7643 | 1136 |
"/ use first arg if # is not given (i.e. no number before s,p ...) |
| 5964 | 1137 |
"/ <n> replace by a newLine character |
1138 |
"/ <t> replace by a tab character |
|
|
6834
eba4b58b8692
fixed expandMacros (%-escape)
Claus Gittinger <cg@exept.de>
parents:
6810
diff
changeset
|
1139 |
"/ %X the X character itself |
| 5964 | 1140 |
|
|
8999
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1141 |
|in out c fmt nr arg s1 s2 peekc| |
| 5964 | 1142 |
|
1143 |
in := self readStream. |
|
|
9019
bf8954c06aef
expandPlaceHolders: care for being Unicode
Claus Gittinger <cg@exept.de>
parents:
9007
diff
changeset
|
1144 |
out := WriteStream on:(self species uninitializedNew:self size). |
| 5964 | 1145 |
|
1146 |
[in atEnd] whileFalse:[ |
|
|
8999
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1147 |
c := in next. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1148 |
c == $% ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1149 |
c := in next. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1150 |
out nextPut:c |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1151 |
] ifFalse:[c ~~ $< ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1152 |
out nextPut:c. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1153 |
] ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1154 |
peekc := in peek. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1155 |
[peekc == $<] whileTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1156 |
out nextPut:$<. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1157 |
peekc := in nextPeek. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1158 |
]. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1159 |
peekc == $n ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1160 |
peekc := in nextPeek. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1161 |
peekc == $> ifTrue:[ |
|
9007
ef6b316c040c
#expandMacrosWithArguments: fix '><'
Stefan Vogel <sv@exept.de>
parents:
8999
diff
changeset
|
1162 |
in next. |
|
ef6b316c040c
#expandMacrosWithArguments: fix '><'
Stefan Vogel <sv@exept.de>
parents:
8999
diff
changeset
|
1163 |
out cr. |
|
8999
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1164 |
] ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1165 |
out nextPutAll:'<n'. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1166 |
] |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1167 |
] ifFalse:[peekc == $t ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1168 |
peekc := in nextPeek. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1169 |
peekc == $> ifTrue:[ |
|
9007
ef6b316c040c
#expandMacrosWithArguments: fix '><'
Stefan Vogel <sv@exept.de>
parents:
8999
diff
changeset
|
1170 |
in next. |
|
ef6b316c040c
#expandMacrosWithArguments: fix '><'
Stefan Vogel <sv@exept.de>
parents:
8999
diff
changeset
|
1171 |
out tab. |
|
8999
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1172 |
] ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1173 |
out nextPutAll:'<t'. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1174 |
] |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1175 |
] ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1176 |
peekc isDigit ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1177 |
"start an argument expansion ..." |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1178 |
nr := Integer readFrom:in onError:nil. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1179 |
nr isNil ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1180 |
"this cannot happen (there is at least one digit)" |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1181 |
self error:'invalid format' mayProceed:true. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1182 |
^ self |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1183 |
]. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1184 |
fmt := in next. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1185 |
(fmt ~~ $? and:[in peek ~~ $>]) ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1186 |
out nextPut:$<. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1187 |
nr printOn:out. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1188 |
out nextPut:fmt. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1189 |
] ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1190 |
(nr between:1 and:argArray size) ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1191 |
arg := argArray at:nr. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1192 |
] ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1193 |
arg := '' |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1194 |
]. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1195 |
|
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1196 |
fmt == $p ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1197 |
"expand with args printString" |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1198 |
arg printOn:out. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1199 |
] ifFalse:[fmt == $s ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1200 |
"expand with arg itself" |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1201 |
arg isText ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1202 |
out contentsSpecies isText ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1203 |
out := (WriteStream on:Text new) nextPutAll:out contents; yourself. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1204 |
]. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1205 |
out nextPutAll:arg. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1206 |
] ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1207 |
out nextPutAll:arg asString string. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1208 |
] |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1209 |
] ifFalse:[fmt == $? ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1210 |
s1 := in upTo:$:. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1211 |
s2 := in nextUpTo:$>. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1212 |
arg == true ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1213 |
out nextPutAll:s1 |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1214 |
] ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1215 |
out nextPutAll:s2 |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1216 |
]. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1217 |
] ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1218 |
"what does VW do here ?" |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1219 |
self error:'invalid format' mayProceed:true. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1220 |
^ self |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1221 |
]]]. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1222 |
c := in next. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1223 |
c ~~ $> ifTrue:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1224 |
"what does VW do here ?" |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1225 |
self error:'invalid format' mayProceed:true. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1226 |
^ self |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1227 |
] |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1228 |
|
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1229 |
]. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1230 |
] ifFalse:[ |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1231 |
out nextPut:$<. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1232 |
]. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1233 |
]]. |
|
f3fda66ea51e
#expandMacrosWithArguments: - be mor friendly to xml strings.
Stefan Vogel <sv@exept.de>
parents:
8971
diff
changeset
|
1234 |
]]. |
| 5964 | 1235 |
]. |
1236 |
^ out contents |
|
1237 |
||
1238 |
" |
|
| 8874 | 1239 |
'hello <1s> how are you' expandMacrosWith:(OperatingSystem getLoginName) |
1240 |
'one plus one is <1p>' expandMacrosWith:2 |
|
| 5964 | 1241 |
" |
1242 |
||
1243 |
"Modified: / 18.6.1998 / 16:04:46 / cg" |
|
| 9190 | 1244 |
o\' includesMatchCharacters |
1245 |
" |
|
1246 |
||
1247 |
"Modified: 2.4.1997 / 18:12:34 / cg" |
|
| 8632 | 1248 |
! |
1249 |
||
| 9190 | 1250 |
includesSeparator |
1251 |
"return true, if the receiver contains any whitespace characters" |
|
1252 |
||
1253 |
^ (self indexOfSeparator ~~ 0) |
|
1254 |
||
1255 |
" |
|
1256 |
'hello world' includesSeparator |
|
1257 |
'helloworld' includesSeparator |
|
1258 |
" |
|
1259 |
! |
|
1260 |
||
1261 |
includesUnescapedMatchCharacters |
|
1262 |
"return true if includesMatchCharacters |
|
| 8874 | 1263 |
"return true if the receiver includes any meta characters (i.e. $* or $#) |
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1264 |
for match operations; false if not. |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1265 |
Here, do not care for $\ escapes" |
| 608 | 1266 |
|
1267 |
^ self includesAny:'*#[' |
|
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1268 |
|
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1269 |
" |
| 8874 | 1270 |
'*foo' includesMatchCharacters |
1271 |
'\*foo' includesMatchCharacters |
|
1272 |
'\*foo' includesUnescapedMatchCharacters |
|
1273 |
'*foo' includesMatchCharacters |
|
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1274 |
'\\*foo' includesMatchCharacters |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1275 |
'foo*' includesMatchCharacters |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1276 |
'foo\*' includesMatchCharacters |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1277 |
'foo\' includesMatchCharacters |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1278 |
" |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1279 |
|
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1280 |
"Modified: 2.4.1997 / 18:12:34 / cg" |
| 9190 | 1281 |
esUnescapedMatchCharacters |
1282 |
'foo\' includesUnescapedMatchCharacters |
|
1283 |
" |
|
1284 |
||
1285 |
"Modified: 2.4.1997 / 17:08:52 / cg" |
|
1286 |
"Created: 2.4.1997 / 17:23:26 / cg" |
|
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1287 |
! |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1288 |
|
| 9190 | 1289 |
indexOfControlCharacterStartingAt:startIndex |
1290 |
includesUnescapedMatchCharacters |
|
| 8874 | 1291 |
"return true if the receiver really includes any meta characters (i.e. $* or $#) |
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1292 |
for match operations; false if not. |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1293 |
Here, care for $\ escapes" |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1294 |
|
|
5342
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
1295 |
|idx sz specialChars escape| |
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1296 |
|
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1297 |
idx := 1. |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1298 |
sz := self size. |
|
5342
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
1299 |
specialChars := '*#[\'. |
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
1300 |
(escape := self class matchEscapeCharacter) ~~ $\ ifTrue:[ |
| 9191 | 1301 |
specialChars := specialChars copy. |
1302 |
specialChars at:specialChars size put:escape |
|
|
5342
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
1303 |
]. |
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
1304 |
|
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1305 |
[true] whileTrue:[ |
| 9191 | 1306 |
idx := self indexOfAny:specialChars startingAt:idx. |
1307 |
idx == 0 ifTrue:[^ false]. |
|
1308 |
(self at:idx) == escape ifFalse:[^ true]. |
|
1309 |
idx := idx + 2. |
|
1310 |
idx > sz ifTrue:[^ false]. |
|
| 8874 | 1311 |
]. |
1312 |
||
1313 |
" |
|
1314 |
'*foo' includesUnescapedMatchCharacters |
|
1315 |
'\*foo' includesUnescapedMatchCharacters |
|
1316 |
'\\foo' includesUnescapedMatchCharacters |
|
1317 |
'\\\$foo' includesUnescapedMatchCharacters |
|
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1318 |
'*foo' includesUnescapedMatchCharacters |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1319 |
'\\*foo' includesUnescapedMatchCharacters |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1320 |
'foo*' includesUnescapedMatchCharacters |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1321 |
'foo\*' includesUnescapedMatchCharacters |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1322 |
'foo\' includesUnescapedMatchCharacters |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1323 |
" |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1324 |
|
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1325 |
"Modified: 2.4.1997 / 17:08:52 / cg" |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1326 |
"Created: 2.4.1997 / 17:23:26 / cg" |
| 9190 | 1327 |
ld' aindexOfControlCharacterStartingAt:startIndex |
|
3668
76271941f044
added #indexOfControlCharacterStartingAt:
Claus Gittinger <cg@exept.de>
parents:
3656
diff
changeset
|
1328 |
"return the index of the next control character; |
| 4117 | 1329 |
starting the search at startIndex, searching forward; |
|
3668
76271941f044
added #indexOfControlCharacterStartingAt:
Claus Gittinger <cg@exept.de>
parents:
3656
diff
changeset
|
1330 |
that is a character with asciiValue < 32. |
|
76271941f044
added #indexOfControlCharacterStartingAt:
Claus Gittinger <cg@exept.de>
parents:
3656
diff
changeset
|
1331 |
Return 0 if none is found." |
|
76271941f044
added #indexOfControlCharacterStartingAt:
Claus Gittinger <cg@exept.de>
parents:
3656
diff
changeset
|
1332 |
|
|
76271941f044
added #indexOfControlCharacterStartingAt:
Claus Gittinger <cg@exept.de>
parents:
3656
diff
changeset
|
1333 |
|start "{ Class: SmallInteger }"
|
| 9190 | 1334 |
mySize "{ Class: SmallInteger }
|
|
3668
76271941f044
added #indexOfControlCharacterStartingAt:
Claus Gittinger <cg@exept.de>
parents:
3656
diff
changeset
|
1335 |
|
| 8541 | 1336 |
indexOfNonSeparator |
1337 |
"return the index of the first non-whitespace character. |
|
1338 |
return 0 if no non-separator was found" |
|
1339 |
||
1340 |
^ self indexOfNonSeparatorStartingAt:1. |
|
1341 |
||
1342 |
" |
|
| 8874 | 1343 |
' hello world' indexOfNonSeparator |
1344 |
' ' indexOfNonSeparator |
|
1345 |
'a ' indexOfNonSeparator |
|
1346 |
'abc' indexOfNonSeparator |
|
1347 |
' ' indexOfNonSeparator |
|
| 9190 | 1348 |
'' indexOfNonSepa |
1349 |
' ' indexOfNonSeparatorStartingAt:1 |
|
1350 |
'a ' indexOfNonSeparatorStartingAt:2 |
|
1351 |
" |
|
1352 |
||
1353 |
" |
|
1354 |
|s index1 index2| |
|
1355 |
s := ' foo bar baz'. |
|
1356 |
index1 := s indexOfNonSeparatorStartingAt:1. |
|
1357 |
index2 := s indexOfSeparatorStartingAt:index1. |
|
1358 |
s copyFrom:index1 to:index2 - 1 |
|
| 8541 | 1359 |
" |
1360 |
! |
|
1361 |
||
| 9190 | 1362 |
indexOfSeparator |
1363 |
"return the index of the first whitespaindexOfNonSeparatorStartingAt:startIndex |
|
| 4117 | 1364 |
"return the index of the next non-whitespace character, |
1365 |
starting the search at startIndex, searching forward; |
|
1366 |
return 0 if no non-separator was found" |
|
| 608 | 1367 |
|
1368 |
|start "{ Class: SmallInteger }"
|
|
1369 |
mySize "{ Class: SmallInteger }"|
|
|
1370 |
||
1371 |
start := startIndex. |
|
1372 |
mySize := self size. |
|
1373 |
||
1374 |
start to:mySize do:[:index | |
|
| 9191 | 1375 |
(self at:index) isSeparator ifFalse:[^ index] |
| 608 | 1376 |
]. |
1377 |
^ 0 |
|
1378 |
||
1379 |
" |
|
| 8874 | 1380 |
' hello world' indexOfNonSeparatorStartingAt:1 |
1381 |
' ' indexOfNonSeparatorStartingAt:1 |
|
1382 |
'a ' indexOfNonSeparatorStartingAt:2 |
|
| 4117 | 1383 |
" |
1384 |
||
| 608 | 1385 |
" |
1386 |
|s index1 index2| |
|
1387 |
s := ' foo bar baz'. |
|
1388 |
index1 := s indexOfNonSeparatorStartingAt:1. |
|
1389 |
index2 := s indexOfSeparatorStartingAt:index1. |
|
1390 |
s copyFrom:index1 to:index2 - 1 |
|
1391 |
" |
|
| 9190 | 1392 |
d' indexOfSeparatorStartingAt:3 |
| 8874 | 1393 |
'hello world ' indexOfSeparatorStartingAt:3 |
1394 |
'hello world ' indexOfSeparatorStartingAt:6 |
|
1395 |
'hello world ' indexOfSeparatorStartingAt:7 |
|
1396 |
'helloworld ' indexOfSeparatorStartingAt:7 |
|
1397 |
'helloworld' indexOfSeparatorStartingAt:7 |
|
| 4117 | 1398 |
" |
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1399 |
! |
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
1400 |
|
| 4117 | 1401 |
lastIndexOfSeparator |
1402 |
"return the last index of the whitespace character. |
|
| 9190 | 1403 |
(iindexOfSeparatorStartingAt:startIndex |
1404 |
"return the index of the next whitespace character, |
|
1405 |
starting the search at startIndex, searching forward; |
|
1406 |
return 0 if no separator was found" |
|
1407 |
||
1408 |
|start "{ Class: SmallInteger }"
|
|
1409 |
mySize "{ Class: SmallInteger }
|
|
1410 |
||
1411 |
lastIndexOfSeparator |
|
1412 |
"return the last index of the whitespace character. |
|
| 4117 | 1413 |
(i.e. start the search at the end and search backwards); |
1414 |
Returns 0 if no separator is found." |
|
1415 |
||
1416 |
^ self lastIndexOfSeparatorStartingAt:(self size) |
|
1417 |
||
1418 |
" |
|
| 8874 | 1419 |
'hello world' lastIndexOfSeparator |
1420 |
'helloworld' lastIndexOfSeparator |
|
1421 |
'hel lo wor ld' lastIndexOfSeparator |
|
1422 |
'hel ' lastIndexOfSeparator 6 |
|
| 4117 | 1423 |
" |
1424 |
! |
|
1425 |
||
1426 |
lastIndexOfSeparatorStartingAt:startIndex |
|
| 9190 | 1427 |
"return the index of the previlloworld' lastIndexOfSeparatorStartingAt:7 |
| 8874 | 1428 |
' helloworld' lastIndexOfSeparatorStartingAt:7 |
| 4117 | 1429 |
" |
| 608 | 1430 |
! ! |
1431 |
||
1432 |
!CharacterArray methodsFor:'comparing'! |
|
1433 |
||
|
6742
292775fda600
avoid endless recursion when comparing int < string
Claus Gittinger <cg@exept.de>
parents:
6655
diff
changeset
|
1434 |
< aString |
| 608 | 1435 |
"Compare the receiver with the argument and return true if the |
|
6742
292775fda600
avoid endless recursion when comparing int < string
Claus Gittinger <cg@exept.de>
parents:
6655
diff
changeset
|
1436 |
receiver is greater than the argument. Otherwise return false. |
| 8874 | 1437 |
This comparison is based on the elements ascii code - |
| 9190 | 1438 |
i.e. upper/lowercase & upper/lowercase & natilastIndexOfSeparatorStartingAt:startIndex |
1439 |
"return the index of the previous whitespace character, |
|
1440 |
starting the search at startIndex (and searching backwards); |
|
1441 |
returns 0 if no separator was found" |
|
1442 |
||
1443 |
|start "{ Class: SmallInteger }.
|
|
|
6742
292775fda600
avoid endless recursion when comparing int < string
Claus Gittinger <cg@exept.de>
parents:
6655
diff
changeset
|
1444 |
n := mySize min:otherSize. |
|
292775fda600
avoid endless recursion when comparing int < string
Claus Gittinger <cg@exept.de>
parents:
6655
diff
changeset
|
1445 |
|
|
292775fda600
avoid endless recursion when comparing int < string
Claus Gittinger <cg@exept.de>
parents:
6655
diff
changeset
|
1446 |
1 to:n do:[:index | |
| 9191 | 1447 |
c1 := self at:index. |
1448 |
c2 := aString at:index. |
|
1449 |
(c1 == c2 or:[c1 = c2]) ifFalse:[^ c1 < c2]. |
|
|
6742
292775fda600
avoid endless recursion when comparing int < string
Claus Gittinger <cg@exept.de>
parents:
6655
diff
changeset
|
1450 |
]. |
|
292775fda600
avoid endless recursion when comparing int < string
Claus Gittinger <cg@exept.de>
parents:
6655
diff
changeset
|
1451 |
^ mySize < otherSize |
| 608 | 1452 |
! |
1453 |
||
1454 |
= aString |
|
1455 |
"Compare the receiver with the argument and return true if the |
|
1456 |
receiver is equal to the argument. Otherwise return false. |
|
| 1252 | 1457 |
|
| 9191 | 1458 |
This compare does NOT ignore case differences, |
1459 |
< aString |
|
| 9190 | 1460 |
"Compare the receiver with the argument and return true if the |
1461 |
receiver is greater than the argument. Otherwise return false. |
|
1462 |
This comparison is based on the elements ascii code - |
|
1463 |
i.e. upper/lowercase & upper/lowercase & national characters are NOT treated specially." |
|
1464 |
||
1465 |
|mySize "{ Class: SmallInteger }"
|
|
1466 |
otherSize "{ Class: SmallInteger }"
|
|
1467 |
n "{ Class: SmallInteger } otherSize := aString size.
|
|
1468 |
mySize == otherSize ifFalse:[^ false]. |
|
1469 |
||
1470 |
1 to:mySize do:[:index | |
|
| 9191 | 1471 |
(self at:index) = (aString at:index) ifFalse:[^ false]. |
| 9190 | 1472 |
]. |
1473 |
^ true |
|
1474 |
||
1475 |
" |
|
1476 |
'foo' = 'Foo' |
|
1477 |
'foo' = 'bar' |
|
1478 |
'foo' = 'foo' |
|
1479 |
'foo' = aString |
|
1480 |
"Compare the receiver with the argument and return true if the |
|
1481 |
receiver is equal to the argument. Otherwise return false. |
|
1482 |
||
| 8874 | 1483 |
This compare does NOT ignore case differences, |
| 608 | 1484 |
therefore 'foo' = 'Foo' will return false. |
1485 |
Since this is incompatible to ST-80 (at least, V2.x) , this may change." |
|
1486 |
||
1487 |
|mySize "{ Class: SmallInteger }"
|
|
1488 |
otherSize | |
|
1489 |
||
|
5963
23462e06dcff
compare with Texts fixed (should compare equal if strings are)
Claus Gittinger <cg@exept.de>
parents:
5946
diff
changeset
|
1490 |
(aString isString or:[aString species == self species]) ifFalse:[ |
| 9191 | 1491 |
^ false |
|
5963
23462e06dcff
compare with Texts fixed (should compare equal if strings are)
Claus Gittinger <cg@exept.de>
parents:
5946
diff
changeset
|
1492 |
]. |
| 608 | 1493 |
mySize := self size. |
1494 |
otherSize := aString size. |
|
1495 |
mySize == otherSize ifFalse:[^ false]. |
|
1496 |
||
1497 |
1 to:mySize do:[:index | |
|
| 9191 | 1498 |
(self at:index) = (aString at:index) ifFalse:[^ false]. |
| 608 | 1499 |
]. |
1500 |
^ true |
|
1501 |
||
1502 |
" |
|
| 8874 | 1503 |
'foo' = 'Foo' |
1504 |
'foo' = 'bar' |
|
1505 |
'foo' = 'foo' |
|
1506 |
'foo' = 'foo' asText |
|
1507 |
'foo' asText = 'foo' |
|
1508 |
'foo' asText = 'foo' asText |
|
| 608 | 1509 |
" |
| 1252 | 1510 |
|
1511 |
"Modified: 22.4.1996 / 15:53:58 / cg" |
|
| 608 | 1512 |
! |
| 9191 | 1513 |
. |
1514 |
> aString |
|
| 608 | 1515 |
"Compare the receiver with the argument and return true if the |
1516 |
receiver is greater than the argument. Otherwise return false. |
|
| 8874 | 1517 |
This comparison is based on the elements ascii code - |
| 1252 | 1518 |
i.e. upper/lowercase & upper/lowercase & national characters are NOT treated specially." |
| 608 | 1519 |
|
1520 |
|mySize "{ Class: SmallInteger }"
|
|
| 8874 | 1521 |
otherSize "{ Class: SmallInteger }"
|
| 9190 | 1522 |
n "{ Class: SmallInteger }f size.
|
| 5929 | 1523 |
otherSize := aString string size. |
1524 |
n := mySize min:otherSize. |
|
1525 |
||
1526 |
1 to:n do:[:index | |
|
| 9191 | 1527 |
c1 := (self at:index) asLowercase. |
1528 |
c2 := (aString at:index) asLowercase. |
|
1529 |
c1 > c2 ifTrue:[^ 1]. |
|
1530 |
c1 < c2 ifTrue:[^ -1]. |
|
| 5929 | 1531 |
]. |
1532 |
mySize > otherSize ifTrue:[^ 1]. |
|
| 9190 | 1533 |
mySize < otherSize ifTrucompareCaselessWith:aString |
1534 |
"Compare the receiver against the argument, ignoreing case. |
|
1535 |
Return 1 if the receiver is greater, 0 if equal and -1 if less than the argument. |
|
1536 |
||
| 8874 | 1537 |
This comparison is based on the elements ascii code - |
| 9190 | 1538 |
i.e. national characters are NOT treated specially. |
1539 |
'foo' compareWith: 'Foo' will return 0" |
|
| 608 | 1540 |
|
1541 |
|mySize "{ Class: SmallInteger }"
|
|
| 8874 | 1542 |
otherSize "{ Class: SmallInteger }"
|
| 9190 | 1543 |
n "{ Class: SmallInteger }eger }"
|
| 608 | 1544 |
c1 c2| |
1545 |
||
1546 |
mySize := self size. |
|
|
5739
4fd3be7ccc75
oops some relational ops did accept a non-string arg
Claus Gittinger <cg@exept.de>
parents:
5732
diff
changeset
|
1547 |
otherSize := aString string size. |
| 608 | 1548 |
n := mySize min:otherSize. |
1549 |
||
1550 |
1 to:n do:[:index | |
|
| 9191 | 1551 |
c1 := self at:index. |
1552 |
c2 := aString at:index. |
|
1553 |
c1 > c2 ifTrue:[^ 1]. |
|
1554 |
c1 < c2 ifTrue:[^ -1]. |
|
| 608 | 1555 |
]. |
1556 |
mySize > otherSize ifTrue:[^ 1]. |
|
1557 |
mySize < otherSize ifTrue:[^ -1]. |
|
1558 |
^ 0 |
|
| 1252 | 1559 |
|
1560 |
"Modified: 22.4.1996 / 15:56:07 / cg" |
|
| 608 | 1561 |
! |
1562 |
||
| 9190 | 1563 |
hashcompareWith:aString |
1564 |
"Compare the receiver with the argument and return 1 if the receiver is |
|
1565 |
greater, 0 if equal and -1 if less than the argument. |
|
1566 |
This comparison is based on the elements ascii code - |
|
1567 |
i.e. upper/lowercase & national characters are NOT treated specially. |
|
1568 |
'foo' compareWith: 'Foo' will return 1. |
|
1569 |
while 'foo' sameAs:'Foo' will return true" |
|
1570 |
||
1571 |
|mySize "{ Class: SmallInteger }"
|
|
1572 |
otherSize "{ Class: SmallInteger }"
|
|
1573 |
n "{ Class: SmallInteger }This speeds up Set and Dictionary by a factor of 10!
|
|
1574 |
*/ |
|
1575 |
val *= 31415821; |
|
1576 |
RETURN ( __mkSmallInteger(val & 0x3fffffff)); |
|
1577 |
%} |
|
1578 |
||
1579 |
! |
|
1580 |
||
1581 |
sameAs:aString |
|
1582 |
"Compare the receiver with the argument like =, but ignore case differences. |
|
1583 |
Return true or false." |
|
1584 |
||
1585 |
|mySize "{ Class: SmallInteger }"
|
|
1586 |
otherSize c1 c2| |
|
1587 |
||
1588 |
self == aString ifTruehash |
|
| 608 | 1589 |
"return an integer useful as a hash-key" |
1590 |
||
1591 |
%{ /* NOCONTEXT */
|
|
1592 |
||
| 3985 | 1593 |
REGISTER unsigned g, val; |
| 608 | 1594 |
REGISTER unsigned char *cp, *cp0; |
1595 |
int l; |
|
1596 |
||
| 1133 | 1597 |
cp = __stringVal(self); |
1598 |
l = __stringSize(self); |
|
| 3987 | 1599 |
if (__qClass(self) != @global(String)) {
|
| 9191 | 1600 |
int n = __OBJS2BYTES__(__intVal(__ClassInstPtr(__qClass(self))->c_ninstvars)); |
1601 |
||
1602 |
cp += n; |
|
1603 |
l -= n; |
|
| 608 | 1604 |
} |
1605 |
||
1606 |
/* |
|
| 8874 | 1607 |
* this is the dragon-book algorithm |
| 608 | 1608 |
*/ |
| 3985 | 1609 |
|
1610 |
val = 0; |
|
1611 |
switch (l) {
|
|
1612 |
default: |
|
| 9191 | 1613 |
for (cp0 = cp, cp += l - 1; cp >= cp0; cp--) {
|
1614 |
val = (val << 4) + *cp; |
|
1615 |
if (g = (val & 0xF0000000)) {
|
|
1616 |
val ^= g >> 24; |
|
1617 |
val ^= g; |
|
1618 |
} |
|
1619 |
} |
|
1620 |
break; |
|
| 3985 | 1621 |
case 7: |
| 9191 | 1622 |
val = cp[6] << 4; |
| 3985 | 1623 |
case 6: |
| 9191 | 1624 |
val = (val + cp[5]) << 4; |
| 3985 | 1625 |
case 5: |
| 9191 | 1626 |
val = (val + cp[4]) << 4; |
| 3985 | 1627 |
case 4: |
| 9191 | 1628 |
val = (val + cp[3]) << 4; |
| 3985 | 1629 |
case 3: |
| 9191 | 1630 |
val = (val + cp[2]) << 4; |
| 3985 | 1631 |
case 2: |
| 9191 | 1632 |
val = (val + cp[1]) << 4; |
| 3985 | 1633 |
case 1: |
| 9191 | 1634 |
val = val + cp[0]; |
| 3985 | 1635 |
case 0: |
| 9191 | 1636 |
break; |
| 608 | 1637 |
} |
1638 |
||
| 3985 | 1639 |
/* |
1640 |
* multiply by large prime to spread values |
|
1641 |
* This speeds up Set and Dictionary by a factor of 10! |
|
| 608 | 1642 |
" |
| 8874 | 1643 |
'foo' sameAs: 'Foo' |
1644 |
'foo' sameAs: 'bar' |
|
1645 |
'foo' sameAs: 'foo' |
|
| 9190 | 1646 |
sameAs:aString |
1647 |
"Compare the receiver with the argument like =, but ignore case differences. |
|
1648 |
Return true or false." |
|
1649 |
||
1650 |
|mySize "{ Class: SmallInteger }
|
|
| 608 | 1651 |
|
| 5242 | 1652 |
sameAs:aString ignoreCase:ignoreCase |
1653 |
"Compare the receiver with the argument. |
|
1654 |
If ignoreCase is true, this is the same as #sameAs:, |
|
1655 |
if false, this is the same as #=." |
|
1656 |
||
1657 |
ignoreCase ifTrue:[ |
|
| 9191 | 1658 |
^ self sameAs:aString |
| 5242 | 1659 |
]. |
1660 |
^ self = aString |
|
1661 |
||
1662 |
" |
|
| 8874 | 1663 |
'foo' sameAs:'Foo' ignoreCase:false |
1664 |
'foo' sameAs:'foo' ignoreCase:true |
|
| 5242 | 1665 |
" |
1666 |
||
1667 |
! |
|
1668 |
||
| 608 | 1669 |
sameCharacters:aString |
| 9190 | 1670 |
"count & return the number of characters which are the sameameCharacters: 'foo' |
| 8874 | 1671 |
'foobarbaz' sameCharacters: 'Foo' |
1672 |
'foobarbaz' sameCharacters: 'baz' |
|
| 608 | 1673 |
" |
|
5510
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1674 |
! |
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1675 |
|
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1676 |
sameEmphasisAs:aStringOrText |
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1677 |
"compare the receivers and the arguments emphasis" |
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1678 |
|
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1679 |
^ self emphasis = aStringOrText emphasis |
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1680 |
|
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1681 |
" |
| 8874 | 1682 |
'hello' asText sameEmphasisAs: 'hello' |
1683 |
'hello' asText sameEmphasisAs: 'hello' asText |
|
| 9190 | 1684 |
sameCharacters:aString |
1685 |
"count & return the number of characters which are the same |
|
1686 |
(ignoring case and emphasis) in the receiver and the argument, aString." |
|
1687 |
||
1688 |
|n "{ Class: SmallInteger }
|
|
1689 |
||
1690 |
sameEmphasisAs:aStringOrText |
|
1691 |
"compare the receivers and the arguments emphasis" |
|
1692 |
||
1693 |
^ self emphasis = aStringOrText emphasis |
|
1694 |
||
1695 |
" |
|
1696 |
'hello' asText sameEmphasisAs: 'hello' |
|
1697 |
'hello' asText sameEmphasisAs: 'hello' asText |
|
| 8874 | 1698 |
'hello' asText allBold sameEmphasisAs: 'hello' |
1699 |
'hello' asText allBold sameEmphasisAs: 'fooba' asText allBold |
|
| 9190 | 1700 |
'hello' asText allBold sameEm 'fooba' asText allBold |
1701 |
'hello' sameStringAndEmphasisAs: 'fooba' asText allItalic |
|
1702 |
" |
|
1703 |
! ! |
|
1704 |
||
1705 |
!CharacterArray methodsFor:'converting'! |
|
1706 |
||
1707 |
asArrayOfSubstrings |
|
1708 |
"return an array of substrings from the receiver, interpreting |
|
1709 |
separators (i.e. spaces & newlines) as word-delimiters. |
|
1710 |
This is a compatibility method - the actual work is done in |
|
1711 |
asCollectionOfWords." |
|
1712 |
||
| 9191 | 1713 |
^ self asCollectionOfWords asArray |
1714 |
sameStringAndEmphasisAs:aStringOrText |
|
|
5739
4fd3be7ccc75
oops some relational ops did accept a non-string arg
Claus Gittinger <cg@exept.de>
parents:
5732
diff
changeset
|
1715 |
"compare both emphasis and string of the receiver and the argument" |
|
5510
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1716 |
|
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1717 |
aStringOrText isString ifFalse:[^ false]. |
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1718 |
(self string = aStringOrText string) ifFalse:[^ false]. |
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1719 |
self hasChangeOfEmphasis = aStringOrText hasChangeOfEmphasis ifFalse:[^ false]. |
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1720 |
^ self emphasis = aStringOrText emphasis |
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1721 |
|
|
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1722 |
" |
| 8874 | 1723 |
'hello' asText sameEmphasisAs: 'hello' |
1724 |
'hello' asText sameEmphasisAs: 'hello' asText |
|
1725 |
'hello' asText allBold sameEmphasisAs: 'hello' |
|
1726 |
'hello' asText allBold sameEmphasisAs: 'fooba' asText allBold |
|
1727 |
'hello' asText allBold sameEmphasisAs: 'fooba' asText allItalic |
|
1728 |
||
1729 |
'hello' sameEmphasisAs: 'hello' asText |
|
1730 |
'hello' sameEmphasisAs: 'hello' asText allBold |
|
1731 |
'hello' sameEmphasisAs: 'fooba' |
|
1732 |
'hello' sameEmphasisAs: 'fooba' asText |
|
1733 |
'hello' sameEmphasisAs: 'fooba' asText allBold |
|
1734 |
'hello' sameEmphasisAs: 'fooba' asText allItalic |
|
1735 |
||
1736 |
'hello' asText sameStringAndEmphasisAs: 'hello' |
|
1737 |
'hello' asText sameStringAndEmphasisAs: 'hello' asText |
|
1738 |
'hello' asText allBold sameStringAndEmphasisAs: 'hello' |
|
1739 |
'hello' asText allBold sameStringAndEmphasisAs: 'fooba' asText allBold |
|
1740 |
'hello' asText allBold sameStringAndEmphasisAs: 'fooba' asText allItalic |
|
1741 |
||
1742 |
'hello' sameStringAndEmphasisAs: 'hello' asText |
|
1743 |
'hello' sameStringAndEmphasisAs: 'hello' asText allBold |
|
1744 |
'hello' sameStringAndEmphasisAs: 'fooba' |
|
1745 |
'hello' sameStringAndEmphasisAs: 'fooba' asText |
|
1746 |
'hello' sameStringAndEmphasisAs: 'fooba' asText allBold |
|
1747 |
'hello' sameStringAndEmphasisAs: 'fooba' asText allItalic |
|
|
5510
a261b8f5d66f
moved #sameEmphasisAs: and #sameStringAndEmphasisAs:
Claus Gittinger <cg@exept.de>
parents:
5487
diff
changeset
|
1748 |
" |
| 9190 | 1749 |
|
| 608 | 1750 |
|
1751 |
asCollectionOfLines |
|
| 8874 | 1752 |
"return a collection containing the lines (separated by cr) |
| 608 | 1753 |
of the receiver. If multiple cr's occur in a row, the result will |
1754 |
contain empty strings." |
|
1755 |
||
1756 |
^ self asCollectionOfSubstringsSeparatedBy:Character cr |
|
1757 |
||
1758 |
" |
|
1759 |
'1 one\2 two\3 three\4 four\5 five' withCRs asCollectionOfLines |
|
| 8874 | 1760 |
'1 one\\\\2 two\3 three' withCRs asCollectionOfLines |
| 608 | 1761 |
" |
1762 |
! |
|
1763 |
||
1764 |
asCollectionOfSubstringsSeparatedBy:aCharacter |
|
| 9190 | 1765 |
"re! |
1766 |
||
1767 |
asCollectionOfSubstringsSeparatedByAll:aSeparatorString |
|
1768 |
"return a collection containing the lines (separated by aSeparatorString) |
|
1769 |
of the receiver. If aSeparatorString occurs multiple times in a row, |
|
| 608 | 1770 |
the result will contain empty strings." |
1771 |
||
| 9190 | 1772 |
^ self asCollectionOfSubCollectionsSeparatedByAll:aSeparatorString |
1773 |
||
1774 |
" |
|
1775 |
'1::2::3::4::5::' asCollectionOfSubstrin |
|
| 608 | 1776 |
|
|
7339
20969b5770c9
asCollectionOfSubstringsSeparatedByAll:
Claus Gittinger <cg@exept.de>
parents:
7318
diff
changeset
|
1777 |
asCollectionOfSubstringsSeparatedByAll:aSeparatorString |
| 8874 | 1778 |
"return a collection containing the lines (separated by aSeparatorString) |
1779 |
of the receiver. If aSeparatorString occurs multiple times in a row, |
|
|
7339
20969b5770c9
asCollectionOfSubstringsSeparatedByAll:
Claus Gittinger <cg@exept.de>
parents:
7318
diff
changeset
|
1780 |
the result will contain empty strings." |
|
20969b5770c9
asCollectionOfSubstringsSeparatedByAll:
Claus Gittinger <cg@exept.de>
parents:
7318
diff
changeset
|
1781 |
|
|
8852
9f5984eff2c2
use asCollectionOfSubCollectionsSeparatedBy*
Claus Gittinger <cg@exept.de>
parents:
8794
diff
changeset
|
1782 |
^ self asCollectionOfSubCollectionsSeparatedByAll:aSeparatorString |
|
9f5984eff2c2
use asCollectionOfSubCollectionsSeparatedBy*
Claus Gittinger <cg@exept.de>
parents:
8794
diff
changeset
|
1783 |
|
|
9f5984eff2c2
use asCollectionOfSubCollectionsSeparatedBy*
Claus Gittinger <cg@exept.de>
parents:
8794
diff
changeset
|
1784 |
" |
| 8874 | 1785 |
'1::2::3::4::5::' asCollectionOfSubstringsSeparatedByAll:'::' |
|
7339
20969b5770c9
asCollectionOfSubstringsSeparatedByAll:
Claus Gittinger <cg@exept.de>
parents:
7318
diff
changeset
|
1786 |
" |
|
20969b5770c9
asCollectionOfSubstringsSeparatedByAll:
Claus Gittinger <cg@exept.de>
parents:
7318
diff
changeset
|
1787 |
! |
|
20969b5770c9
asCollectionOfSubstringsSeparatedByAll:
Claus Gittinger <cg@exept.de>
parents:
7318
diff
changeset
|
1788 |
|
| 608 | 1789 |
asCollectionOfSubstringsSeparatedByAny:aCollectionOfSeparators |
| 9190 | 1790 |
"return a ':' |
1791 |
'hello:world:isnt:this nice' asCollectionOfSubstringsSeparatedByAny:(Array with:$: with:Character space) |
|
1792 |
'hello:world:isnt:this nice' asCollectionOfSubstringsSeparatedByAny:': ' |
|
1793 |
'h1e2l3l4o' asCollectionOfSubstringsSeparatedByAny:($1 to: $9) |
|
1794 |
" |
|
1795 |
! |
|
1796 |
||
1797 |
asCollectionOfWords |
|
1798 |
"return a collection containing the words (separated by whitespace) |
|
1799 |
of the receiver. Multiple occurrencasCollectionOfSubstringsSeparatedByAny:aCollectionOfSeparators |
|
| 608 | 1800 |
"return a collection containing the words (separated by any character |
1801 |
from aCollectionOfSeparators) of the receiver. |
|
1802 |
This allows breaking up strings using any character as separator." |
|
1803 |
||
|
8852
9f5984eff2c2
use asCollectionOfSubCollectionsSeparatedBy*
Claus Gittinger <cg@exept.de>
parents:
8794
diff
changeset
|
1804 |
^ self asCollectionOfSubCollectionsSeparatedByAny:aCollectionOfSeparators |
| 608 | 1805 |
|
1806 |
" |
|
1807 |
'hello:world:isnt:this nice' asCollectionOfSubstringsSeparatedByAny:#($:) |
|
| 8874 | 1808 |
'hello:world:isnt:this nice' asCollectionOfSubstringsSeparatedByAny:':' |
1809 |
'hello:world:isnt:this nice' asCollectionOfSubstringsSeparatedByAny:(Array with:$: with:Character space) |
|
1810 |
'hello:world:isnt:this nice' asCollectionOfSubstringsSeparatedByAny:': ' |
|
1811 |
'h1e2l3l4o' asCollectionOfSubstringsSeparatedByAny:($1 to: $9) |
|
| 608 | 1812 |
" |
1813 |
! |
|
1814 |
||
1815 |
asCollectionOfWords |
|
| 8874 | 1816 |
"return a collection containing the words (separated by whitespace) |
1817 |
of the receiver. Multiple occurrences of whitespace characters will |
|
| 608 | 1818 |
be treated like one - i.e. whitespace is skipped." |
1819 |
||
| 7797 | 1820 |
|words| |
| 608 | 1821 |
|
1822 |
words := OrderedCollection new. |
|
| 7797 | 1823 |
self asCollectionOfWordsDo:[:w | words add:w]. |
| 608 | 1824 |
^ words |
1825 |
||
1826 |
" |
|
1827 |
'hello world isnt this nice' asCollectionOfWords |
|
1828 |
' hello world isnt this nice ' asCollectionOfWords |
|
1829 |
'hello' asCollectionOfWords |
|
1830 |
'' asCollectionOfWords |
|
1831 |
' ' asCollectionOfWords |
|
| 9049 | 1832 |
' foo bar__baz__bla__ bar ' asCollectionOfWords |
1833 |
' foo __bar__baz__bla__ bar ' asCollectionOfWords |
|
| 608 | 1834 |
" |
| 9190 | 1835 |
isasCollectionOfWordsDo:aBlock |
| 8874 | 1836 |
"evaluate aBlock for each word (separated by whitespace) of the receiver. |
1837 |
Multiple occurrences of whitespace characters will be treated like one |
|
| 7797 | 1838 |
- i.e. whitespace is skipped. |
1839 |
Returns the number of words (i.e. the number of invocations of aBlock)." |
|
1840 |
||
1841 |
|count "{ Class:SmallInteger }"
|
|
| 8874 | 1842 |
start "{ Class:SmallInteger }"
|
1843 |
stop "{ Class:SmallInteger }"
|
|
| 7797 | 1844 |
mySize "{ Class:SmallInteger }"|
|
1845 |
||
1846 |
count := 0. |
|
1847 |
start := 1. |
|
1848 |
mySize := self size. |
|
1849 |
[start <= mySize] whileTrue:[ |
|
| 9191 | 1850 |
start := self indexOfNonSeparatorStartingAt:start. |
1851 |
start == 0 ifTrue:[ |
|
1852 |
^ count |
|
1853 |
]. |
|
1854 |
stop := self indexOfSeparatorStartingAt:start. |
|
1855 |
stop == 0 ifTrue:[ |
|
1856 |
aBlock value:(self copyFrom:start to:mySize). |
|
1857 |
^ count + 1 |
|
1858 |
]. |
|
1859 |
aBlock value:(self copyFrom:start to:(stop - 1)). |
|
1860 |
start := stop. |
|
1861 |
count := count + 1 |
|
| 7797 | 1862 |
]. |
1863 |
^ count |
|
1864 |
||
1865 |
" |
|
1866 |
'hello world isnt this nice' asCollectionOfWordsDo:[:w | Transcript showCR:w] |
|
1867 |
' hello world isnt this nice ' asCollectionOfWordsDo:[:w | Transcript showCR:w] |
|
1868 |
'hello' asCollectionOfWordsDo:[:w | Transcript showCR:w] |
|
1869 |
'' asCollectionOfWordsDo:[:w | Transcript showCR:w] |
|
1870 |
' ' asCollectionOfWordsDo:[:w | Transcript showCR:w] |
|
1871 |
" |
|
| 9190 | 1872 |
|
| 608 | 1873 |
|
| 3060 | 1874 |
asFixedPoint |
1875 |
"read a fixedPoint number from the receiver. |
|
| 8874 | 1876 |
Notice, that errors may occur during the read, |
| 7423 | 1877 |
so you better setup some signal handler when using this method." |
| 3060 | 1878 |
|
1879 |
^ FixedPoint readFromString:self |
|
1880 |
||
1881 |
" |
|
| 8874 | 1882 |
'0.123' asFixedPoint |
1883 |
'12345' asFixedPoint |
|
| 3060 | 1884 |
'(1/5)' asFixedPoint |
1885 |
'foo' asFixedPoint |
|
| 9190 | 1886 |
Object errorSignal handle:[:ex | ex r '3.14157' asFixedPoint:2 |
1887 |
'foo' asFixedPoint:2 |
|
1888 |
" |
|
1889 |
||
1890 |
"Modified: / 25.10.1997 / 15:21:57 / cg" |
|
| 3060 | 1891 |
! |
1892 |
||
| 9190 | 1893 |
asFloat |
1894 |
"read a float number from the receiver. |
|
1895 |
Notice, that errors may occur during the read, |
|
1896 |
so you better setup some exception handler when using this method." |
|
1897 |
||
1898 |
^ (Number readFromString:self) asFloat |
|
1899 |
||
1900 |
" |
|
1901 |
'0.123' asFloat |
|
1902 |
'12345' asFloat |
|
1903 |
'(1/5)' asFloat |
|
1904 |
Object errorSignal handle:[:ex | ex returasFixedPoint:scale |
|
| 3060 | 1905 |
"read a fixedPoint number with scale number of post-decimal digits |
1906 |
from the receiver. Scale controls the number of displayed digits, |
|
1907 |
not the number of actually valid digits. |
|
| 8874 | 1908 |
Notice, that errors may occur during the read, |
| 7423 | 1909 |
so you better setup some signal handler when using this method." |
| 3060 | 1910 |
|
1911 |
^ (FixedPoint readFromString:self) scale:scale |
|
1912 |
||
1913 |
" |
|
| 8874 | 1914 |
'0.123' asFixedPoint:2 |
1915 |
'123456' asFixedPoint:2 |
|
1916 |
('3.14157' asFixedPoint:1) asFixedPoint:5
|
|
1917 |
'3.14157' asFixedPoint:2 |
|
1918 |
'foo' asFixedPoint:2 |
|
| 3060 | 1919 |
" |
1920 |
||
1921 |
"Modified: / 25.10.1997 / 15:21:57 / cg" |
|
| 9190 | 1922 |
|
1923 |
||
1924 |
aasFloat |
|
| 608 | 1925 |
"read a float number from the receiver. |
| 8874 | 1926 |
Notice, that errors may occur during the read, |
| 7422 | 1927 |
so you better setup some exception handler when using this method." |
| 608 | 1928 |
|
1929 |
^ (Number readFromString:self) asFloat |
|
1930 |
||
1931 |
" |
|
| 8874 | 1932 |
'0.123' asFloat |
| 608 | 1933 |
'12345' asFloat |
1934 |
'(1/5)' asFloat |
|
| 8874 | 1935 |
Object errorSignal handle:[:ex | ex return:0] do:['foo' asFloat] |
| 608 | 1936 |
" |
1937 |
! |
|
1938 |
||
1939 |
asInteger |
|
| 9190 | 1940 |
"convert the receiver into an integer. |
| 8874 | 1941 |
Notice, that errors may occur during the read, |
| 9190 | 1942 |
so you better setup some exception handler when using this method. |
1943 |
Also notice, that this method here is more strict than the code found |
|
1944 |
in other smalltalks. |
|
1945 |
For less strict integer reading, use Integer readFrom:aString" |
|
| 608 | 1946 |
|
1947 |
^ Integer readFromString:self |
|
1948 |
||
1949 |
" |
|
1950 |
'12345678901234567890' asInteger |
|
1951 |
'-1234' asInteger |
|
| 9190 | 1952 |
|
1953 |
The following raises an error: |
|
1954 |
'0.123' asInteger <- reader stops at ., returning 0 here |
|
1955 |
||
1956 |
whereas the less strict readFrom does not: |
|
1957 |
Integer readFrom:'0.123' <- reader stops at ., returning 0 here |
|
1958 |
||
1959 |
'0.123' asInteger |
|
| 608 | 1960 |
'0.123' asNumber <- returns what you expect |
| 8874 | 1961 |
Object errorSignal handle:[:ex | ex return:0] do:['foo' asInteger] |
| 9190 | 1962 |
|
1963 |
" |
|
1964 |
:c |
|
1965 |
asLowercase |
|
| 608 | 1966 |
"return a copy of myself in lowercase letters" |
1967 |
||
|
8003
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
1968 |
|newStr c bitsPerCharacter |
| 9190 | 1969 |
mySize "{ Class: SmallInteger } ].
|
1970 |
newString at:1 put:firstCharAsLowercase. |
|
1971 |
^ newString |
|
| 608 | 1972 |
|
1973 |
" |
|
| 8874 | 1974 |
'HelloWorld' asLowercase |
1975 |
'HelloWorld' asLowercaseFirst |
|
| 608 | 1976 |
" |
1977 |
! |
|
1978 |
||
| 9190 | 1979 |
asNumber |
1980 |
"read a number from the receiver. |
|
1981 |
Notice, that (in contrast to ST-80) errors may occur during the read, |
|
1982 |
so you better setup some signal handler when using this method. |
|
1983 |
Also notice, that this is meant to read end-user numbers from a string; |
|
1984 |
it does not handle smalltalk nuasLowercaseFirst |
|
|
8003
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
1985 |
"return a copy of myself where the first character is converted to lowercase." |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
1986 |
|
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
1987 |
|newString firstChar firstCharAsLowercase| |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
1988 |
|
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
1989 |
firstChar := (self at:1). |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
1990 |
firstCharAsLowercase := firstChar asLowercase. |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
1991 |
firstChar == firstCharAsLowercase ifTrue:[ ^ self]. |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
1992 |
|
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
1993 |
firstCharAsLowercase bitsPerCharacter > self bitsPerCharacter ifTrue:[ |
| 9191 | 1994 |
newString := firstCharAsLowercase stringSpecies fromString:self. |
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
1995 |
] ifFalse:[ |
| 9191 | 1996 |
newString := self stringSpecies fromString:self. |
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
1997 |
]. |
|
8003
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
1998 |
newString at:1 put:firstCharAsLowercase. |
| 608 | 1999 |
^ newString |
2000 |
||
2001 |
" |
|
| 8874 | 2002 |
'HelloWorld' asLowercase |
2003 |
'HelloWorld' asLowercaseFirst |
|
| 608 | 2004 |
" |
| 9190 | 2005 |
|
2006 |
||
2007 |
asNumber |
|
| 608 | 2008 |
"read a number from the receiver. |
| 8874 | 2009 |
Notice, that (in contrast to ST-80) errors may occur during the read, |
| 608 | 2010 |
so you better setup some signal handler when using this method. |
| 9135 | 2011 |
Also notice, that this is meant to read end-user numbers from a string; |
2012 |
it does not handle smalltalk numbers (i.e. radix). |
|
2013 |
To read a smalltalk number, use Number >> readSmalltalkFrom:. |
|
| 608 | 2014 |
This may change if ANSI specifies it." |
2015 |
||
2016 |
"/ ST-80 behavior: |
|
2017 |
"/ ^ Number readFromString:self onError:0 |
|
2018 |
||
2019 |
^ Number readFromString:self |
|
2020 |
||
2021 |
" |
|
2022 |
'123' asNumber |
|
2023 |
'123.567' asNumber |
|
2024 |
'(5/6)' asNumber |
|
2025 |
'foo' asNumber |
|
| 8874 | 2026 |
Object errorSignal handle:[:ex | ex returnWith:0] do:['foo' asNumber] |
| 608 | 2027 |
" |
| 9190 | 2028 |
'DasNumberFromFormatString:ignored |
| 608 | 2029 |
"read a number from the receiver, ignoring any nonDigit characters. |
2030 |
This is typically used to convert from strings which include |
|
2031 |
dollar-signs or millenium digits. However, this method also ignores |
|
2032 |
the decimal point (if any) and therefore should be used with care." |
|
2033 |
||
2034 |
|tempString| |
|
2035 |
||
2036 |
tempString := self collect:[:char | char isDigit]. |
|
2037 |
^ Number readFromString:tempString onError:0 |
|
2038 |
||
2039 |
" |
|
2040 |
'USD 123' asNumberFromFormatString:'foo' |
|
2041 |
'DM 123' asNumberFromFormatString:'foo' |
|
2042 |
'123' asNumberFromFormatString:'foo' |
|
2043 |
'123.567' asNumberFromFormatString:'foo' |
|
2044 |
'(5/6)' asNumberFromFormatString:'foo' |
|
2045 |
'foo' asNumberFromFormatString:'foo' |
|
2046 |
" |
|
| 9190 | 2047 |
|
| 608 | 2048 |
|
| 7977 | 2049 |
asSingleByteStringIfPossible |
2050 |
"if possible, return the receiver converted to a 'normal' string. |
|
2051 |
It is only possible, if there are no characters with codePoints above 255 in the receiver." |
|
2052 |
||
2053 |
self bitsPerCharacter == 8 ifTrue:[^ self]. |
|
|
8102
e0537422e2d3
Use the ANSI-blessed #codePoint instead of deprecated #asciiValue
Stefan Vogel <sv@exept.de>
parents:
8083
diff
changeset
|
2054 |
(self contains:[:char | char codePoint > 255]) ifFalse:[^ self asSingleByteString]. |
| 9190 | 2055 |
^ |
| 9191 | 2056 |
] ifFalse:[ |
2057 |
newString at:idx put:replacementCharacter |
|
2058 |
]. |
|
| 9190 | 2059 |
]. |
2060 |
^ newString |
|
2061 |
||
2062 |
"Created: 30.6.1997 / 13:02:14 / cg" |
|
| 7977 | 2063 |
! |
2064 |
||
| 9190 | 2065 |
asSoundexCode |
2066 |
"return a soundex string or nil. |
|
2067 |
Soundex returns similar codes for similar sounding words, making it a useful |
|
2068 |
tool when searching for words where the correct spelling is unknown. |
|
2069 |
(read Knuth or search the web if you dont know what a soundex code is). |
|
2070 |
Caveat: 'similar soundingasSingleByteStringReplaceInvalidWith:replacementCharacter |
|
| 2732 | 2071 |
"return the receiver converted to a 'normal' string, |
2072 |
with invalid characters replaced by replacementCharacter. |
|
2073 |
Can be used to convert from 16-bit strings to 8-bit strings |
|
2074 |
and replace characters above code-255 with some replacement." |
|
2075 |
||
2076 |
|newString| |
|
2077 |
||
2078 |
newString := String new:(self size). |
|
2079 |
1 to:self size do:[:idx | |
|
| 9191 | 2080 |
|char| |
2081 |
||
2082 |
char := self at:idx. |
|
2083 |
char codePoint <= 16rFF ifTrue:[ |
|
2084 |
newString at:idx put:char |
|
2085 |
] ifFalse:[ |
|
2086 |
newString at:idx put:replacementCharacter |
|
2087 |
]. |
|
| 2732 | 2088 |
]. |
2089 |
^ newString |
|
2090 |
||
2091 |
"Created: 30.6.1997 / 13:02:14 / cg" |
|
| 9190 | 2092 |
lltalasSoundexCode |
|
8003
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2093 |
"return a soundex string or nil. |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2094 |
Soundex returns similar codes for similar sounding words, making it a useful |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2095 |
tool when searching for words where the correct spelling is unknown. |
|
8862
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
2096 |
(read Knuth or search the web if you dont know what a soundex code is). |
|
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
2097 |
Caveat: 'similar sounding words' means: 'similar sounding in english'." |
| 5843 | 2098 |
|
| 6643 | 2099 |
|s d ch last lch n codes sc| |
| 5843 | 2100 |
|
2101 |
s := self readStream. |
|
2102 |
s skipSeparators. |
|
2103 |
s atEnd ifTrue:[ |
|
| 9191 | 2104 |
^ nil |
| 5843 | 2105 |
]. |
2106 |
ch := s next. |
|
2107 |
ch isLetter ifFalse:[ |
|
| 9191 | 2108 |
^ nil |
| 5843 | 2109 |
]. |
2110 |
n := 0. |
|
2111 |
||
2112 |
codes := Dictionary new. |
|
|
8862
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
2113 |
codes atAll:'bpfv' put:$1. |
|
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
2114 |
codes atAll:'cskgjqxz' put:$2. |
|
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
2115 |
codes atAll:'dt' put:$3. |
| 5843 | 2116 |
codes at:$l put:$4. |
|
8862
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
2117 |
codes atAll:'nm' put:$5. |
| 5843 | 2118 |
codes at:$r put:$6. |
2119 |
||
|
8222
f2c454a9a038
replaced '' writeStream by String writeStream
Claus Gittinger <cg@exept.de>
parents:
8124
diff
changeset
|
2120 |
d := String writeStream. |
| 8874 | 2121 |
d nextPut:(ch asUppercase). |
| 5843 | 2122 |
|
2123 |
[s atEnd] whileFalse:[ |
|
| 9191 | 2124 |
ch := s next. |
2125 |
lch := ch asLowercase. |
|
2126 |
lch = last ifFalse:[ |
|
2127 |
last := lch. |
|
2128 |
||
2129 |
sc := codes at:ch ifAbsent:nil. |
|
2130 |
sc notNil ifTrue:[ |
|
2131 |
n < 3 ifTrue:[ |
|
2132 |
d nextPut:sc. |
|
2133 |
n := n + 1. |
|
2134 |
] |
|
2135 |
] ifFalse:[ |
|
| 5843 | 2136 |
"/ ch isLetter ifFalse:[ |
2137 |
"/ "/ something else - ignore it |
|
2138 |
"/ ] ifTrue:[ |
|
| 8874 | 2139 |
"/ "/ else its a vowel and we ignore it |
| 5843 | 2140 |
"/ ] |
| 9191 | 2141 |
]. |
2142 |
] |
|
| 5843 | 2143 |
]. |
2144 |
[ n < 3 ] whileTrue:[ |
|
| 9191 | 2145 |
d nextPut:$0. |
2146 |
n := n + 1. |
|
| 5843 | 2147 |
]. |
2148 |
||
2149 |
^ d contents |
|
2150 |
||
2151 |
" |
|
| 8874 | 2152 |
'claus' asSoundexCode |
2153 |
'clause' asSoundexCode |
|
2154 |
'close' asSoundexCode |
|
2155 |
'smalltalk' asSoundexCode |
|
2156 |
'smaltalk' asSoundexCode |
|
2157 |
'smaltak' asSoundexCode |
|
2158 |
'smaltok' asSoundexCode |
|
2159 |
'smoltok' asSoundexCode |
|
2160 |
'aa' asSoundexCode |
|
2161 |
'by' asSoundexCode |
|
2162 |
'bab' asSoundexCode |
|
2163 |
'bob' asSoundexCode |
|
2164 |
'bop' asSoundexCode |
|
| 5843 | 2165 |
" |
| 9190 | 2166 |
|
|
1375
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2167 |
|
| 1434 | 2168 |
asSymbol |
2169 |
"return a unique symbol with the name taken from my characters. |
|
|
8049
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
2170 |
The receiver must be a singleByte-String. |
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
2171 |
TwoByte- and FourByteSymbols are (currently ?) not allowed." |
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
2172 |
|
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
2173 |
|str| |
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
2174 |
|
| 9190 | 2175 |
str := self stf ifTrue:[ |
2176 |
^ s asSymbolIfInterned |
|
2177 |
]. |
|
2178 |
^ nil. |
|
2179 |
||
2180 |
"Created: 22.5.1996 / 16:37:04 / cg" |
|
| 1434 | 2181 |
! |
2182 |
||
| 9190 | 2183 |
asText |
2184 |
"return a Text-object (collection of lines) from myself." |
|
2185 |
||
2186 |
"this test allows for small non-gui apps to be built without libbasic2 (where Text is)" |
|
2187 |
Text isNil ifTrue:[^ self]. |
|
2188 |
asSymbolIfInterned |
|
|
1435
3ee1fde2aa58
allow asSymbolIfInterned for Texts
Claus Gittinger <cg@exept.de>
parents:
1434
diff
changeset
|
2189 |
"if a symbol with the receivers characters is already known, return it. |
|
3ee1fde2aa58
allow asSymbolIfInterned for Texts
Claus Gittinger <cg@exept.de>
parents:
1434
diff
changeset
|
2190 |
Otherwise, return nil. This can be used to query for an existing |
|
3ee1fde2aa58
allow asSymbolIfInterned for Texts
Claus Gittinger <cg@exept.de>
parents:
1434
diff
changeset
|
2191 |
symbol and is the same as |
| 9191 | 2192 |
self knownAsSymbol ifTrue:[self asSymbol] ifFalse:[nil] |
|
8049
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
2193 |
but slightly faster, since the symbol lookup operation is only performed once. |
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
2194 |
The receiver must be a singleByte-String. |
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
2195 |
TwoByte- and FourByteSymbols are (currently ?) not allowed." |
|
1435
3ee1fde2aa58
allow asSymbolIfInterned for Texts
Claus Gittinger <cg@exept.de>
parents:
1434
diff
changeset
|
2196 |
|
| 6642 | 2197 |
|s| |
2198 |
||
2199 |
||
2200 |
s := self string. |
|
2201 |
s ~~ self ifTrue:[ |
|
2202 |
^ s asSymbolIfInterned |
|
2203 |
]. |
|
2204 |
^ nil. |
|
|
1435
3ee1fde2aa58
allow asSymbolIfInterned for Texts
Claus Gittinger <cg@exept.de>
parents:
1434
diff
changeset
|
2205 |
|
|
3ee1fde2aa58
allow asSymbolIfInterned for Texts
Claus Gittinger <cg@exept.de>
parents:
1434
diff
changeset
|
2206 |
"Created: 22.5.1996 / 16:37:04 / cg" |
| 9190 | 2207 |
at:i) asLowercase. |
| 9191 | 2208 |
]. |
2209 |
c bitsPerCharacter > bitsPerCharacter ifTrue:[ |
|
2210 |
newStr := c stringSpecies fromString:newStr. |
|
2211 |
]. |
|
2212 |
newStr at:i put:c |
|
| 9190 | 2213 |
]. |
2214 |
^ newStr |
|
2215 |
||
2216 |
" |
|
2217 |
'helloWorld' asTitlecase |
|
2218 |
'HelloWorld' asTitlecase |
|
2219 |
'HELLOWORLD' asTitlecaseasTitlecase |
|
| 8027 | 2220 |
"return a version of the receiver, where the first character is converted to titlecase, |
2221 |
and everything else to lowercase. |
|
2222 |
See the comment in Character on what titlecase is." |
|
2223 |
||
2224 |
|newStr c bitsPerCharacter |
|
2225 |
mySize "{ Class: SmallInteger }" |
|
|
2226 |
||
2227 |
mySize := self size. |
|
2228 |
newStr := self species new:mySize. |
|
2229 |
bitsPerCharacter := newStr bitsPerCharacter. |
|
2230 |
||
2231 |
1 to:mySize do:[:i | |
|
| 9191 | 2232 |
i == 1 ifTrue:[ |
2233 |
c := (self at:i) asTitlecase. |
|
2234 |
] ifFalse:[ |
|
2235 |
c := (self at:i) asLowercase. |
|
2236 |
]. |
|
2237 |
c bitsPerCharacter > bitsPerCharacter ifTrue:[ |
|
2238 |
newStr := c stringSpecies fromString:newStr. |
|
2239 |
]. |
|
2240 |
newStr at:i put:c |
|
| 8027 | 2241 |
]. |
2242 |
^ newStr |
|
2243 |
||
2244 |
" |
|
| 8874 | 2245 |
'helloWorld' asTitlecase |
2246 |
'HelloWorld' asTitlecase |
|
2247 |
'HELLOWORLD' asTitlecase |
|
2248 |
'helloworld' asTitlecase |
|
| 8027 | 2249 |
" |
| 9190 | 2250 |
stChaasTitlecaseFirst |
| 8023 | 2251 |
"return a version of the receiver, where the first character is converted to titlecase. |
| 8024 | 2252 |
Titlecase is much like uppercase for most characters, with the exception of some combined |
2253 |
(2-character glyphs), which consist of an upper- and lower-case characters. |
|
2254 |
If the first character is already titlecase, or there is no titlecasepercase for it, return the |
|
| 8023 | 2255 |
receiver." |
2256 |
||
| 8026 | 2257 |
" |
| 8874 | 2258 |
For example, in Unicode, character U+01F3 is LATIN SMALL LETTER DZ. |
2259 |
(Let us write this compound character using ASCII as 'dz'.) |
|
2260 |
This character uppercases to character U+01F1, LATIN CAPITAL LETTER DZ. |
|
2261 |
(Which is basically 'DZ'.) |
|
2262 |
But it titlecases to to character U+01F2, LATIN CAPITAL LETTER D WITH SMALL LETTER Z. |
|
| 8026 | 2263 |
(Which we can write 'Dz'.) |
2264 |
||
2265 |
character uppercase titlecase |
|
2266 |
--------- --------- --------- |
|
2267 |
dz DZ Dz |
|
2268 |
" |
|
2269 |
||
| 8023 | 2270 |
|newString firstChar firstCharAsTitlecase| |
2271 |
||
2272 |
firstChar := (self at:1). |
|
| 8025 | 2273 |
firstCharAsTitlecase := firstChar asTitlecase. |
| 8023 | 2274 |
firstChar == firstCharAsTitlecase ifTrue:[ ^ self]. |
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
2275 |
|
|
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
2276 |
firstCharAsTitlecase bitsPerCharacter > self bitsPerCharacter ifTrue:[ |
| 9191 | 2277 |
newString := firstCharAsTitlecase stringSpecies fromString:self. |
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
2278 |
] ifFalse:[ |
| 9191 | 2279 |
newString := self stringSpecies fromString:self. |
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
2280 |
]. |
| 8023 | 2281 |
newString at:1 put:firstCharAsTitlecase. |
2282 |
^ newString |
|
2283 |
||
2284 |
" |
|
| 8874 | 2285 |
'helloWorld' asTitlecaseFirst |
2286 |
'HelloWorld' asTitlecaseFirst |
|
| 8023 | 2287 |
" |
| 9190 | 2288 |
|
| 8710 | 2289 |
|
| 5776 | 2290 |
asURL |
2291 |
"return an URL-object from myself." |
|
2292 |
||
2293 |
^ URL fromString:self |
|
2294 |
||
2295 |
" |
|
| 8874 | 2296 |
'http://www.exept.de:80/index.html' asURL host |
2297 |
'http://www.exept.de:80/index.html' asURL port |
|
| 9190 | 2298 |
'http://www.exept.de:80/index.html' asURL |
|
8295
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2299 |
|
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2300 |
asUnicode32String |
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2301 |
"thats not really true - characters above ascii 16r7F may need special treatment" |
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2302 |
|
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2303 |
^ ((Unicode32String new:self size) replaceFrom:1 to:self size with:self startingAt:1) |
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2304 |
! |
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2305 |
|
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2306 |
asUnicodeString |
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2307 |
"thats not really true - characters above ascii 16r7F may need special treatment" |
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2308 |
|
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2309 |
^ ((UnicodeString new:self size) replaceFrom:1 to:self size with:self startingAt:1) |
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2310 |
! |
|
8433d819226b
Define #asUnicode16String and #asUnicode32String in CharacterArray
Stefan Vogel <sv@exept.de>
parents:
8222
diff
changeset
|
2311 |
|
| 608 | 2312 |
asUppercase |
2313 |
"return a copy of myself in uppercase letters" |
|
2314 |
||
| 9190 | 2315 |
|newStr c bitsPerChara:c |
2316 |
]. |
|
2317 |
^ newStr |
|
2318 |
||
2319 |
" |
|
2320 |
'helloWorld' asUppercase |
|
2321 |
'helloWorld' asUppercaseFirst |
|
2322 |
(Character value:16rB5) asString asUppercase -- needs 16 bits !! |
|
2323 |
" |
|
2324 |
! |
|
2325 |
||
2326 |
asUppercaseFirst |
|
2327 |
"return a version of the receiver, where the first character is converted to uppercase. |
|
2328 |
If the first character is already uppercase, or there is no uppercase for it, return the |
|
2329 |
asUppercase |
|
2330 |
"return a copy of myself in uppercase letters" |
|
2331 |
||
|
8003
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2332 |
|newStr c bitsPerCharacter |
| 608 | 2333 |
mySize "{ Class: SmallInteger }" |
|
2334 |
||
2335 |
mySize := self size. |
|
|
8954
12e701d7b5ac
allow for receiver to be empty in asLowercase and asUppercase.
Claus Gittinger <cg@exept.de>
parents:
8938
diff
changeset
|
2336 |
mySize == 0 ifTrue:[^ self]. |
| 608 | 2337 |
newStr := self species new:mySize. |
|
8003
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2338 |
bitsPerCharacter := newStr bitsPerCharacter. |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2339 |
|
| 608 | 2340 |
1 to:mySize do:[:i | |
|
8954
12e701d7b5ac
allow for receiver to be empty in asLowercase and asUppercase.
Claus Gittinger <cg@exept.de>
parents:
8938
diff
changeset
|
2341 |
c := (self at:i) asUppercase. |
|
12e701d7b5ac
allow for receiver to be empty in asLowercase and asUppercase.
Claus Gittinger <cg@exept.de>
parents:
8938
diff
changeset
|
2342 |
c bitsPerCharacter > bitsPerCharacter ifTrue:[ |
|
12e701d7b5ac
allow for receiver to be empty in asLowercase and asUppercase.
Claus Gittinger <cg@exept.de>
parents:
8938
diff
changeset
|
2343 |
newStr := c stringSpecies fromString:newStr. |
|
12e701d7b5ac
allow for receiver to be empty in asLowercase and asUppercase.
Claus Gittinger <cg@exept.de>
parents:
8938
diff
changeset
|
2344 |
]. |
|
12e701d7b5ac
allow for receiver to be empty in asLowercase and asUppercase.
Claus Gittinger <cg@exept.de>
parents:
8938
diff
changeset
|
2345 |
newStr at:i put:c |
| 608 | 2346 |
]. |
2347 |
^ newStr |
|
2348 |
||
2349 |
" |
|
| 8874 | 2350 |
'helloWorld' asUppercase |
2351 |
'helloWorld' asUppercaseFirst |
|
| 9190 | 2352 |
(Character value:16rB5) asString asUppercase -- needs 16 bits irstCharAsUppeasUppercaseFirst |
|
8003
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2353 |
"return a version of the receiver, where the first character is converted to uppercase. |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2354 |
If the first character is already uppercase, or there is no uppercase for it, return the |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2355 |
receiver." |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2356 |
|
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2357 |
|newString firstChar firstCharAsUppercase| |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2358 |
|
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2359 |
firstChar := (self at:1). |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2360 |
firstCharAsUppercase := firstChar asUppercase. |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2361 |
firstChar == firstCharAsUppercase ifTrue:[ ^ self]. |
|
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2362 |
|
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
2363 |
firstCharAsUppercase bitsPerCharacter > self bitsPerCharacter ifTrue:[ |
| 9191 | 2364 |
newString := firstCharAsUppercase stringSpecies fromString:self. |
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
2365 |
] ifFalse:[ |
| 9191 | 2366 |
newString := self stringSpecies fromString:self. |
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
2367 |
]. |
|
8003
6e690f4b79f1
asLowercase/asUppercase and friends care for 16-bit chars
Claus Gittinger <cg@exept.de>
parents:
7995
diff
changeset
|
2368 |
newString at:1 put:firstCharAsUppercase. |
| 608 | 2369 |
^ newString |
2370 |
||
2371 |
" |
|
| 8874 | 2372 |
'helloWorld' asUppercase |
2373 |
'helloWorld' asUppercaseFirst |
|
2374 |
'HelloWorld' asUppercaseFirst |
|
| 608 | 2375 |
" |
| 9190 | 2376 |
#bold) |
2377 |
" |
|
2378 |
||
2379 |
"Modified: 28.6.1997 / 00:13:17 / cg" |
|
| 608 | 2380 |
! |
2381 |
||
| 9190 | 2382 |
,, aString |
2383 |
"concatenate with a newLine in between" |
|
2384 |
||
2385 |
^ (self copyWith:Character cr) , aString |
|
2386 |
||
2387 |
" |
|
2388 |
hello ,, world |
|
2389 |
'hello' ,, 'world' |
|
2390 |
" |
|
2391 |
! |
|
2392 |
||
2393 |
concatenate:string1 and:string2 |
|
2394 |
"return the concatenation of myself and the arguments, string1 and string2. |
|
2395 |
This is equivalent to self , string1 , string2 |
|
2396 |
- generated by compiler when such a construct is detected and the receiver |
|
2397 |
is known to be a strin, aStringOrCharacter |
|
|
2720
98c21d9a07c0
allow mixed strings to be concatenated using comma
Claus Gittinger <cg@exept.de>
parents:
2692
diff
changeset
|
2398 |
"redefined to allow characters and mixed strings to be appended. |
|
1050
2f14bbd792a9
, now accepts a character-argument
Claus Gittinger <cg@exept.de>
parents:
1028
diff
changeset
|
2399 |
This is nonStandard, but convenient" |
|
2f14bbd792a9
, now accepts a character-argument
Claus Gittinger <cg@exept.de>
parents:
1028
diff
changeset
|
2400 |
|
|
2720
98c21d9a07c0
allow mixed strings to be concatenated using comma
Claus Gittinger <cg@exept.de>
parents:
2692
diff
changeset
|
2401 |
|myWidth otherWidth| |
|
98c21d9a07c0
allow mixed strings to be concatenated using comma
Claus Gittinger <cg@exept.de>
parents:
2692
diff
changeset
|
2402 |
|
|
1050
2f14bbd792a9
, now accepts a character-argument
Claus Gittinger <cg@exept.de>
parents:
1028
diff
changeset
|
2403 |
aStringOrCharacter isCharacter ifTrue:[ |
| 9191 | 2404 |
^ self copyWith:aStringOrCharacter |
|
1050
2f14bbd792a9
, now accepts a character-argument
Claus Gittinger <cg@exept.de>
parents:
1028
diff
changeset
|
2405 |
]. |
|
1396
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2406 |
aStringOrCharacter isText ifTrue:[ |
| 9191 | 2407 |
^ aStringOrCharacter concatenateFromString:self |
|
1396
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2408 |
]. |
|
2720
98c21d9a07c0
allow mixed strings to be concatenated using comma
Claus Gittinger <cg@exept.de>
parents:
2692
diff
changeset
|
2409 |
aStringOrCharacter isString ifTrue:[ |
| 9191 | 2410 |
(otherWidth := aStringOrCharacter bitsPerCharacter) ~~ (myWidth := self bitsPerCharacter) ifTrue:[ |
2411 |
otherWidth > myWidth ifTrue:[ |
|
2412 |
^ (aStringOrCharacter species fromString:self) , aStringOrCharacter |
|
2413 |
]. |
|
2414 |
^ self , (self species fromString:aStringOrCharacter) |
|
2415 |
]. |
|
|
2720
98c21d9a07c0
allow mixed strings to be concatenated using comma
Claus Gittinger <cg@exept.de>
parents:
2692
diff
changeset
|
2416 |
]. |
|
1050
2f14bbd792a9
, now accepts a character-argument
Claus Gittinger <cg@exept.de>
parents:
1028
diff
changeset
|
2417 |
^ super , aStringOrCharacter |
|
2f14bbd792a9
, now accepts a character-argument
Claus Gittinger <cg@exept.de>
parents:
1028
diff
changeset
|
2418 |
|
|
2f14bbd792a9
, now accepts a character-argument
Claus Gittinger <cg@exept.de>
parents:
1028
diff
changeset
|
2419 |
" |
| 8874 | 2420 |
'hello' , $1 |
2421 |
'hello' , '1' |
|
2422 |
'hello' , (' world' asText allBold)
|
|
2423 |
'hello' , (JISEncodedString fromString:' world') |
|
|
2720
98c21d9a07c0
allow mixed strings to be concatenated using comma
Claus Gittinger <cg@exept.de>
parents:
2692
diff
changeset
|
2424 |
(JISEncodedString fromString:'hello') , ' world' |
| 1411 | 2425 |
|
| 1422 | 2426 |
Transcript showCR: |
| 9191 | 2427 |
(Text string:'hello' emphasis:#italic) , (Text string:' world' emphasis:#bold) |
|
1050
2f14bbd792a9
, now accepts a character-argument
Claus Gittinger <cg@exept.de>
parents:
1028
diff
changeset
|
2428 |
" |
|
2f14bbd792a9
, now accepts a character-argument
Claus Gittinger <cg@exept.de>
parents:
1028
diff
changeset
|
2429 |
|
|
2720
98c21d9a07c0
allow mixed strings to be concatenated using comma
Claus Gittinger <cg@exept.de>
parents:
2692
diff
changeset
|
2430 |
"Modified: 28.6.1997 / 00:13:17 / cg" |
| 9190 | 2431 |
|
2432 |
||
2433 |
,,, aString |
|
| 7020 | 2434 |
"concatenate with a newLine in between" |
| 6965 | 2435 |
|
2436 |
^ (self copyWith:Character cr) , aString |
|
2437 |
||
2438 |
" |
|
2439 |
hello ,, world |
|
| 7020 | 2440 |
'hello' ,, 'world' |
| 6965 | 2441 |
" |
2442 |
! |
|
2443 |
||
| 608 | 2444 |
concatenate:string1 and:string2 |
2445 |
"return the concatenation of myself and the arguments, string1 and string2. |
|
2446 |
This is equivalent to self , string1 , string2 |
|
2447 |
- generated by compiler when such a construct is detected and the receiver |
|
2448 |
is known to be a string." |
|
2449 |
||
2450 |
^ self , string1 , string2 |
|
2451 |
! |
|
2452 |
||
2453 |
concatenate:string1 and:string2 and:string3 |
|
2454 |
"return the concatenation of myself and the string arguments. |
|
2455 |
This is equivalent to self , string1 , string2 , string3 |
|
2456 |
- generated by compiler when such a construct is detected and the receiver |
|
2457 |
is known to be a string." |
|
2458 |
||
2459 |
^ self , string1 , string2 , string3 |
|
|
8040
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2460 |
! |
|
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2461 |
|
| 8347 | 2462 |
copyReplaceString:subString withString:newString |
2463 |
"return a copy of the receiver, with all sequences of subString replaced |
|
2464 |
by newString (i.e. slice in the newString in place of the oldString)." |
|
2465 |
||
2466 |
|tmpStream idx idx1| |
|
2467 |
||
2468 |
tmpStream := WriteStream on:(self class new). |
|
2469 |
idx := 1. |
|
2470 |
[idx ~~ 0] whileTrue:[ |
|
| 9191 | 2471 |
idx1 := idx. |
2472 |
idx := self indexOfSubCollection:subString startingAt:idx. |
|
2473 |
idx ~~ 0 ifTrue:[ |
|
2474 |
tmpStream nextPutAll:(self copyFrom:idx1 to:idx-1). |
|
2475 |
tmpStream nextPutAll:newString. |
|
2476 |
idx := idx + subString size |
|
2477 |
] |
|
| 8347 | 2478 |
]. |
2479 |
tmpStream nextPutAll:(self copyFrom:idx1). |
|
2480 |
^ tmpStream contents |
|
2481 |
||
2482 |
" |
|
| 8874 | 2483 |
'12345678901234567890' replString:'123' withString:'OneTwoThree' |
2484 |
'12345678901234567890' replString:'123' withString:'*' |
|
2485 |
'12345678901234567890' replString:'234' withString:'foo' |
|
| 8347 | 2486 |
|
2487 |
('a string with spaces' replChar:$ withString:' foo ')
|
|
| 9191 | 2488 |
replString:'foo' withString:'bar' |
| 8347 | 2489 |
" |
2490 |
||
2491 |
"Modified: / 31-05-1999 / 12:33:59 / cg" |
|
2492 |
"Created: / 12-05-2004 / 12:00:00 / cg" |
|
2493 |
! |
|
2494 |
||
|
8040
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2495 |
copyWith:aCharacter |
|
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2496 |
"return a new string containing the receivers characters |
| 8874 | 2497 |
and the single new character, aCharacter. |
|
8040
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2498 |
This is different from concatentation, which expects another string |
|
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2499 |
as argument, but equivalent to copy-and-addLast. |
| 8874 | 2500 |
The code below cares for different width characters |
|
8040
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2501 |
(i.e. when appending a 16bit char to an 8bit string)" |
|
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2502 |
|
|
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2503 |
|sz newString| |
|
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2504 |
|
|
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2505 |
aCharacter bitsPerCharacter > self bitsPerCharacter ifTrue:[ |
| 9191 | 2506 |
sz := self size. |
2507 |
newString := aCharacter stringSpecies new:sz + 1. |
|
2508 |
newString replaceFrom:1 to:sz with:self startingAt:1. |
|
2509 |
newString at:sz+1 put:aCharacter. |
|
2510 |
^ newString. |
|
|
8040
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2511 |
]. |
|
f10cca439066
copyWith for mixed-width string/character
Claus Gittinger <cg@exept.de>
parents:
8037
diff
changeset
|
2512 |
^ super copyWith:aCharacter |
| 608 | 2513 |
! ! |
2514 |
||
2515 |
!CharacterArray methodsFor:'displaying'! |
|
2516 |
||
| 2762 | 2517 |
displayOn:aGC x:x y:y from:start to:stop |
2518 |
"display the receiver on a GC" |
|
2519 |
||
2520 |
"q&d hack" |
|
2521 |
||
2522 |
(self copyFrom:start to:stop) displayOn:aGC x:x y:y opaque:false |
|
2523 |
||
2524 |
"Modified: 12.5.1996 / 12:49:33 / cg" |
|
2525 |
! |
|
2526 |
||
| 7082 | 2527 |
displayOn:aGc x:x y:y opaque:opaque |
| 2762 | 2528 |
"display the receiver in a graphicsContext - this method allows |
2529 |
strings to be used like DisplayObjects." |
|
2530 |
||
2531 |
|s| |
|
2532 |
||
2533 |
s := self string. |
|
| 7082 | 2534 |
opaque ifTrue:[ |
| 9191 | 2535 |
aGc displayOpaqueString:s x:x y:y. |
| 2762 | 2536 |
] ifFalse:[ |
| 9191 | 2537 |
aGc displayString:s x:x y:y. |
| 2762 | 2538 |
]. |
| 1381 | 2539 |
|
2540 |
"Modified: 11.5.1996 / 14:42:48 / cg" |
|
2541 |
! |
|
2542 |
||
| 2762 | 2543 |
displayOpaqueOn:aGC x:x y:y from:start to:stop |
2544 |
"display the receiver on a GC" |
|
2545 |
||
2546 |
"q&d hack" |
|
2547 |
||
2548 |
(self copyFrom:start to:stop) displayOn:aGC x:x y:y opaque:true |
|
2549 |
||
2550 |
"Created: 12.5.1996 / 12:29:37 / cg" |
|
2551 |
"Modified: 12.5.1996 / 12:49:19 / cg" |
|
| 608 | 2552 |
! ! |
2553 |
||
|
1375
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2554 |
!CharacterArray methodsFor:'emphasis'! |
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2555 |
|
| 5732 | 2556 |
allBold |
2557 |
"return a test object representing the receiver, but all boldified" |
|
2558 |
||
| 8874 | 2559 |
"this test allows for small non-gui apps to be built without libbasic2 (where Text is)" |
| 6187 | 2560 |
Text isNil ifTrue:[^ self]. |
| 5732 | 2561 |
^ self asText allBold |
2562 |
||
2563 |
" |
|
2564 |
Transcript showCR:'hello' asText allBold |
|
2565 |
Transcript showCR:'hello' allBold |
|
2566 |
" |
|
2567 |
! |
|
2568 |
||
|
5898
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2569 |
allItalic |
|
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2570 |
"return a test object representing the receiver, but all in italic" |
|
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2571 |
|
| 8874 | 2572 |
"this test allows for small non-gui apps to be built without libbasic2 (where Text is)" |
| 6187 | 2573 |
Text isNil ifTrue:[^ self]. |
|
5898
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2574 |
^ self asText allItalic |
|
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2575 |
|
|
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2576 |
" |
|
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2577 |
Transcript showCR:'hello' asText allItalic |
|
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2578 |
Transcript showCR:'hello' allItalic |
|
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2579 |
" |
|
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2580 |
! |
|
7bc4f1593b4e
allItalic now understood by CharacterArray
Claus Gittinger <cg@exept.de>
parents:
5848
diff
changeset
|
2581 |
|
| 6228 | 2582 |
colorizeAllWith:aColor |
2583 |
^ self asText colorizeAllWith:aColor |
|
2584 |
||
2585 |
" |
|
2586 |
Transcript showCR:('hello' colorizeAllWith:Color red)
|
|
2587 |
Transcript showCR:('world' colorizeAllWith:Color green darkened)
|
|
2588 |
" |
|
2589 |
! |
|
2590 |
||
|
1396
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2591 |
emphasis |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2592 |
"return the emphasis. |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2593 |
Since characterArrays do not hold any emphasis information, |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2594 |
nil (no emphasis) is returned here." |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2595 |
|
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2596 |
^ RunArray new:self size withAll:nil |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2597 |
|
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2598 |
"Created: 14.5.1996 / 13:58:58 / cg" |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2599 |
! |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2600 |
|
| 6655 | 2601 |
emphasis:emphasisCollection |
2602 |
^ self asText emphasis:emphasisCollection |
|
2603 |
||
2604 |
" |
|
2605 |
Transcript showCR:('hello' emphasis:#(bold bold bold bold bold))
|
|
2606 |
" |
|
2607 |
! |
|
2608 |
||
|
1375
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2609 |
emphasisAt:characterIndex |
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2610 |
"return the emphasis at some index. |
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2611 |
Since characterArrays do not hold any emphasis information, |
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2612 |
nil (no emphasis) is returned here." |
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2613 |
|
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2614 |
^ nil |
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2615 |
|
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2616 |
"Created: 11.5.1996 / 14:13:27 / cg" |
|
1396
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2617 |
! |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2618 |
|
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2619 |
emphasisCollection |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2620 |
"return the emphasis. |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2621 |
Since characterArrays do not hold any emphasis information, |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2622 |
nil (no emphasis) is returned here." |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2623 |
|
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2624 |
^ RunArray new:(self size) |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2625 |
|
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2626 |
"Created: 14.5.1996 / 13:58:58 / cg" |
|
5b5386311dbd
concatenating mixed texts & strings
Claus Gittinger <cg@exept.de>
parents:
1393
diff
changeset
|
2627 |
"Modified: 14.5.1996 / 15:02:29 / cg" |
| 3404 | 2628 |
! |
2629 |
||
2630 |
emphasizeAllWith:emphasis |
|
2631 |
^ self asText emphasizeAllWith:emphasis |
|
2632 |
||
2633 |
" |
|
2634 |
Transcript showCR:('hello' emphasizeAllWith:#bold)
|
|
| 3599 | 2635 |
Transcript showCR:('hello' emphasizeAllWith:(#color -> Color red))
|
2636 |
Transcript showCR:('hello' emphasizeAllWith:(#color -> Color red))
|
|
2637 |
" |
|
2638 |
||
2639 |
"Modified: / 17.6.1998 / 12:51:44 / cg" |
|
| 5097 | 2640 |
! |
2641 |
||
2642 |
makeSelectorBoldIn:aClass |
|
2643 |
"the receiver represents some source code for |
|
2644 |
a method in aClass. |
|
2645 |
Change myself to boldify the selector. |
|
2646 |
Not yet implemented (could easily use the syntaxHighlighter |
|
2647 |
for this ...)" |
|
2648 |
||
2649 |
^ self |
|
2650 |
||
2651 |
"Modified: / 13.12.1999 / 21:49:11 / cg" |
|
2652 |
"Created: / 13.12.1999 / 21:49:24 / cg" |
|
| 6653 | 2653 |
! |
2654 |
||
2655 |
withoutEmphasis:emphasisToRemove |
|
2656 |
^ self |
|
|
1375
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2657 |
! ! |
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
2658 |
|
| 7265 | 2659 |
!CharacterArray methodsFor:'encoding & decoding'! |
|
991
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
2660 |
|
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
2661 |
decodeFrom:encodingSymbol |
| 8874 | 2662 |
"given the receiver encoded as described by encodingSymbol, |
|
8124
ca774158b317
code cleanup (removed old obsolete encoding stuff)
Claus Gittinger <cg@exept.de>
parents:
8102
diff
changeset
|
2663 |
convert it into internal ST/X (unicode) encoding and return a corresponding CharacterArray." |
|
ca774158b317
code cleanup (removed old obsolete encoding stuff)
Claus Gittinger <cg@exept.de>
parents:
8102
diff
changeset
|
2664 |
|
|
ca774158b317
code cleanup (removed old obsolete encoding stuff)
Claus Gittinger <cg@exept.de>
parents:
8102
diff
changeset
|
2665 |
|myEncoding encoder| |
| 7950 | 2666 |
|
2667 |
encodingSymbol isNil ifTrue:[^ self]. |
|
2668 |
myEncoding := self encoding. |
|
2669 |
encodingSymbol == myEncoding ifTrue:[^ self]. |
|
2670 |
||
2671 |
encoder := CharacterEncoder encoderToEncodeFrom:(self encoding) into:encodingSymbol. |
|
|
8124
ca774158b317
code cleanup (removed old obsolete encoding stuff)
Claus Gittinger <cg@exept.de>
parents:
8102
diff
changeset
|
2672 |
encoder isNil ifTrue:[^ self]. |
|
ca774158b317
code cleanup (removed old obsolete encoding stuff)
Claus Gittinger <cg@exept.de>
parents:
8102
diff
changeset
|
2673 |
^ encoder decodeString:self. |
|
991
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
2674 |
! |
|
f0b45a4b82d2
first attempt in encoding/decoding for different character representations (mac/msdos etc.)
Claus Gittinger <cg@exept.de>
parents:
986
diff
changeset
|
2675 |
|
| 9190 | 2676 |
poilers. |
2677 |
A major advantage of rot13 over rot(N) for other N is that it |
|
2678 |
is self-inverse, so the same code crot13 |
|
| 8874 | 2679 |
"Usenet: from `rotate alphabet 13 places'] |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2680 |
The simple Caesar-cypher encryption that replaces each English |
| 8874 | 2681 |
letter with the one 13 places forward or back along the alphabet, |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2682 |
so that 'The butler did it!!' becomes 'Gur ohgyre qvq vg!!' |
| 8874 | 2683 |
Most Usenet news reading and posting programs include a rot13 feature. |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2684 |
It is used to enclose the text in a sealed wrapper that the reader must choose |
| 8874 | 2685 |
to open -- e.g., for posting things that might offend some readers, or spoilers. |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2686 |
A major advantage of rot13 over rot(N) for other N is that it |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2687 |
is self-inverse, so the same code can be used for encoding and decoding." |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2688 |
|
| 8874 | 2689 |
^ self species |
| 9191 | 2690 |
streamContents:[:aStream | |
2691 |
self do:[:char | |
|
2692 |
aStream nextPut:char rot13 ]] |
|
| 8874 | 2693 |
|
2694 |
" |
|
2695 |
'hello world' rot13 |
|
2696 |
'hello world' rot13 rot13 |
|
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2697 |
" |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2698 |
! |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2699 |
|
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2700 |
utf8Decoded |
| 7950 | 2701 |
"Interpreting myself as an UTF-8 representation, decode and return the decoded string." |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2702 |
|
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2703 |
|in out is16Bit c| |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2704 |
|
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2705 |
is16Bit := false. |
|
8222
f2c454a9a038
replaced '' writeStream by String writeStream
Claus Gittinger <cg@exept.de>
parents:
8124
diff
changeset
|
2706 |
out := WriteStream on:(String uninitializedNew:self size). |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2707 |
in := self readStream. |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2708 |
[in atEnd] whileFalse:[ |
| 9191 | 2709 |
c := Character utf8DecodeFrom:in. |
2710 |
is16Bit ifFalse:[ |
|
2711 |
c codePoint > 16rFF ifTrue:[ |
|
2712 |
out := WriteStream with:(UnicodeString fromString:out contents). |
|
2713 |
is16Bit := true. |
|
2714 |
]. |
|
2715 |
]. |
|
2716 |
out nextPut:c. |
|
| 8874 | 2717 |
]. |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2718 |
^ out contents |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2719 |
|
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2720 |
" |
| 8874 | 2721 |
#[16rC8 16rA0] asString utf8Decoded |
2722 |
(Character value:16r220) utf8Encoded |
|
2723 |
(Character value:16r220) utf8Encoded utf8Decoded |
|
2724 |
||
2725 |
(Character value:16r800) utf8Encoded |
|
2726 |
(Character value:16r220) utf8Encoded utf8Decoded |
|
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2727 |
" |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2728 |
|
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2729 |
"test: |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2730 |
|
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2731 |
|utf8Encoding original readBack| |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2732 |
|
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2733 |
1 to:16rFFFF do:[:ascii | |
| 9191 | 2734 |
original := (Character value:ascii) asString. |
2735 |
utf8Encoding := original utf8Encoded. |
|
2736 |
readBack := utf8Encoding utf8Decoded. |
|
2737 |
readBack = original ifFalse:[ |
|
2738 |
self halt |
|
2739 |
] |
|
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2740 |
] |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2741 |
" |
| 9190 | 2742 |
nextutf8DecodedWithTwoByteCharactersReplacedBy:replacementCharacter |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2743 |
"Interpreting myself as an UTF-8 representation, decode and return |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2744 |
the decoded string. Suppress all 2-byte (above 16rFF) characters, |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2745 |
and replace them with replacementCharacter" |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2746 |
|
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2747 |
|in out c| |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2748 |
|
|
8222
f2c454a9a038
replaced '' writeStream by String writeStream
Claus Gittinger <cg@exept.de>
parents:
8124
diff
changeset
|
2749 |
out := WriteStream on:(String uninitializedNew:self size). |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2750 |
in := self readStream. |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2751 |
[in atEnd] whileFalse:[ |
| 9191 | 2752 |
c := Character utf8DecodeFrom:in. |
2753 |
c codePoint > 16rFF ifTrue:[ |
|
2754 |
c := replacementCharacter |
|
2755 |
]. |
|
2756 |
out nextPut:c. |
|
| 8874 | 2757 |
]. |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2758 |
^ out contents |
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2759 |
|
|
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2760 |
" |
| 8874 | 2761 |
(Character value:16r220) utf8Encoded |
| 9191 | 2762 |
utf8DecodedWithTwoByteCharactersReplacedBy:(Character space) |
|
6835
bdb2cb03200c
added utf8Decode with 8-bit chars only
Claus Gittinger <cg@exept.de>
parents:
6834
diff
changeset
|
2763 |
" |
| 9190 | 2764 |
|
2765 |
||
2766 |
centerPaddedTo:size with:padCharacter |
|
2767 |
"return a new string of length size, which contains the receiver |
|
2768 |
centered (i.e. padded on both sides). |
|
2769 |
Characters are filled with padCharacter. |
|
| 8874 | 2770 |
If the receivers size is equal or greater than the length argument, |
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2771 |
the original receiver is returned unchanged." |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2772 |
|
| 9190 | 2773 |
|len s| |
2774 |
||
2775 |
len := self size. |
|
2776 |
(len < |
|
2777 |
" |
|
2778 |
'foo' centerPaddedTo:11 with:$. |
|
2779 |
'fooBar' centerPaddedTo:5 with:$. |
|
2780 |
123 printString centerPaddedTo:10 with:$. |
|
2781 |
(' ' , 123 printString) centerPaddedTo:10 with:$.
|
|
2782 |
(Float pi printString) centerPaddedTo:15 with:(Character space) |
|
2783 |
(Float pi printString) centerPaddedTo:15 with:$- |
|
2784 |
(' ' , Float pi class name) centerPaddedTo:15 with:$.
|
|
2785 |
" |
|
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2786 |
! |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2787 |
|
| 9190 | 2788 |
decimalPaddedTo:size and:afterPeriod at:decimalCharacter |
2789 |
"return a new string of overall length size, whicenterPaddedTo:size with:padCharacter |
|
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2790 |
"return a new string of length size, which contains the receiver |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2791 |
centered (i.e. padded on both sides). |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2792 |
Characters are filled with padCharacter. |
| 8874 | 2793 |
If the receivers size is equal or greater than the length argument, |
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2794 |
the original receiver is returned unchanged." |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2795 |
|
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2796 |
|len s| |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2797 |
|
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2798 |
len := self size. |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2799 |
(len < size) ifTrue:[ |
| 9191 | 2800 |
s := self species new:size withAll:padCharacter. |
2801 |
s replaceFrom:(size - len) // 2 + 1 with:self. |
|
2802 |
^ s |
|
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2803 |
] |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2804 |
|
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2805 |
" |
| 8874 | 2806 |
'foo' centerPaddedTo:11 with:$. |
2807 |
'fooBar' centerPaddedTo:5 with:$. |
|
2808 |
123 printString centerPaddedTo:10 with:$. |
|
2809 |
(' ' , 123 printString) centerPaddedTo:10 with:$.
|
|
2810 |
(Float pi printString) centerPaddedTo:15 with:(Character space) |
|
2811 |
(Float pi printString) centerPaddedTo:15 with:$- |
|
2812 |
(' ' , Float pi class name) centerPaddedTo:15 with:$.
|
|
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2813 |
" |
| 9190 | 2814 |
123.1decimalPaddedTo:size and:afterPeriod at:decimalCharacter |
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2815 |
"return a new string of overall length size, which contains the receiver |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2816 |
aligned at the decimal-period column and afterPeriod characters to the right |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2817 |
of the period. The periodCharacter is passed as arguments (allowing for US and European formats |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2818 |
to be padded). |
| 8874 | 2819 |
If the receivers size is equal or greater than the length argument, |
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2820 |
the original receiver is returned unchanged. |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2821 |
(sounds complicated ? -> see examples below)." |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2822 |
|
| 8874 | 2823 |
^ self |
| 9191 | 2824 |
decimalPaddedTo:size |
2825 |
and:afterPeriod |
|
2826 |
at:decimalCharacter |
|
2827 |
withLeft:(Character space) |
|
2828 |
right:$0 |
|
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2829 |
|
| 8874 | 2830 |
" |
2831 |
'123' decimalPaddedTo:10 and:3 at:$. -> ' 123 ' |
|
2832 |
'123' decimalPaddedTo:10 and:3 at:$. -> ' 123.000' |
|
2833 |
'123.' decimalPaddedTo:10 and:3 at:$. -> ' 123.000' |
|
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2834 |
'123.1' decimalPaddedTo:10 and:3 at:$. -> ' 123.100' |
| 8874 | 2835 |
'123.1' decimalPaddedTo:10 and:3 at:$. -> ' 123.1 ' |
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2836 |
'123.123' decimalPaddedTo:10 and:3 at:$. -> ' 123.123' |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2837 |
" |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2838 |
|
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2839 |
"Created: 23.12.1995 / 13:11:52 / cg" |
| 9190 | 2840 |
: 23.decimalPaddedTo:size and:afterPeriod at:decimalCharacter withLeft:leftPadChar right:rightPadChar |
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2841 |
"return a new string of overall length size, which contains the receiver |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2842 |
aligned at the decimal-period column and afterPeriod characters to the right |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2843 |
of the period. |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2844 |
Characters on the left are filled with leftPadChar. |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2845 |
If rightPadChar is nil, characters on the right are filled with leftPadCharacter too; |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2846 |
otherwise, if missing, a decimal point is added and right characters filled with this. |
| 8874 | 2847 |
If the receivers size is equal or greater than the length argument, |
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2848 |
the original receiver is returned unchanged. |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2849 |
(sounds complicated ? -> see examples below)." |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2850 |
|
|
5230
655c21ffe6cc
care for negative size of rest-string when decimalPadding
Claus Gittinger <cg@exept.de>
parents:
5221
diff
changeset
|
2851 |
|s idx n rest| |
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2852 |
|
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2853 |
idx := self indexOf:decimalCharacter. |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2854 |
idx == 0 ifTrue:[ |
| 9191 | 2855 |
"/ |
2856 |
"/ no decimal point found; adjust string to the left of the period column |
|
2857 |
"/ |
|
2858 |
rightPadChar isNil ifTrue:[ |
|
2859 |
s := self , (self species new:afterPeriod + 1 withAll:leftPadChar) |
|
2860 |
] ifFalse:[ |
|
2861 |
s:= self , decimalCharacter asString , (self species new:afterPeriod withAll:rightPadChar). |
|
2862 |
]. |
|
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2863 |
] ifFalse:[ |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2864 |
|
| 9191 | 2865 |
"/ the number of after-decimalPoint characters |
2866 |
n := self size - idx. |
|
2867 |
rest := afterPeriod - n. |
|
2868 |
rest > 0 ifTrue:[ |
|
2869 |
s := (self species new:rest withAll:(rightPadChar ? leftPadChar)). |
|
2870 |
] ifFalse:[ |
|
2871 |
s := '' |
|
2872 |
]. |
|
2873 |
s := self , s. |
|
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2874 |
]. |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2875 |
|
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2876 |
^ s leftPaddedTo:size with:leftPadChar |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2877 |
|
| 8874 | 2878 |
" |
2879 |
'123' decimalPaddedTo:10 and:3 at:$. withLeft:(Character space) right:nil -> ' 123 ' |
|
2880 |
'123' decimalPaddedTo:10 and:3 at:$. withLeft:(Character space) right:$0 -> ' 123.000' |
|
2881 |
'123.' decimalPaddedTo:10 and:3 at:$. withLeft:(Character space) right:$0 -> ' 123.000' |
|
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2882 |
'123.1' decimalPaddedTo:10 and:3 at:$. withLeft:(Character space) right:$0 -> ' 123.100' |
| 8874 | 2883 |
'123.1' decimalPaddedTo:10 and:3 at:$. withLeft:(Character space) right:nil -> ' 123.1 ' |
|
815
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2884 |
'123.123' decimalPaddedTo:10 and:3 at:$. withLeft:(Character space) right:$0 -> ' 123.123' |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2885 |
" |
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2886 |
|
|
997a02c135e7
support for decimal padding added
Claus Gittinger <cg@exept.de>
parents:
759
diff
changeset
|
2887 |
"Modified: 23.12.1995 / 13:08:18 / cg" |
| 9190 | 2888 |
|
| 608 | 2889 |
|
2890 |
paddedTo:newSize |
|
2891 |
"return a new string consisting of the receivers characters, |
|
2892 |
plus spaces up to length. |
|
| 8874 | 2893 |
If the receivers size is equal or greater than the length argument, |
| 608 | 2894 |
the original receiver is returned unchanged." |
2895 |
||
2896 |
^ self paddedTo:newSize with:(Character space) |
|
2897 |
||
2898 |
" |
|
| 8874 | 2899 |
'foo' paddedTo:10 |
2900 |
123 printString paddedTo:10 |
|
| 608 | 2901 |
" |
| 64 | 2902 |
! ! |
2903 |
||
| 328 | 2904 |
!CharacterArray methodsFor:'pattern matching'! |
2905 |
||
| 2551 | 2906 |
compoundMatch:aString |
| 9190 | 2907 |
"like match, but thfied: / 16.12.1999 / 01:22:08 / cg" |
2908 |
! |
|
2909 |
||
2910 |
compoundMatch:aString ignoreCase:ignoreCase |
|
2911 |
"like match, but the receiver may be a compound match pattern, |
|
2912 |
consisting of multiple simple patterns, separated by semicolons. |
|
2913 |
This is usable with fileName pattern fields." |
|
2914 |
||
2915 |
|matchers| |
|
2916 |
||
2917 |
matchers := self asCollectionOfSubstringsSeparatedBy:$;. |
|
2918 |
matchers do:[:aPattern | |
|
| 9191 | 2919 |
(aPattern match:aString igcompoundMatch:aString |
| 3253 | 2920 |
"like match, but the receiver may be a compound match pattern, |
| 2551 | 2921 |
consisting of multiple simple patterns, separated by semicolons. |
2922 |
This is usable with fileName pattern fields." |
|
2923 |
||
|
5124
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2924 |
^self compoundMatch:aString ignoreCase:false |
|
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2925 |
|
|
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2926 |
" |
| 8874 | 2927 |
'f*' match:'foo' |
2928 |
'b*' match:'foo' |
|
2929 |
'f*;b*' match:'foo' |
|
2930 |
'f*;b*' match:'bar' |
|
2931 |
'f*;b*' compoundMatch:'foo' |
|
2932 |
'f*;b*' compoundMatch:'bar' |
|
|
5124
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2933 |
" |
|
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2934 |
|
|
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2935 |
"Modified: / 30.1.1998 / 11:40:18 / stefan" |
|
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2936 |
"Modified: / 16.12.1999 / 01:22:08 / cg" |
| 9190 | 2937 |
h:'focompoundMatch:aString ignoreCase:ignoreCase |
|
5124
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2938 |
"like match, but the receiver may be a compound match pattern, |
|
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2939 |
consisting of multiple simple patterns, separated by semicolons. |
|
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2940 |
This is usable with fileName pattern fields." |
|
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2941 |
|
| 2551 | 2942 |
|matchers| |
2943 |
||
2944 |
matchers := self asCollectionOfSubstringsSeparatedBy:$;. |
|
2945 |
matchers do:[:aPattern | |
|
| 9191 | 2946 |
(aPattern match:aString ignoreCase:ignoreCase) ifTrue:[^ true]. |
| 2551 | 2947 |
]. |
2948 |
^ false. |
|
2949 |
||
2950 |
" |
|
| 8874 | 2951 |
'f*' match:'foo' |
2952 |
'b*' match:'foo' |
|
2953 |
'f*;b*' match:'foo' |
|
2954 |
'f*;b*' match:'bar' |
|
2955 |
'f*;b*' compoundMatch:'foo' |
|
2956 |
'f*;b*' compoundMatch:'bar' |
|
2957 |
'f*;b*' compoundMatch:'Foo' ignoreCase:true |
|
| 2551 | 2958 |
" |
2959 |
||
| 3253 | 2960 |
"Modified: / 15.4.1997 / 15:50:33 / cg" |
2961 |
"Modified: / 30.1.1998 / 11:40:18 / stefan" |
|
|
5124
727f88b16fb3
case-insensitive version of compoundMatch: added
Claus Gittinger <cg@exept.de>
parents:
5097
diff
changeset
|
2962 |
"Created: / 16.12.1999 / 01:21:35 / cg" |
| 9190 | 2963 |
|
| 328 | 2964 |
|
2965 |
findMatchString:matchString startingAt:index |
|
2966 |
"like findString, but allowing match patterns. |
|
2967 |
find matchstring, starting at index. if found, return the index; |
|
2968 |
if not found, return 0." |
|
2969 |
||
| 9190 | 2970 |
^ self findMatchString:matchString startingAt:index ignoreCaseartingAt:3 ignoreCase:true ifAbsent:0 |
2971 |
" |
|
2972 |
||
2973 |
"Modified: 13.9.1997 / 06:31:22 / cg" |
|
| 328 | 2974 |
! |
2975 |
||
| 9190 | 2976 |
includesMatchString:matchString |
2977 |
"like includesString, but allowing match patterns. |
|
2978 |
find matchstring; if found, return true, otherwise return false" |
|
2979 |
||
2980 |
^ (self findMatchString:findMatchString:matchString startingAt:index ignoreCase:ignoreCase ifAbsent:exceptionBlock |
|
| 328 | 2981 |
"like findString, but allowing match patterns. |
2982 |
find matchstring, starting at index. if found, return the index; |
|
2983 |
if not found, return the result of evaluating exceptionBlock. |
|
2984 |
This is a q&d hack - not very efficient" |
|
2985 |
||
| 2916 | 2986 |
|firstChar firstSet |
| 328 | 2987 |
startIndex "{ Class: SmallInteger }"
|
2988 |
matchSize "{ Class: SmallInteger }"
|
|
2989 |
mySize "{ Class: SmallInteger }"
|
|
|
8032
af9606171d86
string match for 16bit characters (do not use #== to compare characters)
Claus Gittinger <cg@exept.de>
parents:
8027
diff
changeset
|
2990 |
realMatchString lcChar ucChar| |
| 328 | 2991 |
|
2992 |
matchSize := matchString size. |
|
2993 |
matchSize == 0 ifTrue:[^ index]. "empty string matches" |
|
2994 |
||
2995 |
realMatchString := matchString. |
|
2996 |
(realMatchString endsWith:$*) ifFalse:[ |
|
| 9191 | 2997 |
realMatchString := realMatchString , '*'. |
2998 |
matchSize := matchSize + 1 |
|
| 328 | 2999 |
]. |
3000 |
||
3001 |
mySize := self size. |
|
3002 |
firstChar := realMatchString at:1. |
|
|
5342
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
3003 |
firstChar == self class matchEscapeCharacter ifTrue:[ |
| 9191 | 3004 |
firstChar := realMatchString at:2. |
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
3005 |
]. |
| 328 | 3006 |
|
3007 |
firstChar asString includesMatchCharacters ifTrue:[ |
|
| 9191 | 3008 |
index to:mySize do:[:col | |
3009 |
(realMatchString match:self from:col to:mySize ignoreCase:ignoreCase) |
|
3010 |
ifTrue:[^ col] |
|
3011 |
]. |
|
3012 |
^ exceptionBlock value. |
|
| 328 | 3013 |
]. |
|
8032
af9606171d86
string match for 16bit characters (do not use #== to compare characters)
Claus Gittinger <cg@exept.de>
parents:
8027
diff
changeset
|
3014 |
|
|
af9606171d86
string match for 16bit characters (do not use #== to compare characters)
Claus Gittinger <cg@exept.de>
parents:
8027
diff
changeset
|
3015 |
lcChar := firstChar asLowercase. |
|
af9606171d86
string match for 16bit characters (do not use #== to compare characters)
Claus Gittinger <cg@exept.de>
parents:
8027
diff
changeset
|
3016 |
ucChar := firstChar asUppercase. |
|
af9606171d86
string match for 16bit characters (do not use #== to compare characters)
Claus Gittinger <cg@exept.de>
parents:
8027
diff
changeset
|
3017 |
(ignoreCase and:[ lcChar ~= ucChar]) ifTrue:[ |
| 9191 | 3018 |
firstSet := Array with:ucChar with:lcChar. |
3019 |
startIndex := self indexOfAny:firstSet startingAt:index. |
|
| 2916 | 3020 |
] ifFalse:[ |
| 9191 | 3021 |
startIndex := self indexOf:firstChar startingAt:index. |
| 2916 | 3022 |
]. |
| 328 | 3023 |
[startIndex == 0] whileFalse:[ |
| 9191 | 3024 |
(realMatchString match:self from:startIndex to:mySize ignoreCase:ignoreCase) |
3025 |
ifTrue:[^ startIndex]. |
|
3026 |
firstSet notNil ifTrue:[ |
|
3027 |
startIndex := self indexOfAny:firstSet startingAt:(startIndex + 1). |
|
3028 |
] ifFalse:[ |
|
3029 |
startIndex := self indexOf:firstChar startingAt:(startIndex + 1). |
|
3030 |
]. |
|
| 328 | 3031 |
]. |
3032 |
^ exceptionBlock value |
|
3033 |
||
3034 |
" |
|
3035 |
'one two three four' findMatchString:'o[nu]' |
|
3036 |
'one two three four' findMatchString:'o[nu]' startingAt:3 |
|
| 8874 | 3037 |
'one two three four one' findMatchString:'ONE' startingAt:3 ignoreCase:true ifAbsent:0 |
| 328 | 3038 |
" |
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
3039 |
|
| 2916 | 3040 |
"Modified: 13.9.1997 / 06:31:22 / cg" |
| 9190 | 3041 |
llo world' includesMatchString:'h*' caseSensitive:false |
3042 |
||
3043 |
'hello world' includesMatchString:'h[aeiou]llo' caseSensitive:true |
|
3044 |
'hello world' includesMatchString:'h[aeiou]llo' caseSensitive:false |
|
3045 |
||
3046 |
'hello world' includesMatchString:'wor*' caseSensitive:true |
|
3047 |
'hello world' includesMatchString:'wor*' caseSensitive:false |
|
3048 |
||
3049 |
'hello world' includesMatchString:'woR*' caseSensitive:true includesMatchString:matchString caseSensitive:caseSensitive |
|
| 9058 | 3050 |
"like includesString, but allowing match patterns. |
3051 |
find matchstring; if found, return true, otherwise return false" |
|
3052 |
||
3053 |
^ (self findMatchString:matchString startingAt:1 ignoreCase:caseSensitive not ifAbsent:0) ~~ 0 |
|
3054 |
||
3055 |
" |
|
3056 |
'hello world' includesMatchString:'h*' caseSensitive:true |
|
3057 |
'hello world' includesMatchString:'h*' caseSensitive:false |
|
3058 |
'Hello world' includesMatchString:'h*' caseSensitive:true |
|
3059 |
'Hello world' includesMatchString:'h*' caseSensitive:false |
|
3060 |
||
3061 |
'hello world' includesMatchString:'h[aeiou]llo' caseSensitive:true |
|
3062 |
'hello world' includesMatchString:'h[aeiou]llo' caseSensitive:false |
|
3063 |
||
3064 |
'hello world' includesMatchString:'wor*' caseSensitive:true |
|
3065 |
'hello world' includesMatchString:'wor*' caseSensitive:false |
|
3066 |
||
3067 |
'hello world' includesMatchString:'woR*' caseSensitive:true |
|
3068 |
'hello world' includesMatchString:'woR*' caseSensitive:false |
|
3069 |
" |
|
| 9190 | 3070 |
:'*f'match:aString |
| 8874 | 3071 |
"return true if aString matches self, where self may contain meta-match |
| 328 | 3072 |
characters $* (to match any string) or $# (to match any character). |
3073 |
or [...] to match a set of characters. |
|
| 345 | 3074 |
Lower/uppercase are considered different. |
| 8874 | 3075 |
NOTICE: match-meta character interpretation is like in unix-matching, |
| 9191 | 3076 |
NOT the ST-80 meaning." |
| 328 | 3077 |
|
3078 |
^ self match:aString from:1 to:aString size ignoreCase:false |
|
3079 |
||
3080 |
" |
|
| 8874 | 3081 |
'\*f*' match:'f' |
3082 |
'\*f*' match:'*f' |
|
3083 |
'*\*f*' match:'*f' |
|
3084 |
'*f*' match:'*f' |
|
3085 |
'*ute*' match:'computer' |
|
3086 |
'*uter' match:'computer' |
|
3087 |
'uter*' match:'computer' |
|
3088 |
'*ute*' match:'' |
|
3089 |
'[abcd]*' match:'computer' |
|
3090 |
'[abcd]*' match:'komputer' |
|
3091 |
'*some*compl*ern*' match:'this is some more complicated pattern match' |
|
3092 |
'*some*compl*ern*' match:'this is another complicated pattern match' |
|
| 3537 | 3093 |
'*-hh' match:'anton-h' |
3094 |
" |
|
3095 |
||
3096 |
"Modified: / 9.6.1998 / 18:50:00 / cg" |
|
| 9190 | 3097 |
match:aString from:start to:stop ignoreCase:ignoreCase |
| 8874 | 3098 |
"return true if part of aString matches myself, |
3099 |
where self may contain meta-match |
|
| 328 | 3100 |
characters $* (to match any string) or $# (to match any character) |
3101 |
or [...] to match a set of characters. |
|
| 345 | 3102 |
If ignoreCase is true, lower/uppercase are considered the same. |
| 8874 | 3103 |
NOTICE: match-meta character interpretation is like in unix-matching, |
| 9191 | 3104 |
NOT the ST-80 meaning." |
| 328 | 3105 |
|
3106 |
|matchScanArray| |
|
3107 |
||
3108 |
" |
|
3109 |
keep the matchScanArray from the most recent match - |
|
3110 |
avoids parsing the pattern over-and over if multiple searches |
|
3111 |
are done with the same pattern. |
|
3112 |
" |
|
3113 |
(PreviousMatch notNil |
|
3114 |
and:[PreviousMatch key = self]) ifTrue:[ |
|
| 9191 | 3115 |
matchScanArray := PreviousMatch value |
| 328 | 3116 |
] ifFalse:[ |
| 9191 | 3117 |
matchScanArray := self class matchScanArrayFrom:self. |
3118 |
matchScanArray isNil ifTrue:[ |
|
3119 |
'CharacterArray [info]: invalid matchpattern:''' infoPrint. self infoPrint. ''' comparing for equality.' infoPrintCR. |
|
3120 |
^ self = aString |
|
| 3926 | 3121 |
"/ ^ false |
| 9191 | 3122 |
]. |
3123 |
PreviousMatch := self -> matchScanArray. |
|
| 328 | 3124 |
]. |
3125 |
||
3126 |
^ self class |
|
| 9191 | 3127 |
matchScan:matchScanArray |
3128 |
from:1 to:matchScanArray size |
|
3129 |
with:aString |
|
3130 |
from:start to:stop |
|
3131 |
ignoreCase:ignoreCase |
|
| 8874 | 3132 |
|
3133 |
" |
|
3134 |
'*ute*' match:'12345COMPUTER' from:1 to:5 ignoreCase:true |
|
3135 |
'*ute*' match:'12345COMPUTER' from:6 to:13 ignoreCase:true |
|
| 328 | 3136 |
" |
| 1429 | 3137 |
|
| 3926 | 3138 |
"Modified: / 10.11.1998 / 21:43:46 / cg" |
| 9190 | 3139 |
compmatch:aString ignoreCase:ignoreCase |
| 8874 | 3140 |
"return true if aString matches self, where self may contain meta-match |
| 608 | 3141 |
characters $* (to match any string) or $# (to match any character) |
3142 |
or [...] to match a set of characters. |
|
3143 |
If ignoreCase is true, lower/uppercase are considered the same. |
|
| 8874 | 3144 |
NOTICE: match-meta character interpretation is like in unix-matching, |
| 9191 | 3145 |
NOT the ST-80 meaning." |
| 608 | 3146 |
|
3147 |
^ self match:aString from:1 to:aString size ignoreCase:ignoreCase |
|
3148 |
||
3149 |
" |
|
| 8874 | 3150 |
'*ute*' match:'COMPUTER' ignoreCase:true |
3151 |
'*uter' match:'COMPUTER' ignoreCase:false |
|
3152 |
'[abcd]*' match:'computer' ignoreCase:false |
|
3153 |
'[abcd]*' match:'Computer' ignoreCase:false |
|
3154 |
'[a-k]*' match:'komputer' ignoreCase:false |
|
3155 |
'[a-k]*' match:'zomputer' ignoreCase:false |
|
3156 |
'[a-k]*' match:'Komputer' ignoreCase:false |
|
3157 |
'[a-k]*' match:'Komputer' ignoreCase:true |
|
3158 |
'*some*compl*ern*' match:'this is some more complicated pattern match' ignoreCase:true |
|
3159 |
'*some*compl*ern*' match:'this is another complicated pattern match' ignoreCase:true |
|
| 608 | 3160 |
|
3161 |
Time millisecondsToRun:[ |
|
| 9191 | 3162 |
Symbol allInstancesDo:[:sym | |
3163 |
'[ab]*' match:sym ignoreCase:false |
|
3164 |
] |
|
| 8874 | 3165 |
]. |
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
3166 |
Time millisecondsToRun:[ |
| 9191 | 3167 |
Symbol allInstancesDo:[:sym | |
3168 |
'*at:*' match:sym ignoreCase:false |
|
3169 |
] |
|
| 8874 | 3170 |
]. |
| 608 | 3171 |
" |
|
2519
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
3172 |
|
|
07674cea76dd
allow escaping a * when doing a match
Claus Gittinger <cg@exept.de>
parents:
2452
diff
changeset
|
3173 |
"Modified: 2.4.1997 / 17:28:58 / cg" |
| 9190 | 3174 |
|
| 608 | 3175 |
|
3176 |
article |
|
3177 |
"return an article string for the receiver." |
|
3178 |
||
3179 |
|firstChar| |
|
3180 |
||
| 8874 | 3181 |
firstChar := (self at:1) asLowercase. |
| 5848 | 3182 |
((firstChar isVowel and:[firstChar ~~ $u]) or:[firstChar == $x]) ifTrue:[ |
| 9191 | 3183 |
^ 'an' |
| 608 | 3184 |
]. |
3185 |
^ 'a' |
|
| 5848 | 3186 |
|
3187 |
" |
|
| 9191 | 3188 |
'uboot' article. |
3189 |
'xmas' article. |
|
3190 |
'alarm' article. |
|
3191 |
'baby' article. |
|
| 5848 | 3192 |
" |
| 608 | 3193 |
! |
3194 |
||
|
8049
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
3195 |
basicStoreString |
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
3196 |
"return a String for storing myself" |
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
3197 |
|
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
3198 |
|s n index| |
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
3199 |
|
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
3200 |
n := self occurrencesOf:$'. |
|
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
3201 |
n == 0 ifFalse:[ |
| 9191 | 3202 |
s := String new:(n |
| 9190 | 3203 |
|
3204 |
basicStoreString |
|
3205 |
"return a String for storing myself" |
|
3206 |
||
3207 |
|s n index| |
|
3208 |
||
3209 |
n := self occurrencesOf:$'. |
|
3210 |
n == 0 ifFalse:[ |
|
| 9191 | 3211 |
s := String new:(n + 2 + self size). |
3212 |
s at:1 put:$'. |
|
3213 |
index := 2. |
|
3214 |
self do:[:thisChar | |
|
3215 |
(thisChar == $') ifTrue:[ |
|
3216 |
s at:index put:thisChar. |
|
3217 |
index := inde |
|
|
8049
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
3218 |
|
| 608 | 3219 |
displayString |
|
8049
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
3220 |
"return a string to display the receiver - use storeString to have quotes around." |
| 608 | 3221 |
|
3222 |
^ self storeString |
|
3223 |
! |
|
3224 |
||
3225 |
printOn:aStream |
|
3226 |
"print the receiver on aStream" |
|
3227 |
||
3228 |
aStream nextPutAll:self |
|
3229 |
! |
|
3230 |
||
3231 |
printString |
|
3232 |
"return a string for printing - thats myself" |
|
3233 |
||
3234 |
^ self |
|
| 3578 | 3235 |
! |
3236 |
||
3237 |
printWithQuotesDoubledOn:aStream |
|
3238 |
"put the raw storeString of myself on aStream" |
|
3239 |
||
3240 |
self do:[:thisChar | |
|
| 9191 | 3241 |
( |
| 9190 | 3242 |
|
3243 |
printWithQuotesDoubledOn:aStream |
|
3244 |
"put the raw storeString of myself on aStream" |
|
3245 |
||
3246 |
self do:[:thisChar | |
|
| 9191 | 3247 |
(thisChar == $') ifTrue:[aStream nextPut:thisChar]. |
3248 |
aStream nextPut:thisChar |
|
| 3578 | 3249 |
] |
3250 |
||
3251 |
"Modified: / 15.6.1998 / 17:21:17 / cg" |
|
3252 |
"Created: / 15.6.1998 / 17:22:13 / cg" |
|
|
8862
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
3253 |
|
| 8874 | 3254 |
isUnarySelector |
|
8862
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
3255 |
"Answer true if the receiver contains only chars in an ANSI unary method selector, false otherwise." |
|
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
3256 |
|
|
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
3257 |
^ (self first isLetter or:[ self first = $_ ]) |
|
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
3258 |
and:[ self allSatisfy: [ :chr | chr isLetterOrDigit ]] |
|
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
3259 |
|
|
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
3260 |
"Modified: / 20-04-2005 / 12:20:43 / cg" |
|
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
3261 |
! ! |
|
6f550a144c6a
isUnarySelector/isBinarySelector -> CharacterArray and fixed
Claus Gittinger <cg@exept.de>
parents:
8852
diff
changeset
|
3262 |
|
| 9190 | 3263 |
!CharacterArray methodsFor:'quer |
| 608 | 3264 |
|
| 4429 | 3265 |
bitsPerCharacter |
| 8874 | 3266 |
"return the underlying strings bitsPerCharacter |
| 4429 | 3267 |
(i.e. is it a regular String or a TwoByteString)" |
3268 |
||
| 8457 | 3269 |
|string max| |
3270 |
||
3271 |
(string := self string) ~~ self ifTrue:[ |
|
| 9191 | 3272 |
^ string bitsPerCharacter |
| 8457 | 3273 |
]. |
3274 |
||
3275 |
max := 8. |
|
3276 |
self do:[:eachCharacter | |
|
| 9191 | 3277 |
max := max max:(eachCharacter bitsPerCharact |
| 8457 | 3278 |
|
3279 |
contains8BitCharacters |
|
| 8874 | 3280 |
"return true, if the underlying string contains 8BitCharacters (or widers) |
| 8457 | 3281 |
(i.e. if it is non-ascii)" |
3282 |
||
3283 |
|string| |
|
3284 |
||
3285 |
(string := self string) ~~ self ifTrue:[ |
|
| 8899 | 3286 |
^ string contains8BitCharacters |
| 8457 | 3287 |
]. |
| 8900 | 3288 |
^ self contains:[:aCharacter | aCharacter codePoint > 16r7F ]. |
| 8457 | 3289 |
|
3290 |
" |
|
| 8874 | 3291 |
'hello' contains8BitCharacters |
3292 |
'hello' asText allBold contains8BitCharacters |
|
| 9191 | 3293 |
" |
3294 |
||
| 4429 | 3295 |
|
| 608 | 3296 |
encoding |
| 1311 | 3297 |
"return the strings encoding, as a symbol. |
| 7982 | 3298 |
Here, by default, we assume unicode-encoding. |
3299 |
Notice, that iso8859-1 is a true subset of unicode, |
|
| 8874 | 3300 |
and that singleByteStrings are therefore both unicode AND |
| 7982 | 3301 |
8859-1 encoded." |
3302 |
||
3303 |
^ #'unicode' |
|
| 608 | 3304 |
! |
3305 |
||
|
1375
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
3306 |
hasChangeOfEmphasis |
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
3307 |
^ false |
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
3308 |
|
| 1381 | 3309 |
"Created: 12.5.1996 / 12:31:39 / cg" |
|
1375
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
3310 |
! |
|
f035bb7468f5
additions for new Text class
Claus Gittinger <cg@exept.de>
parents:
1333
diff
changeset
|
3311 |
|
|
7995
8ddbb403d67b
+hasIcon and hasImage for LabelAndIcon compatibility
Claus Gittinger <cg@exept.de>
parents:
7991
diff
changeset
|
3312 |
hasIcon |
| 9191 | 3313 |
"for LabelAndIcon compatibility" |
3314 |
||
|
7995
8ddbb403d67b
+hasIcon and hasImage for LabelAndIcon compatibility
Claus Gittinger <cg@exept.de>
parents:
7991
diff
changeset
|
3315 |
|
| 1388 | 3316 |
heightOn:aGC |
3317 |
"return the size of the recevier in device units if displayed on aGC" |
|
3318 |
||
| 6528 | 3319 |
^ (aGC font onDevice:aGC device) heightOf:self |
| 1388 | 3320 |
|
3321 |
" |
|
| 8874 | 3322 |
'hello world' heightOn:(View new) |
| 1388 | 3323 |
" |
3324 |
||
3325 |
"Created: 12.5.1996 / 20:09:29 / cg" |
|
3326 |
"Modified: 12.5.1996 / 20:32:05 / cg" |
|
3327 |
! |
|
3328 |
||
| 608 | 3329 |
isString |
3330 |
"return true, if the receiver is some kind of string; |
|
3331 |
true is returned here - redefinition of Object>>isString." |
|
3332 |
||
3333 |
^ true |
|
| 1235 | 3334 |
! |
3335 |
||
| 8350 | 3336 |
isValidSmalltalkIdentifier |
| 9190 | 3337 |
foo_bar_' isValidSmalltalkIdentifier |
3338 |
'foo ' isValidSmalltalkIdentifier |
|
3339 |
' foo' isValidSmalltalkIdentifier |
|
3340 |
" |
|
3341 |
! |
|
3342 |
||
3343 |
knownAsSymbol |
|
3344 |
"for now, only single character strings are allowed as symbols. |
|
3345 |
This method is redefined in String." |
|
3346 |
||
3347 |
^ false |
|
3348 |
! |
|
3349 |
||
3350 |
leftIndent |
|
3351 |
"if the receiver starts with spaces, return the number of spaces |
|
3352 |
at the left - otherwise, return 0. |
|
3353 |
If the receiver consistisValidSmalltalkIdentifier |
|
| 8350 | 3354 |
"return true, if the receivers characters make up a valid smalltalk identifier" |
3355 |
||
3356 |
|scanner tok| |
|
3357 |
||
3358 |
scanner := Compiler new. |
|
3359 |
scanner source:(self readStream). |
|
3360 |
tok := scanner nextToken. |
|
3361 |
tok ~~ #Identifier ifTrue:[ |
|
| 9191 | 3362 |
^ false |
| 8350 | 3363 |
]. |
3364 |
scanner tokenPosition == 1 ifFalse:[^ false]. |
|
3365 |
^ scanner sourceStream atEnd. |
|
3366 |
||
3367 |
" |
|
| 8874 | 3368 |
'foo' isValidSmalltalkIdentifier |
3369 |
'1foo' isValidSmalltalkIdentifier |
|
3370 |
'_foo' isValidSmalltalkIdentifier |
|
3371 |
'_foo_bar_' isValidSmalltalkIdentifier |
|
3372 |
'foo ' isValidSmalltalkIdentifier |
|
3373 |
' foo' isValidSmalltalkIdentifier |
|
| 8350 | 3374 |
" |
| 9190 | 3375 |
|
3376 |
' ' leftIndent |
|
3377 |
" |
|
3378 |
||
3379 |
"Modified: 20.4.1996 / 19:28:43 / cg" |
|
| 8350 | 3380 |
! |
3381 |
||
| 9190 | 3382 |
stringSpecies |
3383 |
"return the underlying strings bitsPerChaleftIndent |
|
| 1235 | 3384 |
"if the receiver starts with spaces, return the number of spaces |
3385 |
at the left - otherwise, return 0. |
|
3386 |
If the receiver consists of spaces only, return the receivers size." |
|
3387 |
||
3388 |
|index "{Class: SmallInteger }"
|
|
| 9190 | 3389 |
end "{Class: SmallInteger }
|
| 1388 | 3390 |
|
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
3391 |
stringSpecies |
| 8874 | 3392 |
"return the underlying strings bitsPerCharacter |
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
3393 |
(i.e. is it a regular String or a TwoByteString)" |
|
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
3394 |
|
|
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
3395 |
|string| |
|
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
3396 |
|
|
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
3397 |
(string := self string) == self ifTrue:[^ self class]. |
|
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
3398 |
^ string stringSpecies |
|
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
3399 |
|
|
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
3400 |
" |
| 8874 | 3401 |
'hello' stringSpecies |
| 9190 | 3402 |
'hello' asText allBold stringSp |
|
8037
8cce25b1059c
care for wide chars in asXXXFirst
Claus Gittinger <cg@exept.de>
parents:
8032
diff
changeset
|
3403 |
|
| 6653 | 3404 |
widthFrom:startIndex to:endIndex on:aGC |
3405 |
"return ths size of part of the recevier in device units if displayed on aGC" |
|
3406 |
||
3407 |
^ (aGC font onDevice:aGC device) widthOf:self from:startIndex to:endIndex |
|
3408 |
||
3409 |
" |
|
| 8874 | 3410 |
'hello world' widthFrom:1 to:5 on:(View new) |
3411 |
'hello' widthOn:(View new) |
|
| 6653 | 3412 |
" |
| 9190 | 3413 |
|
| 6653 | 3414 |
|
| 1388 | 3415 |
widthOn:aGC |
3416 |
"return ths size of the recevier in device units if displayed on aGC" |
|
3417 |
||
| 6528 | 3418 |
^ (aGC font onDevice:aGC device) widthOf:self |
| 1388 | 3419 |
|
3420 |
" |
|
| 8874 | 3421 |
'hello world' widthOn:(View new) |
| 1388 | 3422 |
" |
3423 |
||
3424 |
"Created: 12.5.1996 / 20:09:29 / cg" |
|
| 2554 | 3425 |
"Modified: 17.4.1997 / 12:50:23 / cg" |
| 608 | 3426 |
! ! |
3427 |
||
| 9190 | 3428 |
!Charac |
| 7349 | 3429 |
|
3430 |
allRegexMatches: rxString |
|
3431 |
"return a collection of substrings in the receiver, which match the regular expression in rxString" |
|
3432 |
||
3433 |
^ rxString asRegex matchesIn: self |
|
3434 |
||
3435 |
" |
|
| 8874 | 3436 |
'1234 abcd 3456 defg' allRegexMatches:'[0-9]+' |
| 7349 | 3437 |
|
3438 |
'[0-9]+' asRegex matchesIn:'1234 abcd 3456 defg' |
|
3439 |
" |
|
3440 |
! |
|
3441 |
||
3442 |
asRegex |
|
| 9190 | 3443 |
"Compile t |
3444 |
||
3445 |
asRegex |
|
| 8874 | 3446 |
"Compile the receiver as a regex matcher. |
| 7349 | 3447 |
May raise RxParser>>syntaxErrorSignal or RxParser>>compilationErrorSignal. |
| 8874 | 3448 |
This is a part of the Regular Expression Matcher package, |
| 9191 | 3449 |
(c) 1996, 1999 Vassili Bykov. |
| 9190 | 3450 |
Refer to `documentation' protocol of RxParser class f |
| 7349 | 3451 |
|
3452 |
asRegexIgnoringCase |
|
| 8874 | 3453 |
"Compile the receiver as a regex matcher. |
| 7349 | 3454 |
May raise RxParser>>syntaxErrorSignal or RxParser>>compilationErrorSignal. |
| 8874 | 3455 |
This is a part of the Regular Expression Matcher package, |
| 9191 | 3456 |
(c) 1996, 1999 Vassili Bykov. |
| 7349 | 3457 |
Refer to `documentation' protocol of RxParser class for details." |
3458 |
||
3459 |
^ Regex::RxParser preferredMatcherClass |
|
| 9191 | 3460 |
for: (Regex::RxP |
| 7349 | 3461 |
|
3462 |
copyWithRegex: rxString matchesReplacedWith: aString |
|
3463 |
^ rxString asRegex copy: self replacingMatchesWith: aString |
|
3464 |
! |
|
3465 |
||
3466 |
copyWithRegex: rxString matchesTranslatedUsing: aBlock |
|
3467 |
^ rxString asRegex copy: self translatingMatchesUsing: aBlock |
|
3468 |
! |
|
3469 |
||
3470 |
matchesRegex: regexString |
|
| 8874 | 3471 |
"Test if the receiver matches a regex. |
| 7349 | 3472 |
May raise RxParser>>regexErrorSignal or child signals. |
| 9190 | 3473 |
This is a part of the Regular Expr |
3474 |
||
3475 |
matchesRegex: regexString |
|
3476 |
"Test if the receiver matches a regex. |
|
3477 |
May raise RxParser>>regexErrorSignal or child signals. |
|
| 8874 | 3478 |
This is a part of the Regular Expression Matcher package, |
| 9191 | 3479 |
(c) 1996, 1999 Vassili Bykov. |
| 9190 | 3480 |
Refer to `d |
| 7349 | 3481 |
|
3482 |
matchesRegexIgnoringCase: regexString |
|
| 8874 | 3483 |
"Test if the receiver matches a regex. |
| 7349 | 3484 |
May raise RxParser>>regexErrorSignal or child signals. |
| 8874 | 3485 |
This is a part of the Regular Expression Matcher package, |
| 9191 | 3486 |
(c) 1996, 1999 Vassili Bykov. |
| 7349 | 3487 |
Refer to `documentation' protocol of RxParser class for details." |
3488 |
||
3489 |
^regexString asRegexIgnoringCase matches: self |
|
3490 |
||
3491 |
" |
|
| 9190 | 3492 |
'hElLo w |
| 7349 | 3493 |
|
3494 |
prefixMatchesRegex: regexString |
|
| 8874 | 3495 |
"Test if the receiver's prefix matches a regex. |
| 7349 | 3496 |
May raise RxParser class>>regexErrorSignal or child signals. |
| 8874 | 3497 |
This is a part of the Regular Expression Matcher package, |
| 9191 | 3498 |
(c) 1996, 1999 Vassili Bykov. |
| 7349 | 3499 |
Refer to `documentation' protocol of RxParser class for details." |
3500 |
||
3501 |
^regexString asRegex matchesPrefix: self |
|
3502 |
! |
|
3503 |
||
3504 |
prefixMatchesRegexIgnoringCase: regexString |
|
| 9190 | 3505 |
" |
3506 |
||
3507 |
prefixMatchesRegexIgnoringCase: regexString |
|
| 8874 | 3508 |
"Test if the receiver's prefix matches a regex. |
| 7349 | 3509 |
May raise RxParser class>>regexErrorSignal or child signals. |
| 8874 | 3510 |
This is a part of the Regular Expression Matcher package, |
| 9191 | 3511 |
(c) 1996, 1999 Vassili Bykov. |
| 7349 | 3512 |
Refer to `documentation' protocol of RxParser class for details." |
3513 |
||
| 9190 | 3514 |
^regexString asRegexIgn |
| 7349 | 3515 |
|
3516 |
regex: rxString matchesCollect: aBlock |
|
3517 |
"for all substrings in the receiver which match the regular expression in rxString, evaluate aBlock, |
|
3518 |
and collect the returned values." |
|
3519 |
||
3520 |
^ rxString asRegex matchesIn: self collect: aBlock |
|
3521 |
||
3522 |
" |
|
| 8874 | 3523 |
'hello world' regex:'\w+' matchesCollect:[:each | each asUppercase ]. |
| 9190 | 3524 |
'1234 hello 456 world' regex:'\d+' matchesCollect:[: |
| 7349 | 3525 |
|
3526 |
regex:rxString matchesDo: aBlock |
|
3527 |
"for all substrings in the receiver which match the regular expression in rxString, evaluate aBlock" |
|
3528 |
||
3529 |
^ rxString asRegex matchesIn: self do: aBlock |
|
3530 |
||
3531 |
" |
|
3532 |
'hello world' regex:'\w+' matchesDo:[:each | Transcript showCR:each ]. |
|
3533 |
" |
|
3534 |
! ! |
|
3535 |
||
| 608 | 3536 |
!CharacterArray methodsFor:'special string converting'! |
3537 |
||
3538 |
chopTo:maxLen |
|
| 9190 | 3539 |
"if the receivers size is less or equal to ma15 |
3540 |
'1234567890123456' chopTo:15 |
|
3541 |
'aShortString' chopTo:15 |
|
3542 |
'aVeryLongNameForAStringThatShouldBeShortened' chopTo:15 |
|
3543 |
" |
|
3544 |
! |
|
3545 |
||
3546 |
contractAtBeginningTo:maxLen |
|
3547 |
"if the receivers size is less or equal to maxLen, return it. |
|
3548 |
Otherwise, return a copy of the receiver, where some characters |
|
3549 |
at the beginning have been chopTo:maxLen |
|
| 608 | 3550 |
"if the receivers size is less or equal to maxLen, return it. |
| 8874 | 3551 |
Otherwise, return a copy of the receiver, where some characters |
| 608 | 3552 |
in the middle have been removed for a total string length |
3553 |
of maxLen." |
|
3554 |
||
3555 |
|sz n1 n2| |
|
3556 |
||
3557 |
(sz := self size) > maxLen ifTrue:[ |
|
| 9191 | 3558 |
n1 := n2 := maxLen // 2. |
3559 |
maxLen odd ifTrue:[ |
|
3560 |
n2 := n1 + 1 |
|
3561 |
]. |
|
3562 |
^ (self copyFrom:1 to:n1) , (self copyFrom:sz - n2 + 1) |
|
| 608 | 3563 |
] |
3564 |
||
3565 |
" |
|
| 8874 | 3566 |
'12345678901234' chopTo:15 |
3567 |
'123456789012345' chopTo:15 |
|
3568 |
'1234567890123456' chopTo:15 |
|
3569 |
'aShortString' chopTo:15 |
|
3570 |
'aVeryLongNameForAStringThatShouldBeShortened' chopTo:15 |
|
| 608 | 3571 |
" |
| 9190 | 3572 |
ctAtBcontractAtBeginningTo:maxLen |
| 608 | 3573 |
"if the receivers size is less or equal to maxLen, return it. |
| 8874 | 3574 |
Otherwise, return a copy of the receiver, where some characters |
| 608 | 3575 |
at the beginning have been replaced by '...' for a total string length |
3576 |
of maxLen. Can be used to abbreviate long entries in tables." |
|
3577 |
||
3578 |
|sz| |
|
3579 |
||
3580 |
(sz := self size) > maxLen ifTrue:[ |
|
| 9191 | 3581 |
^ '...' , (self copyFrom:(sz - (maxLen - 4))) |
| 608 | 3582 |
] |
3583 |
||
3584 |
" |
|
| 8874 | 3585 |
'12345678901234' contractAtBeginningTo:15 |
3586 |
'123456789012345' contractAtBeginningTo:15 |
|
3587 |
'1234567890123456' contractAtBeginningTo:15 |
|
3588 |
'aShortString' contractAtBeginningTo:15 |
|
| 608 | 3589 |
'aVeryLongNameForAStringThatShouldBeShortened' contractAtBeginningTo:15 |
3590 |
" |
|
| 9190 | 3591 |
5 |
3592 |
contractAtEndTo:maxLen |
|
| 608 | 3593 |
"if the receivers size is less or equal to maxLen, return it. |
| 8874 | 3594 |
Otherwise, return a copy of the receiver, where some characters |
| 608 | 3595 |
at the end have been replaced by '...' for a total string length |
3596 |
of maxLen. Can be used to abbreviate long entries in tables." |
|
3597 |
||
3598 |
|sz| |
|
3599 |
||
3600 |
(sz := self size) > maxLen ifTrue:[ |
|
| 9191 | 3601 |
^ self copyReplaceFrom:maxLen - 3 |
3602 |
with:'...' |
|
| 608 | 3603 |
] |
3604 |
||
3605 |
" |
|
| 8874 | 3606 |
'12345678901234' contractAtEndTo:15 |
3607 |
'123456789012345' contractAtEndTo:15 |
|
3608 |
'1234567890123456' contractAtEndTo:15 |
|
3609 |
'aShortString' contractAtEndTo:15 |
|
3610 |
'aVeryLongNameForAStringThatShouldBeShortened' contractAtEndTo:15 |
|
| 608 | 3611 |
" |
| 9190 | 3612 |
|
| 9191 | 3613 |
contractTo:maxLen |
| 608 | 3614 |
"if the receivers size is less or equal to maxLen, return it. |
| 8874 | 3615 |
Otherwise, return a copy of the receiver, where some characters |
| 608 | 3616 |
in the middle have been replaced by '...' for a total string length |
3617 |
of maxLen. Can be used to abbreviate long entries in tables." |
|
3618 |
||
3619 |
|sz "{ SmallInteger }"
|
|
| 9190 | 3620 |
halfSize "{ SmallInteger } '%%(1)0 gives %(1)0' expandPlaceholdersWith:#(123)
|
3621 |
'%%10 gives %10' expandPlaceholdersWith:#(123) |
|
3622 |
'%%(10) gives %(10)' expandPlaceholdersWith:#(123) |
|
3623 |
'%test gives %1' expandPlaceholdersWith:#(123) |
|
3624 |
" |
|
3625 |
||
3626 |
" |
|
3627 |
|dict| |
|
3628 |
||
3629 |
dict := Dictionary new. |
|
3630 |
dict at:1 put:'one'. |
|
3631 |
dict at:$a put:'AAAAA'. |
|
3632 |
dict at:$b put:[ Time now ]. |
|
3633 |
'hello %1 %a %b' expanexpandPlaceholdersWith:argArrayOrDictionary |
|
| 608 | 3634 |
"return a copy of the receiver, where all %i escapes are |
|
4410
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3635 |
replaced by corresponding arguments' printStrings from the argArrayOrDictionary. |
| 608 | 3636 |
I.e. 'hello %1; how is %2' expandPlaceholdersWith:#('world' 'this') results
|
3637 |
in the new string 'hello world; how is this'. |
|
|
4410
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3638 |
As an extension, the argument may also be a dictionary, providing |
| 7313 | 3639 |
values for symbolic keys. |
| 4842 | 3640 |
In this case, %a .. %z and %(...) are also allowed. |
|
4410
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3641 |
(%1..%9 require a numeric key in the dictionary, however) |
| 608 | 3642 |
To get a '%' character, use a '%%'-escape. |
| 7313 | 3643 |
To get an integer-indexed placeHolder followed by another digit, |
3644 |
or an index > 9, you must use %(digit). |
|
| 608 | 3645 |
See also bindWith:... for VisualAge compatibility." |
3646 |
||
|
8049
ccd0e8f26439
printing - migrating towards unicodeSupport in ST-code
Claus Gittinger <cg@exept.de>
parents:
8040
diff
changeset
|
3647 |
|expandedString next v key keyAsSymbol |
| 608 | 3648 |
idx "{ SmallInteger }"
|
| 4842 | 3649 |
idx2 "{ SmallInteger }"
|
| 608 | 3650 |
start "{ SmallInteger }"
|
3651 |
stop "{ SmallInteger }"|
|
|
3652 |
||
|
1026
455c9f609f4d
care for 16bit strings in expandPlaceHolders
Claus Gittinger <cg@exept.de>
parents:
1025
diff
changeset
|
3653 |
expandedString := self species new:0. |
| 608 | 3654 |
stop := self size. |
3655 |
start := 1. |
|
3656 |
[start <= stop] whileTrue:[ |
|
|
8938
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3657 |
idx := self indexOf:$% startingAt:start. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3658 |
(idx == 0 or:[idx == stop]) ifTrue:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3659 |
^ expandedString , (self copyFrom:start to:stop) |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3660 |
]. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3661 |
"found a %" |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3662 |
expandedString := expandedString , (self copyFrom:start to:(idx - 1)). |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3663 |
next := self at:(idx + 1). |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3664 |
(next == $%) ifTrue:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3665 |
expandedString := expandedString , '%' |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3666 |
] ifFalse:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3667 |
(next between:$1 and:$9) ifTrue:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3668 |
v := argArrayOrDictionary at:(next digitValue) ifAbsent:nil |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3669 |
] ifFalse:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3670 |
next == $( ifTrue:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3671 |
idx2 := self indexOf:$) startingAt:idx+2. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3672 |
key := self copyFrom:idx+2 to:idx2-1. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3673 |
idx := idx2 - 1. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3674 |
keyAsSymbol := key asSymbolIfInterned. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3675 |
(keyAsSymbol notNil and:[ argArrayOrDictionary includesKey:keyAsSymbol ]) ifTrue:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3676 |
v := argArrayOrDictionary at:keyAsSymbol |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3677 |
] ifFalse:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3678 |
(key conform:[:each | each isDigit]) ifTrue:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3679 |
key := Number readFrom:key onError:nil. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3680 |
]. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3681 |
v := argArrayOrDictionary at:key ifAbsent:nil |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3682 |
]. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3683 |
] ifFalse:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3684 |
argArrayOrDictionary isSequenceable ifFalse:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3685 |
"Arrays etc. can be only indexed with integers, not with characters or strings" |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3686 |
v := argArrayOrDictionary at:next ifAbsent:nil. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3687 |
v isNil ifTrue:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3688 |
v := argArrayOrDictionary at:next asString asSymbol ifAbsent:nil. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3689 |
]. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3690 |
]. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3691 |
v isNil ifTrue:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3692 |
v := String with:$% with:next. "No match, keep original sequence" |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3693 |
]. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3694 |
] |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3695 |
]. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3696 |
v isNil |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3697 |
ifTrue:[v := ''] |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3698 |
ifFalse:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3699 |
v isBlock ifTrue:[ |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3700 |
v := v value |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3701 |
]]. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3702 |
expandedString := expandedString , v printString |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3703 |
]. |
|
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3704 |
start := idx + 2 |
| 608 | 3705 |
]. |
3706 |
^ expandedString |
|
3707 |
||
3708 |
" |
|
| 8874 | 3709 |
'hello %1' expandPlaceholdersWith:#('world')
|
3710 |
'hello %1; how is %2' expandPlaceholdersWith:#('world' 'this')
|
|
3711 |
'hello %2; how is %1' expandPlaceholdersWith:#('world' 'this')
|
|
3712 |
'%1 plus %2 gives %3 ' expandPlaceholdersWith:#(4 5 9) |
|
3713 |
'%%(1)0 gives %(1)0' expandPlaceholdersWith:#(123) |
|
3714 |
'%%10 gives %10' expandPlaceholdersWith:#(123) |
|
3715 |
'%%(10) gives %(10)' expandPlaceholdersWith:#(123) |
|
|
8938
860bbcbd3bd8
Fix #expandPlaceHoldersWith:, when a %a is encountered and and an Array
Stefan Vogel <sv@exept.de>
parents:
8913
diff
changeset
|
3716 |
'%test gives %1' expandPlaceholdersWith:#(123) |
| 608 | 3717 |
" |
|
1026
455c9f609f4d
care for 16bit strings in expandPlaceHolders
Claus Gittinger <cg@exept.de>
parents:
1025
diff
changeset
|
3718 |
|
|
4410
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3719 |
" |
|
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3720 |
|dict| |
|
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3721 |
|
|
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3722 |
dict := Dictionary new. |
|
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3723 |
dict at:1 put:'one'. |
|
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3724 |
dict at:$a put:'AAAAA'. |
|
8794
a5df33935142
nindWith supports blocks in the value dictionary
Claus Gittinger <cg@exept.de>
parents:
8710
diff
changeset
|
3725 |
dict at:$b put:[ Time now ]. |
| 8874 | 3726 |
'hello %1 %a %b' expandPlaceholdersWith:dict |
|
4410
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3727 |
" |
|
2bdc0c89a42e
enhanced #exapndPlaceHolders, to allow for a dictionary arg
Claus Gittinger <cg@exept.de>
parents:
4409
diff
changeset
|
3728 |
|
|
2737
9f29aecea39d
care for % at end in expandPlaceHolders.
Claus Gittinger <cg@exept.de>
parents:
2736
diff
changeset
|
3729 |
"Modified: 1.7.1997 / 00:53:24 / cg" |
| 9190 | 3730 |
ello\nworld\na\n\tnice\n\t\tstring' withEscapes |
3731 |
'hello\tworld\n' withEscapes |
|
3732 |
'hello\010world' withEscapes |
|
3733 |
'hello\r\nworld' withEscapes |
|
3734 |
" |
|
3735 |
||
3736 |
"Modified: 12.5.1996 / 12:53:34 / cg" |
|
| 608 | 3737 |
! |
3738 |
||
| 9190 | 3739 |
withMatchEscapes |
3740 |
"return a copy of the receiver withwithEscapes |
|
| 608 | 3741 |
"return a new string consisting of receivers characters |
3742 |
with all \X-character escapes replaced by corresponding-characters. |
|
3743 |
(similar to the way C-language Strings are converted). |
|
3744 |
The following escapes are supported: |
|
| 9191 | 3745 |
\r return character |
3746 |
\n newline character |
|
3747 |
\b backspace character |
|
3748 |
\f formfeed character |
|
3749 |
\t tab character |
|
3750 |
\e escape character |
|
3751 |
\\ the \ character itself |
|
3752 |
\nnn three digit octal number defining the characters ascii value |
|
3753 |
\other other |
|
| 608 | 3754 |
|
3755 |
Notice, that \' is NOT a valid escape, since the general syntax of |
|
3756 |
string constants is not affected by this method. |
|
3757 |
||
3758 |
Although easily implementable, this is NOT done automatically |
|
3759 |
by the compiler (due to a lack of a language standard for this). |
|
3760 |
However, the compiler may detect sends ot #withEscapes to string literals |
|
3761 |
and place a modified string constant into the binary/byte-code. |
|
3762 |
Therefore, no runtime penalty will be payed for using these escapes. |
|
3763 |
(not in pre 2.11 versions) |
|
3764 |
" |
|
3765 |
||
3766 |
|sz "{ SmallInteger }"
|
|
3767 |
newSize "{ SmallInteger }"
|
|
3768 |
srcIdx "{ SmallInteger }"
|
|
3769 |
dstIdx "{ SmallInteger }"
|
|
3770 |
val "{ SmallInteger }"
|
|
| 8874 | 3771 |
newString next hasEmphasis e| |
| 359 | 3772 |
|
3773 |
" |
|
| 608 | 3774 |
first, count the number of escapes, to allow preallocation |
3775 |
of the new string ... |
|
3776 |
(it is faster to scan the string twice than to reallocate it multiple |
|
3777 |
times in a WriteStream) |
|
3778 |
" |
|
3779 |
sz := newSize := self size. |
|
3780 |
srcIdx := 1. |
|
3781 |
[(srcIdx := self indexOf:$\ startingAt:srcIdx) ~~ 0] whileTrue:[ |
|
| 9191 | 3782 |
srcIdx == sz ifFalse:[ |
3783 |
newSize := newSize - 1. |
|
3784 |
srcIdx := srcIdx + 1. |
|
3785 |
next := self at:srcIdx. |
|
3786 |
next == $0 ifTrue:[ |
|
3787 |
[srcIdx < sz and:[next isDigit]] whileTrue:[ |
|
3788 |
newSize := newSize - 1. srcIdx := srcIdx + 1. next := self at:srcIdx. |
|
3789 |
] |
|
3790 |
]. |
|
3791 |
]. |
|
3792 |
srcIdx := srcIdx + 1. |
|
| 608 | 3793 |
]. |
3794 |
||
3795 |
newSize == sz ifTrue:[ |
|
| 9191 | 3796 |
^ self |
| 608 | 3797 |
]. |
3798 |
||
3799 |
newString := self species new:newSize. |
|
| 1381 | 3800 |
|
3801 |
hasEmphasis := self hasChangeOfEmphasis. |
|
3802 |
||
| 608 | 3803 |
" |
3804 |
copy over, replace escapes |
|
3805 |
" |
|
3806 |
srcIdx := dstIdx := 1. |
|
3807 |
[srcIdx <= sz] whileTrue:[ |
|
| 9191 | 3808 |
next := self at:srcIdx. |
3809 |
hasEmphasis ifTrue:[ |
|
3810 |
e := self emphasisAt:srcIdx |
|
3811 |
]. |
|
3812 |
srcIdx := srcIdx + 1. |
|
3813 |
next == $\ ifTrue:[ |
|
3814 |
srcIdx <= sz ifTrue:[ |
|
3815 |
next := self at:srcIdx. |
|
3816 |
srcIdx := srcIdx + 1. |
|
3817 |
next == $r ifTrue:[ |
|
3818 |
next := Character return |
|
3819 |
] ifFalse:[ |
|
3820 |
next == $n ifTrue:[ |
|
3821 |
next := Character nl |
|
3822 |
] ifFalse:[ |
|
3823 |
next == $b ifTrue:[ |
|
3824 |
next := Character backspace |
|
3825 |
] ifFalse:[ |
|
3826 |
next == $f ifTrue:[ |
|
3827 |
next := Character newPage |
|
3828 |
] ifFalse:[ |
|
3829 |
next == $t ifTrue:[ |
|
3830 |
next := Character tab |
|
3831 |
] ifFalse:[ |
|
3832 |
next == $e ifTrue:[ |
|
3833 |
next := Character esc |
|
3834 |
] ifFalse:[ |
|
3835 |
next == $0 ifTrue:[ |
|
3836 |
val := 0. |
|
3837 |
[next notNil and:[next isDigit]] whileTrue:[ |
|
3838 |
val := val * 8 + next digitValue. |
|
3839 |
srcIdx <= sz ifTrue:[ |
|
3840 |
next := self at:srcIdx. |
|
3841 |
srcIdx := srcIdx + 1. |
|
3842 |
] ifFalse:[ |
|
3843 |
next := nil |
|
3844 |
] |
|
3845 |
]. |
|
3846 |
next := Character value:val. |
|
3847 |
] |
|
3848 |
] |
|
3849 |
] |
|
3850 |
] |
|
3851 |
] |
|
3852 |
] |
|
3853 |
]. |
|
3854 |
]. |
|
3855 |
]. |
|
3856 |
newString at:dstIdx put:next. |
|
3857 |
hasEmphasis ifTrue:[ |
|
3858 |
newString emphasisAt:dstIdx put:e |
|
3859 |
]. |
|
3860 |
dstIdx := dstIdx + 1. |
|
| 608 | 3861 |
]. |
3862 |
^ newString |
|
3863 |
||
3864 |
" |
|
| 8874 | 3865 |
'hello world' withEscapes |
3866 |
'hello\world' withEscapes |
|
3867 |
'hello\world\' withEscapes |
|
3868 |
'hello world\' withEscapes |
|
3869 |
'hello\tworld' withEscapes |
|
3870 |
'hello\nworld\na\n\tnice\n\t\tstring' withEscapes |
|
3871 |
'hello\tworld\n' withEscapes |
|
3872 |
'hello\010world' withEscapes |
|
3873 |
'hello\r\nworld' withEscapes |
|
| 608 | 3874 |
" |
| 1008 | 3875 |
|
| 1381 | 3876 |
"Modified: 12.5.1996 / 12:53:34 / cg" |
| 9190 | 3877 |
contewithMatchEscapes |
| 4117 | 3878 |
"return a copy of the receiver with all match characters escaped |
3879 |
by $\ characters (to be usable as a match string). |
|
3880 |
Return the receiver, if there are none." |
|
3881 |
||
|
5342
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
3882 |
|in out c escape| |
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
3883 |
|
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
3884 |
escape := self class matchEscapeCharacter. |
| 4117 | 3885 |
|
3886 |
in := self readStream. |
|
| 7986 | 3887 |
out := WriteStream on:(self species new:self size). |
| 4117 | 3888 |
[in atEnd] whileFalse:[ |
| 9191 | 3889 |
c := in next. |
3890 |
(c == escape or:['*[#' includes:c]) ifTrue:[ |
|
3891 |
out nextPut:$\. |
|
3892 |
]. |
|
3893 |
out nextPut:c. |
|
| 4117 | 3894 |
]. |
3895 |
^ out contents. |
|
3896 |
||
3897 |
" |
|
| 8874 | 3898 |
'*foo' withMatchEscapes |
3899 |
'\*foo' withMatchEscapes |
|
3900 |
'*foo' withMatchEscapes |
|
3901 |
'\\*foo' withMatchEscapes |
|
3902 |
'foo*' withMatchEscapes |
|
3903 |
'foo\*' withMatchEscapes |
|
3904 |
'foo\' withMatchEscapes |
|
3905 |
'f*o*o' withMatchEscapes |
|
| 4117 | 3906 |
" |
3907 |
||
3908 |
"Modified: 2.4.1997 / 18:13:04 / cg" |
|
| 9190 | 3909 |
67890withTabs |
|
735
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
3910 |
"return a string consisting of the receivers characters |
|
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
3911 |
where leading spaces are replaced by tabulator characters (assuming 8-col tabs). |
|
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
3912 |
Notice: if the receiver does not contain any tabs, it is returned unchanged; |
|
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
3913 |
otherwise a new string is returned. |
| 8874 | 3914 |
Limitation: only the very first spaces are replaced |
| 9191 | 3915 |
(i.e. if the receiver contains newLine characters, |
3916 |
no tabs are inserted after those lineBreaks)" |
|
| 8874 | 3917 |
|
3918 |
|idx "{ SmallInteger }"
|
|
3919 |
nTabs "{ SmallInteger }"
|
|
| 608 | 3920 |
newString| |
3921 |
||
3922 |
idx := self findFirst:[:c | (c ~~ Character space)]. |
|
3923 |
nTabs := (idx-1) // 8. |
|
|
5815
4acdd87a77d4
// can return negative numbers.
Stefan Vogel <sv@exept.de>
parents:
5795
diff
changeset
|
3924 |
nTabs <= 0 ifTrue:[^ self]. |
| 608 | 3925 |
|
3926 |
"any tabs" |
|
3927 |
newString := self class new:(self size - (nTabs * 7)). |
|
3928 |
newString atAll:(1 to:nTabs) put:(Character tab). |
|
3929 |
newString replaceFrom:(nTabs + 1) with:self startingAt:(nTabs * 8 + 1). |
|
3930 |
^ newString |
|
3931 |
||
3932 |
" |
|
| 8874 | 3933 |
'12345678901234567890' withTabs |
3934 |
' 8901234567890' withTabs |
|
3935 |
' 901234567890' withTabs |
|
| 608 | 3936 |
' 67890' withTabs |
3937 |
' 7890' withTabs |
|
3938 |
' 890' withTabs |
|
3939 |
" |
|
| 9190 | 3940 |
wwithTabsExpanded |
|
735
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
3941 |
"return a string consisting of the receivers characters, |
| 8874 | 3942 |
where all tabulator characters are expanded into spaces (assuming 8-col tabs). |
|
735
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
3943 |
Notice: if the receiver does not contain any tabs, it is returned unchanged; |
|
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
3944 |
otherwise a new string is returned. |
|
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
3945 |
This does handle multiline strings." |
| 608 | 3946 |
|
|
5741
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3947 |
^ self withTabsExpanded:8 |
|
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3948 |
|
|
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3949 |
" |
| 8874 | 3950 |
('1' , Character tab asString , 'x') withTabsExpanded
|
3951 |
('12345' , Character tab asString , 'x') withTabsExpanded
|
|
3952 |
('123456' , Character tab asString , 'x') withTabsExpanded
|
|
3953 |
('1234567' , Character tab asString , 'x') withTabsExpanded
|
|
3954 |
('12345678' , Character tab asString , 'x') withTabsExpanded
|
|
3955 |
('123456789' , Character tab asString , 'x') withTabsExpanded
|
|
|
5741
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3956 |
|
|
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3957 |
(String with:Character tab |
| 9191 | 3958 |
with:Character tab |
3959 |
with:$1) withTabsExpanded |
|
|
5741
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3960 |
|
|
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3961 |
(String with:Character tab |
| 9191 | 3962 |
with:$1 |
3963 |
with:Character tab |
|
3964 |
with:$2) withTabsExpanded |
|
|
5741
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3965 |
|
|
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3966 |
(String with:Character tab |
| 9191 | 3967 |
with:$1 |
3968 |
with:Character cr |
|
3969 |
with:Character tab |
|
3970 |
with:$2) withTabsExpanded |
|
|
5741
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3971 |
" |
|
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3972 |
|
|
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3973 |
"Modified: 12.5.1996 / 13:05:10 / cg" |
| 9190 | 3974 |
9' , withTabsExpanded:numSpaces |
|
5741
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3975 |
"return a string consisting of the receivers characters, |
| 8874 | 3976 |
where all tabulator characters are expanded into spaces (assuming numSpaces-col tabs). |
|
5741
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3977 |
Notice: if the receiver does not contain any tabs, it is returned unchanged; |
|
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3978 |
otherwise a new string is returned. |
|
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3979 |
This does handle multiline strings." |
|
3a6086774d38
added #withTabsExpanded: (i.e. arg to specify tab-column width)
Claus Gittinger <cg@exept.de>
parents:
5739
diff
changeset
|
3980 |
|
| 8874 | 3981 |
|col "{ SmallInteger }"
|
| 1381 | 3982 |
str ch |
3983 |
dstIdx "{ SmallInteger }"
|
|
3984 |
newSz "{ SmallInteger }"
|
|
3985 |
sz "{ SmallInteger }"
|
|
3986 |
hasEmphasis e| |
|
| 608 | 3987 |
|
3988 |
(self includes:(Character tab)) ifFalse:[^ self]. |
|
| 1381 | 3989 |
|
3990 |
sz := self size. |
|
3991 |
||
3992 |
"/ count the new size first, instead of |
|
3993 |
"/ multiple resizing (better for large strings) |
|
3994 |
||
3995 |
col := 1. newSz := 0. |
|
3996 |
1 to:sz do:[:srcIdx | |
|
| 9191 | 3997 |
ch := self at:srcIdx. |
3998 |
ch == Character tab ifFalse:[ |
|
3999 |
col := col + 1. |
|
4000 |
newSz := newSz + 1. |
|
4001 |
ch == Character cr ifTrue:[ |
|
4002 |
col := 1 |
|
4003 |
]. |
|
4004 |
] ifTrue:[ |
|
4005 |
(col \\ numSpaces) to:numSpaces do:[:ii | |
|
4006 |
newSz := newSz + 1. |
|
4007 |
col := col + 1 |
|
4008 |
]. |
|
4009 |
] |
|
| 608 | 4010 |
]. |
| 1381 | 4011 |
|
4012 |
str := self species new:newSz. |
|
4013 |
||
4014 |
hasEmphasis := self hasChangeOfEmphasis. |
|
4015 |
||
4016 |
col := 1. dstIdx := 1. |
|
4017 |
1 to:sz do:[:srcIdx | |
|
| 9191 | 4018 |
ch := self at:srcIdx. |
4019 |
||
4020 |
ch == Character tab ifFalse:[ |
|
4021 |
col := col + 1. |
|
4022 |
ch == Character cr ifTrue:[ |
|
4023 |
col := 1 |
|
4024 |
]. |
|
4025 |
hasEmphasis ifTrue:[ |
|
4026 |
e := self emphasisAt:srcIdx. |
|
4027 |
str emphasisAt:dstIdx put:e |
|
4028 |
]. |
|
4029 |
str at:dstIdx put:ch. |
|
4030 |
dstIdx := dstIdx + 1 |
|
4031 |
] ifTrue:[ |
|
4032 |
(col \\ numSpaces) to:numSpaces do:[:ii | |
|
4033 |
str at:dstIdx put:Character space. |
|
4034 |
dstIdx := dstIdx + 1. |
|
4035 |
col := col + 1 |
|
4036 |
]. |
|
4037 |
] |
|
| 1381 | 4038 |
]. |
4039 |
^ str |
|
| 608 | 4040 |
|
4041 |
" |
|
| 8874 | 4042 |
('1' , Character tab asString , 'x') withTabsExpanded
|
4043 |
('1' , Character tab asString , 'x') withTabsExpanded:4
|
|
4044 |
('12345' , Character tab asString , 'x') withTabsExpanded
|
|
4045 |
('123456' , Character tab asString , 'x') withTabsExpanded
|
|
4046 |
('1234567' , Character tab asString , 'x') withTabsExpanded
|
|
4047 |
('12345678' , Character tab asString , 'x') withTabsExpanded
|
|
4048 |
('123456789' , Character tab asString , 'x') withTabsExpanded
|
|
|
735
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
4049 |
|
| 608 | 4050 |
(String with:Character tab |
| 9191 | 4051 |
with:Character tab |
4052 |
with:$1) withTabsExpanded |
|
| 608 | 4053 |
|
4054 |
(String with:Character tab |
|
| 9191 | 4055 |
with:$1 |
4056 |
with:Character tab |
|
4057 |
with:$2) withTabsExpanded |
|
|
735
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
4058 |
|
|
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
4059 |
(String with:Character tab |
| 9191 | 4060 |
with:$1 |
4061 |
with:Character cr |
|
4062 |
with:Character tab |
|
4063 |
with:$2) withTabsExpanded |
|
| 608 | 4064 |
" |
|
735
362ce9e28d89
expandTabs now handles multiline strings
Claus Gittinger <cg@exept.de>
parents:
678
diff
changeset
|
4065 |
|
| 1381 | 4066 |
"Modified: 12.5.1996 / 13:05:10 / cg" |
| 9190 | 4067 |
' foo' withoutLeadingSeparators |
4068 |
' ' withoutLeadingSeparators |
|
4069 |
'foo' withoutLeadingSeparators |
|
4070 |
(' ' , Character tab asString , ' foo ') withoutLeadingSeparators inspect
|
|
4071 |
" |
|
| 608 | 4072 |
! |
4073 |
||
| 9190 | 4074 |
withoutMatchEscapes |
4075 |
"return a copy of the receiver with all $\ removed or |
|
4076 |
the receivwithoutLeadingSeparators |
|
| 608 | 4077 |
"return a copy of myself without leading separators. |
4078 |
Notice: this does remove tabs, newline or any other whitespace. |
|
4079 |
Returns an empty string, if the receiver consist only of whitespace." |
|
4080 |
||
4081 |
|index| |
|
4082 |
||
4083 |
index := self indexOfNonSeparatorStartingAt:1. |
|
4084 |
index ~~ 0 ifTrue:[ |
|
| 9191 | 4085 |
index == 1 ifTrue:[ |
4086 |
^ self |
|
4087 |
]. |
|
4088 |
^ self copyFrom:index |
|
| 608 | 4089 |
]. |
4090 |
^ '' |
|
4091 |
||
4092 |
" |
|
| 8874 | 4093 |
' foo ' withoutLeadingSeparators |
4094 |
'foo ' withoutLeadingSeparators |
|
4095 |
' foo' withoutLeadingSeparators |
|
4096 |
' ' withoutLeadingSeparators |
|
4097 |
'foo' withoutLeadingSeparators |
|
4098 |
(' ' , Character tab asString , ' foo ') withoutLeadingSeparators inspect
|
|
| 608 | 4099 |
" |
| 9190 | 4100 |
withoutMatchEscapes |
| 4117 | 4101 |
"return a copy of the receiver with all $\ removed or |
4102 |
the receiver, if there are none." |
|
4103 |
||
|
5342
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
4104 |
|in out c escape| |
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
4105 |
|
|
5010cea0ffd4
match-escape character definition in one place
Claus Gittinger <cg@exept.de>
parents:
5316
diff
changeset
|
4106 |
escape := self class matchEscapeCharacter. |
| 4117 | 4107 |
|
4108 |
in := self readStream. |
|
|
8222
f2c454a9a038
replaced '' writeStream by String writeStream
Claus Gittinger <cg@exept.de>
parents:
8124
diff
changeset
|
4109 |
out := self species writeStream. |
| 4117 | 4110 |
[in atEnd] whileFalse:[ |
| 9191 | 4111 |
c := in next. |
4112 |
c == escape ifTrue:[ |
|
4113 |
in atEnd ifFalse:[ |
|
4114 |
c := in next. |
|
4115 |
] |
|
4116 |
]. |
|
4117 |
out nextPut:c. |
|
| 4117 | 4118 |
]. |
4119 |
^ out contents. |
|
4120 |
||
4121 |
" |
|
| 8874 | 4122 |
'*foo' withoutMatchEscapes |
4123 |
'\*foo' withoutMatchEscapes |
|
4124 |
'*foo' withoutMatchEscapes |
|
4125 |
'\\*foo' withoutMatchEscapes |
|
4126 |
'foo*' withoutMatchEscapes |
|
4127 |
'foo\*' withoutMatchEscapes |
|
4128 |
'foo\' withoutMatchEscapes |
|
4129 |
'f\*o\*o' withoutMatchEscapes |
|
| 4117 | 4130 |
" |
4131 |
||
4132 |
"Modified: 30.6.1997 / 13:40:23 / cg" |
|
| 9190 | 4133 |
ators inspect |
| 7247 | 4134 |
" |
4135 |
! |
|
4136 |
||
| 9190 | 4137 |
withoutSpaces |
4138 |
"return a copy of myself without leading and trailing spaces. |
|
4139 |
(but spaces in-between are preserved) |
|
4140 |
Notice: this does NOT remove tabs, newline or any other whitespace. |
|
4141 |
Use withoutSeparators for this." |
|
4142 |
||
4143 |
|startIndex "{ Class: SmallInteger }"
|
|
4144 |
enwithoutSeparators |
|
| 608 | 4145 |
"return a copy of myself without leading and trailing whitespace. |
| 4156 | 4146 |
(but whiteSpace in-between is preserved) |
| 608 | 4147 |
Whitespace is space, tab, newline, formfeed. |
4148 |
Use withoutSpaces, if you want to remove spaces only." |
|
4149 |
||
4150 |
|startIndex "{ Class: SmallInteger }"
|
|
| 8874 | 4151 |
endIndex "{ Class: SmallInteger }"
|
| 608 | 4152 |
sz| |
4153 |
||
4154 |
sz := self size. |
|
4155 |
startIndex := 1. |
|
4156 |
endIndex := sz. |
|
4157 |
||
4158 |
[(startIndex < endIndex) and:[(self at:startIndex) isSeparator]] whileTrue:[ |
|
| 9191 | 4159 |
startIndex := startIndex + 1 |
| 608 | 4160 |
]. |
4161 |
[(endIndex > 1) and:[(self at:endIndex) isSeparator]] whileTrue:[ |
|
| 9191 | 4162 |
endIndex := endIndex - 1 |
| 608 | 4163 |
]. |
4164 |
startIndex > endIndex ifTrue:[ |
|
| 9191 | 4165 |
^ '' |
| 608 | 4166 |
]. |
4167 |
((startIndex == 1) and:[endIndex == sz]) ifTrue:[ |
|
| 9191 | 4168 |
^ self |
| 608 | 4169 |
]. |
4170 |
^ self copyFrom:startIndex to:endIndex |
|
4171 |
||
4172 |
" |
|
| 8874 | 4173 |
' foo ' withoutSeparators |
4174 |
' foo' withoutSeparators |
|
4175 |
'foo ' withoutSeparators |
|
4176 |
' ' withoutSeparators |
|
4177 |
(' foo' , Character tab asString , ' ') withoutSeparators inspect
|
|
| 608 | 4178 |
" |
| 9190 | 4179 |
ab aswithoutSpaces |
| 608 | 4180 |
"return a copy of myself without leading and trailing spaces. |
| 4156 | 4181 |
(but spaces in-between are preserved) |
| 608 | 4182 |
Notice: this does NOT remove tabs, newline or any other whitespace. |
4183 |
Use withoutSeparators for this." |
|
4184 |
||
4185 |
|startIndex "{ Class: SmallInteger }"
|
|
| 8874 | 4186 |
endIndex "{ Class: SmallInteger }"
|
| 608 | 4187 |
sz| |
4188 |
||
4189 |
sz := self size. |
|
4190 |
startIndex := 1. |
|
4191 |
endIndex := sz. |
|
4192 |
||
4193 |
[(startIndex < endIndex) and:[(self at:startIndex) == Character space]] whileTrue:[ |
|
| 9191 | 4194 |
startIndex := startIndex + 1 |
| 608 | 4195 |
]. |
4196 |
[(endIndex > 1) and:[(self at:endIndex) == Character space]] whileTrue:[ |
|
| 9191 | 4197 |
endIndex := endIndex - 1 |
| 608 | 4198 |
]. |
4199 |
startIndex > endIndex ifTrue:[ |
|
| 9191 | 4200 |
^ '' |
| 608 | 4201 |
]. |
4202 |
((startIndex == 1) and:[endIndex == sz]) ifTrue:[ |
|
| 9191 | 4203 |
^ self |
| 608 | 4204 |
]. |
4205 |
^ self copyFrom:startIndex to:endIndex |
|
4206 |
||
4207 |
" |
|
| 8874 | 4208 |
' foo ' withoutSpaces |
4209 |
'foo ' withoutSpaces |
|
4210 |
' foo' withoutSpaces |
|
4211 |
' ' withoutSpaces |
|
4212 |
'a b' withoutSpaces |
|
4213 |
(' foo' , Character tab asString , ' ') withoutSpaces inspect
|
|
| 608 | 4214 |
" |
| 9190 | 4215 |
atorswithoutTrailingSeparators |
|
915
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4216 |
"return a copy of myself without trailing separators. |
|
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4217 |
Notice: this does remove tabs, newline or any other whitespace. |
|
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4218 |
Returns an empty string, if the receiver consist only of whitespace." |
|
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4219 |
|
|
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4220 |
|index| |
|
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4221 |
|
|
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4222 |
index := self size. |
|
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4223 |
[index ~~ 0] whileTrue:[ |
| 9191 | 4224 |
(self at:index) isSeparator ifFalse:[ |
4225 |
^ self copyTo:index |
|
4226 |
]. |
|
4227 |
index := index - 1 |
|
|
915
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4228 |
]. |
|
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4229 |
^ '' |
|
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4230 |
|
|
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4231 |
" |
| 8874 | 4232 |
' foo ' withoutTrailingSeparators |
4233 |
'foo ' withoutTrailingSeparators |
|
4234 |
' foo' withoutTrailingSeparators |
|
4235 |
' ' withoutTrailingSeparators |
|
4236 |
'foo' withoutTrailingSeparators |
|
4237 |
(' ' , Character tab asString , ' foo ') withoutTrailingSeparators inspect
|
|
4238 |
(' foo' , Character tab asString) withoutTrailingSeparators inspect
|
|
|
915
1ac1c7e942f4
added withoutTrailingSeparators
Claus Gittinger <cg@exept.de>
parents:
828
diff
changeset
|
4239 |
" |
| 9190 | 4240 |
|
| 608 | 4241 |
|
4242 |
findString:subString ifAbsent:exceptionBlock |
|
4243 |
"find a substring. If found, return the index; |
|
4244 |
if not found, return the result of evaluating exceptionBlock." |
|
4245 |
||
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4246 |
^ self indexOfSubCollection:subString startingAt:1 ifAbsent:exceptionBlock caseSensitive:true |
| 608 | 4247 |
! |
4248 |
||
4249 |
findString:subString startingAt:index |
|
| 9190 | 4250 |
"find a substring, |
4251 |
||
4252 |
findString:subString startingAt:index |
|
| 608 | 4253 |
"find a substring, starting at index. if found, return the index; |
4254 |
if not found, return 0." |
|
4255 |
||
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4256 |
^ self indexOfSubCollection:subString startingAt:index ifAbsent:0 caseSensitive:true |
| 608 | 4257 |
|
4258 |
" |
|
| 9190 | 4259 |
'hello yello' fin |
| 608 | 4260 |
|
4261 |
findString:subString startingAt:index ifAbsent:exceptionBlock |
|
4262 |
"find a substring, starting at index. if found, return the index; |
|
4263 |
if not found, return the result of evaluating exceptionBlock." |
|
4264 |
||
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4265 |
^ self indexOfSubCollection:subString startingAt:index ifAbsent:exceptionBlock caseSensitive:true |
| 608 | 4266 |
! |
4267 |
||
4268 |
includesString:aString |
|
| 9190 | 4269 |
"return true, if a substring is contained in the |
4270 |
||
4271 |
includesString:aString |
|
| 608 | 4272 |
"return true, if a substring is contained in the receiver" |
4273 |
||
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4274 |
^ (self indexOfSubCollection:aString startingAt:1 ifAbsent:0 caseSensitive:true) ~~ 0 |
| 608 | 4275 |
|
4276 |
" |
|
| 8874 | 4277 |
'hello world' includesString:'hel' |
4278 |
'hello world' includesString:'rld' |
|
| 9190 | 4279 |
'hello world' includesStrin' caseSensitive:true |
4280 |
'hello world' includesString:'WOR' caseSensitive:false |
|
| 608 | 4281 |
" |
4282 |
! |
|
4283 |
||
| 9190 | 4284 |
indexOfSubCollection:subString startingAt:index ifAbsent:exceptionBlock |
4285 |
"find a substring, starting at index. if found, return the index; |
|
4286 |
if not found, return the result of evaluating exceptionBlock. |
|
4287 |
This is a q&d hack - not very efficient" |
|
| 9191 | 4288 |
|
4289 |
includesString:aString caseSensitive:caseSensitive |
|
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4290 |
"return true, if a substring is contained in the receiver" |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4291 |
|
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4292 |
^ (self indexOfSubCollection:aString startingAt:1 ifAbsent:0 caseSensitive:caseSensitive) ~~ 0 |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4293 |
|
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4294 |
" |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4295 |
'hello world' includesString:'hel' caseSensitive:true |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4296 |
'hello world' includesString:'HEL' caseSensitive:true |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4297 |
'hello world' includesString:'HEL' caseSensitive:false |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4298 |
|
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4299 |
'hello world' includesString:'wor' caseSensitive:true |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4300 |
'hello world' includesString:'WOR' caseSensitive:true |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4301 |
'hello world' includesString:'WOR' caseSensitive:false |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4302 |
" |
| 9190 | 4303 |
^ startIndex |
4304 |
]. |
|
4305 |
caseSensitive ifTrue:[ |
|
4306 |
startIndex := self indexOf:firstChar startingAt:(startIndex + 1) |
|
4307 |
] ifFalse:[ |
|
4308 |
startIndex := self findFirst:[:c | c sameAs:firstChar] startingAt:(startIndex + 1). |
|
4309 |
]. |
|
4310 |
]. |
|
4311 |
^ exceptionBlock value |
|
4312 |
||
4313 |
"Modified: 23.2.1996 / 15:35:15 / cg" |
|
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4314 |
! |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4315 |
|
| 9190 | 4316 |
restAfter:keywindexOfSubCollection:subString startingAt:index ifAbsent:exceptionBlock caseSensitive:caseSensitive |
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4317 |
"find a substring, starting at index. if found, return the index; |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4318 |
if not found, return the result of evaluating exceptionBlock. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4319 |
This is a q&d hack - not very efficient" |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4320 |
|
| 608 | 4321 |
|firstChar found |
4322 |
startIndex "{ Class: SmallInteger }"
|
|
4323 |
subSize "{ Class: SmallInteger }"
|
|
4324 |
mySize "{ Class: SmallInteger }"
|
|
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4325 |
runIdx "{ Class: SmallInteger }"
|
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4326 |
tester| |
| 608 | 4327 |
|
4328 |
subSize := subString size. |
|
|
6122
a98958f1c85f
handle non-string arg to indexOfSubString:
Claus Gittinger <cg@exept.de>
parents:
6119
diff
changeset
|
4329 |
subSize == 0 ifTrue:[ "empty string matches" |
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4330 |
subString isString ifFalse:[ |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4331 |
self error:'non string argument' mayProceed:true. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4332 |
]. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4333 |
^ index |
|
6122
a98958f1c85f
handle non-string arg to indexOfSubString:
Claus Gittinger <cg@exept.de>
parents:
6119
diff
changeset
|
4334 |
]. |
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4335 |
tester := caseSensitive ifTrue:[ [:c1 :c2 | c1 = c2 ] ] ifFalse:[ [:c1 :c2 | c1 sameAs: c2 ] ]. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4336 |
|
| 608 | 4337 |
mySize := self size. |
4338 |
firstChar := subString at:1. |
|
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4339 |
caseSensitive ifTrue:[ |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4340 |
startIndex := self indexOf:firstChar startingAt:index. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4341 |
] ifFalse:[ |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4342 |
startIndex := self findFirst:[:c | c sameAs:firstChar] startingAt:index. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4343 |
]. |
| 608 | 4344 |
[startIndex == 0] whileFalse:[ |
|
9087
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4345 |
runIdx := startIndex. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4346 |
found := true. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4347 |
1 to:subSize do:[:i | |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4348 |
runIdx > mySize ifTrue:[ |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4349 |
found := false |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4350 |
] ifFalse:[ |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4351 |
(tester value:(subString at:i) value:(self at:runIdx)) ifFalse:[ |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4352 |
found := false |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4353 |
] |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4354 |
]. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4355 |
runIdx := runIdx + 1 |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4356 |
]. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4357 |
found ifTrue:[ |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4358 |
^ startIndex |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4359 |
]. |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4360 |
caseSensitive ifTrue:[ |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4361 |
startIndex := self indexOf:firstChar startingAt:(startIndex + 1) |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4362 |
] ifFalse:[ |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4363 |
startIndex := self findFirst:[:c | c sameAs:firstChar] startingAt:(startIndex + 1). |
|
9e28a6168468
case-insensitive search cleanup
Claus Gittinger <cg@exept.de>
parents:
9058
diff
changeset
|
4364 |
]. |
| 608 | 4365 |
]. |
4366 |
^ exceptionBlock value |
|
|
1005
4507487e0be6
more on encoding/decoding (still unfinished)
Claus Gittinger <cg@exept.de>
parents:
997
diff
changeset
|
4367 |
|
|
4507487e0be6
more on encoding/decoding (still unfinished)
Claus Gittinger <cg@exept.de>
parents:
997
diff
changeset
|
4368 |
"Modified: 23.2.1996 / 15:35:15 / cg" |
| 9190 | 4369 |
AfterrestAfter:keyword withoutSeparators:strip |
|
651
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4370 |
"compare the left of the receiver with keyword, |
| 8874 | 4371 |
if it matches return the right. |
|
651
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4372 |
Finally, if strip is true, remove whiteSpace. |
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4373 |
This method is used to match and extract lines of the form: |
| 9191 | 4374 |
something: rest |
|
759
908363ce8a32
interest is written with one 'r' (shame on me)
Claus Gittinger <cg@exept.de>
parents:
735
diff
changeset
|
4375 |
where we are interested in rest, but only if the receiver string |
| 8874 | 4376 |
begins with something. |
|
651
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4377 |
|
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4378 |
You may wonder why such a specialized method exists here |
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4379 |
- this is so common when processing mailboxes, |
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4380 |
rcs files, nntp/pop3 responses, that is was considered worth |
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4381 |
a special method here to avoid having the code below a hundred |
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4382 |
times in variuos places." |
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4383 |
|
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4384 |
|rest| |
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4385 |
|
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4386 |
(self startsWith:keyword) ifTrue:[ |
| 9191 | 4387 |
rest := self copyFrom:(keyword size + 1). |
4388 |
strip ifTrue:[ |
|
4389 |
rest := rest withoutSeparators |
|
4390 |
]. |
|
4391 |
^ rest |
|
|
651
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4392 |
]. |
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4393 |
^ nil |
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4394 |
|
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4395 |
" |
| 8874 | 4396 |
'foo: hello world' restAfter:'foo:' withoutSeparators:true |
4397 |
'funny: something' restAfter:'foo:' withoutSeparators:true |
|
4398 |
||
4399 |
'foo: hello world ' restAfter:'foo:' withoutSeparators:true |
|
4400 |
'foo: hello world ' restAfter:'foo:' withoutSeparators:false |
|
|
651
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4401 |
" |
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4402 |
|
|
2093b018dc3a
added restAfter:withoutSeparators:
Claus Gittinger <cg@exept.de>
parents:
650
diff
changeset
|
4403 |
"Created: 25.11.1995 / 11:04:18 / cg" |
| 9190 | 4404 |
|
4405 |
||
4406 |
"Created: 12.5.1996 / 15:46:40 / cg" |
|
4407 |
continuesWith:aString startingAt:startIndex |
|
| 1613 | 4408 |
"return true, if the receiver beginning at startIndex |
4409 |
contains the characters in aString." |
|
4410 |
||
4411 |
|sz "{Class: SmallInteger }"
|
|
| 9190 | 4412 |
idx "{Class: SmallInteger }
|
| 9191 | 4413 |
tally := tally + 1. |
4414 |
start := stop |
|
4415 |
] |
|
| 64 | 4416 |
]. |
4417 |
^ tally |
|
4418 |
||
| 92 | 4419 |
" |
4420 |
'hello world isnt this nice' countWords' |
|
4421 |
" |
|
| 64 | 4422 |
! |
4423 |
||
|
7958
fba655ee2aa2
endsWith: fixed for character argument
Claus Gittinger <cg@exept.de>
parents:
7950
diff
changeset
|
4424 |
endsWith:aStringOrCharacter |
|
fba655ee2aa2
endsWith: fixed for character argument
Claus Gittinger <cg@exept.de>
parents:
7950
diff
changeset
|
4425 |
"return true, if the receiver ends with something, aStringOrCharacter." |
| 1381 | 4426 |
|
4427 |
|s| |
|
4428 |
||
4429 |
(s := self string) ~~ self ifTrue:[ |
|
| 9191 | 4430 |
^ s endsWith:aStringOrCharacter |
| 1381 | 4431 |
]. |
| 9190 | 4432 |
aStringOrCharacter isCcountWords |
4433 |
"return the number of words, which are separated by separators" |
|
4434 |
||
4435 |
|tally "{ Class: SmallInteger }"
|
|
4436 |
start "{
|
|
4437 |
||
4438 |
endsWith:aStringOrCharacter |
|
4439 |
"return true, if the receiver ends with something, aStringOrCharacter." |
|
4440 |
||
4441 |
|s| |
|
4442 |
||
4443 |
(s := self string) ~~ self ifTrue:[ |
|
| 9191 | 4444 |
^ s endsWith:aStringOrCharacter |
| 9190 | 4445 |
]. |
|
7958
fba655ee2aa2
endsWith: fixed for character argument
Claus Gittinger <cg@exept.de>
parents:
7950
diff
changeset
|
4446 |
aStringOrCharacter isCharacter ifTrue:[ |
| 9191 | 4447 |
^ self last = aStringOrCharacter |
|
7958
fba655ee2aa2
endsWith: fixed for character argument
Claus Gittinger <cg@exept.de>
parents:
7950
diff
changeset
|
4448 |
]. |
|
fba655ee2aa2
endsWith: fixed for character argument
Claus Gittinger <cg@exept.de>
parents:
7950
diff
changeset
|
4449 |
^ super endsWith:aStringOrCharacter |
| 1381 | 4450 |
|
4451 |
" |
|
| 8874 | 4452 |
'hello world' endsWith:'world' |
4453 |
'hello world' asText allBold endsWith:'world' |
|
| 1381 | 4454 |
" |
4455 |
||
4456 |
"Modified: 12.5.1996 / 15:49:18 / cg" |
|
4457 |
! |
|
| 9191 | 4458 |
|
4459 |
||
| 9190 | 4460 |
|
4461 |
isBinarySelector |
|
4462 |
"treating the receiver as a message selector, return true if its a binary selector" |
|
4463 |
||
4464 |
|binopChars| |
|
4465 |
||
4466 |
(self size > 3) ifTrue:[^ false]. |
|
4467 |
||
4468 |
binopChars := Scanner binarySelectorCharacters. |
|
4469 |
^ (self conform:[:char | (binopChars includes:char)]) |
|
4470 |
||
4471 |
" |
|
4472 |
'foo:bar:' isBinarySelector |
|
4473 |
#foo:bar: isBinarySelector |
|
4474 |
'hello' isBinarySelector |
|
4475 |
'+' isBinarySelector |
|
4476 |
'|' isBinarySelector |
|
4477 |
'?' isBinarySelsBinarySelector |
|
4478 |
'->' isBinarySelector |
|
4479 |
'<->' isBinarySelector |
|
4480 |
'::' isBinarySelector |
|
4481 |
" |
|
4482 |
||
4483 |
"Modified: 4.1.1997 / 14:16:14 / cg" |
|
4484 |
! |
|
4485 |
||
4486 |
isBlank |
|
4487 |
"return true, if the receiver contains spaces only" |
|
4488 |
||
4489 |
self do:[:char | |
|
| 9191 | 4490 |
char ~~ Character space ifTrue:[^ false]. |
| 9190 | 4491 |
]. |
4492 |
^ true |
|
4493 |
! |
|
4494 |
||
4495 |
isNumeric |
|
4496 |
"return true, if the receiver is some numeric word; |
|
4497 |
i.e. consists only of digits." |
|
| 64 | 4498 |
|
| 608 | 4499 |
self size == 0 ifTrue:[ |
| 9191 | 4500 |
^ false |
| 64 | 4501 |
]. |
| 9190 | 4502 |
self do:[:char |isBinarySelector |
| 7482 | 4503 |
"treating the receiver as a message selector, return true if its a binary selector" |
4504 |
||
4505 |
|binopChars| |
|
4506 |
||
4507 |
(self size > 3) ifTrue:[^ false]. |
|
4508 |
||
4509 |
binopChars := Scanner binarySelectorCharacters. |
|
4510 |
^ (self conform:[:char | (binopChars includes:char)]) |
|
4511 |
||
4512 |
" |
|
| 8874 | 4513 |
'foo:bar:' isBinarySelector |
4514 |
#foo:bar: isBinarySelector |
|
4515 |
'hello' isBinarySelector |
|
4516 |
'+' isBinarySelector |
|
4517 |
'|' isBinarySelector |
|
4518 |
'?' isBinarySelector |
|
4519 |
':' isBinarySelector |
|
4520 |
'a:' isBinarySelector |
|
4521 |
'->' isBinarySelector |
|
4522 |
'<->' isBinarySelector |
|
4523 |
'::' isBinarySelector |
|
| 7482 | 4524 |
" |
4525 |
||
4526 |
"Modified: 4.1.1997 / 14:16:14 / cg" |
|
| 9190 | 4527 |
ocmprt' levenshteinTo:'computer' |
4528 |
'computer' levenshteinTo:'computer' |
|
4529 |
'ocmputer' levenshteinTo:'computer' |
|
4530 |
'cmputer' levenshteinTo:'computer' |
|
4531 |
'computer' levenshteinTo:'cmputer' |
|
4532 |
'Computer' levenshteinTo:'computer' |
|
| 8525 | 4533 |
" |
4534 |
! |
|
4535 |
||
| 9190 | 4536 |
levenshteinTo:aString s:substWeight k:kbdTypoWeight c:caseWeight i:insrtWeight d:deleteWeight |
4537 |
"parametrized levenshtein. |
|
4538 |
return the levenshtein distance to the argument, aString; |
|
4539 |
this value corrensponds to the number of replacements that halevenshteinTo:aString |
|
| 64 | 4540 |
"return the levenshtein distance to the argument, aString; |
| 7318 | 4541 |
this value corresponds to the number of replacements that have to be |
| 64 | 4542 |
made to get aString from the receiver. |
| 196 | 4543 |
See IEEE transactions on Computers 1976 Pg 172 ff." |
| 64 | 4544 |
|
| 327 | 4545 |
" |
| 7318 | 4546 |
in the following, we assume that ommiting a character |
| 327 | 4547 |
is less of an error than inserting an extra character. |
| 7318 | 4548 |
Therefore the different insertion (i) and deletion (d) values. |
4549 |
s: substitution weight |
|
|
8646
5b40b09db73b
Remove unused method #levenshteinTo:s:c:i:d:
Stefan Vogel <sv@exept.de>
parents:
8632
diff
changeset
|
4550 |
k: keyboard weight (typing a nearby key) |
| 7318 | 4551 |
c: case weight |
4552 |
i: insertion of extra character weight |
|
4553 |
d: delete of a character weight |
|
4554 |
" |
|
4555 |
||
4556 |
^ self levenshteinTo:aString s:4 k:2 c:1 i:2 d:6 |
|
4557 |
||
4558 |
" |
|
| 8874 | 4559 |
'computer' levenshteinTo:'computer' |
4560 |
'cOmputer' levenshteinTo:'computer' |
|
4561 |
'cOmpuTer' levenshteinTo:'computer' |
|
4562 |
'cimputer' levenshteinTo:'computer' |
|
4563 |
'cumputer' levenshteinTo:'computer' |
|
4564 |
||
4565 |
'cmputer' levenshteinTo:'computer' |
|
4566 |
'coomputer' levenshteinTo:'computer' |
|
4567 |
||
4568 |
'ocmprt' levenshteinTo:'computer' |
|
| 154 | 4569 |
'computer' levenshteinTo:'computer' |
4570 |
'ocmputer' levenshteinTo:'computer' |
|
4571 |
'cmputer' levenshteinTo:'computer' |
|
| 327 | 4572 |
'computer' levenshteinTo:'cmputer' |
| 154 | 4573 |
'Computer' levenshteinTo:'computer' |
4574 |
" |
|
| 9190 | 4575 |
ifTrulevenshteinTo:aString s:substWeight k:kbdTypoWeight c:caseWeight i:insrtWeight d:deleteWeight |
| 8874 | 4576 |
"parametrized levenshtein. |
| 7318 | 4577 |
return the levenshtein distance to the argument, aString; |
4578 |
this value corrensponds to the number of replacements that have to be |
|
4579 |
made to get aString from the receiver. |
|
| 8874 | 4580 |
The arguments are the costs for |
| 9191 | 4581 |
s:substitution, |
4582 |
k:keyboard type (substitution), |
|
4583 |
c:case-change, |
|
4584 |
i:insertion |
|
4585 |
d:deletion |
|
| 7318 | 4586 |
of a character. |
| 196 | 4587 |
See IEEE transactions on Computers 1976 Pg 172 ff" |
| 64 | 4588 |
|
4589 |
|d "delta matrix" |
|
| 196 | 4590 |
len1 "{ Class: SmallInteger }"
|
4591 |
len2 "{ Class: SmallInteger }"
|
|
4592 |
dim "{ Class: SmallInteger }"
|
|
| 8874 | 4593 |
prevRow row col |
| 196 | 4594 |
dimPlus1 "{ Class: SmallInteger }"
|
| 64 | 4595 |
min pp c1 c2| |
4596 |
||
4597 |
len1 := self size. |
|
4598 |
len2 := aString size. |
|
4599 |
||
4600 |
"create the help-matrix" |
|
4601 |
||
4602 |
dim := len1 max:len2. |
|
4603 |
dimPlus1 := dim + 1. |
|
4604 |
||
4605 |
d := Array new:dimPlus1. |
|
4606 |
1 to:dimPlus1 do:[:i | |
|
| 9191 | 4607 |
d at:i put:(Array new:dimPlus1) |
| 64 | 4608 |
]. |
4609 |
||
4610 |
"init help-matrix" |
|
4611 |
||
4612 |
(d at:1) at:1 put:0. |
|
4613 |
row := d at:1. |
|
4614 |
1 to:dim do:[:j | |
|
| 9191 | 4615 |
row at:(j + 1) put:( (row at:j) + insrtWeight ) |
| 64 | 4616 |
]. |
4617 |
||
4618 |
1 to:dim do:[:i | |
|
| 9191 | 4619 |
(d at:(i + 1)) at:1 put:( ((d at:i) at:1) + deleteWeight ) |
| 64 | 4620 |
]. |
4621 |
||
4622 |
1 to:len1 do:[:i | |
|
| 9191 | 4623 |
c1 := self at:i. |
4624 |
1 to:len2 do:[:j | |
|
4625 |
c2 := aString at:j. |
|
4626 |
(c1 == c2) ifTrue:[ |
|
4627 |
pp := 0 |
|
4628 |
] ifFalse:[ |
|
4629 |
(c1 asLowercase == c2 asLowercase) ifTrue:[ |
|
4630 |
pp := caseWeight |
|
4631 |
] ifFalse:[ |
|
4632 |
pp := substWeight. |
|
4633 |
substWeight ~~ kbdTypoWeight ifTrue:[ |
|
4634 |
(DoWhatIMeanSupport isKey:c1 asLowercase nextTo:c2 asLowercase) ifTrue:[ |
|
4635 |
pp := kbdTypoWeight. |
|
4636 |
]. |
|
4637 |
]. |
|
4638 |
] |
|
4639 |
]. |
|
4640 |
prevRow := d at:i. |
|
4641 |
row := d at:(i + 1). |
|
4642 |
col := j + 1. |
|
4643 |
min := (prevRow at:j) + pp. |
|
4644 |
min := min min:( (row at:j) + insrtWeight). |
|
4645 |
min := min min:( (prevRow at:col) + deleteWeight). |
|
4646 |
row at:col put: min |
|
4647 |
] |
|
| 64 | 4648 |
]. |
4649 |
||
4650 |
^ (d at:(len1 + 1)) at:(len2 + 1) |
|
| 9190 | 4651 |
++' nnumArgs |
| 608 | 4652 |
"treating the receiver as a message selector, return how many arguments would it take" |
4653 |
||
|
9038
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4654 |
|binopChars| |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4655 |
|
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4656 |
(self size <= Scanner maxBinarySelectorSize) ifTrue:[ |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4657 |
binopChars := Scanner binarySelectorCharacters. |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4658 |
(self contains:[:someChar | (binopChars includes:someChar) not]) ifFalse:[ |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4659 |
^ 1 |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4660 |
]. |
| 608 | 4661 |
]. |
4662 |
^ self occurrencesOf:$: |
|
4663 |
||
4664 |
" |
|
|
9038
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4665 |
'foo:bar:' numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4666 |
#foo:bar: numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4667 |
'hello' numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4668 |
'+' numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4669 |
'++' numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4670 |
'+++' numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4671 |
'|' numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4672 |
'?' numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4673 |
'_' numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4674 |
'_:' numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4675 |
'_:_:' numArgs |
|
d75b89570ec0
numArgs fixed (binOps can now be longer)
Claus Gittinger <cg@exept.de>
parents:
9019
diff
changeset
|
4676 |
'<->' numArgs |
| 608 | 4677 |
" |
|
2051
08fd6a693ee3
ask Scanner for valid binarySelector characters in numArgs
Claus Gittinger <cg@exept.de>
parents:
1978
diff
changeset
|
4678 |
|
|
08fd6a693ee3
ask Scanner for valid binarySelector characters in numArgs
Claus Gittinger <cg@exept.de>
parents:
1978
diff
changeset
|
4679 |
"Modified: 4.1.1997 / 14:16:14 / cg" |
| 9190 | 4680 |
#fpartsIfSelector |
| 608 | 4681 |
"treat the receiver as a message selector, return a collection of parts." |
4682 |
||
| 9190 | 4683 |
|idx1 "{ Class: SmallInteger }
|
| 8874 | 4684 |
" |
4685 |
! |
|
4686 |
||
| 9190 | 4687 |
startsWith:aString |
4688 |
"return true, if the receiver starts with something, aString. |
|
4689 |
If the argument is empty, true is returned." |
|
4690 |
||
4691 |
|s| |
|
4692 |
||
4693 |
(s := self string) ~~ self ifTrue:[ |
|
| 9191 | 4694 |
^ s startsWith:aString |
| 9190 | 4695 |
]. |
4696 |
^ super startsWith:aString |
|
4697 |
||
4698 |
" |
|
4699 |
'hello world' startsWith:'hello' |
|
4700 |
'hello world' asText allBold startsWith:'hello' |
|
4701 |
'hello world' asText allBold startsWith:'' |
|
4702 |
" |
|
4703 |
||
4704 |
"Created: 12.5.199spellAgainst: aString |
|
| 8874 | 4705 |
"return an integer between 0 and 100 indicating how similar |
| 154 | 4706 |
the argument is to the receiver. No case conversion is done. |
4707 |
This algorithm is much simpler (but also less exact) than the |
|
4708 |
levenshtein distance. Experiment which is better for your |
|
4709 |
application." |
|
| 64 | 4710 |
|
4711 |
| i1 "{ Class: SmallInteger }"
|
|
4712 |
i2 "{ Class: SmallInteger }"
|
|
| 154 | 4713 |
next1 "{ Class: SmallInteger }"
|
4714 |
next2 "{ Class: SmallInteger }"
|
|
| 64 | 4715 |
size1 "{ Class: SmallInteger }"
|
4716 |
size2 "{ Class: SmallInteger }"
|
|
4717 |
score "{ Class: SmallInteger }"
|
|
4718 |
maxLen "{ Class: SmallInteger }" |
|
|
4719 |
||
4720 |
size1 := self size. |
|
4721 |
size2 := aString size. |
|
| 154 | 4722 |
maxLen := size1 max:size2. |
| 64 | 4723 |
score := 0. |
4724 |
i1 := i2 := 1. |
|
4725 |
[i1 <= size1 and: [i2 <= size2]] whileTrue:[ |
|
| 9191 | 4726 |
next1 := i1 + 1. |
4727 |
next2 := i2 + 1. |
|
4728 |
(self at:i1) == (aString at:i2) ifTrue: [ |
|
4729 |
score := score+1. |
|
4730 |
i1 := next1. |
|
4731 |
i2 := next2 |
|
4732 |
] ifFalse: [ |
|
4733 |
(i2 < size2 and: [(self at:i1) == (aString at:next2)]) ifTrue: [ |
|
4734 |
i2 := next2 |
|
4735 |
] ifFalse: [ |
|
4736 |
(i1 < size1 and: [(self at:next1) == (aString at:i2)]) ifTrue: [ |
|
4737 |
i1 := next1 |
|
4738 |
] ifFalse: [ |
|
4739 |
i1 := next1. |
|
4740 |
i2 := next2 |
|
4741 |
] |
|
4742 |
] |
|
4743 |
] |
|
| 64 | 4744 |
]. |
4745 |
||
| 1124 | 4746 |
score == maxLen ifTrue: [^ 100]. |
| 154 | 4747 |
^ 100 * score // maxLen |
| 64 | 4748 |
|
| 8874 | 4749 |
" |
4750 |
'Smalltalk' spellAgainst: 'Smalltlak' |
|
4751 |
'Smalltalk' spellAgainst: 'smalltlak' |
|
4752 |
'Smalltalk' spellAgainst: 'smalltalk' |
|
4753 |
'Smalltalk' spellAgainst: 'smalltlk' |
|
4754 |
'Smalltalk' spellAgainst: 'Smalltolk' |
|
| 154 | 4755 |
" |
| 9190 | 4756 |
|
4757 |
||
4758 |
sstartsWith:aString |
|
|
4389
3655a0fea430
definite (true) return for zero-length arg in #startsWith:
Claus Gittinger <cg@exept.de>
parents:
4294
diff
changeset
|
4759 |
"return true, if the receiver starts with something, aString. |
|
3655a0fea430
definite (true) return for zero-length arg in #startsWith:
Claus Gittinger <cg@exept.de>
parents:
4294
diff
changeset
|
4760 |
If the argument is empty, true is returned." |
| 1381 | 4761 |
|
4762 |
|s| |
|
4763 |
||
4764 |
(s := self string) ~~ self ifTrue:[ |
|
| 9191 | 4765 |
^ s startsWith:aString |
| 1381 | 4766 |
]. |
4767 |
^ super startsWith:aString |
|
4768 |
||
4769 |
" |
|
| 8874 | 4770 |
'hello world' startsWith:'hello' |
4771 |
'hello world' asText allBold startsWith:'hello' |
|
4772 |
'hello world' asText allBold startsWith:'' |
|
| 1381 | 4773 |
" |
4774 |
||
4775 |
"Created: 12.5.1996 / 15:46:40 / cg" |
|
4776 |
"Modified: 12.5.1996 / 15:49:24 / cg" |
|
| 64 | 4777 |
! ! |
4778 |
||
| 9190 | 4779 |
!CharacterArray |
|
8395
6bd97113cb4c
Generalize visitor pattern and define #visit...:with: -methods instead
Stefan Vogel <sv@exept.de>
parents:
8376
diff
changeset
|
4780 |
|
|
6bd97113cb4c
Generalize visitor pattern and define #visit...:with: -methods instead
Stefan Vogel <sv@exept.de>
parents:
8376
diff
changeset
|
4781 |
acceptVisitor:aVisitor with:aParameter |
|
6bd97113cb4c
Generalize visitor pattern and define #visit...:with: -methods instead
Stefan Vogel <sv@exept.de>
parents:
8376
diff
changeset
|
4782 |
|
|
6bd97113cb4c
Generalize visitor pattern and define #visit...:with: -methods instead
Stefan Vogel <sv@exept.de>
parents:
8376
diff
changeset
|
4783 |
^ aVisitor visitString:self with:aParameter |
|
6bd97113cb4c
Generalize visitor pattern and define #visit...:with: -methods instead
Stefan Vogel <sv@exept.de>
parents:
8376
diff
changeset
|
4784 |
! ! |
|
6bd97113cb4c
Generalize visitor pattern and define #visit...:with: -methods instead
Stefan Vogel <sv@exept.de>
parents:
8376
diff
changeset
|
4785 |
|
|
1849
9f35f5934ec6
added msdos-latin1 en/decoder (code = #msdos850)
Claus Gittinger <cg@exept.de>
parents:
1669
diff
changeset
|
4786 |
!CharacterArray class methodsFor:'documentation'! |
| 629 | 4787 |
|
4788 |
version |
|
| 9191 | 4789 |
^ '$Header: /cvs/stx/stx/libbasic/CharacterArray.st,v 1.334 2006-02-23 18:42:30 cg Exp $' |
| 9190 | 4790 |
! ! |
4791 |
||
4792 |
CharacacceptVisitor:aVisitor with:aParameter |
|
4793 |
||
4794 |
^ aVisitor visitString:self with:aParameter |
|
4795 |
! ! |
|
4796 |
||
4797 |
!CharacterArray class methodsFor:'documentation'! |
|
4798 |
||
4799 |
version |
|
| 9191 | 4800 |
^ '$Header: /cvs/stx/stx/libbasic/CharacterArray.st,v 1.334 2006-02-23 18:42:30 cg Exp $' |
| 629 | 4801 |
! ! |
| 6809 | 4802 |
|
|
2728
7569e45a4d15
raise error signal if anything goes wrong while decoding.
Claus Gittinger <cg@exept.de>
parents:
2721
diff
changeset
|
4803 |
CharacterArray initialize! |