1699
|
1 |
"{ Encoding: utf8 }"
|
|
2 |
|
1447
|
3 |
"{ Package: 'stx:goodies/regression' }"
|
229
|
4 |
|
272
|
5 |
"{ NameSpace: RegressionTests }"
|
|
6 |
|
229
|
7 |
TestCase subclass:#CharacterEncoderTests
|
|
8 |
instanceVariableNames:''
|
|
9 |
classVariableNames:''
|
|
10 |
poolDictionaries:''
|
1549
|
11 |
category:'tests-Regression-Collections'
|
229
|
12 |
!
|
|
13 |
|
|
14 |
!CharacterEncoderTests class methodsFor:'documentation'!
|
|
15 |
|
|
16 |
documentation
|
|
17 |
"
|
|
18 |
documentation to be added.
|
|
19 |
|
|
20 |
[author:]
|
272
|
21 |
Claus Gittinger (cg@alan)
|
229
|
22 |
|
|
23 |
[instance variables:]
|
|
24 |
|
|
25 |
[class variables:]
|
|
26 |
|
|
27 |
[see also:]
|
|
28 |
|
|
29 |
"
|
|
30 |
!
|
|
31 |
|
|
32 |
history
|
|
33 |
"Created: / 11.2.2004 / 12:25:54 / cg"
|
|
34 |
! !
|
|
35 |
|
|
36 |
!CharacterEncoderTests methodsFor:'helpers'!
|
|
37 |
|
|
38 |
xTestReversability:encoderClass
|
|
39 |
self xTestReversability:encoderClass skip:#()
|
|
40 |
|
|
41 |
"
|
|
42 |
self new xTestReversability: (CharacterEncoder::ISO8859_1)
|
|
43 |
"
|
|
44 |
!
|
|
45 |
|
231
|
46 |
xTestReversability:encoderClass ignoreInvalid:ignoreInvalid
|
|
47 |
self xTestReversability:encoderClass skip:#() ignoreInvalid:ignoreInvalid
|
|
48 |
|
|
49 |
"
|
|
50 |
self new xTestReversability: (CharacterEncoder::ISO8859_1)
|
|
51 |
"
|
|
52 |
!
|
|
53 |
|
|
54 |
xTestReversability:encoderClass range:range skip:skippedCodes
|
|
55 |
self
|
272
|
56 |
xTestReversability:encoderClass
|
|
57 |
range:range
|
|
58 |
skip:skippedCodes
|
|
59 |
ignoreInvalid:false
|
231
|
60 |
|
|
61 |
"
|
|
62 |
self new xTestReversability: (CharacterEncoder::ISO8859_1)
|
|
63 |
"
|
|
64 |
!
|
|
65 |
|
|
66 |
xTestReversability:encoderClass range:range skip:skippedCodes ignoreInvalid:ignoreInvalid
|
229
|
67 |
|encoder|
|
|
68 |
|
|
69 |
encoder := encoderClass new.
|
231
|
70 |
range do:[:code |
|
272
|
71 |
|unicode code2 invalid|
|
229
|
72 |
|
272
|
73 |
(skippedCodes includes:code) ifTrue:[
|
|
74 |
self should:[ encoder decode:code ] raise:CharacterEncoderError.
|
|
75 |
] ifFalse:[
|
|
76 |
ignoreInvalid ifTrue:[
|
|
77 |
invalid := false.
|
|
78 |
CharacterEncoderError handle:[:ex |
|
|
79 |
invalid := true.
|
|
80 |
] do:[
|
|
81 |
unicode := encoder decode:code.
|
|
82 |
].
|
|
83 |
invalid ifFalse:[
|
|
84 |
code2 := encoder encode:unicode.
|
|
85 |
self assert:(code == code2).
|
|
86 |
].
|
|
87 |
] ifFalse:[
|
|
88 |
unicode := encoder decode:code.
|
|
89 |
code2 := encoder encode:unicode.
|
|
90 |
self assert:(code == code2).
|
|
91 |
]
|
|
92 |
]
|
229
|
93 |
].
|
|
94 |
|
|
95 |
"
|
|
96 |
self new xTestReversability: (CharacterEncoder::ISO8859_1)
|
231
|
97 |
|
272
|
98 |
self new
|
|
99 |
xTestReversability:(CharacterEncoder::JIS0208)
|
|
100 |
range:(16r8140 to:16rEAA4)
|
|
101 |
skip:#()
|
|
102 |
ignoreInvalid:true
|
231
|
103 |
"
|
|
104 |
!
|
|
105 |
|
|
106 |
xTestReversability:encoderClass skip:skippedCodes
|
272
|
107 |
self
|
|
108 |
xTestReversability:encoderClass
|
|
109 |
range:(encoderClass minCode to:encoderClass maxCode)
|
|
110 |
skip:skippedCodes
|
231
|
111 |
|
|
112 |
"
|
|
113 |
self new xTestReversability: (CharacterEncoder::ISO8859_1)
|
|
114 |
self new xTestReversability: (CharacterEncoder::JIS0208)
|
|
115 |
"
|
|
116 |
!
|
|
117 |
|
|
118 |
xTestReversability:encoderClass skip:skippedCodes ignoreInvalid:ignoreInvalid
|
272
|
119 |
self
|
|
120 |
xTestReversability:encoderClass
|
|
121 |
range:(encoderClass minCode to:encoderClass maxCode)
|
|
122 |
skip:skippedCodes
|
|
123 |
ignoreInvalid:ignoreInvalid
|
231
|
124 |
|
|
125 |
"
|
|
126 |
self new xTestReversability: (CharacterEncoder::ISO8859_1)
|
230
|
127 |
self new xTestReversability: (CharacterEncoder::JIS0208)
|
229
|
128 |
"
|
|
129 |
! !
|
|
130 |
|
|
131 |
!CharacterEncoderTests methodsFor:'initialize / release'!
|
|
132 |
|
|
133 |
setUp
|
|
134 |
"common setup - invoked before testing."
|
|
135 |
|
|
136 |
super setUp
|
|
137 |
!
|
|
138 |
|
|
139 |
tearDown
|
|
140 |
"common cleanup - invoked after testing."
|
|
141 |
|
|
142 |
super tearDown
|
|
143 |
! !
|
|
144 |
|
|
145 |
!CharacterEncoderTests methodsFor:'tests'!
|
|
146 |
|
|
147 |
test1
|
1427
|
148 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_1).
|
|
149 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_2).
|
|
150 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_3) skip:#( 16rA5 16rAE 16rBE 16rC3 16rD0 16rE3 16rF0 ).
|
|
151 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_4).
|
|
152 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_5).
|
|
153 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_6) skip:#( 16rA1 16rA2 16rA3 16rA5 16rA6 16rA7
|
1447
|
154 |
16rA8 16rA9 16rAA 16rAB 16rAE 16rAF
|
|
155 |
16rB0 16rB1 16rB2 16rB3 16rB4 16rB5 16rB6 16rB7
|
|
156 |
16rB8 16rB9 16rBA 16rBC 16rBD 16rBE
|
|
157 |
16rC0
|
|
158 |
16rDB 16rDC 16rDD 16rDE 16rDF
|
|
159 |
16rF3 16rF4 16rF5 16rF6 16rF7
|
|
160 |
16rF8 16rF9 16rFA 16rFB 16rFC 16rFD 16rFE 16rFF
|
|
161 |
).
|
1427
|
162 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_7) skip:#( 16rAE 16rD2 16rFF ).
|
|
163 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_8) skip:#( 16rA1 16rBF
|
1447
|
164 |
16rC0 16rC1 16rC2 16rC3 16rC4 16rC5 16rC6 16rC7
|
|
165 |
16rC8 16rC9 16rCA 16rCB 16rCC 16rCD 16rCE 16rCF
|
|
166 |
16rD0 16rD1 16rD2 16rD3 16rD4 16rD5 16rD6 16rD7
|
|
167 |
16rD8 16rD9 16rDA 16rDB 16rDC 16rDD 16rDE
|
|
168 |
16rFB 16rFC 16rFF
|
|
169 |
).
|
1427
|
170 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_9).
|
|
171 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_10).
|
|
172 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_11) skip:#( 16rDB 16rDC 16rDD 16rDE 16rFC 16rFD 16rFE 16rFF ).
|
|
173 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_13).
|
|
174 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_14).
|
|
175 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_15).
|
|
176 |
self xTestReversability: (CharacterEncoderImplementations::ISO8859_16).
|
|
177 |
self xTestReversability: (CharacterEncoderImplementations::KOI8_U).
|
|
178 |
self xTestReversability: (CharacterEncoderImplementations::KOI8_R).
|
229
|
179 |
|
1427
|
180 |
"/ self xTestReversability: (CharacterEncoderImplementations::GSM0338) skip:( 16r80 to: 16rFF).
|
229
|
181 |
|
1427
|
182 |
"/ self xTestReversability: (CharacterEncoderImplementations::JIS0208) ignoreInvalid:true.
|
231
|
183 |
|
229
|
184 |
"
|
|
185 |
self new test1
|
|
186 |
"
|
232
|
187 |
!
|
|
188 |
|
|
189 |
testUTF8
|
|
190 |
|encoder|
|
|
191 |
|
|
192 |
encoder := CharacterEncoder encoderToEncodeFrom:#unicode into:#utf8.
|
|
193 |
|
|
194 |
"/ reversibility
|
1427
|
195 |
(0 to:16r1FFFF) do:[:eachCodePoint |
|
1699
|
196 |
|s1 s2 s3|
|
232
|
197 |
|
1699
|
198 |
s1 := (Character value:eachCodePoint) asString.
|
|
199 |
s2 := encoder encodeString:s1.
|
|
200 |
s3 := encoder decodeString:s2.
|
|
201 |
self assert:(s1 = s3).
|
232
|
202 |
].
|
|
203 |
|
233
|
204 |
"/ 00 .. 7F -> 0xxxxxxx
|
|
205 |
#[16r00 16r01 16r02 16r04 16r08 16r10 16r20 16r40
|
1699
|
206 |
16r03 16r07 16r0F 16r1F 16r3F 16r7F]
|
233
|
207 |
do:[:eachCodePoint |
|
1699
|
208 |
|s1 s2 s3|
|
233
|
209 |
|
1699
|
210 |
s1 := (Character value:eachCodePoint) asString.
|
|
211 |
s2 := encoder encodeString:s1.
|
|
212 |
s3 := encoder decodeString:s2.
|
|
213 |
self assert:(s1 = s2).
|
|
214 |
self assert:(s2 = s3).
|
|
215 |
self assert:(s2 size == 1).
|
233
|
216 |
].
|
|
217 |
|
|
218 |
"/ 80 .. 7FF -> 110xxxxx 10xxxxxx
|
|
219 |
#(16r80 16r100 16r200 16r400
|
1699
|
220 |
16r0FF 16r1FF 16r3FF 16r7FF)
|
233
|
221 |
do:[:eachCodePoint |
|
1699
|
222 |
|s1 s2 s3|
|
233
|
223 |
|
1699
|
224 |
s1 := (Character value:eachCodePoint) asString.
|
|
225 |
s2 := encoder encodeString:s1.
|
|
226 |
self assert:(s2 size == 2).
|
|
227 |
self assert:((s2 first codePoint bitAnd:2r11100000) == 2r11000000).
|
|
228 |
self assert:((s2 second codePoint bitAnd:2r11000000) == 2r10000000).
|
|
229 |
s3 := encoder decodeString:s2.
|
|
230 |
self assert:(s1 = s3).
|
233
|
231 |
].
|
|
232 |
|
|
233 |
"/ 800 .. FFFF -> 1110xxxx 10xxxxxx 10xxxxxx
|
|
234 |
#(16r800 16r1000 16r2000 16r4000 16r8000
|
1699
|
235 |
16r0FFF 16r1FFF 16r3FFF 16r7FFF 16rFFFF)
|
233
|
236 |
do:[:eachCodePoint |
|
1699
|
237 |
|s1 s2 s3|
|
233
|
238 |
|
1699
|
239 |
s1 := (Character value:eachCodePoint) asString.
|
|
240 |
s2 := encoder encodeString:s1.
|
|
241 |
self assert:(s2 size == 3).
|
|
242 |
self assert:((s2 first codePoint bitAnd:2r11110000) == 2r11100000).
|
|
243 |
self assert:((s2 second codePoint bitAnd:2r11000000) == 2r10000000).
|
|
244 |
self assert:((s2 third codePoint bitAnd:2r11000000) == 2r10000000).
|
|
245 |
s3 := encoder decodeString:s2.
|
|
246 |
self assert:(s1 = s3).
|
233
|
247 |
].
|
|
248 |
|
|
249 |
"/ 10000 .. 1FFFFF -> 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
|
|
250 |
#(16r10000 16r20000 16r40000 16r80000 16r10000
|
1699
|
251 |
16r1FFFF 16r3FFFF 16r7FFFF 16rFFFFF 16r1FFFFF)
|
233
|
252 |
do:[:eachCodePoint |
|
1699
|
253 |
|s1 s2 s3|
|
233
|
254 |
|
1699
|
255 |
s1 := (Character value:eachCodePoint) asString.
|
|
256 |
s2 := encoder encodeString:s1.
|
|
257 |
self assert:(s2 size == 4).
|
|
258 |
self assert:((s2 first codePoint bitAnd:2r11111000) == 2r11110000).
|
|
259 |
self assert:((s2 second codePoint bitAnd:2r11000000) == 2r10000000).
|
|
260 |
self assert:((s2 third codePoint bitAnd:2r11000000) == 2r10000000).
|
|
261 |
self assert:((s2 fourth codePoint bitAnd:2r11000000) == 2r10000000).
|
|
262 |
s3 := encoder decodeString:s2.
|
|
263 |
self assert:(s1 = s3).
|
233
|
264 |
].
|
|
265 |
|
|
266 |
"/ 200000 .. 3FFFFFF -> 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
|
|
267 |
#(16r200000 16r400000 16r800000 16r1000000 16r2000000
|
1699
|
268 |
16r3FFFFF 16r7FFFFF 16r0FFFFFF 16r1FFFFFF 16r3FFFFFF)
|
233
|
269 |
do:[:eachCodePoint |
|
1699
|
270 |
|s1 s2 s3|
|
233
|
271 |
|
1699
|
272 |
s1 := (Character value:eachCodePoint) asString.
|
|
273 |
s2 := encoder encodeString:s1.
|
|
274 |
self assert:(s2 size == 5).
|
|
275 |
self assert:((s2 first codePoint bitAnd:2r11111100) == 2r11111000).
|
|
276 |
self assert:((s2 second codePoint bitAnd:2r11000000) == 2r10000000).
|
|
277 |
self assert:((s2 third codePoint bitAnd:2r11000000) == 2r10000000).
|
|
278 |
self assert:((s2 fourth codePoint bitAnd:2r11000000) == 2r10000000).
|
|
279 |
self assert:((s2 fifth codePoint bitAnd:2r11000000) == 2r10000000).
|
|
280 |
s3 := encoder decodeString:s2.
|
|
281 |
self assert:(s1 = s3).
|
233
|
282 |
].
|
|
283 |
|
1427
|
284 |
"/ ST/X limitation: only 30 bit integers (to avoid largeInteger codePoint)
|
233
|
285 |
|
|
286 |
"/ 4000000 .. 7FFFFFFF -> 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
|
|
287 |
#(16r4000000 16r8000000 16r10000000 16r20000000 "16r40000000"
|
1699
|
288 |
16r7FFFFFF 16r0FFFFFFF 16r1FFFFFFF "16r3FFFFFFF 16r7FFFFFFF")
|
272
|
289 |
do:[:eachCodePoint |
|
1699
|
290 |
|s1 s2 s3|
|
233
|
291 |
|
1699
|
292 |
s1 := (Character value:eachCodePoint) asString.
|
|
293 |
s2 := encoder encodeString:s1.
|
|
294 |
self assert:(s2 size == 6).
|
|
295 |
self assert:((s2 first codePoint bitAnd:2r11111110) == 2r11111100).
|
|
296 |
self assert:((s2 second codePoint bitAnd:2r11000000) == 2r10000000).
|
|
297 |
self assert:((s2 third codePoint bitAnd:2r11000000) == 2r10000000).
|
|
298 |
self assert:((s2 fourth codePoint bitAnd:2r11000000) == 2r10000000).
|
|
299 |
self assert:((s2 fifth codePoint bitAnd:2r11000000) == 2r10000000).
|
|
300 |
self assert:((s2 sixth codePoint bitAnd:2r11000000) == 2r10000000).
|
|
301 |
s3 := encoder decodeString:s2.
|
|
302 |
self assert:(s1 = s3).
|
233
|
303 |
].
|
|
304 |
|
232
|
305 |
"
|
|
306 |
self new testUTF8
|
|
307 |
"
|
1699
|
308 |
!
|
|
309 |
|
|
310 |
testUTF8_2
|
|
311 |
|encoder s|
|
|
312 |
|
|
313 |
encoder := CharacterEncoder encoderToEncodeFrom:#unicode into:#utf8.
|
|
314 |
|
|
315 |
self assert:(s := encoder encodeString:(Character value:16r1FFFFF)) asByteArray = #[16rF7 16rBF 16rBF 16rBF ].
|
|
316 |
self assert:( encoder decodeString:s ) first = (Character value:16r1FFFFF).
|
|
317 |
|
|
318 |
self assert:(s := encoder encodeString:(Character value:16r200000)) asByteArray = #[16rF8 16r88 16r80 16r80 16r80 ].
|
|
319 |
self assert:( encoder decodeString:s ) first = (Character value:16r200000).
|
|
320 |
|
|
321 |
self assert:( encoder encodeString:(Character value:16r2FFFFF)) asByteArray = #[16rF8 16r8B 16rBF 16rBF 16rBF ].
|
|
322 |
self assert:( encoder encodeString:(Character value:16r3FFFFF)) asByteArray = #[16rF8 16r8F 16rBF 16rBF 16rBF ].
|
|
323 |
self assert:( encoder encodeString:(Character value:16r4FFFFF)) asByteArray = #[16rF8 16r93 16rBF 16rBF 16rBF ].
|
|
324 |
self assert:( encoder encodeString:(Character value:16r5FFFFF)) asByteArray = #[16rF8 16r97 16rBF 16rBF 16rBF ].
|
|
325 |
self assert:( encoder encodeString:(Character value:16r6FFFFF)) asByteArray = #[16rF8 16r9B 16rBF 16rBF 16rBF ].
|
|
326 |
self assert:( encoder encodeString:(Character value:16r7FFFFF)) asByteArray = #[16rF8 16r9F 16rBF 16rBF 16rBF ].
|
|
327 |
self assert:( encoder encodeString:(Character value:16r800000)) asByteArray = #[16rF8 16rA0 16r80 16r80 16r80 ].
|
|
328 |
|
|
329 |
self assert:( encoder encodeString:(Character value:16r3FFFFFF)) asByteArray = #[16rFB 16rBF 16rBF 16rBF 16rBF ].
|
|
330 |
self assert:( encoder encodeString:(Character value:16r4000000)) asByteArray = #[16rFC 16r84 16r80 16r80 16r80 16r80 ].
|
|
331 |
self assert:( encoder encodeString:(Character value:16r7FFFFFF)) asByteArray = #[16rFC 16r87 16rBF 16rBF 16rBF 16rBF ].
|
|
332 |
self assert:( encoder encodeString:(Character value:16rFFFFFFF)) asByteArray = #[16rFC 16r8F 16rBF 16rBF 16rBF 16rBF ].
|
|
333 |
|
|
334 |
self assert:(s := encoder encodeString:(Character value:16r3FFFFFFF)) asByteArray = #[16rFC 16rBF 16rBF 16rBF 16rBF 16rBF ].
|
|
335 |
self assert:( encoder decodeString:s ) first = (Character value:16r3FFFFFFF).
|
|
336 |
|
|
337 |
"/ STX <<only>> supports characters up to 31 bit.
|
|
338 |
"/ self assert:( encoder encodeString:(Character value:16r7FFFFFFF)) asByteArray = #[16rFD 16rBF 16rBF 16rBF 16rBF 16rBF ].
|
|
339 |
|
|
340 |
|
|
341 |
"
|
|
342 |
self new testUTF8_2
|
|
343 |
"
|
229
|
344 |
! !
|
|
345 |
|
|
346 |
!CharacterEncoderTests class methodsFor:'documentation'!
|
|
347 |
|
|
348 |
version
|
|
349 |
^ '$Header$'
|
|
350 |
! !
|
1427
|
351 |
|