hg/stx-libbasic: comparison String.st

equal deleted inserted replaced

-:196706395bbc
+:4118d61ddba0
 ^ self == String
 "Modified: 23.4.1996 / 16:00:38 / cg"
 ! !
 ^ self copy changeClassTo:ImmutableString
 !
 asLowercase
-"a tuned version. Some apps call this very heavily"
+"a tuned version for Strings with size < 255. Some apps call this very heavily.
+We can do this for 8-bit strings, since the mapping is well known and lowercase chars
-%{  /* NOCONTEXT */
+fit in one byte also."
-#if 0
+%{  /* NOCONTEXT */
 #ifndef __SCHTEAM__
 REGISTER OBJ slf = self;
-if (__qClass(slf) == String) {
+if (__isStringLike(slf)) {
-	char quickBuffer[256];
+char quickBuffer[256];
-	int sz = __stringSize(slf);
+int sz = __stringSize(slf);
-	if (sz < (sizeof(quickBuffer)-1)) {
+if (sz < (sizeof(quickBuffer)-1)) {
-	    REGISTER int i = 0;
+REGISTER int i = 0;
-	    int anyChange = 0;
+int anyChange = 0;
-	    REGISTER unsigned char *cp = __stringVal(slf);
+REGISTER unsigned char *cp = __stringVal(slf);
-	    // fast advance
+// fast advance
-	    // all uppercase chars are in the ranges 0x41 .. 0x5A (A..Z)
+// all uppercase chars are in the ranges 0x41 .. 0x5A (A..Z)
-	    // or 0xC0 .. 0xDF.
+// or 0xC0 .. 0xDF.
-	    // I.e. they have the 0x20 bit clear.
+// I.e. they have the 0x20 bit clear.
-	    // Thus, we can fast skip over lowercase, spaces and some puctuation,
+// Thus, we can fast skip over lowercase, spaces and some punctuation,
-	    // if all bytes of a word have the x20 bit set.
+// if all bytes of a word have the x20 bit set.
-	    //
 #if __POINTER_SIZE__ == 8
-	    for (; i<(sz-8); i+=8) {
+for (; i < (sz-8); i += 8) {
-		unsigned INT eightChars;
+unsigned INT eightChars = *(unsigned INT *)(cp+i);
+if ((eightChars & 0x2020202020202020ULL) != 0x2020202020202020ULL) goto convert;
-		eightChars = *((unsigned INT *)(cp));
+*(unsigned INT *)(quickBuffer+i) = eightChars;
-		if ((eightChars & 0x2020202020202020ULL) != 0x2020202020202020ULL) break;
+}
-		*((unsigned INT *)(&quickBuffer[i])) = eightChars;
-	    }
 #endif
-	    for (; i<(sz-4); i+=4) {
+for (; i < (sz-4); i += 4) {
-		unsigned int fourChars;
+unsigned int fourChars = *(unsigned int *)(cp+i);
+if ((fourChars & 0x20202020U) != 0x20202020U) break;
-		fourChars = *((unsigned int *)(cp));
+*(unsigned int *)(quickBuffer+i) = fourChars;
-		if ((fourChars & 0x20202020U) != 0x20202020U) break;
+}
-		*((unsigned int *)(&quickBuffer[i])) = fourChars;
+convert:
-	    }
+for (; i<sz; i++) {
+unsigned char ch = cp[i];
-	    for (; i<sz; i++) {
-		unsigned char ch = cp[i];
+quickBuffer[i] = ch;
+if ((ch & 0x60) == 0x40) {
-		quickBuffer[i] = ch;
+if (ch >= 'A' && ch <= 'Z') {
-		if ((ch & 0x60) == 0x40) {
+quickBuffer[i] = ch - 'A' + 'a';
-		    if (ch >= 'A') {
+anyChange = 1;
-			if (ch <= 'Z') {
+} else {
-			    quickBuffer[i] = ch - 'A' + 'a';
+// deal with national latin1 characters
-			    anyChange = 1;
+if (ch >= 0xC0 && ch <= 0xDE && ch != 0xD7) {
-			} else {
+quickBuffer[i] = ch + 0x20;
-			    // deal with national latin1 characters
+anyChange = 1;
-			    if (ch >= 0xC0) {
+}
-				if (ch <= 0xDE) {
+}
-				    if (ch != 0xD7) {
+}
-					quickBuffer[i] = ch + 0x20;
+}
-					anyChange = 1;
+if (! anyChange) {
-				    }
+RETURN(slf);
-				}
+}
-			    }
+quickBuffer[i] = '\0';
-			}
+RETURN (__MKSTRING_L(quickBuffer, i));
-		    }
+}
-		}
-	    }
-	    quickBuffer[i] = '\0';
-	    if (! anyChange) {
-		RETURN(slf);
-	    }
-	    RETURN (__MKSTRING_L(quickBuffer, i));
-	}
 }
 #endif /* ! __SCHTEAM__ */
-#endif
 %}.
 ^ super asLowercase
+"
+'Hello WORLD' asLowercase
+(String new:300) asLowercase
+"
 !
 asPackageId
 "given a package-string as receiver, return a packageId object.
 packageIds hide the details of module/directory handling inside the path.
 */
 if (__qIsStringLike(self)) {
 char *cp1 = (char *) __stringVal(self);
 int l1 = __stringSize(self);
 int l2;
-char *cp2;
+char *cp2 = 0;
 int sz;
 OBJ newString;
 char character;
 if (__isCharacter(aStringOrCharacter)) {
 character = __intVal(__characterVal(aStringOrCharacter));
 l2 = 1;
 cp2 = &character;
 } else if (__isStringLike(aStringOrCharacter)) {
 l2 = __stringSize(aStringOrCharacter);
-cp2 = (char *) __stringVal(aStringOrCharacter);
 } else
 goto out;
 sz = OHDR_SIZE + l1 + l2 + 1;
 __qNew(newString, sz);      /* OBJECT ALLOCATION */
+cp1 = (char *) __stringVal(self);
+if (cp2 == 0)
+cp2 = (char *) __stringVal(aStringOrCharacter);
 if (newString != nil) {
 REGISTER unsigned char *dstp;
 __InstPtr(newString)->o_class = String;
 __qSTORE(newString, String);
 dstp = __stringVal(newString);
-# ifdef bcopy4
+# if defined(bcopy4)
 /* knowing that allocation is 4-byte aligned and
 * size rounded up to next 4-byte, the first copy
 * can be done word-wise.
 * that speeds up size-10-string , size-10-string
 * by 10% on a P5/200.
 if (l1 & 3) nw++;
 bcopy4(cp1, dstp, nw);
 dstp += l1;
 }
-# else
+# elif defined(FAST_MEMCPY)
-#  ifdef FAST_MEMCPY
 memcpy(dstp, cp1, l1);
 dstp += l1;
-#  else
+# else
 while (l1 >= 4) {
 *(int *)dstp = *(int *)cp1;
 dstp += 4; cp1 += 4;
 l1 -= 4;
 }
 while (l1--) *dstp++ = *cp1++;
-#  endif
 # endif
 # ifdef bcopy4
 if (((INT)dstp & 3) == 0) {
 int nw = l2 >> 2;

branch	jv
changeset 20131	4118d61ddba0
parent 20083	196706395bbc
parent 20126	8341bd725f11
child 20136	8e3509beb85d