diff options
Diffstat (limited to 'generic/tclStringObj.c')
| -rw-r--r-- | generic/tclStringObj.c | 327 |
1 files changed, 244 insertions, 83 deletions
diff --git a/generic/tclStringObj.c b/generic/tclStringObj.c index 91c0e74..b486106 100644 --- a/generic/tclStringObj.c +++ b/generic/tclStringObj.c @@ -39,15 +39,6 @@ #include "tclStringRep.h" /* - * Set COMPAT to 1 to restore the shimmering patterns to those of Tcl 8.5. - * This is an escape hatch in case the changes have some unexpected unwelcome - * impact on performance. If things go well, this mechanism can go away when - * post-8.6 development begins. - */ - -#define COMPAT 0 - -/* * Prototypes for functions defined later in this file: */ @@ -418,6 +409,15 @@ Tcl_GetCharLength( int numChars; /* + * Quick, no-shimmer return for short string reps. + */ + + if ((objPtr->bytes) && (objPtr->length < 2)) { + /* 0 bytes -> 0 chars; 1 byte -> 1 char */ + return objPtr->length; + } + + /* * Optimize the case where we're really dealing with a bytearray object * without string representation; we don't need to convert to a string to * perform the get-length operation. @@ -445,18 +445,6 @@ Tcl_GetCharLength( if (numChars == -1) { TclNumUtfChars(numChars, objPtr->bytes, objPtr->length); stringPtr->numChars = numChars; - -#if COMPAT - if (numChars < objPtr->length) { - /* - * Since we've just computed the number of chars, and not all UTF - * chars are 1-byte long, go ahead and populate the unicode - * string. - */ - - FillUnicodeRep(objPtr); - } -#endif } return numChars; } @@ -1173,11 +1161,7 @@ Tcl_AppendUnicodeToObj( * objPtr's string rep. */ - if (stringPtr->hasUnicode -#if COMPAT - && stringPtr->numChars > 0 -#endif - ) { + if (stringPtr->hasUnicode) { AppendUnicodeToUnicodeRep(objPtr, unicode, length); } else { AppendUnicodeToUtfRep(objPtr, unicode, length); @@ -1283,11 +1267,7 @@ Tcl_AppendObjToObj( * appendObjPtr and append it. */ - if (stringPtr->hasUnicode -#if COMPAT - && stringPtr->numChars > 0 -#endif - ) { + if (stringPtr->hasUnicode) { /* * If appendObjPtr is not of the "String" type, don't convert it. */ @@ -1320,11 +1300,7 @@ Tcl_AppendObjToObj( AppendUtfToUtfRep(objPtr, bytes, length); - if (numChars >= 0 && appendNumChars >= 0 -#if COMPAT - && appendNumChars == length -#endif - ) { + if (numChars >= 0 && appendNumChars >= 0) { stringPtr->numChars = numChars + appendNumChars; } } @@ -1448,14 +1424,6 @@ AppendUnicodeToUtfRep( if (stringPtr->numChars != -1) { stringPtr->numChars += numChars; } - -#if COMPAT - /* - * Invalidate the unicode rep. - */ - - stringPtr->hasUnicode = 0; -#endif } /* @@ -2328,7 +2296,7 @@ Tcl_AppendFormatToObj( } } - Tcl_GetStringFromObj(segment, &segmentNumBytes); + TclGetStringFromObj(segment, &segmentNumBytes); if (segmentNumBytes > limit) { if (allocSegment) { Tcl_DecrRefCount(segment); @@ -2641,6 +2609,235 @@ TclGetStringStorage( *sizePtr = stringPtr->allocated; return objPtr->bytes; } + +/* + *--------------------------------------------------------------------------- + * + * TclStringCatObjv -- + * + * Performs the [string cat] function. + * + * Results: + * A standard Tcl result. + * + * Side effects: + * Writes to *objPtrPtr the address of Tcl_Obj that is concatenation + * of all objc values in objv. + * + *--------------------------------------------------------------------------- + */ + +int +TclStringCatObjv( + Tcl_Interp *interp, + int inPlace, + int objc, + Tcl_Obj * const objv[], + Tcl_Obj **objPtrPtr) +{ + Tcl_Obj *objPtr, *objResultPtr, * const *ov; + int oc, length = 0, binary = 1, first = 0; + int allowUniChar = 1, requestUniChar = 0; + + /* assert (objc >= 2) */ + + /* + * Analyze to determine what representation result should be. + * GOALS: Avoid shimmering & string rep generation. + * Produce pure bytearray when possible. + * Error on overflow. + */ + + ov = objv, oc = objc; + while (oc-- && (binary || allowUniChar)) { + objPtr = *ov++; + + if (objPtr->bytes) { + /* Value has a string rep. */ + if (objPtr->length) { + /* + * Non-empty string rep. Not a pure bytearray, so we + * won't create a pure bytearray + */ + binary = 0; + if ((objPtr->typePtr) && (objPtr->typePtr != &tclStringType)) { + /* Prevent shimmer of non-string types. */ + allowUniChar = 0; + } + } + } else { + /* assert (objPtr->typePtr != NULL) -- stork! */ + if (TclIsPureByteArray(objPtr)) { + allowUniChar = 0; + } else { + binary = 0; + if (objPtr->typePtr == &tclStringType) { + /* Have a pure Unicode value; ask to preserve it */ + requestUniChar = 1; + } else { + /* Have another type; prevent shimmer */ + allowUniChar = 0; + } + } + } + } + + if (binary) { + /* Result will be pure byte array. Pre-size it */ + ov = objv; oc = objc; + while (oc-- && (length >= 0)) { + objPtr = *ov++; + + if (objPtr->bytes == NULL) { + int numBytes; + + Tcl_GetByteArrayFromObj(objPtr, &numBytes); + if (length == 0) { + first = objc - oc - 1; + } + length += numBytes; + } + } + } else if (allowUniChar && requestUniChar) { + /* Result will be pure Tcl_UniChar array. Pre-size it. */ + ov = objv; oc = objc; + while (oc-- && (length >= 0)) { + objPtr = *ov++; + + if ((objPtr->bytes == NULL) || (objPtr->length)) { + int numChars; + + Tcl_GetUnicodeFromObj(objPtr, &numChars); + if (length == 0) { + first = objc - oc - 1; + } + length += numChars; + } + } + } else { + /* Result will be concat of string reps. Pre-size it. */ + ov = objv; oc = objc; + while (oc-- && (length >= 0)) { + int numBytes; + + objPtr = *ov++; + + Tcl_GetStringFromObj(objPtr, &numBytes); + if ((length == 0) && numBytes) { + first = objc - oc - 1; + } + length += numBytes; + } + } + + if (length < 0) { + if (interp) { + Tcl_SetObjResult(interp, Tcl_ObjPrintf( + "max size for a Tcl value (%d bytes) exceeded", INT_MAX)); + Tcl_SetErrorCode(interp, "TCL", "MEMORY", NULL); + } + return TCL_ERROR; + } + + if (length == 0) { + /* Total length of zero means every value has length zero */ + *objPtrPtr = objv[0]; + return TCL_OK; + } + + objv += first; objc -= first; + + if (binary) { + /* Efficiently produce a pure byte array result */ + unsigned char *dst; + + if (inPlace && !Tcl_IsShared(*objv)) { + int start; + + objResultPtr = *objv++; objc--; + Tcl_GetByteArrayFromObj(objResultPtr, &start); + dst = Tcl_SetByteArrayLength(objResultPtr, length) + start; + } else { + objResultPtr = Tcl_NewByteArrayObj(NULL, length); + dst = Tcl_SetByteArrayLength(objResultPtr, length); + } + while (objc--) { + Tcl_Obj *objPtr = *objv++; + + if (objPtr->bytes == NULL) { + int more; + unsigned char *src = Tcl_GetByteArrayFromObj(objPtr, &more); + memcpy(dst, src, (size_t) more); + dst += more; + } + } + } else if (allowUniChar && requestUniChar) { + /* Efficiently produce a pure Tcl_UniChar array result */ + Tcl_UniChar *dst; + + if (inPlace && !Tcl_IsShared(*objv)) { + int start; + + objResultPtr = *objv++; objc--; + + /* Ugly interface! Force resize of the unicode array. */ + Tcl_GetUnicodeFromObj(objResultPtr, &start); + Tcl_InvalidateStringRep(objResultPtr); + Tcl_SetObjLength(objResultPtr, length); + dst = Tcl_GetUnicode(objResultPtr) + start; + } else { + Tcl_UniChar ch = 0; + + /* Ugly interface! No scheme to init array size. */ + objResultPtr = Tcl_NewUnicodeObj(&ch, 0); + Tcl_SetObjLength(objResultPtr, length); + dst = Tcl_GetUnicode(objResultPtr); + } + while (objc--) { + Tcl_Obj *objPtr = *objv++; + + if ((objPtr->bytes == NULL) || (objPtr->length)) { + int more; + Tcl_UniChar *src = Tcl_GetUnicodeFromObj(objPtr, &more); + memcpy(dst, src, more * sizeof(Tcl_UniChar)); + dst += more; + } + } + } else { + /* Efficiently concatenate string reps */ + char *dst; + + if (inPlace && !Tcl_IsShared(*objv)) { + int start; + + objResultPtr = *objv++; objc--; + + Tcl_GetStringFromObj(objResultPtr, &start); + Tcl_SetObjLength(objResultPtr, length); + dst = Tcl_GetString(objResultPtr) + start; + if (length > start) { + TclFreeIntRep(objResultPtr); + } + } else { + objResultPtr = Tcl_NewObj(); + Tcl_SetObjLength(objResultPtr, length); + dst = Tcl_GetString(objResultPtr); + } + while (objc--) { + Tcl_Obj *objPtr = *objv++; + + if ((objPtr->bytes == NULL) || (objPtr->length)) { + int more; + char *src = Tcl_GetStringFromObj(objPtr, &more); + memcpy(dst, src, (size_t) more); + dst += more; + } + } + } + *objPtrPtr = objResultPtr; + return TCL_OK; +} + /* *--------------------------------------------------------------------------- * @@ -2873,7 +3070,6 @@ DupStringInternalRep( String *srcStringPtr = GET_STRING(srcPtr); String *copyStringPtr = NULL; -#if COMPAT==0 if (srcStringPtr->numChars == -1) { /* * The String struct in the source value holds zero useful data. Don't @@ -2916,41 +3112,6 @@ DupStringInternalRep( */ copyStringPtr->allocated = copyPtr->bytes ? copyPtr->length : 0; -#else /* COMPAT!=0 */ - /* - * If the src obj is a string of 1-byte Utf chars, then copy the string - * rep of the source object and create an "empty" Unicode internal rep for - * the new object. Otherwise, copy Unicode internal rep, and invalidate - * the string rep of the new object. - */ - - if (srcStringPtr->hasUnicode && srcStringPtr->numChars > 0) { - /* - * Copy the full allocation for the Unicode buffer. - */ - - copyStringPtr = stringAlloc(srcStringPtr->maxChars); - copyStringPtr->maxChars = srcStringPtr->maxChars; - memcpy(copyStringPtr->unicode, srcStringPtr->unicode, - srcStringPtr->numChars * sizeof(Tcl_UniChar)); - copyStringPtr->unicode[srcStringPtr->numChars] = 0; - copyStringPtr->allocated = 0; - } else { - copyStringPtr = stringAlloc(0); - copyStringPtr->unicode[0] = 0; - copyStringPtr->maxChars = 0; - - /* - * Tricky point: the string value was copied by generic object - * management code, so it doesn't contain any extra bytes that might - * exist in the source object. - */ - - copyStringPtr->allocated = copyPtr->length; - } - copyStringPtr->numChars = srcStringPtr->numChars; - copyStringPtr->hasUnicode = srcStringPtr->hasUnicode; -#endif /* COMPAT==0 */ SET_STRING(copyPtr, copyStringPtr); copyPtr->typePtr = &tclStringType; @@ -3056,7 +3217,7 @@ ExtendStringRepWithUnicode( */ int i, origLength, size = 0; - char *dst, buf[TCL_UTF_MAX]; + char *dst; String *stringPtr = GET_STRING(objPtr); if (numChars < 0) { @@ -3082,7 +3243,7 @@ ExtendStringRepWithUnicode( } for (i = 0; i < numChars && size >= 0; i++) { - size += Tcl_UniCharToUtf((int) unicode[i], buf); + size += TclUtfCount(unicode[i]); } if (size < 0) { Tcl_Panic("max size for a Tcl value (%d bytes) exceeded", INT_MAX); |