summaryrefslogtreecommitdiffstats
path: root/generic/tclEncoding.c
diff options
context:
space:
mode:
Diffstat (limited to 'generic/tclEncoding.c')
-rw-r--r--generic/tclEncoding.c199
1 files changed, 64 insertions, 135 deletions
diff --git a/generic/tclEncoding.c b/generic/tclEncoding.c
index fd0386c..cfc8caa 100644
--- a/generic/tclEncoding.c
+++ b/generic/tclEncoding.c
@@ -620,14 +620,14 @@ TclInitEncodingSubsystem(void)
* code to duplicate the structure of a table encoding here.
*/
- dataPtr = (TableEncodingData *)ckalloc(sizeof(TableEncodingData));
+ dataPtr = (TableEncodingData *)Tcl_Alloc(sizeof(TableEncodingData));
memset(dataPtr, 0, sizeof(TableEncodingData));
dataPtr->fallback = '?';
size = 256*(sizeof(unsigned short *) + sizeof(unsigned short));
- dataPtr->toUnicode = (unsigned short **)ckalloc(size);
+ dataPtr->toUnicode = (unsigned short **)Tcl_Alloc(size);
memset(dataPtr->toUnicode, 0, size);
- dataPtr->fromUnicode = (unsigned short **)ckalloc(size);
+ dataPtr->fromUnicode = (unsigned short **)Tcl_Alloc(size);
memset(dataPtr->fromUnicode, 0, size);
dataPtr->toUnicode[0] = (unsigned short *) (dataPtr->toUnicode + 256);
@@ -704,70 +704,6 @@ TclFinalizeEncodingSubsystem(void)
/*
*-------------------------------------------------------------------------
*
- * Tcl_GetDefaultEncodingDir --
- *
- * Legacy public interface to retrieve first directory in the encoding
- * searchPath.
- *
- * Results:
- * The directory pathname, as a string, or NULL for an empty encoding
- * search path.
- *
- * Side effects:
- * None.
- *
- *-------------------------------------------------------------------------
- */
-
-#if !defined(TCL_NO_DEPRECATED) && TCL_MAJOR_VERSION < 9
-const char *
-Tcl_GetDefaultEncodingDir(void)
-{
- int numDirs;
- Tcl_Obj *first, *searchPath = Tcl_GetEncodingSearchPath();
-
- TclListObjLength(NULL, searchPath, &numDirs);
- if (numDirs == 0) {
- return NULL;
- }
- Tcl_ListObjIndex(NULL, searchPath, 0, &first);
-
- return TclGetString(first);
-}
-
-/*
- *-------------------------------------------------------------------------
- *
- * Tcl_SetDefaultEncodingDir --
- *
- * Legacy public interface to set the first directory in the encoding
- * search path.
- *
- * Results:
- * None.
- *
- * Side effects:
- * Modifies the encoding search path.
- *
- *-------------------------------------------------------------------------
- */
-
-void
-Tcl_SetDefaultEncodingDir(
- const char *path)
-{
- Tcl_Obj *searchPath = Tcl_GetEncodingSearchPath();
- Tcl_Obj *directory = Tcl_NewStringObj(path, -1);
-
- searchPath = Tcl_DuplicateObj(searchPath);
- Tcl_ListObjReplace(NULL, searchPath, 0, 0, 1, &directory);
- Tcl_SetEncodingSearchPath(searchPath);
-}
-#endif
-
-/*
- *-------------------------------------------------------------------------
- *
* Tcl_GetEncoding --
*
* Given the name of a encoding, find the corresponding Tcl_Encoding
@@ -877,9 +813,9 @@ FreeEncoding(
Tcl_DeleteHashEntry(encodingPtr->hPtr);
}
if (encodingPtr->name) {
- ckfree(encodingPtr->name);
+ Tcl_Free(encodingPtr->name);
}
- ckfree(encodingPtr);
+ Tcl_Free(encodingPtr);
}
}
@@ -1067,7 +1003,7 @@ Tcl_CreateEncoding(
const Tcl_EncodingType *typePtr)
/* The encoding type. */
{
- Encoding *encodingPtr = (Encoding *)ckalloc(sizeof(Encoding));
+ Encoding *encodingPtr = (Encoding *)Tcl_Alloc(sizeof(Encoding));
encodingPtr->name = NULL;
encodingPtr->toUtfProc = typePtr->toUtfProc;
encodingPtr->fromUtfProc = typePtr->fromUtfProc;
@@ -1101,7 +1037,7 @@ Tcl_CreateEncoding(
replaceMe->hPtr = NULL;
}
- name = (char *)ckalloc(strlen(typePtr->encodingName) + 1);
+ name = (char *)Tcl_Alloc(strlen(typePtr->encodingName) + 1);
encodingPtr->name = strcpy(name, typePtr->encodingName);
encodingPtr->hPtr = hPtr;
Tcl_SetHashValue(hPtr, encodingPtr);
@@ -1132,12 +1068,13 @@ Tcl_CreateEncoding(
*-------------------------------------------------------------------------
*/
+#undef Tcl_ExternalToUtfDString
char *
Tcl_ExternalToUtfDString(
Tcl_Encoding encoding, /* The encoding for the source string, or NULL
* for the default system encoding. */
const char *src, /* Source string in specified encoding. */
- int srcLen, /* Source string length in bytes, or < 0 for
+ size_t srcLen, /* Source string length in bytes, or -1 for
* encoding-specific string length. */
Tcl_DString *dstPtr) /* Uninitialized or free DString in which the
* converted string is stored. */
@@ -1153,11 +1090,9 @@ Tcl_ExternalToUtfDString(
* Tcl_ExternalToUtfDStringEx --
*
* Convert a source buffer from the specified encoding into UTF-8.
-* The parameter flags controls the behavior, if any of the bytes in
+ * The parameter flags controls the behavior, if any of the bytes in
* the source buffer are invalid or cannot be represented in utf-8.
* Possible flags values:
- * TCL_ENCODING_STOPONERROR: don't replace invalid characters/bytes but
- * return the first error position (Default in Tcl 9.0).
* TCL_ENCODING_NOCOMPLAIN: replace invalid characters/bytes by a default
* fallback character. Always return -1 (Default in Tcl 8.7).
* TCL_ENCODING_MODIFIED: convert NULL bytes to \xC0\x80 in stead of 0x00.
@@ -1176,12 +1111,12 @@ Tcl_ExternalToUtfDString(
*-------------------------------------------------------------------------
*/
-int
+size_t
Tcl_ExternalToUtfDStringEx(
Tcl_Encoding encoding, /* The encoding for the source string, or NULL
* for the default system encoding. */
const char *src, /* Source string in specified encoding. */
- int srcLen, /* Source string length in bytes, or < 0 for
+ size_t srcLen, /* Source string length in bytes, or TCL_INDEX_NONE for
* encoding-specific string length. */
int flags, /* Conversion control flags. */
Tcl_DString *dstPtr) /* Uninitialized or free DString in which the
@@ -1190,7 +1125,8 @@ Tcl_ExternalToUtfDStringEx(
char *dst;
Tcl_EncodingState state;
const Encoding *encodingPtr;
- int dstLen, result, soFar, srcRead, dstWrote, dstChars;
+ int result, soFar, srcRead, dstWrote, dstChars;
+ size_t dstLen;
const char *srcStart = src;
Tcl_DStringInit(dstPtr);
@@ -1204,7 +1140,7 @@ Tcl_ExternalToUtfDStringEx(
if (src == NULL) {
srcLen = 0;
- } else if (srcLen < 0) {
+ } else if (srcLen == TCL_INDEX_NONE) {
srcLen = encodingPtr->lengthProc(src);
}
@@ -1221,7 +1157,7 @@ Tcl_ExternalToUtfDStringEx(
src += srcRead;
if (result != TCL_CONVERT_NOSPACE) {
Tcl_DStringSetLength(dstPtr, soFar);
- return (result == TCL_OK) ? TCL_INDEX_NONE : (int)(src - srcStart);
+ return (result == TCL_OK) ? TCL_INDEX_NONE : (size_t)(src - srcStart);
}
flags &= ~TCL_ENCODING_START;
srcLen -= srcRead;
@@ -1258,8 +1194,8 @@ Tcl_ExternalToUtf(
Tcl_Encoding encoding, /* The encoding for the source string, or NULL
* for the default system encoding. */
const char *src, /* Source string in specified encoding. */
- int srcLen, /* Source string length in bytes, or < 0 for
- * encoding-specific string length. */
+ size_t srcLen, /* Source string length in bytes, or -1
+ * for encoding-specific string length. */
int flags, /* Conversion control flags. */
Tcl_EncodingState *statePtr,/* Place for conversion routine to store state
* information used during a piecewise
@@ -1268,7 +1204,7 @@ Tcl_ExternalToUtf(
* routine under control of flags argument. */
char *dst, /* Output buffer in which converted string is
* stored. */
- int dstLen, /* The maximum length of output buffer in
+ size_t dstLen, /* The maximum length of output buffer in
* bytes. */
int *srcReadPtr, /* Filled with the number of bytes from the
* source string that were converted. This may
@@ -1296,7 +1232,7 @@ Tcl_ExternalToUtf(
if (src == NULL) {
srcLen = 0;
- } else if (srcLen < 0) {
+ } else if (srcLen == TCL_INDEX_NONE) {
srcLen = encodingPtr->lengthProc(src);
}
if (statePtr == NULL) {
@@ -1369,13 +1305,13 @@ Tcl_ExternalToUtf(
*
*-------------------------------------------------------------------------
*/
-
+#undef Tcl_UtfToExternalDString
char *
Tcl_UtfToExternalDString(
Tcl_Encoding encoding, /* The encoding for the converted string, or
* NULL for the default system encoding. */
const char *src, /* Source string in UTF-8. */
- int srcLen, /* Source string length in bytes, or < 0 for
+ size_t srcLen, /* Source string length in bytes, or -1 for
* strlen(). */
Tcl_DString *dstPtr) /* Uninitialized or free DString in which the
* converted string is stored. */
@@ -1395,8 +1331,6 @@ Tcl_UtfToExternalDString(
* the source buffer are invalid or cannot be represented in the
* target encoding.
* Possible flags values:
- * TCL_ENCODING_STOPONERROR: don't replace invalid characters/bytes but
- * return the first error position (Default in Tcl 9.0).
* TCL_ENCODING_NOCOMPLAIN: replace invalid characters/bytes by a default
* fallback character. Always return -1 (Default in Tcl 8.7).
* TCL_ENCODING_MODIFIED: convert NULL bytes to \xC0\x80 in stead of 0x00.
@@ -1415,12 +1349,12 @@ Tcl_UtfToExternalDString(
*-------------------------------------------------------------------------
*/
-int
+size_t
Tcl_UtfToExternalDStringEx(
Tcl_Encoding encoding, /* The encoding for the converted string, or
* NULL for the default system encoding. */
const char *src, /* Source string in UTF-8. */
- int srcLen, /* Source string length in bytes, or < 0 for
+ size_t srcLen, /* Source string length in bytes, or < 0 for
* strlen(). */
int flags, /* Conversion control flags. */
Tcl_DString *dstPtr) /* Uninitialized or free DString in which the
@@ -1429,8 +1363,9 @@ Tcl_UtfToExternalDStringEx(
char *dst;
Tcl_EncodingState state;
const Encoding *encodingPtr;
- int dstLen, result, soFar, srcRead, dstWrote, dstChars;
+ int result, soFar, srcRead, dstWrote, dstChars;
const char *srcStart = src;
+ size_t dstLen;
Tcl_DStringInit(dstPtr);
dst = Tcl_DStringValue(dstPtr);
@@ -1443,7 +1378,7 @@ Tcl_UtfToExternalDStringEx(
if (src == NULL) {
srcLen = 0;
- } else if (srcLen < 0) {
+ } else if (srcLen == TCL_INDEX_NONE) {
srcLen = strlen(src);
}
flags |= TCL_ENCODING_START | TCL_ENCODING_END;
@@ -1459,7 +1394,7 @@ Tcl_UtfToExternalDStringEx(
while (i >= soFar) {
Tcl_DStringSetLength(dstPtr, i--);
}
- return (result == TCL_OK) ? TCL_INDEX_NONE : (int)(src - srcStart);
+ return (result == TCL_OK) ? TCL_INDEX_NONE : (size_t)(src - srcStart);
}
flags &= ~TCL_ENCODING_START;
@@ -1497,8 +1432,8 @@ Tcl_UtfToExternal(
Tcl_Encoding encoding, /* The encoding for the converted string, or
* NULL for the default system encoding. */
const char *src, /* Source string in UTF-8. */
- int srcLen, /* Source string length in bytes, or < 0 for
- * strlen(). */
+ size_t srcLen, /* Source string length in bytes, or -1
+ * for strlen(). */
int flags, /* Conversion control flags. */
Tcl_EncodingState *statePtr,/* Place for conversion routine to store state
* information used during a piecewise
@@ -1507,7 +1442,7 @@ Tcl_UtfToExternal(
* routine under control of flags argument. */
char *dst, /* Output buffer in which converted string
* is stored. */
- int dstLen, /* The maximum length of output buffer in
+ size_t dstLen, /* The maximum length of output buffer in
* bytes. */
int *srcReadPtr, /* Filled with the number of bytes from the
* source string that were converted. This may
@@ -1532,7 +1467,7 @@ Tcl_UtfToExternal(
if (src == NULL) {
srcLen = 0;
- } else if (srcLen < 0) {
+ } else if (srcLen == TCL_INDEX_NONE) {
srcLen = strlen(src);
}
if (statePtr == NULL) {
@@ -1772,7 +1707,7 @@ LoadEncodingFile(
"invalid encoding file \"%s\"", name));
Tcl_SetErrorCode(interp, "TCL", "LOOKUP", "ENCODING", name, NULL);
}
- Tcl_Close(NULL, chan);
+ Tcl_CloseEx(NULL, chan, 0);
return encoding;
}
@@ -1862,7 +1797,7 @@ LoadTableEncoding(
#undef PAGESIZE
#define PAGESIZE (256 * sizeof(unsigned short))
- dataPtr = (TableEncodingData *)ckalloc(sizeof(TableEncodingData));
+ dataPtr = (TableEncodingData *)Tcl_Alloc(sizeof(TableEncodingData));
memset(dataPtr, 0, sizeof(TableEncodingData));
dataPtr->fallback = fallback;
@@ -1874,7 +1809,7 @@ LoadTableEncoding(
*/
size = 256 * sizeof(unsigned short *) + numPages * PAGESIZE;
- dataPtr->toUnicode = (unsigned short **)ckalloc(size);
+ dataPtr->toUnicode = (unsigned short **)Tcl_Alloc(size);
memset(dataPtr->toUnicode, 0, size);
pageMemPtr = (unsigned short *) (dataPtr->toUnicode + 256);
@@ -1883,7 +1818,7 @@ LoadTableEncoding(
for (i = 0; i < numPages; i++) {
int ch;
const char *p;
- int expected = 3 + 16 * (16 * 4 + 1);
+ size_t expected = 3 + 16 * (16 * 4 + 1);
if (Tcl_ReadChars(chan, objPtr, expected, 0) != expected) {
return NULL;
@@ -1935,7 +1870,7 @@ LoadTableEncoding(
}
}
size = 256 * sizeof(unsigned short *) + numPages * PAGESIZE;
- dataPtr->fromUnicode = (unsigned short **)ckalloc(size);
+ dataPtr->fromUnicode = (unsigned short **)Tcl_Alloc(size);
memset(dataPtr->fromUnicode, 0, size);
pageMemPtr = (unsigned short *) (dataPtr->fromUnicode + 256);
@@ -2031,7 +1966,7 @@ LoadTableEncoding(
*/
for (TclDStringClear(&lineString);
- (len = Tcl_Gets(chan, &lineString)) >= 0;
+ (len = Tcl_Gets(chan, &lineString)) != -1;
TclDStringClear(&lineString)) {
const unsigned char *p;
int to, from;
@@ -2125,7 +2060,7 @@ LoadEscapeEncoding(
Tcl_DString lineString;
Tcl_DStringInit(&lineString);
- if (Tcl_Gets(chan, &lineString) < 0) {
+ if (Tcl_Gets(chan, &lineString) == TCL_IO_FAILURE) {
break;
}
line = Tcl_DStringValue(&lineString);
@@ -2167,13 +2102,13 @@ LoadEscapeEncoding(
Tcl_DStringAppend(&escapeData, (char *) &est, sizeof(est));
}
}
- ckfree(argv);
+ Tcl_Free((void *)argv);
Tcl_DStringFree(&lineString);
}
size = offsetof(EscapeEncodingData, subTables)
+ Tcl_DStringLength(&escapeData);
- dataPtr = (EscapeEncodingData *)ckalloc(size);
+ dataPtr = (EscapeEncodingData *)Tcl_Alloc(size);
dataPtr->initLen = strlen(init);
memcpy(dataPtr->init, init, dataPtr->initLen + 1);
dataPtr->finalLen = strlen(final);
@@ -2229,7 +2164,7 @@ LoadEscapeEncoding(
static int
BinaryProc(
- TCL_UNUSED(ClientData),
+ TCL_UNUSED(void *),
const char *src, /* Source string (unknown encoding). */
int srcLen, /* Source string length in bytes. */
int flags, /* Conversion control flags. */
@@ -2287,12 +2222,6 @@ BinaryProc(
*-------------------------------------------------------------------------
*/
-#if TCL_MAJOR_VERSION > 8 || defined(TCL_NO_DEPRECATED)
-# define STOPONERROR !(flags & TCL_ENCODING_NOCOMPLAIN)
-#else
-# define STOPONERROR (flags & TCL_ENCODING_STOPONERROR)
-#endif
-
static int
UtfToUtfProc(
ClientData clientData, /* additional flags, e.g. TCL_ENCODING_MODIFIED */
@@ -2375,7 +2304,7 @@ UtfToUtfProc(
*/
if (flags & TCL_ENCODING_MODIFIED) {
- if ((STOPONERROR) && (flags & TCL_ENCODING_CHAR_LIMIT)) {
+ if (!(flags & TCL_ENCODING_NOCOMPLAIN) && (flags & TCL_ENCODING_CHAR_LIMIT)) {
result = TCL_CONVERT_MULTIBYTE;
break;
}
@@ -2390,7 +2319,7 @@ UtfToUtfProc(
int low;
const char *saveSrc = src;
size_t len = TclUtfToUCS4(src, &ch);
- if ((len < 2) && (ch != 0) && STOPONERROR
+ if ((len < 2) && (ch != 0) && !(flags & TCL_ENCODING_NOCOMPLAIN)
&& (flags & TCL_ENCODING_MODIFIED)) {
result = TCL_CONVERT_SYNTAX;
break;
@@ -2416,7 +2345,7 @@ UtfToUtfProc(
if (((low & ~0x3FF) != 0xDC00) || (ch & 0x400)) {
- if (STOPONERROR) {
+ if (!(flags & TCL_ENCODING_NOCOMPLAIN)) {
result = TCL_CONVERT_UNKNOWN;
src = saveSrc;
break;
@@ -2431,7 +2360,7 @@ UtfToUtfProc(
dst += Tcl_UniCharToUtf(ch, dst);
ch = low;
} else if (!Tcl_UniCharIsUnicode(ch)) {
- if (STOPONERROR) {
+ if (!(flags & TCL_ENCODING_NOCOMPLAIN)) {
result = TCL_CONVERT_UNKNOWN;
src = saveSrc;
break;
@@ -2617,7 +2546,7 @@ UtfToUtf32Proc(
}
len = TclUtfToUCS4(src, &ch);
if (!Tcl_UniCharIsUnicode(ch)) {
- if (STOPONERROR) {
+ if (!(flags & TCL_ENCODING_NOCOMPLAIN)) {
result = TCL_CONVERT_UNKNOWN;
break;
}
@@ -2820,7 +2749,7 @@ UtfToUtf16Proc(
}
len = TclUtfToUCS4(src, &ch);
if (!Tcl_UniCharIsUnicode(ch)) {
- if (STOPONERROR) {
+ if (!(flags & TCL_ENCODING_NOCOMPLAIN)) {
result = TCL_CONVERT_UNKNOWN;
break;
}
@@ -3040,7 +2969,7 @@ TableToUtfProc(
ch = pageZero[byte];
}
if ((ch == 0) && (byte != 0)) {
- if (STOPONERROR) {
+ if (!(flags & TCL_ENCODING_NOCOMPLAIN)) {
result = TCL_CONVERT_SYNTAX;
break;
}
@@ -3156,7 +3085,7 @@ TableFromUtfProc(
word = fromUnicode[(ch >> 8)][ch & 0xFF];
if ((word == 0) && (ch != 0)) {
- if ((STOPONERROR) && (flags & TCL_ENCODING_CHAR_LIMIT)) {
+ if (!(flags & TCL_ENCODING_NOCOMPLAIN) && (flags & TCL_ENCODING_CHAR_LIMIT)) {
result = TCL_CONVERT_UNKNOWN;
break;
}
@@ -3205,7 +3134,7 @@ TableFromUtfProc(
static int
Iso88591ToUtfProc(
- TCL_UNUSED(ClientData),
+ TCL_UNUSED(void *),
const char *src, /* Source string in specified encoding. */
int srcLen, /* Source string length in bytes. */
int flags, /* Conversion control flags. */
@@ -3285,7 +3214,7 @@ Iso88591ToUtfProc(
static int
Iso88591FromUtfProc(
- TCL_UNUSED(ClientData),
+ TCL_UNUSED(void *),
const char *src, /* Source string in UTF-8. */
int srcLen, /* Source string length in bytes. */
int flags, /* Conversion control flags. */
@@ -3344,7 +3273,7 @@ Iso88591FromUtfProc(
|| ((ch >= 0xD800) && (len < 3))
#endif
) {
- if (STOPONERROR) {
+ if (!(flags & TCL_ENCODING_NOCOMPLAIN)) {
result = TCL_CONVERT_UNKNOWN;
break;
}
@@ -3402,11 +3331,11 @@ TableFreeProc(
* Make sure we aren't freeing twice on shutdown. [Bug 219314]
*/
- ckfree(dataPtr->toUnicode);
+ Tcl_Free(dataPtr->toUnicode);
dataPtr->toUnicode = NULL;
- ckfree(dataPtr->fromUnicode);
+ Tcl_Free(dataPtr->fromUnicode);
dataPtr->fromUnicode = NULL;
- ckfree(dataPtr);
+ Tcl_Free(dataPtr);
}
/*
@@ -3571,7 +3500,7 @@ EscapeToUtfProc(
if ((checked == dataPtr->numSubTables + 2)
|| (flags & TCL_ENCODING_END)) {
- if (!STOPONERROR) {
+ if (!!(flags & TCL_ENCODING_NOCOMPLAIN)) {
/*
* Skip the unknown escape sequence.
*/
@@ -3746,7 +3675,7 @@ EscapeFromUtfProc(
if (word == 0) {
state = oldState;
- if (STOPONERROR) {
+ if (!(flags & TCL_ENCODING_NOCOMPLAIN)) {
result = TCL_CONVERT_UNKNOWN;
break;
}
@@ -3779,8 +3708,7 @@ EscapeFromUtfProc(
result = TCL_CONVERT_NOSPACE;
break;
}
- memcpy(dst, subTablePtr->sequence,
- subTablePtr->sequenceLen);
+ memcpy(dst, subTablePtr->sequence, subTablePtr->sequenceLen);
dst += subTablePtr->sequenceLen;
}
}
@@ -3884,7 +3812,7 @@ EscapeFreeProc(
subTablePtr++;
}
}
- ckfree(dataPtr);
+ Tcl_Free(dataPtr);
}
/*
@@ -3998,11 +3926,12 @@ unilen4(
static void
InitializeEncodingSearchPath(
char **valuePtr,
- unsigned int *lengthPtr,
+ size_t *lengthPtr,
Tcl_Encoding *encodingPtr)
{
const char *bytes;
- int i, numDirs, numBytes;
+ int i, numDirs;
+ size_t numBytes;
Tcl_Obj *libPathObj, *encodingObj, *searchPathObj;
TclNewLiteralStringObj(encodingObj, "encoding");
@@ -4035,7 +3964,7 @@ InitializeEncodingSearchPath(
bytes = Tcl_GetStringFromObj(searchPathObj, &numBytes);
*lengthPtr = numBytes;
- *valuePtr = (char *)ckalloc(numBytes + 1);
+ *valuePtr = (char *)Tcl_Alloc(numBytes + 1);
memcpy(*valuePtr, bytes, numBytes + 1);
Tcl_DecrRefCount(searchPathObj);
}
generated by cgit v0.12 at 2024-10-22 07:27:11 (GMT)