summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorhobbs <hobbs>2001-11-20 09:24:55 (GMT)
committerhobbs <hobbs>2001-11-20 09:24:55 (GMT)
commitefc5cb86481e5b94396404ff21921ca044118de8 (patch)
tree824e45bb7d1e175f1821d1ecea11b3269ec69c63
parent1fd083c120730401f70839c5434d9658fab053ec (diff)
downloadtcl-efc5cb86481e5b94396404ff21921ca044118de8.zip
tcl-efc5cb86481e5b94396404ff21921ca044118de8.tar.gz
tcl-efc5cb86481e5b94396404ff21921ca044118de8.tar.bz2
* unix/tclUnixInit.c (TclpSetInitialEncodings): update of patch
from 2001-11-16 that uses the old Tcl encoding check mechanism as a fallback to the original. Also added a TCL_DEFAULT_ENCODING #define (defaults to iso8859-1). Tcl will first try setlocale and nl_langinfo, and if that fails, guess based on certain LANG|LC_* env vars. [Patch #418645]
-rw-r--r--ChangeLog14
-rw-r--r--unix/tclUnixInit.c250
2 files changed, 154 insertions, 110 deletions
diff --git a/ChangeLog b/ChangeLog
index 6f97b02..93cb8ef 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,12 @@
+2001-11-20 Jeff Hobbs <jeffh@ActiveState.com>
+
+ * unix/tclUnixInit.c (TclpSetInitialEncodings): update of patch
+ from 2001-11-16 that uses the old Tcl encoding check mechanism as
+ a fallback to the original. Also added a TCL_DEFAULT_ENCODING
+ #define (defaults to iso8859-1). Tcl will first try setlocale and
+ nl_langinfo, and if that fails, guess based on certain LANG|LC_*
+ env vars. [Patch #418645]
+
2001-11-19 David Gravereaux <davygrvy@pobox.com>
* win/buildall.vc.bat: Added useful comments.
@@ -72,7 +81,7 @@
* generic/tclListObj.c: removed a C++-style comment that
was inadvertently left in the source code.
-
+
2001-11-16 Jeff Hobbs <jeffh@ActiveState.com>
* tests/interp.test:
@@ -97,7 +106,6 @@
code for tclCmdNameType objects to tclObj.c (from tclExecute.c).
This code has nothing to do with bytecodes.
-
2001-11-16 Miguel Sofer <msofer@users.sourceforge.net>
* generic/tclBasic.c:
@@ -107,7 +115,7 @@
* generic/tclProc.c:
* tests/stack.test: consolidation of duplicated code (in
TclExecuteByteCode and EvalObjv); renaming of EvalObjv to
- TclEvalObjv i as itisnot static anymore; restored consistency of
+ TclEvalObjv as it is not static anymore; restored consistency of
level counts between compiled and directly evaled code.
[Bug 480896]
diff --git a/unix/tclUnixInit.c b/unix/tclUnixInit.c
index 3651c50..32ee2ba 100644
--- a/unix/tclUnixInit.c
+++ b/unix/tclUnixInit.c
@@ -7,7 +7,7 @@
* Copyright (c) 1999 by Scriptics Corporation.
* All rights reserved.
*
- * RCS: @(#) $Id: tclUnixInit.c,v 1.27 2001/11/17 00:10:40 hobbs Exp $
+ * RCS: @(#) $Id: tclUnixInit.c,v 1.28 2001/11/20 09:24:55 hobbs Exp $
*/
#include "tclInt.h"
@@ -38,6 +38,16 @@ static Tcl_Encoding binaryEncoding = NULL;
static int libraryPathEncodingFixed = 0;
/*
+ * Tcl tries to use standard and homebrew methods to guess the right
+ * encoding on the platform. However, there is always a final fallback,
+ * and this value is it. Make sure it is a real Tcl encoding.
+ */
+
+#ifndef TCL_DEFAULT_ENCODING
+#define TCL_DEFAULT_ENCODING "iso8859-1"
+#endif
+
+/*
* Default directory in which to look for Tcl library scripts. The
* symbol is defined by Makefile.
*/
@@ -86,8 +96,9 @@ static CONST LocaleTable localeTable[] = {
{"tis620", "tis-620"},
{"turkish8", "cp857"},
{"utf8", "utf-8"},
-#endif
-#else
+#endif /* __hpux */
+#endif /* HAVE_LANGINFO */
+
{"ja_JP.SJIS", "shiftjis"},
{"ja_JP.EUC", "euc-jp"},
{"ja_JP.eucJP", "euc-jp"},
@@ -122,7 +133,7 @@ static CONST LocaleTable localeTable[] = {
{"ru_SU", "iso8859-5"},
{"zh", "cp936"},
-#endif /* !HAVE_LANGINFO */
+
{NULL, NULL}
};
@@ -421,8 +432,8 @@ void
TclpSetInitialEncodings()
{
if (libraryPathEncodingFixed == 0) {
- CONST char *encoding;
- int i;
+ CONST char *encoding = NULL;
+ int i, setSysEncCode = TCL_ERROR;
Tcl_Obj *pathPtr;
/*
@@ -431,126 +442,151 @@ TclpSetInitialEncodings()
* but this does not work on some systems (e.g. Linux/i386 RH 5.0).
*/
#ifdef HAVE_LANGINFO
- Tcl_DString ds;
- int code;
-
- setlocale(LC_CTYPE,"");
- Tcl_DStringInit(&ds);
- Tcl_DStringAppend(&ds, nl_langinfo(CODESET), -1);
- encoding = Tcl_DStringValue(&ds);
-
- Tcl_UtfToLower(Tcl_DStringValue(&ds));
- if (encoding[0] == 'i' && encoding[1] == 's' && encoding[2] == 'o'
- && encoding[3] == '-') {
- char *p, *q;
- /* need to strip extra - from ISO-* encoding */
- for(p = Tcl_DStringValue(&ds)+3, q = Tcl_DStringValue(&ds)+4; *p;
- *p++ = *q++);
- } else if (encoding[0] == 'i' && encoding[1] == 'b'
- && encoding[2] == 'm' && encoding[3] >= '0'
- && encoding[3] <= '9') {
- char *p, *q;
- /* if langinfo reports IBMxxx we should use cpXXX*/
- p = Tcl_DStringValue(&ds);
- *p++='c'; *p++='p';
- for(q=p+1;*p;*p++= *q++);
- } else if (!strcmp(encoding, "ansi_x3.4-1968") || *encoding == '\0') {
+ if (setlocale(LC_CTYPE, "") != NULL) {
+ Tcl_DString ds;
+
/*
- * Use iso8859-1 for empty or 'ansi_x3.4-1968' encoding.
+ * Use a DString so we can overwrite it in name compatability
+ * checks below.
*/
- Tcl_DStringSetLength(&ds, 0);
- Tcl_DStringAppend(&ds, "iso8859-1", -1);
- }
- code = Tcl_SetSystemEncoding(NULL, encoding);
- if (code != TCL_OK) {
- for (i = 0; localeTable[i].lang != NULL; i++) {
- if (strcmp(localeTable[i].lang, encoding) == 0) {
- encoding = localeTable[i].encoding;
- code = Tcl_SetSystemEncoding(NULL, encoding);
- break;
- }
+
+ Tcl_DStringInit(&ds);
+ Tcl_DStringAppend(&ds, nl_langinfo(CODESET), -1);
+ encoding = Tcl_DStringValue(&ds);
+
+ Tcl_UtfToLower(Tcl_DStringValue(&ds));
+#ifdef HAVE_LANGINFO_DEBUG
+ fprintf(stderr, "encoding '%s'", encoding);
+#endif
+ if (encoding[0] == 'i' && encoding[1] == 's' && encoding[2] == 'o'
+ && encoding[3] == '-') {
+ char *p, *q;
+ /* need to strip '-' from iso-* encoding */
+ for(p = Tcl_DStringValue(&ds)+3, q = Tcl_DStringValue(&ds)+4;
+ *p; *p++ = *q++);
+ } else if (encoding[0] == 'i' && encoding[1] == 'b'
+ && encoding[2] == 'm' && encoding[3] >= '0'
+ && encoding[3] <= '9') {
+ char *p, *q;
+ /* if langinfo reports "ibm*" we should use "cp*" */
+ p = Tcl_DStringValue(&ds);
+ *p++ = 'c'; *p++ = 'p';
+ for(q = p+1; *p ; *p++ = *q++);
+ } else if ((*encoding == '\0')
+ || !strcmp(encoding, "ansi_x3.4-1968")) {
+ /* Use iso8859-1 for empty or 'ansi_x3.4-1968' encoding */
+ encoding = "iso8859-1";
}
- }
- if (code != TCL_OK) {
- Tcl_Channel errChannel = Tcl_GetStdChannel(TCL_STDERR);
- if (errChannel) {
- char msg[100];
- sprintf(msg, "Couldn't set encoding to '%.50s'\n", encoding);
- Tcl_Write(errChannel, msg, -1);
+#ifdef HAVE_LANGINFO_DEBUG
+ fprintf(stderr, " ?%s?", encoding);
+#endif
+ setSysEncCode = Tcl_SetSystemEncoding(NULL, encoding);
+ if (setSysEncCode != TCL_OK) {
+ /*
+ * If this doesn't return TCL_OK, the encoding returned by
+ * nl_langinfo or as we translated it wasn't accepted. Do
+ * this fallback check. If this fails, we will enter the
+ * old fallback below.
+ */
+
+ for (i = 0; localeTable[i].lang != NULL; i++) {
+ if (strcmp(localeTable[i].lang, encoding) == 0) {
+ setSysEncCode = Tcl_SetSystemEncoding(NULL,
+ localeTable[i].encoding);
+ break;
+ }
+ }
}
- encoding = "iso8859-1";
- Tcl_SetSystemEncoding(NULL, encoding);
- }
- Tcl_DStringFree(&ds);
-#else
- char *langEnv;
- langEnv = getenv("LC_ALL");
-
- if (langEnv == NULL || langEnv[0] == '\0') {
- langEnv = getenv("LC_CTYPE");
- }
- if (langEnv == NULL || langEnv[0] == '\0') {
- langEnv = getenv("LANG");
+#ifdef HAVE_LANGINFO_DEBUG
+ fprintf(stderr, " => '%s'\n", encoding);
+#endif
+ Tcl_DStringFree(&ds);
}
- if (langEnv == NULL || langEnv[0] == '\0') {
- langEnv = NULL;
+#ifdef HAVE_LANGINFO_DEBUG
+ else {
+ fprintf(stderr, "setlocale returned NULL\n");
}
+#endif
+#endif /* HAVE_LANGINFO */
- encoding = NULL;
- if (langEnv != NULL) {
- for (i = 0; localeTable[i].lang != NULL; i++) {
- if (strcmp(localeTable[i].lang, langEnv) == 0) {
- encoding = localeTable[i].encoding;
- break;
- }
- }
+ if (setSysEncCode != TCL_OK) {
/*
- * There was no mapping in the locale table. If there is an
- * encoding subfield, we can try to guess from that.
+ * Classic fallback check. This tries a homebrew algorithm to
+ * determine what encoding should be used based on env vars.
*/
+ char *langEnv = getenv("LC_ALL");
+ encoding = NULL;
- if (encoding == NULL) {
- char *p;
- for (p = langEnv; *p != '\0'; p++) {
- if (*p == '.') {
- p++;
+ if (langEnv == NULL || langEnv[0] == '\0') {
+ langEnv = getenv("LC_CTYPE");
+ }
+ if (langEnv == NULL || langEnv[0] == '\0') {
+ langEnv = getenv("LANG");
+ }
+ if (langEnv == NULL || langEnv[0] == '\0') {
+ langEnv = NULL;
+ }
+
+ if (langEnv != NULL) {
+ for (i = 0; localeTable[i].lang != NULL; i++) {
+ if (strcmp(localeTable[i].lang, langEnv) == 0) {
+ encoding = localeTable[i].encoding;
break;
}
}
- if (*p != '\0') {
- Tcl_DString ds;
- Tcl_DStringInit(&ds);
- Tcl_DStringAppend(&ds, p, -1);
-
- encoding = Tcl_DStringValue(&ds);
- Tcl_UtfToLower(Tcl_DStringValue(&ds));
- if (Tcl_SetSystemEncoding(NULL, encoding) == TCL_OK) {
+ /*
+ * There was no mapping in the locale table. If there is an
+ * encoding subfield, we can try to guess from that.
+ */
+
+ if (encoding == NULL) {
+ char *p;
+ for (p = langEnv; *p != '\0'; p++) {
+ if (*p == '.') {
+ p++;
+ break;
+ }
+ }
+ if (*p != '\0') {
+ Tcl_DString ds;
+ Tcl_DStringInit(&ds);
+ Tcl_DStringAppend(&ds, p, -1);
+
+ encoding = Tcl_DStringValue(&ds);
+ Tcl_UtfToLower(Tcl_DStringValue(&ds));
+ setSysEncCode = Tcl_SetSystemEncoding(NULL, encoding);
+ if (setSysEncCode != TCL_OK) {
+ encoding = NULL;
+ }
Tcl_DStringFree(&ds);
- goto resetPath;
}
- Tcl_DStringFree(&ds);
- encoding = NULL;
}
+#ifdef HAVE_LANGINFO_DEBUG
+ fprintf(stderr, "encoding fallback check '%s' => '%s'\n",
+ langEnv, encoding);
+#endif
}
- }
- if (encoding == NULL) {
- encoding = "iso8859-1";
- }
-
- Tcl_SetSystemEncoding(NULL, encoding);
+ if (setSysEncCode != TCL_OK) {
+ if (encoding == NULL) {
+ encoding = TCL_DEFAULT_ENCODING;
+ }
- resetPath:
- /*
- * Initialize the C library's locale subsystem. This is required
- * for input methods to work properly on X11. We only do this for
- * LC_CTYPE because that's the necessary one, and we don't want to
- * affect LC_TIME here. The side effect of setting the default locale
- * should be to load any locale specific modules that are needed by X.
- * [BUG: 5422 3345 4236 2522 2521].
- */
+ Tcl_SetSystemEncoding(NULL, encoding);
+ }
- setlocale(LC_CTYPE, "");
-#endif /* HAVE_LANGINFO */
+ /*
+ * Initialize the C library's locale subsystem. This is required
+ * for input methods to work properly on X11. We only do this for
+ * LC_CTYPE because that's the necessary one, and we don't want to
+ * affect LC_TIME here. The side effect of setting the default
+ * locale should be to load any locale specific modules that are
+ * needed by X. [BUG: 5422 3345 4236 2522 2521].
+ * In HAVE_LANGINFO, this call is already done above.
+ */
+#ifndef HAVE_LANGINFO
+ setlocale(LC_CTYPE, "");
+#endif
+ }
/*
* In case the initial locale is not "C", ensure that the numeric