[Rpm-maint] [Suse patch] Convert changelog and i18n headers to current locale
Panu Matilainen
pmatilai at redhat.com
Mon Jun 4 12:56:09 UTC 2007
I can kinda see the point behind this, but it's just scary. Wouldn't
this be a job for iconv or something instead?
Further, as pointed out here by JBJ, it doesn't really solve the problem
of potentially having multiple encodings present in the spec:
https://lists.dulug.duke.edu/pipermail/rpm-devel/2006-March/000931.html
Character encoding + conversion business makes me want to stick my head
into sand (and stay there)...
- Panu -
---
Convert changelog and i18n header elements to current locale.
[#43347], rh#140050
Already in rpm-4.4.7
--- ./lib/formats.c.orig 2005-01-26 04:46:54.000000000 +0000
+++ ./lib/formats.c 2006-03-17 15:27:06.000000000 +0000
@@ -2,6 +2,7 @@
* \file lib/formats.c
*/
+#include <wchar.h>
#include "system.h"
#include "rpmio_internal.h"
#include <rpmlib.h>
@@ -18,6 +19,114 @@
/*@access pgpDig @*/
/*@access pgpDigParams @*/
+static const char * strtolocale(const char *str)
+{
+ wchar_t *wstr, *wp;
+ const unsigned char *cp;
+ char *cc;
+ int state = 0;
+ int c;
+ int ccl, cca, mb_cur_max;
+ size_t l;
+ mbstate_t ps;
+ int strisutf8 = 1;
+ int locisutf8 = 1;
+
+ if (!str)
+ return 0;
+ if (!*str)
+ return str;
+ wstr = (wchar_t *)xmalloc((strlen(str) + 1) * sizeof(*wstr));
+ wp = wstr;
+ cp = (const unsigned char *)str;
+ while ((c = *cp++) != 0) {
+ if (state) {
+ if ((c & 0xc0) != 0x80) {
+ /* encoding error */
+ break;
+ }
+ c = (c & 0x3f) | (state << 6);
+ if (!(state & 0x40000000)) {
+ /* check for overlong sequences */
+ if ((c & 0x820823e0) == 0x80000000)
+ c = 0xfdffffff;
+ else if ((c & 0x020821f0) == 0x02000000)
+ c = 0xfff7ffff;
+ else if ((c & 0x000820f8) == 0x00080000)
+ c = 0xffffd000;
+ else if ((c & 0x0000207c) == 0x00002000)
+ c = 0xffffff70;
+ }
+ } else {
+ /* new sequence */
+ if (c >= 0xfe)
+ c = 0xfffd;
+ else if (c >= 0xfc)
+ c = (c & 0x01) | 0xbffffffc; /* 5 bytes to follow */
+ else if (c >= 0xf8)
+ c = (c & 0x03) | 0xbfffff00; /* 4 */
+ else if (c >= 0xf0)
+ c = (c & 0x07) | 0xbfffc000; /* 3 */
+ else if (c >= 0xe0)
+ c = (c & 0x0f) | 0xbff00000; /* 2 */
+ else if (c >= 0xc2)
+ c = (c & 0x1f) | 0xfc000000; /* 1 */
+ else if (c >= 0xc0)
+ c = 0xfdffffff; /* overlong */
+ else if (c >= 0x80)
+ c = 0xfffd;
+ }
+ state = (c & 0x80000000) ? c : 0;
+ if (state)
+ continue;
+ *wp++ = (wchar_t)c;
+ }
+ if (state) {
+ /* encoding error, assume latin1 */
+ strisutf8 = 0;
+ cp = (const unsigned char *)str;
+ wp = wstr;
+ while ((c = *cp++) != 0) {
+ *wp++ = (wchar_t)c;
+ }
+ }
+ *wp = 0;
+ mb_cur_max = MB_CUR_MAX;
+ memset(&ps, 0, sizeof(ps));
+ cc = xmalloc(mb_cur_max);
+ /* test locale encoding */
+ if (wcrtomb(cc, 0x20ac, &ps) != 3 || memcmp(cc, "\342\202\254", 3))
+ locisutf8 = 0;
+ if (locisutf8 == strisutf8) {
+ wstr = _free(wstr);
+ return str;
+ }
+ str = _free(str);
+ memset(&ps, 0, sizeof(ps));
+ ccl = cca = 0;
+ for (wp = wstr; ; wp++) {
+ l = wcrtomb(cc + ccl, *wp, &ps);
+ if (*wp == 0)
+ break;
+ if (l == (size_t)-1) {
+ if (*wp < (wchar_t)256 && mbsinit(&ps)) {
+ cc[ccl] = *wp;
+ l = 1;
+ } else
+ l = wcrtomb(cc + ccl, (wchar_t)'?', &ps);
+ }
+ if (l == 0 || l == (size_t)-1)
+ continue;
+ ccl += l;
+ if (ccl > cca) {
+ cca = ccl + 16;
+ cc = xrealloc(cc, cca + mb_cur_max);
+ }
+ }
+ wstr = _free(wstr);
+ return (const char *)cc;
+}
+
/**
* Identify type of trigger.
* @param type tag type
@@ -1077,6 +1220,7 @@ static int i18nTag(Header h, int_32 tag,
if (rc && (*data) != NULL) {
*data = xstrdup(*data);
+ *data = strtolocale(*data);
*freeData = 1;
return 0;
}
@@ -1088,6 +1232,56 @@ static int i18nTag(Header h, int_32 tag,
}
/**
+ * Retrieve text and convert to locale.
+ */
+static int localeTag(Header h, int_32 tag, /*@out@*/ rpmTagType * type,
+ /*@out@*/ const void ** data, /*@out@*/ int_32 * count,
+ /*@out@*/ int * freeData)
+{
+ HGE_t hge = (HGE_t)headerGetEntryMinMemory;
+ rpmTagType t;
+ char **d, **d2, *dp;
+ int rc, i, l;
+
+ rc = hge(h, tag, &t, (void **)&d, count);
+ if (!rc || d == NULL || *count == 0) {
+ *freeData = 0;
+ *data = NULL;
+ *count = 0;
+ return 1;
+ }
+ if (type)
+ *type = t;
+ if (t == RPM_STRING_TYPE) {
+ d = (char **)xstrdup((char *)d);
+ d = (char **)strtolocale((char *)d);
+ *freeData = 1;
+ } else if (t == RPM_STRING_ARRAY_TYPE) {
+ l = 0;
+ for (i = 0; i < *count; i++) {
+ d[i] = xstrdup(d[i]);
+ d[i] = (char *)strtolocale(d[i]);
+ l += strlen(d[i]) + 1;
+ }
+ d2 = xmalloc(*count * sizeof(char *) + l);
+ dp = (char *)(d2 + *count);
+ for (i = 0; i < *count; i++) {
+ d2[i] = dp;
+ strcpy(dp, d[i]);
+ dp += strlen(dp) + 1;
+ d[i] = _free(d[i]);
+ }
+ d = _free(d);
+ d = d2;
+ *freeData = 1;
+ } else
+ *freeData = 0;
+ *data = (void **)d;
+ return 0;
+}
+
+
+/**
* Retrieve summary text.
* @param h header
* @retval *type tag type
@@ -1127,6 +1321,20 @@ static int descriptionTag(Header h, /*@o
return i18nTag(h, RPMTAG_DESCRIPTION, type, data, count, freeData);
}
+static int changelognameTag(Header h, /*@out@*/ rpmTagType * type,
+ /*@out@*/ const void ** data, /*@out@*/ int_32 * count,
+ /*@out@*/ int * freeData)
+{
+ return localeTag(h, RPMTAG_CHANGELOGNAME, type, data, count, freeData);
+}
+
+static int changelogtextTag(Header h, /*@out@*/ rpmTagType * type,
+ /*@out@*/ const void ** data, /*@out@*/ int_32 * count,
+ /*@out@*/ int * freeData)
+{
+ return localeTag(h, RPMTAG_CHANGELOGTEXT, type, data, count, freeData);
+}
+
/**
* Retrieve group text.
* @param h header
@@ -1152,6 +1360,8 @@ const struct headerSprintfExtension_s rp
{ HEADER_EXT_TAG, "RPMTAG_GROUP", { groupTag } },
{ HEADER_EXT_TAG, "RPMTAG_DESCRIPTION", { descriptionTag } },
{ HEADER_EXT_TAG, "RPMTAG_SUMMARY", { summaryTag } },
+ { HEADER_EXT_TAG, "RPMTAG_CHANGELOGNAME", { changelognameTag } },
+ { HEADER_EXT_TAG, "RPMTAG_CHANGELOGTEXT", { changelogtextTag } },
{ HEADER_EXT_TAG, "RPMTAG_FILECLASS", { fileclassTag } },
{ HEADER_EXT_TAG, "RPMTAG_FILECONTEXTS", { filecontextsTag } },
{ HEADER_EXT_TAG, "RPMTAG_FILENAMES", { filenamesTag } },
More information about the Rpm-maint
mailing list