Attachment 147018 Details for Bug 225576 – support for non-latin1 filenames in archive

[patch] support for non-latin1 filenames in archive

unzip-5.52-iconv.patch (text/plain), 5.92 KB, created by Dmitry Butskoy on 2007-01-31 13:47:01 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Dmitry Butskoy

Created: 2007-01-31 13:47:01 UTC

Size: 5.92 KB

patch

obsolete

>diff -Nrbu unzip-5.52/fileio.c unzip-5.52-OK/fileio.c
>--- unzip-5.52/fileio.c	2007-01-30 21:12:18.000000000 +0300
>+++ unzip-5.52-OK/fileio.c	2007-01-30 21:12:01.000000000 +0300
>@@ -2078,9 +2078,9 @@
>                 /* translate the text coded in the entry's host-dependent
>                    "extended ASCII" charset into the compiler's (system's)
>                    internal text code page */
>-                Ext_ASCII_TO_Native((char *)G.outbuf, G.pInfo->hostnum,
>-                                    G.pInfo->hostver, G.pInfo->HasUxAtt,
>-                                    FALSE);
>+                Ext_ASCII_TO_Native((char *)G.outbuf, OUTBUFSIZ,
>+				    G.pInfo->hostnum, G.pInfo->hostver,
>+				    G.pInfo->HasUxAtt, FALSE);
> #ifdef WINDLL
>                 /* translate to ANSI (RTL internal codepage may be OEM) */
>                 INTERN_TO_ISO((char *)G.outbuf, (char *)G.outbuf);
>@@ -2160,7 +2160,8 @@
> 
>         /* translate the Zip entry filename coded in host-dependent "extended
>            ASCII" into the compiler's (system's) internal text code page */
>-        Ext_ASCII_TO_Native(G.filename, G.pInfo->hostnum, G.pInfo->hostver,
>+        Ext_ASCII_TO_Native(G.filename, FILNAMSIZ,
>+			    G.pInfo->hostnum, G.pInfo->hostver,
>                             G.pInfo->HasUxAtt, (option == DS_FN_L));
> 
>         if (G.pInfo->lcflag)      /* replace with lowercase filename */
>diff -Nrbu unzip-5.52/unix/unix.c unzip-5.52-OK/unix/unix.c
>--- unzip-5.52/unix/unix.c	2007-01-30 21:12:18.000000000 +0300
>+++ unzip-5.52-OK/unix/unix.c	2007-01-30 21:14:36.000000000 +0300
>@@ -30,6 +30,10 @@
> #define UNZIP_INTERNAL
> #include "unzip.h"
> 
>+#include <locale.h>
>+#include <iconv.h>
>+#include <langinfo.h>
>+
> #ifdef SCO_XENIX
> #  define SYSNDIR
> #else  /* SCO Unix, AIX, DNIX, TI SysV, Coherent 4.x, ... */
>@@ -1671,3 +1675,116 @@
>     }
> }
> #endif /* QLZIP */
>+
>+
>+static struct {
>+	const char *iso_name;
>+	const char *cp_name;
>+} enc_map[] = {
>+	{ "ISO-8859-1", "CP850" },
>+	{ "ISO-8859-2", "CP852" },
>+	{ "ISO-8859-5", "CP866" },
>+	{ "ISO-8859-6", "CP864" },
>+	{ "ISO-8859-7", "CP869" },
>+	{ "ISO-8859-8", "CP862" },
>+	{ "ISO-8859-9", "CP857" },
>+	{ "TIS-620", "CP874" },
>+	{ "EUC-JP", "CP932" },
>+	{ "EUC-KR", "CP949" },
>+	{ "GB2312", "CP936" },
>+	{ "BIG5", "CP950" },
>+};
>+
>+static const char *from_enc = "CP850";
>+static const char *to_enc = "UTF-8";
>+
>+void init_encodings (void) {
>+	const char *saved_locale, *non_utf8;
>+	char *loc, *p;
>+	int i;
>+
>+	loc = getenv ("LC_ALL") ?: getenv ("LANG");
>+	if (!loc || !(loc = strdup (loc)))  return;
>+
>+
>+	/*  save the locale anyway -- we don't know whether program
>+	   do setlocale() itself or not...
>+	*/
>+	saved_locale = setlocale (LC_ALL, NULL);
>+
>+	setlocale (LC_ALL, "");
>+	to_enc = nl_langinfo (CODESET);
>+
>+	/*  The idea to obtain "from_enc" is:
>+	  Remove codeset part of LC_ALL/LANG (leave only language/country),
>+	  and see what charmap would be used. By this "non-utf8" charmap
>+	  we try to guess the correspond OEM/DOS codepage...
>+	*/
>+	p = strchr (loc, '.');
>+	if (p)  *p = '\0';
>+
>+	/*  temporary...  */
>+	setlocale (LC_ALL, loc);
>+	non_utf8 = nl_langinfo (CODESET);
>+
>+	for (i = 0; i < sizeof (enc_map) / sizeof (*enc_map); i++) {
>+
>+	    if (!strcasecmp (non_utf8, enc_map[i].iso_name)) {
>+
>+		from_enc = enc_map[i].cp_name;
>+
>+		/*  need some more work for latin1 ...  */
>+		if (!strcmp (enc_map[i].iso_name, "ISO-8859-1")) {
>+		    char *lang = getenv ("LC_ALL") ?: getenv ("LANG");
>+		    if (lang) {
>+			if (!strncasecmp (lang, "da", 2) ||
>+			    !strncasecmp (lang, "nb", 2) ||
>+			    !strncasecmp (lang, "nn", 2) ||
>+			    !strncasecmp (lang, "no", 2)
>+			)  from_enc = "CP865";
>+			else if (!strncasecmp (lang, "is", 2))
>+				from_enc = "CP861";
>+			else if (!strncasecmp (lang, "lt", 2) ||
>+				 !strncasecmp (lang, "lv", 2)
>+			)  from_enc = "CP775";
>+		    }
>+		}
>+
>+		break;
>+	    }
>+	}
>+
>+	setlocale (LC_ALL, saved_locale);
>+
>+	return;
>+}
>+
>+void oem_to_local (char *str, size_t size) {
>+	iconv_t conv;
>+	size_t in_len, out_len;
>+	char *out, *out_ptr, *in_ptr;
>+
>+
>+	conv = iconv_open (to_enc, from_enc);
>+	if (conv == (iconv_t) -1)  return;	/*  foo on you   */
>+
>+	in_len = strlen (str);
>+	in_ptr = str;
>+
>+	out_len = 4 * in_len + 1;	/*  assume enough even for utf8 ...  */
>+	if (out_len > size)  out_len = size;
>+	out_len--;		/*  for terminating '\0'   */
>+	out = out_ptr = alloca (out_len);
>+
>+	if (iconv (conv, &in_ptr, &in_len, &out_ptr, &out_len) < 0)
>+		return;		/*  foo on you   */
>+
>+	out_len = (out_ptr - out);
>+	if (out_len >= size)
>+		return;		/*  paranoia   */
>+
>+	memcpy (str, out, out_len);
>+	str[out_len] = '\0';
>+
>+	return;
>+}
>diff -Nrbu unzip-5.52/unzip.c unzip-5.52-OK/unzip.c
>--- unzip-5.52/unzip.c	2005-02-27 07:37:14.000000000 +0300
>+++ unzip-5.52-OK/unzip.c	2007-01-30 21:12:01.000000000 +0300
>@@ -668,6 +668,10 @@
> 
>     SETLOCALE(LC_CTYPE,"");
> 
>+#ifdef UNIX
>+    init_encodings ();
>+#endif
>+
> #if (defined(__IBMC__) && defined(__DEBUG_ALLOC__))
>     extern void DebugMalloc(void);
> 
>diff -Nrbu unzip-5.52/unzpriv.h unzip-5.52-OK/unzpriv.h
>--- unzip-5.52/unzpriv.h	2007-01-30 21:12:18.000000000 +0300
>+++ unzip-5.52-OK/unzpriv.h	2007-01-30 21:12:01.000000000 +0300
>@@ -2576,13 +2576,19 @@
>  * All other ports are assumed to code zip entry filenames in ISO 8859-1.
>  */
> #ifndef Ext_ASCII_TO_Native
>-#  define Ext_ASCII_TO_Native(string, hostnum, hostver, isuxatt, islochdr) \
>+#ifndef UNIX
>+#define _OEM_INTERN_SIZE(STR,SIZE)	_OEM_INTERN((STR))
>+#else
>+extern void oem_to_local (char *str, size_t size);
>+#define _OEM_INTERN_SIZE(STR,SIZE)	oem_to_local ((STR), (SIZE))
>+#endif
>+#  define Ext_ASCII_TO_Native(string, size, hostnum, hostver, isuxatt, islochdr) \
>     if (((hostnum) == FS_FAT_ && \
>          !(((islochdr) || (isuxatt)) && \
>            ((hostver) == 25 || (hostver) == 26 || (hostver) == 40))) || \
>         (hostnum) == FS_HPFS_ || \
>         ((hostnum) == FS_NTFS_ && (hostver) == 50)) { \
>-        _OEM_INTERN((string)); \
>+        _OEM_INTERN_SIZE((string),(size)); \
>     } else { \
>         _ISO_INTERN((string)); \
>     }

Actions: View | Diff

Attachments on bug 225576: 147018 | 157351 | 513838 | 515107 | 1033515 | 1097704 | 1097705 | 1097765 | 1097835 | 1098716 | 1098839