Attachment #475062 for bug #620470

Lines 30-35 Link Here

(-)unzip-6.0.orig/unix/unix.c (+102 lines)
30	#define UNZIP_INTERNAL	30	#define UNZIP_INTERNAL
31	#include "unzip.h"	31	#include "unzip.h"
32		32
		33	#include <iconv.h>
		34	#include <langinfo.h>
		35
33	#ifdef SCO_XENIX	36	#ifdef SCO_XENIX
34	# define SYSNDIR	37	# define SYSNDIR
35	#else /* SCO Unix, AIX, DNIX, TI SysV, Coherent 4.x, ... */	38	#else /* SCO Unix, AIX, DNIX, TI SysV, Coherent 4.x, ... */
Lines 1874-1876 Link Here
1874	}	1877	}
1875	}	1878	}
1876	#endif /* QLZIP */	1879	#endif /* QLZIP */
		1880
		1881
		1882	typedef struct {
		1883	char *local_charset;
		1884	char *archive_charset;
		1885	} CHARSET_MAP;
		1886
		1887	/* A mapping of local <-> archive charsets used by default to convert filenames
		1888	* of DOS/Windows Zip archives. Currently very basic. */
		1889	static CHARSET_MAP dos_charset_map[] = {
		1890	{ "ANSI_X3.4-1968", "CP850" },
		1891	{ "ISO-8859-1", "CP850" },
		1892	{ "CP1252", "CP850" },
		1893	{ "UTF-8", "CP866" },
		1894	{ "KOI8-R", "CP866" },
		1895	{ "KOI8-U", "CP866" },
		1896	{ "ISO-8859-5", "CP866" }
		1897	};
		1898
		1899	char OEM_CP[MAX_CP_NAME] = "";
		1900	char ISO_CP[MAX_CP_NAME] = "";
		1901
		1902	/* Try to guess the default value of OEM_CP based on the current locale.
		1903	* ISO_CP is left alone for now. */
		1904	void init_conversion_charsets()
		1905	{
		1906	const char *local_charset;
		1907	int i;
		1908
		1909	/* Make a guess only if OEM_CP not already set. */
		1910	if(*OEM_CP == '\0') {
		1911	local_charset = nl_langinfo(CODESET);
		1912	for(i = 0; i < sizeof(dos_charset_map)/sizeof(CHARSET_MAP); i++)
		1913	if(!strcasecmp(local_charset, dos_charset_map[i].local_charset)) {
		1914	strncpy(OEM_CP, dos_charset_map[i].archive_charset,
		1915	sizeof(OEM_CP));
		1916	break;
		1917	}
		1918	}
		1919	}
		1920
		1921	/* Convert a string from one encoding to the current locale using iconv().
		1922	* Be as non-intrusive as possible. If error is encountered during covertion
		1923	* just leave the string intact. */
		1924	static void charset_to_intern(char string, char from_charset)
		1925	{
		1926	iconv_t cd;
		1927	char s,d, *buf;
		1928	size_t slen, dlen, buflen;
		1929	const char *local_charset;
		1930
		1931	if(*from_charset == '\0')
		1932	return;
		1933
		1934	buf = NULL;
		1935	local_charset = nl_langinfo(CODESET);
		1936
		1937	if((cd = iconv_open(local_charset, from_charset)) == (iconv_t)-1)
		1938	return;
		1939
		1940	slen = strlen(string);
		1941	s = string;
		1942
		1943	/* Make sure OUTBUFSIZ + 1 never ends up smaller than FILNAMSIZ
1944	* as this function also gets called with G.outbuf in fileio.c
1945	*/
1946	buflen = FILNAMSIZ;
1947	if (OUTBUFSIZ + 1 < FILNAMSIZ)
1948	{
1949	buflen = OUTBUFSIZ + 1;
1950	}
1951
1952	d = buf = malloc(buflen);
1953	if(!d)
1954	goto cleanup;
1955
1956	bzero(buf,buflen);
1957	dlen = buflen - 1;
1958
1959	if(iconv(cd, &s, &slen, &d, &dlen) == (size_t)-1)
1960	goto cleanup;
1961	strncpy(string, buf, buflen);
1962
1963	cleanup:
1964	free(buf);
1965	iconv_close(cd);
1966	}
1967
1968	/* Convert a string from OEM_CP to the current locale charset. */
1969	inline void oem_intern(char *string)
1970	{
1971	charset_to_intern(string, OEM_CP);
1972	}
1973
1974	/* Convert a string from ISO_CP to the current locale charset. */
1975	inline void iso_intern(char *string)
1976	{
1977	charset_to_intern(string, ISO_CP);
1978	}




/* wild_dir, dirname, wildname, matchname[], dirnamelen, have_dirname, */
/*    and notfirstcall are used by do_wild().                          */


#define MAX_CP_NAME 25 
   
#ifdef SETLOCALE
#  undef SETLOCALE
#endif
#define SETLOCALE(category, locale) setlocale(category, locale)
#include <locale.h>
   
#ifdef _ISO_INTERN
#  undef _ISO_INTERN
#endif
#define _ISO_INTERN(str1) iso_intern(str1)

#ifdef _OEM_INTERN
#  undef _OEM_INTERN
#endif
#ifndef IZ_OEM2ISO_ARRAY
#  define IZ_OEM2ISO_ARRAY
#endif
#define _OEM_INTERN(str1) oem_intern(str1)

void iso_intern(char *);
void oem_intern(char *);
void init_conversion_charsets(void);
   
#endif /* !__unxcfg_h */

Lines 327-337 Link Here

(-)unzip-6.0.orig/unzip.c (+88 lines)
327	-2 just filenames but allow -h/-t/-z -l long Unix \"ls -l\" format\n\	327	-2 just filenames but allow -h/-t/-z -l long Unix \"ls -l\" format\n\
328	-v verbose, multi-page format\n";	328	-v verbose, multi-page format\n";
329		329
		330	#ifndef UNIX
330	static ZCONST char Far ZipInfoUsageLine3[] = "miscellaneous options:\n\	331	static ZCONST char Far ZipInfoUsageLine3[] = "miscellaneous options:\n\
331	-h print header line -t print totals for listed files or for all\n\	332	-h print header line -t print totals for listed files or for all\n\
332	-z print zipfile comment -T print file times in sortable decimal format\	333	-z print zipfile comment -T print file times in sortable decimal format\
333	\n -C be case-insensitive %s\	334	\n -C be case-insensitive %s\
334	-x exclude filenames that follow from listing\n";	335	-x exclude filenames that follow from listing\n";
		336	#else /* UNIX */
		337	static ZCONST char Far ZipInfoUsageLine3[] = "miscellaneous options:\n\
		338	-h print header line -t print totals for listed files or for all\n\
		339	-z print zipfile comment %c-T%c print file times in sortable decimal format\
		340	\n %c-C%c be case-insensitive %s\
		341	-x exclude filenames that follow from listing\n\
		342	-O CHARSET specify a character encoding for DOS, Windows and OS/2 archives\n\
		343	-I CHARSET specify a character encoding for UNIX and other archives\n";
		344	#endif /* !UNIX */
335	#ifdef MORE	345	#ifdef MORE
336	static ZCONST char Far ZipInfoUsageLine4[] =	346	static ZCONST char Far ZipInfoUsageLine4[] =
337	" -M page output through built-in \"more\"\n";	347	" -M page output through built-in \"more\"\n";
Lines 664-669 Link Here
664	-U use escapes for all non-ASCII Unicode -UU ignore any Unicode fields\n\	674	-U use escapes for all non-ASCII Unicode -UU ignore any Unicode fields\n\
665	-C match filenames case-insensitively -L make (some) names \	675	-C match filenames case-insensitively -L make (some) names \
666	lowercase\n %-42s -V retain VMS version numbers\n%s";	676	lowercase\n %-42s -V retain VMS version numbers\n%s";
		677	#elif (defined UNIX)
		678	static ZCONST char Far UnzipUsageLine4[] = "\
		679	modifiers:\n\
		680	-n never overwrite existing files -q quiet mode (-qq => quieter)\n\
		681	-o overwrite files WITHOUT prompting -a auto-convert any text files\n\
		682	-j junk paths (do not make directories) -aa treat ALL files as text\n\
		683	-U use escapes for all non-ASCII Unicode -UU ignore any Unicode fields\n\
		684	-C match filenames case-insensitively -L make (some) names \
		685	lowercase\n %-42s -V retain VMS version numbers\n%s\
		686	-O CHARSET specify a character encoding for DOS, Windows and OS/2 archives\n\
		687	-I CHARSET specify a character encoding for UNIX and other archives\n\n";
667	#else /* !VMS */	688	#else /* !VMS */
668	static ZCONST char Far UnzipUsageLine4[] = "\	689	static ZCONST char Far UnzipUsageLine4[] = "\
669	modifiers:\n\	690	modifiers:\n\
Lines 802-807 Link Here
802	#endif /* UNICODE_SUPPORT */	823	#endif /* UNICODE_SUPPORT */
803		824
804		825
		826	#ifdef UNIX
		827	init_conversion_charsets();
		828	#endif
		829
805	#if (defined(__IBMC__) && defined(__DEBUG_ALLOC__))	830	#if (defined(__IBMC__) && defined(__DEBUG_ALLOC__))
806	extern void DebugMalloc(void);	831	extern void DebugMalloc(void);
807		832
Lines 1335-1340 Link Here
1335	argc = *pargc;	1360	argc = *pargc;
1336	argv = *pargv;	1361	argv = *pargv;
1337		1362
		1363	#ifdef UNIX
		1364	extern char OEM_CP[MAX_CP_NAME];
		1365	extern char ISO_CP[MAX_CP_NAME];
		1366	#endif
		1367
1338	while (++argv, (--argc > 0 && argv != NULL && *argv == '-')) {	1368	while (++argv, (--argc > 0 && argv != NULL && *argv == '-')) {
1339	s = *argv + 1;	1369	s = *argv + 1;
1340	while ((c = s++) != 0) { / "!= 0": prevent Turbo C warning */	1370	while ((c = s++) != 0) { / "!= 0": prevent Turbo C warning */
Lines 1516-1521 Link Here
1516	}	1546	}
1517	break;	1547	break;
1518	#endif /* MACOS */	1548	#endif /* MACOS */
		1549	#ifdef UNIX
		1550	case ('I'):
		1551	if (negative) {
		1552	Info(slide, 0x401, ((char *)slide,
		1553	"error: encodings can't be negated"));
		1554	return(PK_PARAM);
		1555	} else {
		1556	if(s) { / Handle the -Icharset case */
		1557	/* Assume that charsets can't start with a dash to spot arguments misuse */
		1558	if(*s == '-') {
		1559	Info(slide, 0x401, ((char *)slide,
		1560	"error: a valid character encoding should follow the -I argument"));
		1561	return(PK_PARAM);
		1562	}
		1563	strncpy(ISO_CP, s, sizeof(ISO_CP));
		1564	} else { /* -I charset */
		1565	++argv;
		1566	if(!(--argc > 0 && argv != NULL && *argv != '-')) {
		1567	Info(slide, 0x401, ((char *)slide,
		1568	"error: a valid character encoding should follow the -I argument"));
		1569	return(PK_PARAM);
		1570	}
		1571	s = *argv;
		1572	strncpy(ISO_CP, s, sizeof(ISO_CP));
		1573	}
		1574	while((++s)); / No params straight after charset name */
		1575	}
		1576	break;
		1577	#endif /* ?UNIX */
1519	case ('j'): /* junk pathnames/directory structure */	1578	case ('j'): /* junk pathnames/directory structure */
1520	if (negative)	1579	if (negative)
1521	uO.jflag = FALSE, negative = 0;	1580	uO.jflag = FALSE, negative = 0;
Lines 1591-1596 Link Here
1591	} else	1650	} else
1592	++uO.overwrite_all;	1651	++uO.overwrite_all;
1593	break;	1652	break;
		1653	#ifdef UNIX
		1654	case ('O'):
		1655	if (negative) {
		1656	Info(slide, 0x401, ((char *)slide,
		1657	"error: encodings can't be negated"));
		1658	return(PK_PARAM);
		1659	} else {
		1660	if(s) { / Handle the -Ocharset case */
		1661	/* Assume that charsets can't start with a dash to spot arguments misuse */
		1662	if(*s == '-') {
		1663	Info(slide, 0x401, ((char *)slide,
		1664	"error: a valid character encoding should follow the -I argument"));
		1665	return(PK_PARAM);
		1666	}
		1667	strncpy(OEM_CP, s, sizeof(OEM_CP));
		1668	} else { /* -O charset */
		1669	++argv;
		1670	if(!(--argc > 0 && argv != NULL && *argv != '-')) {
		1671	Info(slide, 0x401, ((char *)slide,
		1672	"error: a valid character encoding should follow the -O argument"));
		1673	return(PK_PARAM);
		1674	}
		1675	s = *argv;
		1676	strncpy(OEM_CP, s, sizeof(OEM_CP));
		1677	}
		1678	while((++s)); / No params straight after charset name */
		1679	}
		1680	break;
		1681	#endif /* ?UNIX */
1594	case ('p'): /* pipes: extract to stdout, no messages */	1682	case ('p'): /* pipes: extract to stdout, no messages */
1595	if (negative) {	1683	if (negative) {
1596	uO.cflag = FALSE;	1684	uO.cflag = FALSE;

Lines 3008-3014 Link Here

(-)unzip-6.0.orig/unzpriv.h (-1 / +1 lines)
3008	!(((islochdr) \|\| (isuxatt)) && \	3008	!(((islochdr) \|\| (isuxatt)) && \
3009	((hostver) == 25 \|\| (hostver) == 26 \|\| (hostver) == 40))) \|\| \	3009	((hostver) == 25 \|\| (hostver) == 26 \|\| (hostver) == 40))) \|\| \
3010	(hostnum) == FS_HPFS_ \|\| \	3010	(hostnum) == FS_HPFS_ \|\| \
3011	((hostnum) == FS_NTFS_ && (hostver) == 50)) { \	3011	((hostnum) == FS_NTFS_ /* && (hostver) == 50 */ )) { \
3012	_OEM_INTERN((string)); \	3012	_OEM_INTERN((string)); \
3013	} else { \	3013	} else { \
3014	_ISO_INTERN((string)); \	3014	_ISO_INTERN((string)); \

Lines 457-462 Link Here

(-)unzip-6.0.orig/zipinfo.c (+62 lines)
457	int tflag_slm=TRUE, tflag_2v=FALSE;	457	int tflag_slm=TRUE, tflag_2v=FALSE;
458	int explicit_h=FALSE, explicit_t=FALSE;	458	int explicit_h=FALSE, explicit_t=FALSE;
459		459
		460	#ifdef UNIX
		461	extern char OEM_CP[MAX_CP_NAME];
		462	extern char ISO_CP[MAX_CP_NAME];
		463	#endif
460		464
461	#ifdef MACOS	465	#ifdef MACOS
462	uO.lflag = LFLAG; /* reset default on each call */	466	uO.lflag = LFLAG; /* reset default on each call */
Lines 501-506 Link Here
501	uO.lflag = 0;	505	uO.lflag = 0;
502	}	506	}
503	break;	507	break;
		508	#ifdef UNIX
		509	case ('I'):
		510	if (negative) {
		511	Info(slide, 0x401, ((char *)slide,
		512	"error: encodings can't be negated"));
		513	return(PK_PARAM);
		514	} else {
		515	if(s) { / Handle the -Icharset case */
		516	/* Assume that charsets can't start with a dash to spot arguments misuse */
		517	if(*s == '-') {
		518	Info(slide, 0x401, ((char *)slide,
		519	"error: a valid character encoding should follow the -I argument"));
		520	return(PK_PARAM);
		521	}
		522	strncpy(ISO_CP, s, sizeof(ISO_CP));
		523	} else { /* -I charset */
		524	++argv;
		525	if(!(--argc > 0 && argv != NULL && *argv != '-')) {
		526	Info(slide, 0x401, ((char *)slide,
		527	"error: a valid character encoding should follow the -I argument"));
		528	return(PK_PARAM);
		529	}
		530	s = *argv;
		531	strncpy(ISO_CP, s, sizeof(ISO_CP));
		532	}
		533	while((++s)); / No params straight after charset name */
		534	}
		535	break;
		536	#endif /* ?UNIX */
504	case 'l': /* longer form of "ls -l" type listing */	537	case 'l': /* longer form of "ls -l" type listing */
505	if (negative)	538	if (negative)
506	uO.lflag = -2, negative = 0;	539	uO.lflag = -2, negative = 0;
Lines 521-526 Link Here
521	G.M_flag = TRUE;	554	G.M_flag = TRUE;
522	break;	555	break;
523	#endif	556	#endif
		557	#ifdef UNIX
		558	case ('O'):
		559	if (negative) {
		560	Info(slide, 0x401, ((char *)slide,
		561	"error: encodings can't be negated"));
		562	return(PK_PARAM);
		563	} else {
		564	if(s) { / Handle the -Ocharset case */
		565	/* Assume that charsets can't start with a dash to spot arguments misuse */
		566	if(*s == '-') {
		567	Info(slide, 0x401, ((char *)slide,
		568	"error: a valid character encoding should follow the -I argument"));
		569	return(PK_PARAM);
		570	}
		571	strncpy(OEM_CP, s, sizeof(OEM_CP));
		572	} else { /* -O charset */
		573	++argv;
		574	if(!(--argc > 0 && argv != NULL && *argv != '-')) {
		575	Info(slide, 0x401, ((char *)slide,
		576	"error: a valid character encoding should follow the -O argument"));
		577	return(PK_PARAM);
		578	}
		579	s = *argv;
		580	strncpy(OEM_CP, s, sizeof(OEM_CP));
		581	}
		582	while((++s)); / No params straight after charset name */
		583	}
		584	break;
		585	#endif /* ?UNIX */
524	case 's': /* default: shorter "ls -l" type listing */	586	case 's': /* default: shorter "ls -l" type listing */
525	if (negative)	587	if (negative)
526	uO.lflag = -2, negative = 0;	588	uO.lflag = -2, negative = 0;

Return to bug 620470

Lines 228-231 Link Here

(-)unzip-6.0.orig/unix/unxcfg.h (+26 lines)
228	/* wild_dir, dirname, wildname, matchname[], dirnamelen, have_dirname, */	228	/* wild_dir, dirname, wildname, matchname[], dirnamelen, have_dirname, */
229	/* and notfirstcall are used by do_wild(). */	229	/* and notfirstcall are used by do_wild(). */
230		230
		231
		232	#define MAX_CP_NAME 25
		233
		234	#ifdef SETLOCALE
		235	# undef SETLOCALE
		236	#endif
		237	#define SETLOCALE(category, locale) setlocale(category, locale)
		238	#include <locale.h>
		239
		240	#ifdef _ISO_INTERN
		241	# undef _ISO_INTERN
		242	#endif
		243	#define _ISO_INTERN(str1) iso_intern(str1)
		244
		245	#ifdef _OEM_INTERN
		246	# undef _OEM_INTERN
		247	#endif
		248	#ifndef IZ_OEM2ISO_ARRAY
		249	# define IZ_OEM2ISO_ARRAY
		250	#endif
		251	#define _OEM_INTERN(str1) oem_intern(str1)
		252
		253	void iso_intern(char *);
		254	void oem_intern(char *);
		255	void init_conversion_charsets(void);
		256
231	#endif /* !__unxcfg_h */	257	#endif /* !__unxcfg_h */