Commit | Line | Data |
---|---|---|
31d289a4 LLB |
1 | From: Giovanni Scafora <giovanni.archlinux.org> |
2 | Subject: unzip files encoded with non-latin, non-unicode file names | |
3 | Last-Update: 2015-02-11 | |
4 | ||
5 | Updated 2015-02-11 by Marc Deslauriers <marc.deslauriers@canonical.com> | |
6 | to fix buffer overflow in charset_to_intern() | |
7 | ||
8 | Index: unzip-6.0/unix/unix.c | |
9 | =================================================================== | |
10 | --- unzip-6.0.orig/unix/unix.c 2015-02-11 08:46:43.675324290 -0500 | |
11 | +++ unzip-6.0/unix/unix.c 2015-02-11 09:18:04.902081319 -0500 | |
12 | @@ -30,6 +30,9 @@ | |
13 | #define UNZIP_INTERNAL | |
14 | #include "unzip.h" | |
15 | ||
16 | +#include <iconv.h> | |
17 | +#include <langinfo.h> | |
18 | + | |
19 | #ifdef SCO_XENIX | |
20 | # define SYSNDIR | |
21 | #else /* SCO Unix, AIX, DNIX, TI SysV, Coherent 4.x, ... */ | |
22 | @@ -1874,3 +1877,102 @@ | |
23 | } | |
24 | } | |
25 | #endif /* QLZIP */ | |
26 | + | |
27 | + | |
28 | +typedef struct { | |
29 | + char *local_charset; | |
30 | + char *archive_charset; | |
31 | +} CHARSET_MAP; | |
32 | + | |
33 | +/* A mapping of local <-> archive charsets used by default to convert filenames | |
34 | + * of DOS/Windows Zip archives. Currently very basic. */ | |
35 | +static CHARSET_MAP dos_charset_map[] = { | |
36 | + { "ANSI_X3.4-1968", "CP850" }, | |
37 | + { "ISO-8859-1", "CP850" }, | |
38 | + { "CP1252", "CP850" }, | |
39 | + { "UTF-8", "CP866" }, | |
40 | + { "KOI8-R", "CP866" }, | |
41 | + { "KOI8-U", "CP866" }, | |
42 | + { "ISO-8859-5", "CP866" } | |
43 | +}; | |
44 | + | |
45 | +char OEM_CP[MAX_CP_NAME] = ""; | |
46 | +char ISO_CP[MAX_CP_NAME] = ""; | |
47 | + | |
48 | +/* Try to guess the default value of OEM_CP based on the current locale. | |
49 | + * ISO_CP is left alone for now. */ | |
50 | +void init_conversion_charsets() | |
51 | +{ | |
52 | + const char *local_charset; | |
53 | + int i; | |
54 | + | |
55 | + /* Make a guess only if OEM_CP not already set. */ | |
56 | + if(*OEM_CP == '\0') { | |
57 | + local_charset = nl_langinfo(CODESET); | |
58 | + for(i = 0; i < sizeof(dos_charset_map)/sizeof(CHARSET_MAP); i++) | |
59 | + if(!strcasecmp(local_charset, dos_charset_map[i].local_charset)) { | |
60 | + strncpy(OEM_CP, dos_charset_map[i].archive_charset, | |
61 | + sizeof(OEM_CP)); | |
62 | + break; | |
63 | + } | |
64 | + } | |
65 | +} | |
66 | + | |
67 | +/* Convert a string from one encoding to the current locale using iconv(). | |
68 | + * Be as non-intrusive as possible. If error is encountered during covertion | |
69 | + * just leave the string intact. */ | |
70 | +static void charset_to_intern(char *string, char *from_charset) | |
71 | +{ | |
72 | + iconv_t cd; | |
73 | + char *s,*d, *buf; | |
74 | + size_t slen, dlen, buflen; | |
75 | + const char *local_charset; | |
76 | + | |
77 | + if(*from_charset == '\0') | |
78 | + return; | |
79 | + | |
80 | + buf = NULL; | |
81 | + local_charset = nl_langinfo(CODESET); | |
82 | + | |
83 | + if((cd = iconv_open(local_charset, from_charset)) == (iconv_t)-1) | |
84 | + return; | |
85 | + | |
86 | + slen = strlen(string); | |
87 | + s = string; | |
88 | + | |
89 | + /* Make sure OUTBUFSIZ + 1 never ends up smaller than FILNAMSIZ | |
90 | + * as this function also gets called with G.outbuf in fileio.c | |
91 | + */ | |
92 | + buflen = FILNAMSIZ; | |
93 | + if (OUTBUFSIZ + 1 < FILNAMSIZ) | |
94 | + { | |
95 | + buflen = OUTBUFSIZ + 1; | |
96 | + } | |
97 | + | |
98 | + d = buf = malloc(buflen); | |
99 | + if(!d) | |
100 | + goto cleanup; | |
101 | + | |
102 | + bzero(buf,buflen); | |
103 | + dlen = buflen - 1; | |
104 | + | |
105 | + if(iconv(cd, &s, &slen, &d, &dlen) == (size_t)-1) | |
106 | + goto cleanup; | |
107 | + strncpy(string, buf, buflen); | |
108 | + | |
109 | + cleanup: | |
110 | + free(buf); | |
111 | + iconv_close(cd); | |
112 | +} | |
113 | + | |
114 | +/* Convert a string from OEM_CP to the current locale charset. */ | |
115 | +inline void oem_intern(char *string) | |
116 | +{ | |
117 | + charset_to_intern(string, OEM_CP); | |
118 | +} | |
119 | + | |
120 | +/* Convert a string from ISO_CP to the current locale charset. */ | |
121 | +inline void iso_intern(char *string) | |
122 | +{ | |
123 | + charset_to_intern(string, ISO_CP); | |
124 | +} | |
125 | Index: unzip-6.0/unix/unxcfg.h | |
126 | =================================================================== | |
127 | --- unzip-6.0.orig/unix/unxcfg.h 2015-02-11 08:46:43.675324290 -0500 | |
128 | +++ unzip-6.0/unix/unxcfg.h 2015-02-11 08:46:43.671324260 -0500 | |
129 | @@ -228,4 +228,30 @@ | |
130 | /* wild_dir, dirname, wildname, matchname[], dirnamelen, have_dirname, */ | |
131 | /* and notfirstcall are used by do_wild(). */ | |
132 | ||
133 | + | |
134 | +#define MAX_CP_NAME 25 | |
135 | + | |
136 | +#ifdef SETLOCALE | |
137 | +# undef SETLOCALE | |
138 | +#endif | |
139 | +#define SETLOCALE(category, locale) setlocale(category, locale) | |
140 | +#include <locale.h> | |
141 | + | |
142 | +#ifdef _ISO_INTERN | |
143 | +# undef _ISO_INTERN | |
144 | +#endif | |
145 | +#define _ISO_INTERN(str1) iso_intern(str1) | |
146 | + | |
147 | +#ifdef _OEM_INTERN | |
148 | +# undef _OEM_INTERN | |
149 | +#endif | |
150 | +#ifndef IZ_OEM2ISO_ARRAY | |
151 | +# define IZ_OEM2ISO_ARRAY | |
152 | +#endif | |
153 | +#define _OEM_INTERN(str1) oem_intern(str1) | |
154 | + | |
155 | +void iso_intern(char *); | |
156 | +void oem_intern(char *); | |
157 | +void init_conversion_charsets(void); | |
158 | + | |
159 | #endif /* !__unxcfg_h */ | |
160 | Index: unzip-6.0/unzip.c | |
161 | =================================================================== | |
162 | --- unzip-6.0.orig/unzip.c 2015-02-11 08:46:43.675324290 -0500 | |
163 | +++ unzip-6.0/unzip.c 2015-02-11 08:46:43.675324290 -0500 | |
164 | @@ -327,11 +327,21 @@ | |
165 | -2 just filenames but allow -h/-t/-z -l long Unix \"ls -l\" format\n\ | |
166 | -v verbose, multi-page format\n"; | |
167 | ||
168 | +#ifndef UNIX | |
169 | static ZCONST char Far ZipInfoUsageLine3[] = "miscellaneous options:\n\ | |
170 | -h print header line -t print totals for listed files or for all\n\ | |
171 | -z print zipfile comment -T print file times in sortable decimal format\ | |
172 | \n -C be case-insensitive %s\ | |
173 | -x exclude filenames that follow from listing\n"; | |
174 | +#else /* UNIX */ | |
175 | +static ZCONST char Far ZipInfoUsageLine3[] = "miscellaneous options:\n\ | |
176 | + -h print header line -t print totals for listed files or for all\n\ | |
177 | + -z print zipfile comment %c-T%c print file times in sortable decimal format\ | |
178 | +\n %c-C%c be case-insensitive %s\ | |
179 | + -x exclude filenames that follow from listing\n\ | |
180 | + -O CHARSET specify a character encoding for DOS, Windows and OS/2 archives\n\ | |
181 | + -I CHARSET specify a character encoding for UNIX and other archives\n"; | |
182 | +#endif /* !UNIX */ | |
183 | #ifdef MORE | |
184 | static ZCONST char Far ZipInfoUsageLine4[] = | |
185 | " -M page output through built-in \"more\"\n"; | |
186 | @@ -664,6 +674,17 @@ | |
187 | -U use escapes for all non-ASCII Unicode -UU ignore any Unicode fields\n\ | |
188 | -C match filenames case-insensitively -L make (some) names \ | |
189 | lowercase\n %-42s -V retain VMS version numbers\n%s"; | |
190 | +#elif (defined UNIX) | |
191 | +static ZCONST char Far UnzipUsageLine4[] = "\ | |
192 | +modifiers:\n\ | |
193 | + -n never overwrite existing files -q quiet mode (-qq => quieter)\n\ | |
194 | + -o overwrite files WITHOUT prompting -a auto-convert any text files\n\ | |
195 | + -j junk paths (do not make directories) -aa treat ALL files as text\n\ | |
196 | + -U use escapes for all non-ASCII Unicode -UU ignore any Unicode fields\n\ | |
197 | + -C match filenames case-insensitively -L make (some) names \ | |
198 | +lowercase\n %-42s -V retain VMS version numbers\n%s\ | |
199 | + -O CHARSET specify a character encoding for DOS, Windows and OS/2 archives\n\ | |
200 | + -I CHARSET specify a character encoding for UNIX and other archives\n\n"; | |
201 | #else /* !VMS */ | |
202 | static ZCONST char Far UnzipUsageLine4[] = "\ | |
203 | modifiers:\n\ | |
204 | @@ -802,6 +823,10 @@ | |
205 | #endif /* UNICODE_SUPPORT */ | |
206 | ||
207 | ||
208 | +#ifdef UNIX | |
209 | + init_conversion_charsets(); | |
210 | +#endif | |
211 | + | |
212 | #if (defined(__IBMC__) && defined(__DEBUG_ALLOC__)) | |
213 | extern void DebugMalloc(void); | |
214 | ||
215 | @@ -1335,6 +1360,11 @@ | |
216 | argc = *pargc; | |
217 | argv = *pargv; | |
218 | ||
219 | +#ifdef UNIX | |
220 | + extern char OEM_CP[MAX_CP_NAME]; | |
221 | + extern char ISO_CP[MAX_CP_NAME]; | |
222 | +#endif | |
223 | + | |
224 | while (++argv, (--argc > 0 && *argv != NULL && **argv == '-')) { | |
225 | s = *argv + 1; | |
226 | while ((c = *s++) != 0) { /* "!= 0": prevent Turbo C warning */ | |
227 | @@ -1516,6 +1546,35 @@ | |
228 | } | |
229 | break; | |
230 | #endif /* MACOS */ | |
231 | +#ifdef UNIX | |
232 | + case ('I'): | |
233 | + if (negative) { | |
234 | + Info(slide, 0x401, ((char *)slide, | |
235 | + "error: encodings can't be negated")); | |
236 | + return(PK_PARAM); | |
237 | + } else { | |
238 | + if(*s) { /* Handle the -Icharset case */ | |
239 | + /* Assume that charsets can't start with a dash to spot arguments misuse */ | |
240 | + if(*s == '-') { | |
241 | + Info(slide, 0x401, ((char *)slide, | |
242 | + "error: a valid character encoding should follow the -I argument")); | |
243 | + return(PK_PARAM); | |
244 | + } | |
245 | + strncpy(ISO_CP, s, sizeof(ISO_CP)); | |
246 | + } else { /* -I charset */ | |
247 | + ++argv; | |
248 | + if(!(--argc > 0 && *argv != NULL && **argv != '-')) { | |
249 | + Info(slide, 0x401, ((char *)slide, | |
250 | + "error: a valid character encoding should follow the -I argument")); | |
251 | + return(PK_PARAM); | |
252 | + } | |
253 | + s = *argv; | |
254 | + strncpy(ISO_CP, s, sizeof(ISO_CP)); | |
255 | + } | |
256 | + while(*(++s)); /* No params straight after charset name */ | |
257 | + } | |
258 | + break; | |
259 | +#endif /* ?UNIX */ | |
260 | case ('j'): /* junk pathnames/directory structure */ | |
261 | if (negative) | |
262 | uO.jflag = FALSE, negative = 0; | |
263 | @@ -1591,6 +1650,35 @@ | |
264 | } else | |
265 | ++uO.overwrite_all; | |
266 | break; | |
267 | +#ifdef UNIX | |
268 | + case ('O'): | |
269 | + if (negative) { | |
270 | + Info(slide, 0x401, ((char *)slide, | |
271 | + "error: encodings can't be negated")); | |
272 | + return(PK_PARAM); | |
273 | + } else { | |
274 | + if(*s) { /* Handle the -Ocharset case */ | |
275 | + /* Assume that charsets can't start with a dash to spot arguments misuse */ | |
276 | + if(*s == '-') { | |
277 | + Info(slide, 0x401, ((char *)slide, | |
278 | + "error: a valid character encoding should follow the -I argument")); | |
279 | + return(PK_PARAM); | |
280 | + } | |
281 | + strncpy(OEM_CP, s, sizeof(OEM_CP)); | |
282 | + } else { /* -O charset */ | |
283 | + ++argv; | |
284 | + if(!(--argc > 0 && *argv != NULL && **argv != '-')) { | |
285 | + Info(slide, 0x401, ((char *)slide, | |
286 | + "error: a valid character encoding should follow the -O argument")); | |
287 | + return(PK_PARAM); | |
288 | + } | |
289 | + s = *argv; | |
290 | + strncpy(OEM_CP, s, sizeof(OEM_CP)); | |
291 | + } | |
292 | + while(*(++s)); /* No params straight after charset name */ | |
293 | + } | |
294 | + break; | |
295 | +#endif /* ?UNIX */ | |
296 | case ('p'): /* pipes: extract to stdout, no messages */ | |
297 | if (negative) { | |
298 | uO.cflag = FALSE; | |
299 | Index: unzip-6.0/unzpriv.h | |
300 | =================================================================== | |
301 | --- unzip-6.0.orig/unzpriv.h 2015-02-11 08:46:43.675324290 -0500 | |
302 | +++ unzip-6.0/unzpriv.h 2015-02-11 08:46:43.675324290 -0500 | |
303 | @@ -3008,7 +3008,7 @@ | |
304 | !(((islochdr) || (isuxatt)) && \ | |
305 | ((hostver) == 25 || (hostver) == 26 || (hostver) == 40))) || \ | |
306 | (hostnum) == FS_HPFS_ || \ | |
307 | - ((hostnum) == FS_NTFS_ && (hostver) == 50)) { \ | |
308 | + ((hostnum) == FS_NTFS_ /* && (hostver) == 50 */ )) { \ | |
309 | _OEM_INTERN((string)); \ | |
310 | } else { \ | |
311 | _ISO_INTERN((string)); \ | |
312 | Index: unzip-6.0/zipinfo.c | |
313 | =================================================================== | |
314 | --- unzip-6.0.orig/zipinfo.c 2015-02-11 08:46:43.675324290 -0500 | |
315 | +++ unzip-6.0/zipinfo.c 2015-02-11 08:46:43.675324290 -0500 | |
316 | @@ -457,6 +457,10 @@ | |
317 | int tflag_slm=TRUE, tflag_2v=FALSE; | |
318 | int explicit_h=FALSE, explicit_t=FALSE; | |
319 | ||
320 | +#ifdef UNIX | |
321 | + extern char OEM_CP[MAX_CP_NAME]; | |
322 | + extern char ISO_CP[MAX_CP_NAME]; | |
323 | +#endif | |
324 | ||
325 | #ifdef MACOS | |
326 | uO.lflag = LFLAG; /* reset default on each call */ | |
327 | @@ -501,6 +505,35 @@ | |
328 | uO.lflag = 0; | |
329 | } | |
330 | break; | |
331 | +#ifdef UNIX | |
332 | + case ('I'): | |
333 | + if (negative) { | |
334 | + Info(slide, 0x401, ((char *)slide, | |
335 | + "error: encodings can't be negated")); | |
336 | + return(PK_PARAM); | |
337 | + } else { | |
338 | + if(*s) { /* Handle the -Icharset case */ | |
339 | + /* Assume that charsets can't start with a dash to spot arguments misuse */ | |
340 | + if(*s == '-') { | |
341 | + Info(slide, 0x401, ((char *)slide, | |
342 | + "error: a valid character encoding should follow the -I argument")); | |
343 | + return(PK_PARAM); | |
344 | + } | |
345 | + strncpy(ISO_CP, s, sizeof(ISO_CP)); | |
346 | + } else { /* -I charset */ | |
347 | + ++argv; | |
348 | + if(!(--argc > 0 && *argv != NULL && **argv != '-')) { | |
349 | + Info(slide, 0x401, ((char *)slide, | |
350 | + "error: a valid character encoding should follow the -I argument")); | |
351 | + return(PK_PARAM); | |
352 | + } | |
353 | + s = *argv; | |
354 | + strncpy(ISO_CP, s, sizeof(ISO_CP)); | |
355 | + } | |
356 | + while(*(++s)); /* No params straight after charset name */ | |
357 | + } | |
358 | + break; | |
359 | +#endif /* ?UNIX */ | |
360 | case 'l': /* longer form of "ls -l" type listing */ | |
361 | if (negative) | |
362 | uO.lflag = -2, negative = 0; | |
363 | @@ -521,6 +554,35 @@ | |
364 | G.M_flag = TRUE; | |
365 | break; | |
366 | #endif | |
367 | +#ifdef UNIX | |
368 | + case ('O'): | |
369 | + if (negative) { | |
370 | + Info(slide, 0x401, ((char *)slide, | |
371 | + "error: encodings can't be negated")); | |
372 | + return(PK_PARAM); | |
373 | + } else { | |
374 | + if(*s) { /* Handle the -Ocharset case */ | |
375 | + /* Assume that charsets can't start with a dash to spot arguments misuse */ | |
376 | + if(*s == '-') { | |
377 | + Info(slide, 0x401, ((char *)slide, | |
378 | + "error: a valid character encoding should follow the -I argument")); | |
379 | + return(PK_PARAM); | |
380 | + } | |
381 | + strncpy(OEM_CP, s, sizeof(OEM_CP)); | |
382 | + } else { /* -O charset */ | |
383 | + ++argv; | |
384 | + if(!(--argc > 0 && *argv != NULL && **argv != '-')) { | |
385 | + Info(slide, 0x401, ((char *)slide, | |
386 | + "error: a valid character encoding should follow the -O argument")); | |
387 | + return(PK_PARAM); | |
388 | + } | |
389 | + s = *argv; | |
390 | + strncpy(OEM_CP, s, sizeof(OEM_CP)); | |
391 | + } | |
392 | + while(*(++s)); /* No params straight after charset name */ | |
393 | + } | |
394 | + break; | |
395 | +#endif /* ?UNIX */ | |
396 | case 's': /* default: shorter "ls -l" type listing */ | |
397 | if (negative) | |
398 | uO.lflag = -2, negative = 0; |