Fix truncation problem when reading in long lines. Found by Koen Kooi.
[oweals/opkg-lede.git] / libopkg / pkg_parse.c
1 /* pkg_parse.c - the opkg package management system
2
3    Steven M. Ayer
4    
5    Copyright (C) 2002 Compaq Computer Corporation
6
7    This program is free software; you can redistribute it and/or
8    modify it under the terms of the GNU General Public License as
9    published by the Free Software Foundation; either version 2, or (at
10    your option) any later version.
11
12    This program is distributed in the hope that it will be useful, but
13    WITHOUT ANY WARRANTY; without even the implied warranty of
14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15    General Public License for more details.
16 */
17
18 #include "includes.h"
19 #include <errno.h>
20 #include <ctype.h>
21    
22 #include "pkg.h"
23 #include "opkg_utils.h"
24 #include "pkg_parse.h"
25 #include "libbb/libbb.h"
26
27 static int
28 is_field(char *type, const char *line)
29 {
30         if (!strncmp(line, type, strlen(type)))
31                 return 1;
32         return 0;
33 }
34
35 static char *
36 parse_simple(char *type, const char *line)
37 {
38         return trim_xstrdup(line + strlen(type) + 1);
39 }
40
41 /*
42  * Parse a comma separated string into an array.
43  */
44 static char **
45 parse_comma_separated(const char *raw, int *count)
46 {
47         char **depends = NULL;
48         const char *start, *end;
49         int line_count = 0;
50
51         /* skip past the "Field:" marker */
52         while (*raw && *raw != ':')
53                 raw++;
54         raw++;
55
56         if (line_is_blank(raw)) {
57                 *count = line_count;
58                 return NULL;
59         }
60
61         while (*raw) {
62                 depends = xrealloc(depends, sizeof(char *) * (line_count + 1));
63         
64                 while (isspace(*raw))
65                         raw++;
66
67                 start = raw;
68                 while (*raw != ',' && *raw)
69                         raw++;
70                 end = raw;
71
72                 while (end > start && isspace(*end))
73                         end--;
74
75                 depends[line_count] = xstrndup(start, end-start);
76
77                 line_count++;
78                 if (*raw == ',')
79                     raw++;
80         }
81
82         *count = line_count;
83         return depends;
84 }
85
86 static void
87 parse_status(pkg_t *pkg, const char *sstr)
88 {
89         char sw_str[64], sf_str[64], ss_str[64];
90
91         if (sscanf(sstr, "Status: %63s %63s %63s",
92                                 sw_str, sf_str, ss_str) != 3) {
93                 fprintf(stderr, "%s: failed to parse Status line for %s\n",
94                                 __FUNCTION__, pkg->name);
95                 return;
96         }
97
98         pkg->state_want = pkg_state_want_from_str(sw_str);
99         pkg->state_flag = pkg_state_flag_from_str(sf_str);
100         pkg->state_status = pkg_state_status_from_str(ss_str);
101 }
102
103 static void
104 parse_conffiles(pkg_t *pkg, const char *cstr)
105 {
106         char file_name[1024], md5sum[35];
107
108         if (sscanf(cstr, "%1023s %34s", file_name, md5sum) != 2) {
109                 fprintf(stderr, "%s: failed to parse Conffiles line for %s\n",
110                                 __FUNCTION__, pkg->name);
111                 return;
112         }
113
114         conffile_list_append(&pkg->conffiles, file_name, md5sum);
115 }
116
117 int
118 parse_version(pkg_t *pkg, const char *vstr)
119 {
120         char *colon;
121
122         if (strncmp(vstr, "Version:", 8) == 0)
123                 vstr += 8;
124
125         while (*vstr && isspace(*vstr))
126                 vstr++;
127
128         colon = strchr(vstr, ':');
129         if (colon) {
130                 errno = 0;
131                 pkg->epoch = strtoul(vstr, NULL, 10);
132                 if (errno) {
133                         fprintf(stderr, "%s: %s: invalid epoch: %s\n",
134                                 __FUNCTION__, pkg->name, strerror(errno));
135                 }
136                 vstr = ++colon;
137         } else {
138                 pkg->epoch= 0;
139         }
140
141         pkg->version= xstrdup(vstr);
142         pkg->revision = strrchr(pkg->version,'-');
143
144         if (pkg->revision)
145                 *pkg->revision++ = '\0';
146
147         return 0;
148 }
149
150 static int
151 pkg_parse_line(pkg_t *pkg, const char *line, uint mask)
152 {
153         /* these flags are a bit hackish... */
154         static int reading_conffiles = 0, reading_description = 0;
155         int ret = 0;
156
157         switch (*line) {
158         case 'A':
159                 if ((mask & PFM_ARCHITECTURE ) && is_field("Architecture", line))
160                         pkg->architecture = parse_simple("Architecture", line);
161                 else if ((mask & PFM_AUTO_INSTALLED) && is_field("Auto-Installed", line)) {
162                         char *tmp = parse_simple("Auto-Installed", line);
163                         if (strcmp(tmp, "yes") == 0)
164                             pkg->auto_installed = 1;
165                         free(tmp);
166                 }
167                 break;
168
169         case 'C':
170                 if ((mask & PFM_CONFFILES) && is_field("Conffiles", line)) {
171                         reading_conffiles = 1;
172                         reading_description = 0;
173                         goto dont_reset_flags;
174                 }
175                 else if ((mask & PFM_CONFLICTS) && is_field("Conflicts", line))
176                         pkg->conflicts_str = parse_comma_separated(line, &pkg->conflicts_count);
177                 break;
178
179         case 'D':
180                 if ((mask & PFM_DESCRIPTION) && is_field("Description", line)) {
181                         pkg->description = parse_simple("Description", line);
182                         reading_conffiles = 0;
183                         reading_description = 1;
184                         goto dont_reset_flags;
185                 } else if ((mask & PFM_DEPENDS) && is_field("Depends", line))
186                         pkg->depends_str = parse_comma_separated(line, &pkg->depends_count);
187                 break;
188
189         case 'E':
190                 if((mask & PFM_ESSENTIAL) && is_field("Essential", line)) {
191                         char *tmp = parse_simple("Essential", line);
192                         if (strcmp(tmp, "yes") == 0)
193                                 pkg->essential = 1;
194                         free(tmp);
195                 }
196                 break;
197
198         case 'F':
199                 if((mask & PFM_FILENAME) && is_field("Filename", line))
200                         pkg->filename = parse_simple("Filename", line);
201                 break;
202
203         case 'I':
204                 if ((mask && PFM_INSTALLED_SIZE) && is_field("Installed-Size", line))
205                         pkg->installed_size = parse_simple("Installed-Size", line);
206                 else if ((mask && PFM_INSTALLED_TIME) && is_field("Installed-Time", line)) {
207                         char *tmp = parse_simple("Installed-Time", line);
208                         pkg->installed_time = strtoul(tmp, NULL, 0);
209                         free (tmp);
210                 }           
211                 break;
212
213         case 'M':
214                 if (mask && PFM_MD5SUM) {
215                         if (is_field("MD5sum:", line))
216                                 pkg->md5sum = parse_simple("MD5sum", line);
217                         /* The old opkg wrote out status files with the wrong
218                         * case for MD5sum, let's parse it either way */
219                         else if (is_field("MD5Sum:", line))
220                                 pkg->md5sum = parse_simple("MD5Sum", line);
221                 } else if((mask & PFM_MAINTAINER) && is_field("Maintainer", line))
222                         pkg->maintainer = parse_simple("Maintainer", line);
223                 break;
224
225         case 'P':
226                 if ((mask & PFM_PACKAGE) && is_field("Package", line)) 
227                         pkg->name = parse_simple("Package", line);
228                 else if ((mask & PFM_PRIORITY) && is_field("Priority", line))
229                         pkg->priority = parse_simple("Priority", line);
230                 else if ((mask & PFM_PROVIDES) && is_field("Provides", line))
231                         pkg->provides_str = parse_comma_separated(line, &pkg->provides_count);
232                 else if ((mask & PFM_PRE_DEPENDS) && is_field("Pre-Depends", line))
233                         pkg->pre_depends_str = parse_comma_separated(line, &pkg->pre_depends_count);
234                 break;
235
236         case 'R':
237                 if ((mask & PFM_RECOMMENDS) && is_field("Recommends", line))
238                         pkg->recommends_str = parse_comma_separated(line, &pkg->recommends_count);
239                 else if ((mask & PFM_REPLACES) && is_field("Replaces", line))
240                         pkg->replaces_str = parse_comma_separated(line, &pkg->replaces_count);
241
242                 break;
243
244         case 'S':
245                 if ((mask & PFM_SECTION) && is_field("Section", line))
246                         pkg->section = parse_simple("Section", line);
247 #ifdef HAVE_SHA256
248                 else if ((mask & PFM_SHA256SUM) && is_field("SHA256sum", line))
249                         pkg->sha256sum = parse_simple("SHA256sum", line);
250 #endif
251                 else if ((mask & PFM_SIZE) && is_field("Size", line))
252                         pkg->size = parse_simple("Size", line);
253                 else if ((mask & PFM_SOURCE) && is_field("Source", line))
254                         pkg->source = parse_simple("Source", line);
255                 else if ((mask & PFM_STATUS) && is_field("Status", line))
256                         parse_status(pkg, line);
257                 else if ((mask & PFM_SUGGESTS) && is_field("Suggests", line))
258                         pkg->suggests_str = parse_comma_separated(line, &pkg->suggests_count);
259                 break;
260
261         case 'T':
262                 if ((mask & PFM_TAGS) && is_field("Tags", line))
263                         pkg->tags = parse_simple("Tags", line);
264                 break;
265
266         case 'V':
267                 if ((mask & PFM_VERSION) && is_field("Version", line))
268                         parse_version(pkg, line);
269                 break;
270
271         case ' ':
272                 if ((mask & PFM_DESCRIPTION) && reading_description) {
273                         pkg->description = xrealloc(pkg->description,
274                                                 strlen(pkg->description)
275                                                 + 1 + strlen(line) + 1);
276                         strcat(pkg->description, "\n");
277                         strcat(pkg->description, (line));
278                         goto dont_reset_flags;
279                 } else if ((mask && PFM_CONFFILES) && reading_conffiles) {
280                         parse_conffiles(pkg, line);
281                         goto dont_reset_flags;
282                 }
283
284                 /* FALLTHROUGH */
285         default:
286                 /* For package lists, signifies end of package. */
287                 if(line_is_blank(line)) {
288                         ret = 1;
289                         break;
290                 }
291         }
292
293         reading_description = 0;
294         reading_conffiles = 0;
295
296 dont_reset_flags:
297
298         return ret;
299 }
300
301 int
302 pkg_parse_from_stream_nomalloc(pkg_t *pkg, FILE *fp, uint mask,
303                                                 char **buf0, size_t buf0len)
304 {
305         int ret, lineno;
306         char *buf, *nl;
307         size_t buflen;
308
309         lineno = 1;
310         ret = 0;
311
312         buflen = buf0len;
313         buf = *buf0;
314         buf[0] = '\0';
315
316         while (1) {
317                 if (fgets(buf, buflen, fp) == NULL) {
318                         if (ferror(fp)) {
319                                 fprintf(stderr, "%s: fgets: %s\n",
320                                         __FUNCTION__, strerror(errno));
321                                 ret = -1;
322                         } else if (strlen(*buf0) == buf0len-1) {
323                                 fprintf(stderr, "%s: missing new line character"
324                                                 " at end of file!\n",
325                                         __FUNCTION__);
326                                 pkg_parse_line(pkg, *buf0, mask);
327                         }
328                         break;
329                 }
330
331                 nl = strchr(buf, '\n');
332                 if (nl == NULL) {
333                         if (strlen(buf) < buflen-1) {
334                                 /*
335                                  * Line could be exactly buflen-1 long and
336                                  * missing a newline, but we won't know until
337                                  * fgets fails to read more data.
338                                  */
339                                 fprintf(stderr, "%s: missing new line character"
340                                                 " at end of file!\n",
341                                         __FUNCTION__);
342                                 pkg_parse_line(pkg, *buf0, mask);
343                                 break;
344                         }
345                         if (buf0len >= EXCESSIVE_LINE_LEN) {
346                                 fprintf(stderr, "%s: excessively long line at "
347                                         "%d. Corrupt file?\n",
348                                         __FUNCTION__, lineno);
349                                 ret = -1;
350                                 break;
351                         }
352
353                         /*
354                          * Realloc and point buf past the data already read,
355                          * at the NULL terminator inserted by fgets.
356                          * |<--------------- buf0len ----------------->|
357                          * |                     |<------- buflen ---->|
358                          * |---------------------|---------------------|
359                          * buf0                   buf
360                          */
361                         buflen = buf0len +1;
362                         buf0len *= 2;
363                         *buf0 = xrealloc(*buf0, buf0len);
364                         buf = *buf0 + buflen -2;
365
366                         continue;
367                 }
368
369                 *nl = '\0';
370
371                 lineno++;
372
373                 if (pkg_parse_line(pkg, *buf0, mask))
374                         break;
375
376                 buf = *buf0;
377                 buflen = buf0len;
378                 buf[0] = '\0';
379         };
380
381         if (pkg->name == NULL) {
382                 /* probably just a blank line */
383                 ret = EINVAL;
384         }
385
386         return ret;
387 }
388
389 int
390 pkg_parse_from_stream(pkg_t *pkg, FILE *fp, uint mask)
391 {
392         int ret;
393         char *buf;
394         const size_t len = 4096;
395
396         buf = xmalloc(len);
397         ret = pkg_parse_from_stream_nomalloc(pkg, fp, mask, &buf, len);
398         free(buf);
399
400         return ret;
401 }