parsing.c

   1 /* config.c: parsing of config files
   2  *
   3  * Copyright (C) 2006 Lars Hjemli
   4  *
   5  * Licensed under GNU General Public License v2
   6  *   (see COPYING for full license text)
   7  */
   8
   9 #include <iconv.h>
  10
  11 #include "cgit.h"
  12
  13 int next_char(FILE *f)
  14 {
  15         int c = fgetc(f);
  16         if (c=='\r') {
  17                 c = fgetc(f);
  18                 if (c!='\n') {
  19                         ungetc(c, f);
  20                         c = '\r';
  21                 }
  22         }
  23         return c;
  24 }
  25
  26 void skip_line(FILE *f)
  27 {
  28         int c;
  29
  30         while((c=next_char(f)) && c!='\n' && c!=EOF)
  31                 ;
  32 }
  33
  34 int read_config_line(FILE *f, char *line, const char **value, int bufsize)
  35 {
  36         int i = 0, isname = 0;
  37
  38         *value = NULL;
  39         while(i<bufsize-1) {
  40                 int c = next_char(f);
  41                 if (!isname && (c=='#' || c==';')) {
  42                         skip_line(f);
  43                         continue;
  44                 }
  45                 if (!isname && isspace(c))
  46                         continue;
  47
  48                 if (c=='=' && !*value) {
  49                         line[i] = 0;
  50                         *value = &line[i+1];
  51                 } else if (c=='\n' && !isname) {
  52                         i = 0;
  53                         continue;
  54                 } else if (c=='\n' || c==EOF) {
  55                         line[i] = 0;
  56                         break;
  57                 } else {
  58                         line[i]=c;
  59                 }
  60                 isname = 1;
  61                 i++;
  62         }
  63         line[i+1] = 0;
  64         return i;
  65 }
  66
  67 int cgit_read_config(const char *filename, configfn fn)
  68 {
  69         static int nesting;
  70         int len;
  71         char line[256];
  72         const char *value;
  73         FILE *f;
  74
  75         /* cancel deeply nested include-commands */
  76         if (nesting > 8)
  77                 return -1;
  78         if (!(f = fopen(filename, "r")))
  79                 return -1;
  80         nesting++;
  81         while((len = read_config_line(f, line, &value, sizeof(line))) > 0)
  82                 (*fn)(line, value);
  83         nesting--;
  84         fclose(f);
  85         return 0;
  86 }
  87
  88 char *convert_query_hexchar(char *txt)
  89 {
  90         int d1, d2;
  91         if (strlen(txt) < 3) {
  92                 *txt = '\0';
  93                 return txt-1;
  94         }
  95         d1 = hextoint(*(txt+1));
  96         d2 = hextoint(*(txt+2));
  97         if (d1<0 || d2<0) {
  98                 strcpy(txt, txt+3);
  99                 return txt-1;
 100         } else {
 101                 *txt = d1 * 16 + d2;
 102                 strcpy(txt+1, txt+3);
 103                 return txt;
 104         }
 105 }
 106
 107 int cgit_parse_query(char *txt, configfn fn)
 108 {
 109         char *t, *value = NULL, c;
 110
 111         if (!txt)
 112                 return 0;
 113
 114         t = txt = xstrdup(txt);
 115
 116         while((c=*t) != '\0') {
 117                 if (c=='=') {
 118                         *t = '\0';
 119                         value = t+1;
 120                 } else if (c=='+') {
 121                         *t = ' ';
 122                 } else if (c=='%') {
 123                         t = convert_query_hexchar(t);
 124                 } else if (c=='&') {
 125                         *t = '\0';
 126                         (*fn)(txt, value);
 127                         txt = t+1;
 128                         value = NULL;
 129                 }
 130                 t++;
 131         }
 132         if (t!=txt)
 133                 (*fn)(txt, value);
 134         return 0;
 135 }
 136
 137 /*
 138  * url syntax: [repo ['/' cmd [ '/' path]]]
 139  *   repo: any valid repo url, may contain '/'
 140  *   cmd:  log | commit | diff | tree | view | blob | snapshot
 141  *   path: any valid path, may contain '/'
 142  *
 143  */
 144 void cgit_parse_url(const char *url)
 145 {
 146         char *cmd, *p;
 147
 148         cgit_repo = NULL;
 149         if (!url || url[0] == '\0')
 150                 return;
 151
 152         cgit_repo = cgit_get_repoinfo(url);
 153         if (cgit_repo) {
 154                 cgit_query_repo = cgit_repo->url;
 155                 return;
 156         }
 157
 158         cmd = strchr(url, '/');
 159         while (!cgit_repo && cmd) {
 160                 cmd[0] = '\0';
 161                 cgit_repo = cgit_get_repoinfo(url);
 162                 if (cgit_repo == NULL) {
 163                         cmd[0] = '/';
 164                         cmd = strchr(cmd + 1, '/');
 165                         continue;
 166                 }
 167
 168                 cgit_query_repo = cgit_repo->url;
 169                 p = strchr(cmd + 1, '/');
 170                 if (p) {
 171                         p[0] = '\0';
 172                         if (p[1])
 173                                 cgit_query_path = trim_end(p + 1, '/');
 174                 }
 175                 cgit_cmd = cgit_get_cmd_index(cmd + 1);
 176                 cgit_query_page = xstrdup(cmd + 1);
 177                 return;
 178         }
 179 }
 180
 181 static char *iconv_msg(char *msg, const char *encoding)
 182 {
 183         iconv_t msg_conv = iconv_open(PAGE_ENCODING, encoding);
 184         size_t inlen = strlen(msg);
 185         char *in;
 186         char *out;
 187         size_t inleft;
 188         size_t outleft;
 189         char *buf;
 190         char *ret;
 191         size_t buf_sz;
 192         int again, fail;
 193
 194         if(msg_conv == (iconv_t)-1)
 195                 return NULL;
 196
 197         buf_sz = inlen * 2;
 198         buf = xmalloc(buf_sz+1);
 199         do {
 200                 in = msg;
 201                 inleft = inlen;
 202
 203                 out = buf;
 204                 outleft = buf_sz;
 205                 iconv(msg_conv, &in, &inleft, &out, &outleft);
 206
 207                 if(inleft == 0) {
 208                         fail = 0;
 209                         again = 0;
 210                 } else if(inleft != 0 && errno == E2BIG) {
 211                         fail = 0;
 212                         again = 1;
 213
 214                         buf_sz *= 2;
 215                         free(buf);
 216                         buf = xmalloc(buf_sz+1);
 217                 } else {
 218                         fail = 1;
 219                         again = 0;
 220                 }
 221         } while(again && !fail);
 222
 223         if(fail) {
 224                 free(buf);
 225                 ret = NULL;
 226         } else {
 227                 buf = xrealloc(buf, out - buf);
 228                 *out = 0;
 229                 ret = buf;
 230         }
 231
 232         iconv_close(msg_conv);
 233
 234         return ret;
 235 }
 236
 237 char *substr(const char *head, const char *tail)
 238 {
 239         char *buf;
 240
 241         buf = xmalloc(tail - head + 1);
 242         strncpy(buf, head, tail - head);
 243         buf[tail - head] = '\0';
 244         return buf;
 245 }
 246
 247 struct commitinfo *cgit_parse_commit(struct commit *commit)
 248 {
 249         struct commitinfo *ret;
 250         char *p = commit->buffer, *t = commit->buffer;
 251
 252         ret = xmalloc(sizeof(*ret));
 253         ret->commit = commit;
 254         ret->author = NULL;
 255         ret->author_email = NULL;
 256         ret->committer = NULL;
 257         ret->committer_email = NULL;
 258         ret->subject = NULL;
 259         ret->msg = NULL;
 260         ret->msg_encoding = NULL;
 261
 262         if (p == NULL)
 263                 return ret;
 264
 265         if (strncmp(p, "tree ", 5))
 266                 die("Bad commit: %s", sha1_to_hex(commit->object.sha1));
 267         else
 268                 p += 46; // "tree " + hex[40] + "\n"
 269
 270         while (!strncmp(p, "parent ", 7))
 271                 p += 48; // "parent " + hex[40] + "\n"
 272
 273         if (!strncmp(p, "author ", 7)) {
 274                 p += 7;
 275                 t = strchr(p, '<') - 1;
 276                 ret->author = substr(p, t);
 277                 p = t;
 278                 t = strchr(t, '>') + 1;
 279                 ret->author_email = substr(p, t);
 280                 ret->author_date = atol(++t);
 281                 p = strchr(t, '\n') + 1;
 282         }
 283
 284         if (!strncmp(p, "committer ", 9)) {
 285                 p += 9;
 286                 t = strchr(p, '<') - 1;
 287                 ret->committer = substr(p, t);
 288                 p = t;
 289                 t = strchr(t, '>') + 1;
 290                 ret->committer_email = substr(p, t);
 291                 ret->committer_date = atol(++t);
 292                 p = strchr(t, '\n') + 1;
 293         }
 294
 295         if (!strncmp(p, "encoding ", 9)) {
 296                 p += 9;
 297                 t = strchr(p, '\n') + 1;
 298                 ret->msg_encoding = substr(p, t);
 299                 p = t;
 300         } else
 301                 ret->msg_encoding = xstrdup(PAGE_ENCODING);
 302
 303         while (*p && (*p != '\n'))
 304                 p = strchr(p, '\n') + 1; // skip unknown header fields
 305
 306         while (*p == '\n')
 307                 p = strchr(p, '\n') + 1;
 308
 309         t = strchr(p, '\n');
 310         if (t) {
 311                 if (*t == '\0')
 312                         ret->subject = "** empty **";
 313                 else
 314                         ret->subject = substr(p, t);
 315                 p = t + 1;
 316
 317                 while (*p == '\n')
 318                         p = strchr(p, '\n') + 1;
 319                 ret->msg = xstrdup(p);
 320         } else
 321                 ret->subject = substr(p, p+strlen(p));
 322
 323         if(strcmp(ret->msg_encoding, PAGE_ENCODING)) {
 324                 t = iconv_msg(ret->subject, ret->msg_encoding);
 325                 if(t) {
 326                         free(ret->subject);
 327                         ret->subject = t;
 328                 }
 329
 330                 t = iconv_msg(ret->msg, ret->msg_encoding);
 331                 if(t) {
 332                         free(ret->msg);
 333                         ret->msg = t;
 334                 }
 335         }
 336
 337         return ret;
 338 }
 339
 340
 341 struct taginfo *cgit_parse_tag(struct tag *tag)
 342 {
 343         void *data;
 344         enum object_type type;
 345         unsigned long size;
 346         char *p, *t;
 347         struct taginfo *ret;
 348
 349         data = read_sha1_file(tag->object.sha1, &type, &size);
 350         if (!data || type != OBJ_TAG) {
 351                 free(data);
 352                 return 0;
 353         }
 354
 355         ret = xmalloc(sizeof(*ret));
 356         ret->tagger = NULL;
 357         ret->tagger_email = NULL;
 358         ret->tagger_date = 0;
 359         ret->msg = NULL;
 360
 361         p = data;
 362
 363         while (p && *p) {
 364                 if (*p == '\n')
 365                         break;
 366
 367                 if (!strncmp(p, "tagger ", 7)) {
 368                         p += 7;
 369                         t = strchr(p, '<') - 1;
 370                         ret->tagger = substr(p, t);
 371                         p = t;
 372                         t = strchr(t, '>') + 1;
 373                         ret->tagger_email = substr(p, t);
 374                         ret->tagger_date = atol(++t);
 375                 }
 376                 p = strchr(p, '\n') + 1;
 377         }
 378
 379         while (p && *p && (*p != '\n'))
 380                 p = strchr(p, '\n') + 1; // skip unknown tag fields
 381
 382         while (p && (*p == '\n'))
 383                 p = strchr(p, '\n') + 1;
 384         if (p && *p)
 385                 ret->msg = xstrdup(p);
 386         free(data);
 387         return ret;
 388 }