sub/find_subfiles.c

   1 #include <dirent.h>
   2 #include <string.h>
   3 #include <stdlib.h>
   4 #include <ctype.h>
   5
   6 #include "mp_msg.h"
   7 #include "path.h"
   8 #include "mpcommon.h"
   9 #include "sub/find_subfiles.h"
  10 #include "sub/sub.h"
  11
  12 static void strcpy_trim(char *d, char *s)
  13 {
  14     // skip leading whitespace
  15     while (*s && isspace(*s)) {
  16         s++;
  17     }
  18     for (;;) {
  19         // copy word
  20         while (*s && !isspace(*s)) {
  21             *d = tolower(*s);
  22             s++; d++;
  23         }
  24         if (*s == 0)
  25             break;
  26         // trim excess whitespace
  27         while (*s && isspace(*s)) {
  28             s++;
  29         }
  30         if (*s == 0)
  31             break;
  32         *d++ = ' ';
  33     }
  34     *d = 0;
  35 }
  36
  37 static void strcpy_strip_ext(char *d, char *s)
  38 {
  39     char *tmp = strrchr(s, '.');
  40     if (!tmp) {
  41         strcpy(d, s);
  42         return;
  43     } else {
  44         strncpy(d, s, tmp-s);
  45         d[tmp-s] = 0;
  46     }
  47     while (*d) {
  48         *d = tolower(*d);
  49         d++;
  50     }
  51 }
  52
  53 static void strcpy_get_ext(char *d, char *s)
  54 {
  55     char *tmp = strrchr(s, '.');
  56     if (!tmp) {
  57         strcpy(d, "");
  58         return;
  59     } else {
  60         strcpy(d, tmp+1);
  61     }
  62 }
  63
  64 static int whiteonly(char *s)
  65 {
  66     while (*s) {
  67         if (!isspace(*s))
  68             return 0;
  69         s++;
  70     }
  71     return 1;
  72 }
  73
  74 typedef struct subfn {
  75     int priority;
  76     char *fname;
  77 } subfn;
  78
  79 static int compare_sub_priority(const void *a, const void *b)
  80 {
  81     if (((const subfn*)a)->priority > ((const subfn*)b)->priority) {
  82         return -1;
  83     } else if (((const subfn*)a)->priority < ((const subfn*)b)->priority) {
  84         return 1;
  85     } else {
  86         return strcoll(((const subfn*)a)->fname, ((const subfn*)b)->fname);
  87     }
  88 }
  89
  90 /**
  91  * @brief Append all the subtitles in the given path matching fname
  92  * @param slist pointer to the subtitles list tallocated
  93  * @param nsub pointer to the number of subtitles
  94  * @param path Look for subtitles in this directory
  95  * @param fname Subtitle filename (pattern)
  96  * @param limit_fuzziness Ignore flag when sub_fuziness == 2
  97  */
  98 static void append_dir_subtitles(struct subfn **slist, int *nsub,
  99                                  struct bstr path, const char *fname,
 100                                  int limit_fuzziness)
 101 {
 102     char *f_fname, *f_fname_noext, *f_fname_trim, *tmp, *tmp_sub_id;
 103     char *tmp_fname_noext, *tmp_fname_trim, *tmp_fname_ext, *tmpresult;
 104
 105     int len, found, i;
 106     char *sub_exts[] = {"utf", "utf8", "utf-8", "sub", "srt", "smi", "rt", "txt", "ssa", "aqt", "jss", "js", "ass", NULL};
 107     FILE *f;
 108
 109     DIR *d;
 110     struct dirent *de;
 111
 112     len =   (strlen(fname) > 256 ? strlen(fname) : 256)
 113           + (path.len      > 256 ? path.len      : 256) + 2;
 114
 115     f_fname       = mp_basename(fname);
 116     f_fname_noext = malloc(len);
 117     f_fname_trim  = malloc(len);
 118
 119     tmp_fname_noext = malloc(len);
 120     tmp_fname_trim  = malloc(len);
 121     tmp_fname_ext   = malloc(len);
 122
 123     tmpresult = malloc(len);
 124
 125     strcpy_strip_ext(f_fname_noext, f_fname);
 126     strcpy_trim(f_fname_trim, f_fname_noext);
 127
 128     /* The code using sub language here is broken - it assumes strict
 129      * "videoname languagename" syntax for the subtitle file, which is
 130      * very unlikely to match especially if language name uses "en,de"
 131      * syntax... */
 132     tmp_sub_id = NULL;
 133 #if 0
 134     if (dvdsub_lang && !whiteonly(dvdsub_lang)) {
 135         tmp_sub_id = malloc(strlen(dvdsub_lang) + 1);
 136         strcpy_trim(tmp_sub_id, dvdsub_lang);
 137     }
 138 #endif
 139
 140     // 0 = nothing
 141     // 1 = any subtitle file
 142     // 2 = any sub file containing movie name
 143     // 3 = sub file containing movie name and the lang extension
 144     char *path0 = bstrdup0(NULL, path);
 145     d = opendir(path0);
 146     talloc_free(path0);
 147     if (d) {
 148         mp_msg(MSGT_SUBREADER, MSGL_INFO, "Load subtitles in %.*s\n", BSTR_P(path));
 149         while ((de = readdir(d))) {
 150             // retrieve various parts of the filename
 151             strcpy_strip_ext(tmp_fname_noext, de->d_name);
 152             strcpy_get_ext(tmp_fname_ext, de->d_name);
 153             strcpy_trim(tmp_fname_trim, tmp_fname_noext);
 154
 155             // does it end with a subtitle extension?
 156             found = 0;
 157 #ifdef CONFIG_ICONV
 158 #ifdef CONFIG_ENCA
 159             for (i = ((sub_cp && strncasecmp(sub_cp, "enca", 4) != 0) ? 3 : 0); sub_exts[i]; i++) {
 160 #else
 161             for (i = (sub_cp ? 3 : 0); sub_exts[i]; i++) {
 162 #endif
 163 #else
 164             for (i = 0; sub_exts[i]; i++) {
 165 #endif
 166                 if (strcasecmp(sub_exts[i], tmp_fname_ext) == 0) {
 167                     found = 1;
 168                     break;
 169                 }
 170             }
 171
 172             // we have a (likely) subtitle file
 173             if (found) {
 174                 int prio = 0;
 175                 if (!prio && tmp_sub_id) {
 176                     sprintf(tmpresult, "%s %s", f_fname_trim, tmp_sub_id);
 177                     if (strcmp(tmp_fname_trim, tmpresult) == 0 && sub_match_fuzziness >= 1) {
 178                         // matches the movie name + lang extension
 179                         prio = 5;
 180                     }
 181                 }
 182                 if (!prio && strcmp(tmp_fname_trim, f_fname_trim) == 0) {
 183                     // matches the movie name
 184                     prio = 4;
 185                 }
 186                 if (!prio && (tmp = strstr(tmp_fname_trim, f_fname_trim)) && sub_match_fuzziness >= 1) {
 187                     // contains the movie name
 188                     tmp += strlen(f_fname_trim);
 189                     if (tmp_sub_id && strstr(tmp, tmp_sub_id)) {
 190                         // with sub_id specified prefer localized subtitles
 191                         prio = 3;
 192                     } else if ((tmp_sub_id == NULL) && whiteonly(tmp)) {
 193                         // without sub_id prefer "plain" name
 194                         prio = 3;
 195                     } else {
 196                         // with no localized subs found, try any else instead
 197                         prio = 2;
 198                     }
 199                 }
 200                 if (!prio) {
 201                     // doesn't contain the movie name
 202                     // don't try in the mplayer subtitle directory
 203                     if (!limit_fuzziness && sub_match_fuzziness >= 2) {
 204                         prio = 1;
 205                     }
 206                 }
 207
 208                 mp_msg(MSGT_SUBREADER, MSGL_DBG2, "Potential sub file: "
 209                        "\"%s\"  Priority: %d\n", de->d_name, prio);
 210                 if (prio) {
 211                     prio += prio;
 212 #ifdef CONFIG_ICONV
 213                     if (i < 3) // prefer UTF-8 coded
 214                         prio++;
 215 #endif
 216                     char *subpath = mp_path_join(*slist, path, BSTR(de->d_name));
 217                     if ((f = fopen(subpath, "rt"))) {
 218                         MP_GROW_ARRAY(*slist, *nsub);
 219                         struct subfn *sub = *slist + (*nsub)++;
 220
 221                         fclose(f);
 222                         sub->priority = prio;
 223                         sub->fname    = subpath;
 224                     } else
 225                         talloc_free(subpath);
 226                 }
 227             }
 228         }
 229         closedir(d);
 230     }
 231
 232     free(tmp_sub_id);
 233
 234     free(f_fname_noext);
 235     free(f_fname_trim);
 236
 237     free(tmp_fname_noext);
 238     free(tmp_fname_trim);
 239     free(tmp_fname_ext);
 240
 241     free(tmpresult);
 242 }
 243
 244 char **find_text_subtitles(const char *fname)
 245 {
 246     struct subfn *slist = talloc_array_ptrtype(NULL, slist, 1);
 247     int n = 0;
 248
 249     // Load subtitles from current media directory
 250     append_dir_subtitles(&slist, &n, mp_dirname(fname), fname, 0);
 251
 252     // Load subtitles in ~/.mplayer/sub limiting sub fuzziness
 253     char *mp_subdir = get_path("sub/");
 254     if (mp_subdir)
 255         append_dir_subtitles(&slist, &n, BSTR(mp_subdir), fname, 1);
 256     free(mp_subdir);
 257
 258     // Sort subs by priority and append them
 259     qsort(slist, n, sizeof(*slist), compare_sub_priority);
 260
 261     char **subnames = talloc_array_ptrtype(NULL, subnames, n);
 262     for (int i = 0; i < n; i++)
 263         subnames[i] = talloc_strdup(subnames, slist[i].fname);
 264
 265     talloc_free(slist);
 266     return subnames;
 267 }
 268
 269 char **find_vob_subtitles(const char *fname)
 270 {
 271     char **vobs = talloc_array_ptrtype(NULL, vobs, 1);
 272     int n = 0;
 273
 274     // Potential vobsub in the media directory
 275     struct bstr bname = BSTR(mp_basename(fname));
 276     int pdot = bstrrchr(bname, '.');
 277     if (pdot >= 0)
 278         bname.len = pdot;
 279     vobs[n++] = mp_path_join(vobs, mp_dirname(fname), bname);
 280
 281     // Potential vobsub in ~/.mplayer/sub
 282     char *mp_subdir = get_path("sub/");
 283     if (mp_subdir) {
 284         MP_GROW_ARRAY(vobs, n);
 285         vobs[n++] = mp_path_join(vobs, BSTR(mp_subdir), bname);
 286     }
 287
 288     free(mp_subdir);
 289     MP_RESIZE_ARRAY(NULL, vobs, n);
 290     return vobs;
 291 }