sub/find_subfiles.c

   1 #include <dirent.h>
   2 #include <string.h>
   3 #include <stdlib.h>
   4 #include <ctype.h>
   5 #include <assert.h>
   6
   7 #include "osdep/io.h"
   8
   9 #include "mp_msg.h"
  10 #include "options.h"
  11 #include "path.h"
  12 #include "mpcommon.h"
  13 #include "sub/find_subfiles.h"
  14 #include "sub/sub.h"
  15
  16 static struct bstr strip_ext(struct bstr str)
  17 {
  18     int dotpos = bstrrchr(str, '.');
  19     if (dotpos < 0)
  20         return str;
  21     return (struct bstr){str.start, dotpos};
  22 }
  23
  24 static struct bstr get_ext(struct bstr s)
  25 {
  26     int dotpos = bstrrchr(s, '.');
  27     if (dotpos < 0)
  28         return (struct bstr){NULL, 0};
  29     return bstr_splice(s, dotpos + 1, s.len);
  30 }
  31
  32 struct subfn {
  33     int priority;
  34     char *fname;
  35 };
  36
  37 static int compare_sub_priority(const void *a, const void *b)
  38 {
  39     const struct subfn *s1 = a;
  40     const struct subfn *s2 = b;
  41     if (s1->priority > s2->priority)
  42         return -1;
  43     if (s1->priority < s2->priority)
  44         return 1;
  45     return strcoll(s1->fname, s2->fname);
  46 }
  47
  48 static struct bstr guess_lang_from_filename(struct bstr name)
  49 {
  50     if (name.len < 2)
  51         return (struct bstr){NULL, 0};
  52
  53     int n = 0;
  54     int i = name.len - 1;
  55
  56     if (name.start[i] == ')' || name.start[i] == ']')
  57         i--;
  58     while (i >= 0 && isalpha(name.start[i])) {
  59         n++;
  60         if (n > 3)
  61             return (struct bstr){NULL, 0};
  62         i--;
  63     }
  64     if (n < 2)
  65         return (struct bstr){NULL, 0};
  66     return (struct bstr){name.start + i + 1, n};
  67 }
  68
  69 struct sub_list {
  70     struct subfn subs[MAX_SUBTITLE_FILES];
  71     int sid;
  72     void *ctx;
  73 };
  74
  75 /**
  76  * @brief Append all the subtitles in the given path matching fname
  77  * @param opts MPlayer options
  78  * @param slist pointer to the subtitles list tallocated
  79  * @param nsub pointer to the number of subtitles
  80  * @param path Look for subtitles in this directory
  81  * @param fname Subtitle filename (pattern)
  82  * @param limit_fuzziness Ignore flag when sub_fuziness == 2
  83  */
  84 static void append_dir_subtitles(struct MPOpts *opts,
  85                                  struct subfn **slist, int *nsub,
  86                                  struct bstr path, const char *fname,
  87                                  int limit_fuzziness)
  88 {
  89     char *sub_exts[] = {"utf", "utf8", "utf-8", "sub", "srt", "smi", "rt", "txt", "ssa", "aqt", "jss", "js", "ass", NULL};
  90     void *tmpmem = talloc_new(NULL);
  91     FILE *f;
  92     assert(strlen(fname) < 1e6);
  93
  94     struct bstr f_fname = bstr(mp_basename(fname));
  95     struct bstr f_fname_noext = bstrdup(tmpmem, strip_ext(f_fname));
  96     bstr_lower(f_fname_noext);
  97     struct bstr f_fname_trim = bstr_strip(f_fname_noext);
  98
  99     // 0 = nothing
 100     // 1 = any subtitle file
 101     // 2 = any sub file containing movie name
 102     // 3 = sub file containing movie name and the lang extension
 103     char *path0 = bstrdup0(tmpmem, path);
 104     DIR *d = opendir(path0);
 105     if (!d)
 106         goto out;
 107     mp_msg(MSGT_SUBREADER, MSGL_INFO, "Load subtitles in %.*s\n", BSTR_P(path));
 108     struct dirent *de;
 109     while ((de = readdir(d))) {
 110         struct bstr dename = bstr(de->d_name);
 111         void *tmpmem2 = talloc_new(tmpmem);
 112
 113         // retrieve various parts of the filename
 114         struct bstr tmp_fname_noext = bstrdup(tmpmem2, strip_ext(dename));
 115         bstr_lower(tmp_fname_noext);
 116         struct bstr tmp_fname_ext = get_ext(dename);
 117         struct bstr tmp_fname_trim = bstr_strip(tmp_fname_noext);
 118
 119         // If it's a .sub, check if there is a .idx with the same name. If
 120         // there is one, it's certainly a vobsub so we skip it.
 121         if (bstrcasecmp(tmp_fname_ext, bstr("sub")) == 0) {
 122             char *idxname = talloc_asprintf(tmpmem2, "%.*s.idx",
 123                                             (int)tmp_fname_noext.len,
 124                                             de->d_name);
 125             char *idx = mp_path_join(tmpmem2, path, bstr(idxname));
 126             f = fopen(idx, "rt");
 127             if (f) {
 128                 fclose(f);
 129                 goto next_sub;
 130             }
 131         }
 132
 133         // does it end with a subtitle extension?
 134 #ifdef CONFIG_ICONV
 135 #ifdef CONFIG_ENCA
 136         int i = (sub_cp && strncasecmp(sub_cp, "enca", 4) != 0) ? 3 : 0;
 137 #else
 138         int i = sub_cp ? 3 : 0;
 139 #endif
 140 #else
 141         int i = 0;
 142 #endif
 143         while (1) {
 144             if (!sub_exts[i])
 145                 goto next_sub;
 146             if (bstrcasecmp(bstr(sub_exts[i]), tmp_fname_ext) == 0)
 147                 break;
 148             i++;
 149         }
 150
 151         // we have a (likely) subtitle file
 152         int prio = 0;
 153         if (opts->sub_lang) {
 154             if (bstr_startswith(tmp_fname_trim, f_fname_trim)) {
 155                 struct bstr lang = guess_lang_from_filename(tmp_fname_trim);
 156                 if (lang.len) {
 157                     for (int n = 0; opts->sub_lang[n]; n++) {
 158                         if (bstr_startswith(lang,
 159                                             bstr(opts->sub_lang[n]))) {
 160                             prio = 4; // matches the movie name + lang extension
 161                             break;
 162                         }
 163                     }
 164                 }
 165             }
 166         }
 167         if (!prio && bstrcmp(tmp_fname_trim, f_fname_trim) == 0)
 168             prio = 3; // matches the movie name
 169         if (!prio && bstr_find(tmp_fname_trim, f_fname_trim) >= 0
 170             && sub_match_fuzziness >= 1)
 171             prio = 2; // contains the movie name
 172         if (!prio) {
 173             // doesn't contain the movie name
 174             // don't try in the mplayer subtitle directory
 175             if (!limit_fuzziness && sub_match_fuzziness >= 2) {
 176                 prio = 1;
 177             }
 178         }
 179
 180         mp_msg(MSGT_SUBREADER, MSGL_DBG2, "Potential sub file: "
 181                "\"%s\"  Priority: %d\n", de->d_name, prio);
 182         if (prio) {
 183             prio += prio;
 184 #ifdef CONFIG_ICONV
 185             if (i < 3) // prefer UTF-8 coded
 186                 prio++;
 187 #endif
 188             char *subpath = mp_path_join(*slist, path, dename);
 189             if ((f = fopen(subpath, "rt"))) {
 190                 MP_GROW_ARRAY(*slist, *nsub);
 191                 struct subfn *sub = *slist + (*nsub)++;
 192
 193                 fclose(f);
 194                 sub->priority = prio;
 195                 sub->fname    = subpath;
 196             } else
 197                 talloc_free(subpath);
 198         }
 199
 200     next_sub:
 201         talloc_free(tmpmem2);
 202     }
 203     closedir(d);
 204
 205  out:
 206     talloc_free(tmpmem);
 207 }
 208
 209 char **find_text_subtitles(struct MPOpts *opts, const char *fname)
 210 {
 211     char **subnames = NULL;
 212     struct subfn *slist = talloc_array_ptrtype(NULL, slist, 1);
 213     int n = 0;
 214
 215     // Load subtitles from current media directory
 216     append_dir_subtitles(opts, &slist, &n, mp_dirname(fname), fname, 0);
 217
 218     // Load subtitles in dirs specified by sub-paths option
 219     if (opts->sub_paths) {
 220         for (int i = 0; opts->sub_paths[i]; i++) {
 221             char *path = mp_path_join(slist, mp_dirname(fname),
 222                                       bstr(opts->sub_paths[i]));
 223             append_dir_subtitles(opts, &slist, &n, bstr(path), fname, 0);
 224         }
 225     }
 226
 227     // Load subtitles in ~/.mplayer/sub limiting sub fuzziness
 228     char *mp_subdir = get_path("sub/");
 229     if (mp_subdir)
 230         append_dir_subtitles(opts, &slist, &n, bstr(mp_subdir), fname, 1);
 231     free(mp_subdir);
 232
 233     // Sort subs by priority and append them
 234     qsort(slist, n, sizeof(*slist), compare_sub_priority);
 235
 236     subnames = talloc_array_ptrtype(NULL, subnames, n);
 237     for (int i = 0; i < n; i++)
 238         subnames[i] = talloc_strdup(subnames, slist[i].fname);
 239
 240     talloc_free(slist);
 241     return subnames;
 242 }
 243
 244 char **find_vob_subtitles(struct MPOpts *opts, const char *fname)
 245 {
 246     char **vobs = talloc_array_ptrtype(NULL, vobs, 1);
 247     int n = 0;
 248
 249     // Potential vobsub in the media directory
 250     struct bstr bname = bstr(mp_basename(fname));
 251     int pdot = bstrrchr(bname, '.');
 252     if (pdot >= 0)
 253         bname.len = pdot;
 254     vobs[n++] = mp_path_join(vobs, mp_dirname(fname), bname);
 255
 256     // Potential vobsubs in directories specified by sub-paths option
 257     if (opts->sub_paths) {
 258         for (int i = 0; opts->sub_paths[i]; i++) {
 259             char *path = mp_path_join(NULL, mp_dirname(fname),
 260                                       bstr(opts->sub_paths[i]));
 261             MP_GROW_ARRAY(vobs, n);
 262             vobs[n++] = mp_path_join(vobs, bstr(path), bname);
 263             talloc_free(path);
 264         }
 265     }
 266
 267     // Potential vobsub in ~/.mplayer/sub
 268     char *mp_subdir = get_path("sub/");
 269     if (mp_subdir) {
 270         MP_GROW_ARRAY(vobs, n);
 271         vobs[n++] = mp_path_join(vobs, bstr(mp_subdir), bname);
 272     }
 273
 274     free(mp_subdir);
 275     MP_RESIZE_ARRAY(NULL, vobs, n);
 276     return vobs;
 277 }