src/tag.c

   1 /*
   2  * Copyright (C) 2003-2010 The Music Player Daemon Project
   3  * http://www.musicpd.org
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  18  */
  19
  20 #include "config.h"
  21 #include "tag.h"
  22 #include "tag_internal.h"
  23 #include "tag_pool.h"
  24 #include "conf.h"
  25 #include "song.h"
  26
  27 #include <glib.h>
  28 #include <assert.h>
  29 #include <stdio.h>
  30 #include <stdlib.h>
  31
  32 /**
  33  * Maximum number of items managed in the bulk list; if it is
  34  * exceeded, we switch back to "normal" reallocation.
  35  */
  36 #define BULK_MAX 64
  37
  38 static struct {
  39 #ifndef NDEBUG
  40         bool busy;
  41 #endif
  42         struct tag_item *items[BULK_MAX];
  43 } bulk;
  44
  45 const char *tag_item_names[TAG_NUM_OF_ITEM_TYPES] = {
  46         [TAG_ARTIST] = "Artist",
  47         [TAG_ARTIST_SORT] = "ArtistSort",
  48         [TAG_ALBUM] = "Album",
  49         [TAG_ALBUM_ARTIST] = "AlbumArtist",
  50         [TAG_ALBUM_ARTIST_SORT] = "AlbumArtistSort",
  51         [TAG_TITLE] = "Title",
  52         [TAG_TRACK] = "Track",
  53         [TAG_NAME] = "Name",
  54         [TAG_GENRE] = "Genre",
  55         [TAG_DATE] = "Date",
  56         [TAG_COMPOSER] = "Composer",
  57         [TAG_PERFORMER] = "Performer",
  58         [TAG_COMMENT] = "Comment",
  59         [TAG_DISC] = "Disc",
  60
  61         /* MusicBrainz tags from http://musicbrainz.org/doc/MusicBrainzTag */
  62         [TAG_MUSICBRAINZ_ARTISTID] = "MUSICBRAINZ_ARTISTID",
  63         [TAG_MUSICBRAINZ_ALBUMID] = "MUSICBRAINZ_ALBUMID",
  64         [TAG_MUSICBRAINZ_ALBUMARTISTID] = "MUSICBRAINZ_ALBUMARTISTID",
  65         [TAG_MUSICBRAINZ_TRACKID] = "MUSICBRAINZ_TRACKID",
  66 };
  67
  68 bool ignore_tag_items[TAG_NUM_OF_ITEM_TYPES];
  69
  70 enum tag_type
  71 tag_name_parse(const char *name)
  72 {
  73         assert(name != NULL);
  74
  75         for (unsigned i = 0; i < TAG_NUM_OF_ITEM_TYPES; ++i) {
  76                 assert(tag_item_names[i] != NULL);
  77
  78                 if (strcmp(name, tag_item_names[i]) == 0)
  79                         return (enum tag_type)i;
  80         }
  81
  82         return TAG_NUM_OF_ITEM_TYPES;
  83 }
  84
  85 enum tag_type
  86 tag_name_parse_i(const char *name)
  87 {
  88         assert(name != NULL);
  89
  90         for (unsigned i = 0; i < TAG_NUM_OF_ITEM_TYPES; ++i) {
  91                 assert(tag_item_names[i] != NULL);
  92
  93                 if (g_ascii_strcasecmp(name, tag_item_names[i]) == 0)
  94                         return (enum tag_type)i;
  95         }
  96
  97         return TAG_NUM_OF_ITEM_TYPES;
  98 }
  99
 100 static size_t items_size(const struct tag *tag)
 101 {
 102         return tag->num_items * sizeof(struct tag_item *);
 103 }
 104
 105 void tag_lib_init(void)
 106 {
 107         const char *value;
 108         int quit = 0;
 109         char *temp;
 110         char *s;
 111         char *c;
 112         enum tag_type type;
 113
 114         /* parse the "metadata_to_use" config parameter below */
 115
 116         /* ignore comments by default */
 117         ignore_tag_items[TAG_COMMENT] = true;
 118
 119         value = config_get_string(CONF_METADATA_TO_USE, NULL);
 120         if (value == NULL)
 121                 return;
 122
 123         memset(ignore_tag_items, true, TAG_NUM_OF_ITEM_TYPES);
 124
 125         if (0 == g_ascii_strcasecmp(value, "none"))
 126                 return;
 127
 128         temp = c = s = g_strdup(value);
 129         while (!quit) {
 130                 if (*s == ',' || *s == '\0') {
 131                         if (*s == '\0')
 132                                 quit = 1;
 133                         *s = '\0';
 134
 135                         c = g_strstrip(c);
 136                         if (*c == 0)
 137                                 continue;
 138
 139                         type = tag_name_parse_i(c);
 140                         if (type == TAG_NUM_OF_ITEM_TYPES)
 141                                 g_error("error parsing metadata item \"%s\"",
 142                                         c);
 143
 144                         ignore_tag_items[type] = false;
 145
 146                         s++;
 147                         c = s;
 148                 }
 149                 s++;
 150         }
 151
 152         g_free(temp);
 153 }
 154
 155 struct tag *tag_new(void)
 156 {
 157         struct tag *ret = g_new(struct tag, 1);
 158         ret->items = NULL;
 159         ret->time = -1;
 160         ret->num_items = 0;
 161         return ret;
 162 }
 163
 164 static void tag_delete_item(struct tag *tag, unsigned idx)
 165 {
 166         assert(idx < tag->num_items);
 167         tag->num_items--;
 168
 169         g_mutex_lock(tag_pool_lock);
 170         tag_pool_put_item(tag->items[idx]);
 171         g_mutex_unlock(tag_pool_lock);
 172
 173         if (tag->num_items - idx > 0) {
 174                 memmove(tag->items + idx, tag->items + idx + 1,
 175                         (tag->num_items - idx) * sizeof(tag->items[0]));
 176         }
 177
 178         if (tag->num_items > 0) {
 179                 tag->items = g_realloc(tag->items, items_size(tag));
 180         } else {
 181                 g_free(tag->items);
 182                 tag->items = NULL;
 183         }
 184 }
 185
 186 void tag_clear_items_by_type(struct tag *tag, enum tag_type type)
 187 {
 188         for (unsigned i = 0; i < tag->num_items; i++) {
 189                 if (tag->items[i]->type == type) {
 190                         tag_delete_item(tag, i);
 191                         /* decrement since when just deleted this node */
 192                         i--;
 193                 }
 194         }
 195 }
 196
 197 void tag_free(struct tag *tag)
 198 {
 199         int i;
 200
 201         assert(tag != NULL);
 202
 203         g_mutex_lock(tag_pool_lock);
 204         for (i = tag->num_items; --i >= 0; )
 205                 tag_pool_put_item(tag->items[i]);
 206         g_mutex_unlock(tag_pool_lock);
 207
 208         if (tag->items == bulk.items) {
 209 #ifndef NDEBUG
 210                 assert(bulk.busy);
 211                 bulk.busy = false;
 212 #endif
 213         } else
 214                 g_free(tag->items);
 215
 216         g_free(tag);
 217 }
 218
 219 struct tag *tag_dup(const struct tag *tag)
 220 {
 221         struct tag *ret;
 222
 223         if (!tag)
 224                 return NULL;
 225
 226         ret = tag_new();
 227         ret->time = tag->time;
 228         ret->num_items = tag->num_items;
 229         ret->items = ret->num_items > 0 ? g_malloc(items_size(tag)) : NULL;
 230
 231         g_mutex_lock(tag_pool_lock);
 232         for (unsigned i = 0; i < tag->num_items; i++)
 233                 ret->items[i] = tag_pool_dup_item(tag->items[i]);
 234         g_mutex_unlock(tag_pool_lock);
 235
 236         return ret;
 237 }
 238
 239 struct tag *
 240 tag_merge(const struct tag *base, const struct tag *add)
 241 {
 242         struct tag *ret;
 243         unsigned n;
 244
 245         assert(base != NULL);
 246         assert(add != NULL);
 247
 248         /* allocate new tag object */
 249
 250         ret = tag_new();
 251         ret->time = add->time > 0 ? add->time : base->time;
 252         ret->num_items = base->num_items + add->num_items;
 253         ret->items = ret->num_items > 0 ? g_malloc(items_size(ret)) : NULL;
 254
 255         g_mutex_lock(tag_pool_lock);
 256
 257         /* copy all items from "add" */
 258
 259         for (unsigned i = 0; i < add->num_items; ++i)
 260                 ret->items[i] = tag_pool_dup_item(add->items[i]);
 261
 262         n = add->num_items;
 263
 264         /* copy additional items from "base" */
 265
 266         for (unsigned i = 0; i < base->num_items; ++i)
 267                 if (!tag_has_type(add, base->items[i]->type))
 268                         ret->items[n++] = tag_pool_dup_item(base->items[i]);
 269
 270         g_mutex_unlock(tag_pool_lock);
 271
 272         assert(n <= ret->num_items);
 273
 274         if (n < ret->num_items) {
 275                 /* some tags were not copied - shrink ret->items */
 276                 assert(n > 0);
 277
 278                 ret->num_items = n;
 279                 ret->items = g_realloc(ret->items, items_size(ret));
 280         }
 281
 282         return ret;
 283 }
 284
 285 struct tag *
 286 tag_merge_replace(struct tag *base, struct tag *add)
 287 {
 288         if (add == NULL)
 289                 return base;
 290
 291         if (base == NULL)
 292                 return add;
 293
 294         struct tag *tag = tag_merge(base, add);
 295         tag_free(base);
 296         tag_free(add);
 297
 298         return tag;
 299 }
 300
 301 const char *
 302 tag_get_value(const struct tag *tag, enum tag_type type)
 303 {
 304         assert(tag != NULL);
 305         assert(type < TAG_NUM_OF_ITEM_TYPES);
 306
 307         for (unsigned i = 0; i < tag->num_items; i++)
 308                 if (tag->items[i]->type == type)
 309                         return tag->items[i]->value;
 310
 311         return NULL;
 312 }
 313
 314 bool tag_has_type(const struct tag *tag, enum tag_type type)
 315 {
 316         return tag_get_value(tag, type) != NULL;
 317 }
 318
 319 bool tag_equal(const struct tag *tag1, const struct tag *tag2)
 320 {
 321         if (tag1 == NULL && tag2 == NULL)
 322                 return true;
 323         else if (!tag1 || !tag2)
 324                 return false;
 325
 326         if (tag1->time != tag2->time)
 327                 return false;
 328
 329         if (tag1->num_items != tag2->num_items)
 330                 return false;
 331
 332         for (unsigned i = 0; i < tag1->num_items; i++) {
 333                 if (tag1->items[i]->type != tag2->items[i]->type)
 334                         return false;
 335                 if (strcmp(tag1->items[i]->value, tag2->items[i]->value)) {
 336                         return false;
 337                 }
 338         }
 339
 340         return true;
 341 }
 342
 343 /**
 344  * Replace invalid sequences with the question mark.
 345  */
 346 static char *
 347 patch_utf8(const char *src, size_t length, const gchar *end)
 348 {
 349         /* duplicate the string, and replace invalid bytes in that
 350            buffer */
 351         char *dest = g_strdup(src);
 352
 353         do {
 354                 dest[end - src] = '?';
 355         } while (!g_utf8_validate(end + 1, (src + length) - (end + 1), &end));
 356
 357         return dest;
 358 }
 359
 360 static char *
 361 fix_utf8(const char *str, size_t length)
 362 {
 363         const gchar *end;
 364         char *temp;
 365         gsize written;
 366
 367         assert(str != NULL);
 368
 369         /* check if the string is already valid UTF-8 */
 370         if (g_utf8_validate(str, length, &end))
 371                 return NULL;
 372
 373         /* no, it's not - try to import it from ISO-Latin-1 */
 374         temp = g_convert(str, length, "utf-8", "iso-8859-1",
 375                          NULL, &written, NULL);
 376         if (temp != NULL)
 377                 /* success! */
 378                 return temp;
 379
 380         /* no, still broken - there's no medication, just patch
 381            invalid sequences */
 382         return patch_utf8(str, length, end);
 383 }
 384
 385 void tag_begin_add(struct tag *tag)
 386 {
 387         assert(!bulk.busy);
 388         assert(tag != NULL);
 389         assert(tag->items == NULL);
 390         assert(tag->num_items == 0);
 391
 392 #ifndef NDEBUG
 393         bulk.busy = true;
 394 #endif
 395         tag->items = bulk.items;
 396 }
 397
 398 void tag_end_add(struct tag *tag)
 399 {
 400         if (tag->items == bulk.items) {
 401                 assert(tag->num_items <= BULK_MAX);
 402
 403                 if (tag->num_items > 0) {
 404                         /* copy the tag items from the bulk list over
 405                            to a new list (which fits exactly) */
 406                         tag->items = g_malloc(items_size(tag));
 407                         memcpy(tag->items, bulk.items, items_size(tag));
 408                 } else
 409                         tag->items = NULL;
 410         }
 411
 412 #ifndef NDEBUG
 413         bulk.busy = false;
 414 #endif
 415 }
 416
 417 static bool
 418 char_is_non_printable(unsigned char ch)
 419 {
 420         return ch < 0x20;
 421 }
 422
 423 static const char *
 424 find_non_printable(const char *p, size_t length)
 425 {
 426         for (size_t i = 0; i < length; ++i)
 427                 if (char_is_non_printable(p[i]))
 428                         return p + i;
 429
 430         return NULL;
 431 }
 432
 433 /**
 434  * Clears all non-printable characters, convert them to space.
 435  * Returns NULL if nothing needs to be cleared.
 436  */
 437 static char *
 438 clear_non_printable(const char *p, size_t length)
 439 {
 440         const char *first = find_non_printable(p, length);
 441         char *dest;
 442
 443         if (first == NULL)
 444                 return NULL;
 445
 446         dest = g_strndup(p, length);
 447
 448         for (size_t i = first - p; i < length; ++i)
 449                 if (char_is_non_printable(dest[i]))
 450                         dest[i] = ' ';
 451
 452         return dest;
 453 }
 454
 455 static char *
 456 fix_tag_value(const char *p, size_t length)
 457 {
 458         char *utf8, *cleared;
 459
 460         utf8 = fix_utf8(p, length);
 461         if (utf8 != NULL) {
 462                 p = utf8;
 463                 length = strlen(p);
 464         }
 465
 466         cleared = clear_non_printable(p, length);
 467         if (cleared == NULL)
 468                 cleared = utf8;
 469         else
 470                 g_free(utf8);
 471
 472         return cleared;
 473 }
 474
 475 static void
 476 tag_add_item_internal(struct tag *tag, enum tag_type type,
 477                       const char *value, size_t len)
 478 {
 479         unsigned int i = tag->num_items;
 480         char *p;
 481
 482         p = fix_tag_value(value, len);
 483         if (p != NULL) {
 484                 value = p;
 485                 len = strlen(value);
 486         }
 487
 488         tag->num_items++;
 489
 490         if (tag->items != bulk.items)
 491                 /* bulk mode disabled */
 492                 tag->items = g_realloc(tag->items, items_size(tag));
 493         else if (tag->num_items >= BULK_MAX) {
 494                 /* bulk list already full - switch back to non-bulk */
 495                 assert(bulk.busy);
 496
 497                 tag->items = g_malloc(items_size(tag));
 498                 memcpy(tag->items, bulk.items,
 499                        items_size(tag) - sizeof(struct tag_item *));
 500         }
 501
 502         g_mutex_lock(tag_pool_lock);
 503         tag->items[i] = tag_pool_get_item(type, value, len);
 504         g_mutex_unlock(tag_pool_lock);
 505
 506         g_free(p);
 507 }
 508
 509 void tag_add_item_n(struct tag *tag, enum tag_type type,
 510                     const char *value, size_t len)
 511 {
 512         if (ignore_tag_items[type])
 513         {
 514                 return;
 515         }
 516         if (!value || !len)
 517                 return;
 518
 519         tag_add_item_internal(tag, type, value, len);
 520 }