src/plugin_common/tags.c

   1 /* plugin_common - Routines common to several plugins
   2  * Copyright (C) 2002,2003,2004,2005,2006  Josh Coalson
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public License
   6  * as published by the Free Software Foundation; either version 2
   7  * of the License, or (at your option) any later version.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * GNU General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public License
  15  * along with this program; if not, write to the Free Software
  16  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
  17  */
  18
  19 #include <stdio.h>
  20 #include <string.h>
  21 #include <stdlib.h>
  22
  23 #include "tags.h"
  24 #include "FLAC/assert.h"
  25 #include "FLAC/metadata.h"
  26
  27
  28 static __inline unsigned local__wide_strlen(const FLAC__uint16 *s)
  29 {
  30         unsigned n = 0;
  31         while(*s++)
  32                 n++;
  33         return n;
  34 }
  35
  36 /*
  37  * also disallows non-shortest-form encodings, c.f.
  38  *   http://www.unicode.org/versions/corrigendum1.html
  39  * and a more clear explanation at the end of this section:
  40  *   http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
  41  */
  42 static __inline unsigned local__utf8len(const FLAC__byte *utf8)
  43 {
  44         FLAC__ASSERT(0 != utf8);
  45         if ((utf8[0] & 0x80) == 0) {
  46                 return 1;
  47         }
  48         else if ((utf8[0] & 0xE0) == 0xC0 && (utf8[1] & 0xC0) == 0x80) {
  49                 if ((utf8[0] & 0xFE) == 0xC0) /* overlong sequence check */
  50                         return 0;
  51                 return 2;
  52         }
  53         else if ((utf8[0] & 0xF0) == 0xE0 && (utf8[1] & 0xC0) == 0x80 && (utf8[2] & 0xC0) == 0x80) {
  54                 if (utf8[0] == 0xE0 && (utf8[1] & 0xE0) == 0x80) /* overlong sequence check */
  55                         return 0;
  56                 /* illegal surrogates check (U+D800...U+DFFF and U+FFFE...U+FFFF) */
  57                 if (utf8[0] == 0xED && (utf8[1] & 0xE0) == 0xA0) /* D800-DFFF */
  58                         return 0;
  59                 if (utf8[0] == 0xEF && utf8[1] == 0xBF && (utf8[2] & 0xFE) == 0xBE) /* FFFE-FFFF */
  60                         return 0;
  61                 return 3;
  62         }
  63         else if ((utf8[0] & 0xF8) == 0xF0 && (utf8[1] & 0xC0) == 0x80 && (utf8[2] & 0xC0) == 0x80 && (utf8[3] & 0xC0) == 0x80) {
  64                 if (utf8[0] == 0xF0 && (utf8[1] & 0xF0) == 0x80) /* overlong sequence check */
  65                         return 0;
  66                 return 4;
  67         }
  68         else if ((utf8[0] & 0xFC) == 0xF8 && (utf8[1] & 0xC0) == 0x80 && (utf8[2] & 0xC0) == 0x80 && (utf8[3] & 0xC0) == 0x80 && (utf8[4] & 0xC0) == 0x80) {
  69                 if (utf8[0] == 0xF8 && (utf8[1] & 0xF8) == 0x80) /* overlong sequence check */
  70                         return 0;
  71                 return 5;
  72         }
  73         else if ((utf8[0] & 0xFE) == 0xFC && (utf8[1] & 0xC0) == 0x80 && (utf8[2] & 0xC0) == 0x80 && (utf8[3] & 0xC0) == 0x80 && (utf8[4] & 0xC0) == 0x80 && (utf8[5] & 0xC0) == 0x80) {
  74                 if (utf8[0] == 0xFC && (utf8[1] & 0xFC) == 0x80) /* overlong sequence check */
  75                         return 0;
  76                 return 6;
  77         }
  78         else {
  79                 return 0;
  80         }
  81 }
  82
  83
  84 static __inline unsigned local__utf8_to_ucs2(const FLAC__byte *utf8, FLAC__uint16 *ucs2)
  85 {
  86         const unsigned len = local__utf8len(utf8);
  87
  88         FLAC__ASSERT(0 != ucs2);
  89
  90         if (len == 1)
  91                 *ucs2 = *utf8;
  92         else if (len == 2)
  93                 *ucs2 = (*utf8 & 0x3F)<<6 | (*(utf8+1) & 0x3F);
  94         else if (len == 3)
  95                 *ucs2 = (*utf8 & 0x1F)<<12 | (*(utf8+1) & 0x3F)<<6 | (*(utf8+2) & 0x3F);
  96         else
  97                 *ucs2 = '?';
  98
  99         return len;
 100 }
 101
 102 static FLAC__uint16 *local__convert_utf8_to_ucs2(const char *src, unsigned length)
 103 {
 104         FLAC__uint16 *out;
 105         unsigned chars = 0;
 106
 107         FLAC__ASSERT(0 != src);
 108
 109         /* calculate length */
 110         {
 111                 const unsigned char *s, *end;
 112                 for (s=(const unsigned char *)src, end=s+length; s<end; chars++) {
 113                         const unsigned n = local__utf8len(s);
 114                         if (n == 0)
 115                                 return 0;
 116                         s += n;
 117                 }
 118                 FLAC__ASSERT(s == end);
 119         }
 120
 121         /* allocate */
 122         out = (FLAC__uint16*)malloc(chars * sizeof(FLAC__uint16));
 123         if (0 == out) {
 124                 FLAC__ASSERT(0);
 125                 return 0;
 126         }
 127
 128         /* convert */
 129         {
 130                 const unsigned char *s = (const unsigned char *)src;
 131                 FLAC__uint16 *u = out;
 132                 for ( ; chars; chars--)
 133                         s += local__utf8_to_ucs2(s, u++);
 134         }
 135
 136         return out;
 137 }
 138
 139 static __inline unsigned local__ucs2len(FLAC__uint16 ucs2)
 140 {
 141         if (ucs2 < 0x0080)
 142                 return 1;
 143         else if (ucs2 < 0x0800)
 144                 return 2;
 145         else
 146                 return 3;
 147 }
 148
 149 static __inline unsigned local__ucs2_to_utf8(FLAC__uint16 ucs2, FLAC__byte *utf8)
 150 {
 151         if (ucs2 < 0x080) {
 152                 utf8[0] = (FLAC__byte)ucs2;
 153                 return 1;
 154         }
 155         else if (ucs2 < 0x800) {
 156                 utf8[0] = 0xc0 | (ucs2 >> 6);
 157                 utf8[1] = 0x80 | (ucs2 & 0x3f);
 158                 return 2;
 159         }
 160         else {
 161                 utf8[0] = 0xe0 | (ucs2 >> 12);
 162                 utf8[1] = 0x80 | ((ucs2 >> 6) & 0x3f);
 163                 utf8[2] = 0x80 | (ucs2 & 0x3f);
 164                 return 3;
 165         }
 166 }
 167
 168 static char *local__convert_ucs2_to_utf8(const FLAC__uint16 *src, unsigned length)
 169 {
 170         char *out;
 171         unsigned len = 0;
 172
 173         FLAC__ASSERT(0 != src);
 174
 175         /* calculate length */
 176         {
 177                 unsigned i;
 178                 for (i = 0; i < length; i++)
 179                         len += local__ucs2len(src[i]);
 180         }
 181
 182         /* allocate */
 183         out = (char*)malloc(len * sizeof(char));
 184         if (0 == out)
 185                 return 0;
 186
 187         /* convert */
 188         {
 189                 unsigned char *u = (unsigned char *)out;
 190                 for ( ; *src; src++)
 191                         u += local__ucs2_to_utf8(*src, u);
 192                 local__ucs2_to_utf8(*src, u);
 193         }
 194
 195         return out;
 196 }
 197
 198
 199 FLAC__bool FLAC_plugin__tags_get(const char *filename, FLAC__StreamMetadata **tags)
 200 {
 201         if(!FLAC__metadata_get_tags(filename, tags))
 202                 if(0 == (*tags = FLAC__metadata_object_new(FLAC__METADATA_TYPE_VORBIS_COMMENT)))
 203                         return false;
 204         return true;
 205 }
 206
 207 FLAC__bool FLAC_plugin__tags_set(const char *filename, const FLAC__StreamMetadata *tags)
 208 {
 209         FLAC__Metadata_Chain *chain;
 210         FLAC__Metadata_Iterator *iterator;
 211         FLAC__StreamMetadata *block;
 212         FLAC__bool got_vorbis_comments = false;
 213         FLAC__bool ok;
 214
 215         if(0 == (chain = FLAC__metadata_chain_new()))
 216                 return false;
 217
 218         if(!FLAC__metadata_chain_read(chain, filename)) {
 219                 FLAC__metadata_chain_delete(chain);
 220                 return false;
 221         }
 222
 223         if(0 == (iterator = FLAC__metadata_iterator_new())) {
 224                 FLAC__metadata_chain_delete(chain);
 225                 return false;
 226         }
 227
 228         FLAC__metadata_iterator_init(iterator, chain);
 229
 230         do {
 231                 if(FLAC__metadata_iterator_get_block_type(iterator) == FLAC__METADATA_TYPE_VORBIS_COMMENT)
 232                         got_vorbis_comments = true;
 233         } while(!got_vorbis_comments && FLAC__metadata_iterator_next(iterator));
 234
 235         if(0 == (block = FLAC__metadata_object_clone(tags))) {
 236                 FLAC__metadata_chain_delete(chain);
 237                 FLAC__metadata_iterator_delete(iterator);
 238                 return false;
 239         }
 240
 241         if(got_vorbis_comments)
 242                 ok = FLAC__metadata_iterator_set_block(iterator, block);
 243         else
 244                 ok = FLAC__metadata_iterator_insert_block_after(iterator, block);
 245
 246         FLAC__metadata_iterator_delete(iterator);
 247
 248         if(ok) {
 249                 FLAC__metadata_chain_sort_padding(chain);
 250                 ok = FLAC__metadata_chain_write(chain, /*use_padding=*/true, /*preserve_file_stats=*/true);
 251         }
 252
 253         FLAC__metadata_chain_delete(chain);
 254
 255         return ok;
 256 }
 257
 258 void FLAC_plugin__tags_destroy(FLAC__StreamMetadata **tags)
 259 {
 260         FLAC__metadata_object_delete(*tags);
 261         *tags = 0;
 262 }
 263
 264 const char *FLAC_plugin__tags_get_tag_utf8(const FLAC__StreamMetadata *tags, const char *name)
 265 {
 266         const int i = FLAC__metadata_object_vorbiscomment_find_entry_from(tags, /*offset=*/0, name);
 267         return (i < 0? 0 : strchr((const char *)tags->data.vorbis_comment.comments[i].entry, '=')+1);
 268 }
 269
 270 FLAC__uint16 *FLAC_plugin__tags_get_tag_ucs2(const FLAC__StreamMetadata *tags, const char *name)
 271 {
 272         const char *utf8 = FLAC_plugin__tags_get_tag_utf8(tags, name);
 273         if(0 == utf8)
 274                 return 0;
 275         return local__convert_utf8_to_ucs2(utf8, strlen(utf8)+1); /* +1 for terminating null */
 276 }
 277
 278 int FLAC_plugin__tags_delete_tag(FLAC__StreamMetadata *tags, const char *name)
 279 {
 280         return FLAC__metadata_object_vorbiscomment_remove_entries_matching(tags, name);
 281 }
 282
 283 int FLAC_plugin__tags_delete_all(FLAC__StreamMetadata *tags)
 284 {
 285         int n = (int)tags->data.vorbis_comment.num_comments;
 286         if(n > 0) {
 287                 if(!FLAC__metadata_object_vorbiscomment_resize_comments(tags, 0))
 288                         n = -1;
 289         }
 290         return n;
 291 }
 292
 293 FLAC__bool FLAC_plugin__tags_add_tag_utf8(FLAC__StreamMetadata *tags, const char *name, const char *value, const char *separator)
 294 {
 295         int i;
 296
 297         FLAC__ASSERT(0 != tags);
 298         FLAC__ASSERT(0 != name);
 299         FLAC__ASSERT(0 != value);
 300
 301         if(separator && (i = FLAC__metadata_object_vorbiscomment_find_entry_from(tags, /*offset=*/0, name)) >= 0) {
 302                 FLAC__StreamMetadata_VorbisComment_Entry *entry = tags->data.vorbis_comment.comments+i;
 303                 const size_t value_len = strlen(value);
 304                 const size_t separator_len = strlen(separator);
 305                 FLAC__byte *new_entry;
 306                 if(0 == (new_entry = (FLAC__byte*)realloc(entry->entry, entry->length + value_len + separator_len + 1)))
 307                         return false;
 308                 memcpy(new_entry+entry->length, separator, separator_len);
 309                 entry->length += separator_len;
 310                 memcpy(new_entry+entry->length, value, value_len);
 311                 entry->length += value_len;
 312                 new_entry[entry->length] = '\0';
 313                 entry->entry = new_entry;
 314         }
 315         else {
 316                 FLAC__StreamMetadata_VorbisComment_Entry entry;
 317                 if(!FLAC__metadata_object_vorbiscomment_entry_from_name_value_pair(&entry, name, value))
 318                         return false;
 319                 FLAC__metadata_object_vorbiscomment_append_comment(tags, entry, /*copy=*/false);
 320         }
 321         return true;
 322 }
 323
 324 FLAC__bool FLAC_plugin__tags_set_tag_ucs2(FLAC__StreamMetadata *tags, const char *name, const FLAC__uint16 *value, FLAC__bool replace_all)
 325 {
 326         FLAC__StreamMetadata_VorbisComment_Entry entry;
 327
 328         FLAC__ASSERT(0 != tags);
 329         FLAC__ASSERT(0 != name);
 330         FLAC__ASSERT(0 != value);
 331
 332         {
 333                 char *utf8 = local__convert_ucs2_to_utf8(value, local__wide_strlen(value)+1); /* +1 for the terminating null */
 334                 if(0 == utf8)
 335                         return false;
 336                 if(!FLAC__metadata_object_vorbiscomment_entry_from_name_value_pair(&entry, name, utf8)) {
 337                         free(utf8);
 338                         return false;
 339                 }
 340                 free(utf8);
 341         }
 342         if(!FLAC__metadata_object_vorbiscomment_replace_comment(tags, entry, replace_all, /*copy=*/false))
 343                 return false;
 344         return true;
 345 }