src/plugin_common/tags.c

   1 /* plugin_common - Routines common to several plugins
   2  * Copyright (C) 2002,2003,2004,2005,2006,2007,2008,2009  Josh Coalson
   3  *
   4  * This library is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU Lesser General Public
   6  * License as published by the Free Software Foundation; either
   7  * version 2.1 of the License, or (at your option) any later version.
   8  *
   9  * This library 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 GNU
  12  * Lesser General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU Lesser General Public
  15  * License along with this library; if not, write to the Free Software
  16  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
  17  */
  18
  19 #if HAVE_CONFIG_H
  20 #  include <config.h>
  21 #endif
  22
  23 #include <stdio.h>
  24 #include <string.h>
  25 #include <stdlib.h>
  26
  27 #include "tags.h"
  28 #include "FLAC/assert.h"
  29 #include "FLAC/metadata.h"
  30 #include "share/alloc.h"
  31
  32 #ifndef FLaC__INLINE
  33 #define FLaC__INLINE
  34 #endif
  35
  36
  37 static FLaC__INLINE size_t local__wide_strlen(const FLAC__uint16 *s)
  38 {
  39         size_t n = 0;
  40         while(*s++)
  41                 n++;
  42         return n;
  43 }
  44
  45 /*
  46  * also disallows non-shortest-form encodings, c.f.
  47  *   http://www.unicode.org/versions/corrigendum1.html
  48  * and a more clear explanation at the end of this section:
  49  *   http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
  50  */
  51 static size_t local__utf8len(const FLAC__byte *utf8)
  52 {
  53         FLAC__ASSERT(0 != utf8);
  54         if ((utf8[0] & 0x80) == 0) {
  55                 return 1;
  56         }
  57         else if ((utf8[0] & 0xE0) == 0xC0 && (utf8[1] & 0xC0) == 0x80) {
  58                 if ((utf8[0] & 0xFE) == 0xC0) /* overlong sequence check */
  59                         return 0;
  60                 return 2;
  61         }
  62         else if ((utf8[0] & 0xF0) == 0xE0 && (utf8[1] & 0xC0) == 0x80 && (utf8[2] & 0xC0) == 0x80) {
  63                 if (utf8[0] == 0xE0 && (utf8[1] & 0xE0) == 0x80) /* overlong sequence check */
  64                         return 0;
  65                 /* illegal surrogates check (U+D800...U+DFFF and U+FFFE...U+FFFF) */
  66                 if (utf8[0] == 0xED && (utf8[1] & 0xE0) == 0xA0) /* D800-DFFF */
  67                         return 0;
  68                 if (utf8[0] == 0xEF && utf8[1] == 0xBF && (utf8[2] & 0xFE) == 0xBE) /* FFFE-FFFF */
  69                         return 0;
  70                 return 3;
  71         }
  72         else if ((utf8[0] & 0xF8) == 0xF0 && (utf8[1] & 0xC0) == 0x80 && (utf8[2] & 0xC0) == 0x80 && (utf8[3] & 0xC0) == 0x80) {
  73                 if (utf8[0] == 0xF0 && (utf8[1] & 0xF0) == 0x80) /* overlong sequence check */
  74                         return 0;
  75                 return 4;
  76         }
  77         else if ((utf8[0] & 0xFC) == 0xF8 && (utf8[1] & 0xC0) == 0x80 && (utf8[2] & 0xC0) == 0x80 && (utf8[3] & 0xC0) == 0x80 && (utf8[4] & 0xC0) == 0x80) {
  78                 if (utf8[0] == 0xF8 && (utf8[1] & 0xF8) == 0x80) /* overlong sequence check */
  79                         return 0;
  80                 return 5;
  81         }
  82         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) {
  83                 if (utf8[0] == 0xFC && (utf8[1] & 0xFC) == 0x80) /* overlong sequence check */
  84                         return 0;
  85                 return 6;
  86         }
  87         else {
  88                 return 0;
  89         }
  90 }
  91
  92
  93 static size_t local__utf8_to_ucs2(const FLAC__byte *utf8, FLAC__uint16 *ucs2)
  94 {
  95         const size_t len = local__utf8len(utf8);
  96
  97         FLAC__ASSERT(0 != ucs2);
  98
  99         if (len == 1)
 100                 *ucs2 = *utf8;
 101         else if (len == 2)
 102                 *ucs2 = (*utf8 & 0x3F)<<6 | (*(utf8+1) & 0x3F);
 103         else if (len == 3)
 104                 *ucs2 = (*utf8 & 0x1F)<<12 | (*(utf8+1) & 0x3F)<<6 | (*(utf8+2) & 0x3F);
 105         else
 106                 *ucs2 = '?';
 107
 108         return len;
 109 }
 110
 111 static FLAC__uint16 *local__convert_utf8_to_ucs2(const char *src, unsigned length)
 112 {
 113         FLAC__uint16 *out;
 114         size_t chars = 0;
 115
 116         FLAC__ASSERT(0 != src);
 117
 118         /* calculate length */
 119         {
 120                 const unsigned char *s, *end;
 121                 for (s=(const unsigned char *)src, end=s+length; s<end; chars++) {
 122                         const unsigned n = local__utf8len(s);
 123                         if (n == 0)
 124                                 return 0;
 125                         s += n;
 126                 }
 127                 FLAC__ASSERT(s == end);
 128         }
 129
 130         /* allocate */
 131         out = safe_malloc_mul_2op_(chars, /*times*/sizeof(FLAC__uint16));
 132         if (0 == out) {
 133                 FLAC__ASSERT(0);
 134                 return 0;
 135         }
 136
 137         /* convert */
 138         {
 139                 const unsigned char *s = (const unsigned char *)src;
 140                 FLAC__uint16 *u = out;
 141                 for ( ; chars; chars--)
 142                         s += local__utf8_to_ucs2(s, u++);
 143         }
 144
 145         return out;
 146 }
 147
 148 static FLaC__INLINE size_t local__ucs2len(FLAC__uint16 ucs2)
 149 {
 150         if (ucs2 < 0x0080)
 151                 return 1;
 152         else if (ucs2 < 0x0800)
 153                 return 2;
 154         else
 155                 return 3;
 156 }
 157
 158 static size_t local__ucs2_to_utf8(FLAC__uint16 ucs2, FLAC__byte *utf8)
 159 {
 160         if (ucs2 < 0x080) {
 161                 utf8[0] = (FLAC__byte)ucs2;
 162                 return 1;
 163         }
 164         else if (ucs2 < 0x800) {
 165                 utf8[0] = 0xc0 | (ucs2 >> 6);
 166                 utf8[1] = 0x80 | (ucs2 & 0x3f);
 167                 return 2;
 168         }
 169         else {
 170                 utf8[0] = 0xe0 | (ucs2 >> 12);
 171                 utf8[1] = 0x80 | ((ucs2 >> 6) & 0x3f);
 172                 utf8[2] = 0x80 | (ucs2 & 0x3f);
 173                 return 3;
 174         }
 175 }
 176
 177 static char *local__convert_ucs2_to_utf8(const FLAC__uint16 *src, unsigned length)
 178 {
 179         char *out;
 180         size_t len = 0, n;
 181
 182         FLAC__ASSERT(0 != src);
 183
 184         /* calculate length */
 185         {
 186                 unsigned i;
 187                 for (i = 0; i < length; i++) {
 188                         n = local__ucs2len(src[i]);
 189                         if(len + n < len) /* overflow check */
 190                                 return 0;
 191                         len += n;
 192                 }
 193         }
 194
 195         /* allocate */
 196         out = safe_malloc_mul_2op_(len, /*times*/sizeof(char));
 197         if (0 == out)
 198                 return 0;
 199
 200         /* convert */
 201         {
 202                 unsigned char *u = (unsigned char *)out;
 203                 for ( ; *src; src++)
 204                         u += local__ucs2_to_utf8(*src, u);
 205                 local__ucs2_to_utf8(*src, u);
 206         }
 207
 208         return out;
 209 }
 210
 211
 212 FLAC__bool FLAC_plugin__tags_get(const char *filename, FLAC__StreamMetadata **tags)
 213 {
 214         if(!FLAC__metadata_get_tags(filename, tags))
 215                 if(0 == (*tags = FLAC__metadata_object_new(FLAC__METADATA_TYPE_VORBIS_COMMENT)))
 216                         return false;
 217         return true;
 218 }
 219
 220 FLAC__bool FLAC_plugin__tags_set(const char *filename, const FLAC__StreamMetadata *tags)
 221 {
 222         FLAC__Metadata_Chain *chain;
 223         FLAC__Metadata_Iterator *iterator;
 224         FLAC__StreamMetadata *block;
 225         FLAC__bool got_vorbis_comments = false;
 226         FLAC__bool ok;
 227
 228         if(0 == (chain = FLAC__metadata_chain_new()))
 229                 return false;
 230
 231         if(!FLAC__metadata_chain_read(chain, filename)) {
 232                 FLAC__metadata_chain_delete(chain);
 233                 return false;
 234         }
 235
 236         if(0 == (iterator = FLAC__metadata_iterator_new())) {
 237                 FLAC__metadata_chain_delete(chain);
 238                 return false;
 239         }
 240
 241         FLAC__metadata_iterator_init(iterator, chain);
 242
 243         do {
 244                 if(FLAC__metadata_iterator_get_block_type(iterator) == FLAC__METADATA_TYPE_VORBIS_COMMENT)
 245                         got_vorbis_comments = true;
 246         } while(!got_vorbis_comments && FLAC__metadata_iterator_next(iterator));
 247
 248         if(0 == (block = FLAC__metadata_object_clone(tags))) {
 249                 FLAC__metadata_chain_delete(chain);
 250                 FLAC__metadata_iterator_delete(iterator);
 251                 return false;
 252         }
 253
 254         if(got_vorbis_comments)
 255                 ok = FLAC__metadata_iterator_set_block(iterator, block);
 256         else
 257                 ok = FLAC__metadata_iterator_insert_block_after(iterator, block);
 258
 259         FLAC__metadata_iterator_delete(iterator);
 260
 261         if(ok) {
 262                 FLAC__metadata_chain_sort_padding(chain);
 263                 ok = FLAC__metadata_chain_write(chain, /*use_padding=*/true, /*preserve_file_stats=*/true);
 264         }
 265
 266         FLAC__metadata_chain_delete(chain);
 267
 268         return ok;
 269 }
 270
 271 void FLAC_plugin__tags_destroy(FLAC__StreamMetadata **tags)
 272 {
 273         FLAC__metadata_object_delete(*tags);
 274         *tags = 0;
 275 }
 276
 277 const char *FLAC_plugin__tags_get_tag_utf8(const FLAC__StreamMetadata *tags, const char *name)
 278 {
 279         const int i = FLAC__metadata_object_vorbiscomment_find_entry_from(tags, /*offset=*/0, name);
 280         return (i < 0? 0 : strchr((const char *)tags->data.vorbis_comment.comments[i].entry, '=')+1);
 281 }
 282
 283 FLAC__uint16 *FLAC_plugin__tags_get_tag_ucs2(const FLAC__StreamMetadata *tags, const char *name)
 284 {
 285         const char *utf8 = FLAC_plugin__tags_get_tag_utf8(tags, name);
 286         if(0 == utf8)
 287                 return 0;
 288         return local__convert_utf8_to_ucs2(utf8, strlen(utf8)+1); /* +1 for terminating null */
 289 }
 290
 291 int FLAC_plugin__tags_delete_tag(FLAC__StreamMetadata *tags, const char *name)
 292 {
 293         return FLAC__metadata_object_vorbiscomment_remove_entries_matching(tags, name);
 294 }
 295
 296 int FLAC_plugin__tags_delete_all(FLAC__StreamMetadata *tags)
 297 {
 298         int n = (int)tags->data.vorbis_comment.num_comments;
 299         if(n > 0) {
 300                 if(!FLAC__metadata_object_vorbiscomment_resize_comments(tags, 0))
 301                         n = -1;
 302         }
 303         return n;
 304 }
 305
 306 FLAC__bool FLAC_plugin__tags_add_tag_utf8(FLAC__StreamMetadata *tags, const char *name, const char *value, const char *separator)
 307 {
 308         int i;
 309
 310         FLAC__ASSERT(0 != tags);
 311         FLAC__ASSERT(0 != name);
 312         FLAC__ASSERT(0 != value);
 313
 314         if(separator && (i = FLAC__metadata_object_vorbiscomment_find_entry_from(tags, /*offset=*/0, name)) >= 0) {
 315                 FLAC__StreamMetadata_VorbisComment_Entry *entry = tags->data.vorbis_comment.comments+i;
 316                 const size_t value_len = strlen(value);
 317                 const size_t separator_len = strlen(separator);
 318                 FLAC__byte *new_entry;
 319                 if(0 == (new_entry = safe_realloc_add_4op_(entry->entry, entry->length, /*+*/value_len, /*+*/separator_len, /*+*/1)))
 320                         return false;
 321                 memcpy(new_entry+entry->length, separator, separator_len);
 322                 entry->length += separator_len;
 323                 memcpy(new_entry+entry->length, value, value_len);
 324                 entry->length += value_len;
 325                 new_entry[entry->length] = '\0';
 326                 entry->entry = new_entry;
 327         }
 328         else {
 329                 FLAC__StreamMetadata_VorbisComment_Entry entry;
 330                 if(!FLAC__metadata_object_vorbiscomment_entry_from_name_value_pair(&entry, name, value))
 331                         return false;
 332                 FLAC__metadata_object_vorbiscomment_append_comment(tags, entry, /*copy=*/false);
 333         }
 334         return true;
 335 }
 336
 337 FLAC__bool FLAC_plugin__tags_set_tag_ucs2(FLAC__StreamMetadata *tags, const char *name, const FLAC__uint16 *value, FLAC__bool replace_all)
 338 {
 339         FLAC__StreamMetadata_VorbisComment_Entry entry;
 340
 341         FLAC__ASSERT(0 != tags);
 342         FLAC__ASSERT(0 != name);
 343         FLAC__ASSERT(0 != value);
 344
 345         {
 346                 char *utf8 = local__convert_ucs2_to_utf8(value, local__wide_strlen(value)+1); /* +1 for the terminating null */
 347                 if(0 == utf8)
 348                         return false;
 349                 if(!FLAC__metadata_object_vorbiscomment_entry_from_name_value_pair(&entry, name, utf8)) {
 350                         free(utf8);
 351                         return false;
 352                 }
 353                 free(utf8);
 354         }
 355         if(!FLAC__metadata_object_vorbiscomment_replace_comment(tags, entry, replace_all, /*copy=*/false))
 356                 return false;
 357         return true;
 358 }