| 1 |
/* |
| 2 |
id3: ID3v2.3 and ID3v2.4 parsing (a relevant subset) |
| 3 |
|
| 4 |
copyright 2006-2023 by the mpg123 project - free software under the terms of the LGPL 2.1 |
| 5 |
see COPYING and AUTHORS files in distribution or http://mpg123.org |
| 6 |
initially written by Thomas Orgis |
| 7 |
|
| 8 |
WIP: Handling of multiple ID3 tags in a stream. |
| 9 |
|
| 10 |
1. With update flag: Add non-unique data, replace unique. |
| 11 |
- Only one TALB, TPE1, etc. |
| 12 |
- Only one TXXX with a certain description. |
| 13 |
- Only one COMM with certain language and description. |
| 14 |
- Only one APIC with certain type and description, generally only one |
| 15 |
of type 1 and 2 each. |
| 16 |
2. Without update flag: wipe whole data and only store new stuff. |
| 17 |
|
| 18 |
BIG BAD BUT: How to properly handle seeks in a stream that make |
| 19 |
the parser encounter the same tags again in random order? Is |
| 20 |
there even a correct way to handle that without storing an |
| 21 |
ordered list of all tags? I could simplify the code here and just |
| 22 |
communicate that a frame should be an update to previous, and |
| 23 |
at which stream position the frame was encountered. But since |
| 24 |
libmpg123 is driven by MPEG frames, there could be multiple |
| 25 |
ID3v2 tags in direct succession treated by the parser without |
| 26 |
the library user being able to interfere. |
| 27 |
|
| 28 |
This is severely fucked. All that complexity also doesn't matter |
| 29 |
in practice, as streams use ICY and individual files have just one |
| 30 |
ID3v2 tag (relevant for libmpg123). It's an academic problem. But |
| 31 |
for seekable files, I could implement some jumping logic to find |
| 32 |
and parse all ID3v2 for once and then set a flag that only jumps |
| 33 |
the frames on seeks. That covers all local disk playback. For |
| 34 |
streams, seeking is no issue (seeking back, at least), so the |
| 35 |
update/replace logic works. |
| 36 |
|
| 37 |
Look at the standard: |
| 38 |
|
| 39 |
------ |
| 40 |
5. Tag location |
| 41 |
|
| 42 |
The default location of an ID3v2 tag is prepended to the audio so |
| 43 |
that players can benefit from the information when the data is |
| 44 |
streamed. It is however possible to append the tag, or make a |
| 45 |
prepend/append combination. When deciding upon where an unembedded |
| 46 |
tag should be located, the following order of preference SHOULD be |
| 47 |
considered. |
| 48 |
|
| 49 |
1. Prepend the tag. |
| 50 |
|
| 51 |
2. Prepend a tag with all vital information and add a second tag at |
| 52 |
the end of the file, before tags from other tagging systems. The |
| 53 |
first tag is required to have a SEEK frame. |
| 54 |
|
| 55 |
3. Add a tag at the end of the file, before tags from other tagging |
| 56 |
systems. |
| 57 |
|
| 58 |
In case 2 and 3 the tag can simply be appended if no other known tags |
| 59 |
are present. The suggested method to find ID3v2 tags are: |
| 60 |
|
| 61 |
1. Look for a prepended tag using the pattern found in section 3.1. |
| 62 |
|
| 63 |
2. If a SEEK frame was found, use its values to guide further |
| 64 |
searching. |
| 65 |
|
| 66 |
3. Look for a tag footer, scanning from the back of the file. |
| 67 |
|
| 68 |
For every new tag that is found, the old tag should be discarded |
| 69 |
unless the update flag in the extended header (section 3.2) is set. |
| 70 |
------ |
| 71 |
|
| 72 |
For seekable streams, I simply need to implement explicit ID3v2 search along |
| 73 |
that recommendation and keep the complete information. Streams that continue |
| 74 |
growing during playback will not recognize added ID3v2 tags. So be it. |
| 75 |
For non-seekable streams, a tag is always parsed when encountered, assuming |
| 76 |
the order of update tags always matches. |
| 77 |
|
| 78 |
First step for the 1.26 release shall be the implementaton of the update |
| 79 |
logic and glossing over the theoretical problem of re-parsing update |
| 80 |
frames in the wrong order by ignoring it. They are not that relevant. |
| 81 |
|
| 82 |
TODO: Cave in and add the missing frames from the spec. Not that far to go. |
| 83 |
But need another data structure to communicate those ... |
| 84 |
*/ |
| 85 |
|
| 86 |
#include "mpg123lib_intern.h" |
| 87 |
#include "id3.h" |
| 88 |
#include "../common/debug.h" |
| 89 |
|
| 90 |
#ifndef NO_ID3V2 /* Only the main parsing routine will always be there. */ |
| 91 |
|
| 92 |
/* We know the usual text frames plus some specifics. */ |
| 93 |
#define KNOWN_FRAMES 5 |
| 94 |
static const char frame_type[KNOWN_FRAMES][5] = { "COMM", "TXXX", "RVA2", "USLT", "APIC" }; |
| 95 |
enum frame_types { unknown = -2, text = -1, comment, extra, rva2, uslt, picture }; |
| 96 |
|
| 97 |
/* UTF support definitions */ |
| 98 |
|
| 99 |
typedef void (*text_converter)(mpg123_string *sb, const unsigned char* source, size_t len, const int noquiet); |
| 100 |
|
| 101 |
static void convert_latin1 (mpg123_string *sb, const unsigned char* source, size_t len, const int noquiet); |
| 102 |
static void convert_utf16bom(mpg123_string *sb, const unsigned char* source, size_t len, const int noquiet); |
| 103 |
static void convert_utf8 (mpg123_string *sb, const unsigned char* source, size_t len, const int noquiet); |
| 104 |
|
| 105 |
static const text_converter text_converters[4] = |
| 106 |
{ |
| 107 |
convert_latin1, |
| 108 |
/* We always check for (multiple) BOM in 16bit unicode. Without BOM, UTF16 BE is the default. |
| 109 |
Errors in encoding are detected anyway. */ |
| 110 |
convert_utf16bom, |
| 111 |
convert_utf16bom, |
| 112 |
convert_utf8 |
| 113 |
}; |
| 114 |
|
| 115 |
static const unsigned int encoding_widths[4] = { 1, 2, 2, 1 }; |
| 116 |
|
| 117 |
/* the code starts here... */ |
| 118 |
|
| 119 |
static void null_id3_links(mpg123_handle *fr) |
| 120 |
{ |
| 121 |
fr->id3v2.title = NULL; |
| 122 |
fr->id3v2.artist = NULL; |
| 123 |
fr->id3v2.album = NULL; |
| 124 |
fr->id3v2.year = NULL; |
| 125 |
fr->id3v2.genre = NULL; |
| 126 |
fr->id3v2.comment = NULL; |
| 127 |
} |
| 128 |
|
| 129 |
void INT123_init_id3(mpg123_handle *fr) |
| 130 |
{ |
| 131 |
fr->id3v2.version = 0; /* nothing there */ |
| 132 |
null_id3_links(fr); |
| 133 |
fr->id3v2.comments = 0; |
| 134 |
fr->id3v2.comment_list = NULL; |
| 135 |
fr->id3v2.texts = 0; |
| 136 |
fr->id3v2.text = NULL; |
| 137 |
fr->id3v2.extras = 0; |
| 138 |
fr->id3v2.extra = NULL; |
| 139 |
fr->id3v2.pictures = 0; |
| 140 |
fr->id3v2.picture = NULL; |
| 141 |
} |
| 142 |
|
| 143 |
/* Managing of the text, comment and extra lists. */ |
| 144 |
|
| 145 |
/* Initialize one element. */ |
| 146 |
static void init_mpg123_text(mpg123_text *txt) |
| 147 |
{ |
| 148 |
mpg123_init_string(&txt->text); |
| 149 |
mpg123_init_string(&txt->description); |
| 150 |
txt->id[0] = 0; |
| 151 |
txt->id[1] = 0; |
| 152 |
txt->id[2] = 0; |
| 153 |
txt->id[3] = 0; |
| 154 |
txt->lang[0] = 0; |
| 155 |
txt->lang[1] = 0; |
| 156 |
txt->lang[2] = 0; |
| 157 |
} |
| 158 |
|
| 159 |
static void init_mpg123_picture(mpg123_picture *pic) |
| 160 |
{ |
| 161 |
mpg123_init_string(&pic->mime_type); |
| 162 |
mpg123_init_string(&pic->description); |
| 163 |
pic->type = 0; |
| 164 |
pic->size = 0; |
| 165 |
pic->data = NULL; |
| 166 |
} |
| 167 |
|
| 168 |
/* Free memory of one element. */ |
| 169 |
static void free_mpg123_text(mpg123_text *txt) |
| 170 |
{ |
| 171 |
mpg123_free_string(&txt->text); |
| 172 |
mpg123_free_string(&txt->description); |
| 173 |
} |
| 174 |
|
| 175 |
static void free_mpg123_picture(mpg123_picture * pic) |
| 176 |
{ |
| 177 |
mpg123_free_string(&pic->mime_type); |
| 178 |
mpg123_free_string(&pic->description); |
| 179 |
if (pic->data != NULL) |
| 180 |
free(pic->data); |
| 181 |
} |
| 182 |
|
| 183 |
/* Free memory of whole list. */ |
| 184 |
#define free_comment(mh) free_id3_text(&((mh)->id3v2.comment_list), &((mh)->id3v2.comments)) |
| 185 |
#define free_text(mh) free_id3_text(&((mh)->id3v2.text), &((mh)->id3v2.texts)) |
| 186 |
#define free_extra(mh) free_id3_text(&((mh)->id3v2.extra), &((mh)->id3v2.extras)) |
| 187 |
#define free_picture(mh) free_id3_picture(&((mh)->id3v2.picture), &((mh)->id3v2.pictures)) |
| 188 |
static void free_id3_text(mpg123_text **list, size_t *size) |
| 189 |
{ |
| 190 |
size_t i; |
| 191 |
for(i=0; i<*size; ++i) free_mpg123_text(&((*list)[i])); |
| 192 |
|
| 193 |
free(*list); |
| 194 |
*list = NULL; |
| 195 |
*size = 0; |
| 196 |
} |
| 197 |
static void free_id3_picture(mpg123_picture **list, size_t *size) |
| 198 |
{ |
| 199 |
size_t i; |
| 200 |
for(i=0; i<*size; ++i) free_mpg123_picture(&((*list)[i])); |
| 201 |
|
| 202 |
free(*list); |
| 203 |
*list = NULL; |
| 204 |
*size = 0; |
| 205 |
} |
| 206 |
|
| 207 |
/* Add items to the list. */ |
| 208 |
|
| 209 |
#define add_comment(mh, l, d) add_id3_text(&((mh)->id3v2.comment_list), &((mh)->id3v2.comments), NULL, l, d) |
| 210 |
#define add_text(mh, id) add_id3_text(&((mh)->id3v2.text), &((mh)->id3v2.texts), id, NULL, NULL) |
| 211 |
#define add_uslt(mh, l, d) add_id3_text(&((mh)->id3v2.text), &((mh)->id3v2.texts), id, l, d) |
| 212 |
#define add_extra(mh, d) add_id3_text(&((mh)->id3v2.extra), &((mh)->id3v2.extras), NULL, NULL, d) |
| 213 |
#define add_picture(mh, t, d) add_id3_picture(&((mh)->id3v2.picture), &((mh)->id3v2.pictures), t, d) |
| 214 |
static mpg123_text *add_id3_text( mpg123_text **list, size_t *size |
| 215 |
, char id[4], char lang[3], mpg123_string *description ) |
| 216 |
{ |
| 217 |
mdebug( "add_id3_text id=%s lang=%s, desc=%s" |
| 218 |
, id ? (char[5]) { id[0], id[1], id[2], id[3], 0 } : "(nil)" |
| 219 |
, lang ? (char[4]) { lang[0], lang[1], lang[2], 0 } : "(nil)" |
| 220 |
, description ? (description->fill ? description->p : "(empty)") : "(nil)" ); |
| 221 |
if(lang && !description) |
| 222 |
return NULL; // no lone language intended |
| 223 |
if(id || description) |
| 224 |
{ |
| 225 |
// Look through list of existing texts and return an existing entry |
| 226 |
// if it should be overwritten. |
| 227 |
for(size_t i=0; i<*size; ++i) |
| 228 |
{ |
| 229 |
mpg123_text *entry = *list+i; |
| 230 |
if(description) |
| 231 |
{ // Overwrite entry with same description and same ID and language. |
| 232 |
if( (!id || !memcmp(id, entry->id, 4)) |
| 233 |
&& (!lang || !memcmp(entry->lang, lang, 3)) |
| 234 |
&& mpg123_same_string(&(entry->description), description) |
| 235 |
) |
| 236 |
return entry; |
| 237 |
} else if(id && !memcmp(id, entry->id, 4)) |
| 238 |
return entry; // Just ovewrite because of same ID. |
| 239 |
mdebug("add_id3_text: entry %zu was no match", i); |
| 240 |
} |
| 241 |
} |
| 242 |
mdebug("add_id3_text: append to list of %zu", *size); |
| 243 |
// Nothing found, add new one. |
| 244 |
mpg123_text *x = INT123_safe_realloc(*list, sizeof(mpg123_text)*(*size+1)); |
| 245 |
if(x == NULL) return NULL; /* bad */ |
| 246 |
|
| 247 |
*list = x; |
| 248 |
*size += 1; |
| 249 |
init_mpg123_text(&((*list)[*size-1])); |
| 250 |
|
| 251 |
return &((*list)[*size-1]); /* Return pointer to the added text. */ |
| 252 |
} |
| 253 |
|
| 254 |
|
| 255 |
static mpg123_picture *add_id3_picture(mpg123_picture **list, size_t *size, char type, mpg123_string *description) |
| 256 |
{ |
| 257 |
if(!description) |
| 258 |
return NULL; |
| 259 |
|
| 260 |
// Return entry to overwrite, if appropriate. |
| 261 |
for(size_t i=0; i<*size; ++i) |
| 262 |
{ |
| 263 |
mpg123_picture *entry = *list+i; |
| 264 |
if( type == entry->type |
| 265 |
&& ( type == 1 || type == 2 || |
| 266 |
mpg123_same_string(&entry->description, description) |
| 267 |
) |
| 268 |
) |
| 269 |
return entry; |
| 270 |
} |
| 271 |
// Append a new one. |
| 272 |
mpg123_picture *x = INT123_safe_realloc(*list, sizeof(mpg123_picture)*(*size+1)); |
| 273 |
if(x == NULL) return NULL; /* bad */ |
| 274 |
|
| 275 |
*list = x; |
| 276 |
*size += 1; |
| 277 |
init_mpg123_picture(&((*list)[*size-1])); |
| 278 |
|
| 279 |
return &((*list)[*size-1]); /* Return pointer to the added picture. */ |
| 280 |
} |
| 281 |
|
| 282 |
/* OK, back to the higher level functions. */ |
| 283 |
|
| 284 |
void INT123_exit_id3(mpg123_handle *fr) |
| 285 |
{ |
| 286 |
free_picture(fr); |
| 287 |
free_comment(fr); |
| 288 |
free_extra(fr); |
| 289 |
free_text(fr); |
| 290 |
} |
| 291 |
|
| 292 |
void INT123_reset_id3(mpg123_handle *fr) |
| 293 |
{ |
| 294 |
INT123_exit_id3(fr); |
| 295 |
INT123_init_id3(fr); |
| 296 |
} |
| 297 |
|
| 298 |
/* Set the id3v2.artist id3v2.title ... links to elements of the array. */ |
| 299 |
void INT123_id3_link(mpg123_handle *fr) |
| 300 |
{ |
| 301 |
size_t i; |
| 302 |
mpg123_id3v2 *v2 = &fr->id3v2; |
| 303 |
debug("linking ID3v2"); |
| 304 |
null_id3_links(fr); |
| 305 |
for(i=0; i<v2->texts; ++i) |
| 306 |
{ |
| 307 |
mpg123_text *entry = &v2->text[i]; |
| 308 |
if (!strncmp("TIT2", entry->id, 4)) v2->title = &entry->text; |
| 309 |
else if(!strncmp("TALB", entry->id, 4)) v2->album = &entry->text; |
| 310 |
else if(!strncmp("TPE1", entry->id, 4)) v2->artist = &entry->text; |
| 311 |
else if(!strncmp("TYER", entry->id, 4)) v2->year = &entry->text; |
| 312 |
else if(!strncmp("TCON", entry->id, 4)) v2->genre = &entry->text; |
| 313 |
} |
| 314 |
for(i=0; i<v2->comments; ++i) |
| 315 |
{ |
| 316 |
mpg123_text *entry = &v2->comment_list[i]; |
| 317 |
if(entry->description.fill == 0 || entry->description.p[0] == 0) |
| 318 |
v2->comment = &entry->text; |
| 319 |
} |
| 320 |
/* When no generic comment found, use the last non-generic one. */ |
| 321 |
if(v2->comment == NULL && v2->comments > 0) |
| 322 |
v2->comment = &v2->comment_list[v2->comments-1].text; |
| 323 |
} |
| 324 |
|
| 325 |
/* |
| 326 |
Store ID3 text data in an mpg123_string; either verbatim copy or |
| 327 |
everything translated to UTF-8 encoding. |
| 328 |
Preserve the zero string separator (I don't need strlen for the total size). |
| 329 |
|
| 330 |
Since we can overwrite strings with ID3 update frames, don't free |
| 331 |
memory, just grow strings. |
| 332 |
*/ |
| 333 |
static void store_id3_text(mpg123_string *sb, unsigned char encoding, unsigned char *source |
| 334 |
, size_t source_size, const int noquiet, const int notranslate) |
| 335 |
{ |
| 336 |
if(sb) // Always overwrite, even with nothing. |
| 337 |
sb->fill = 0; |
| 338 |
if(!source_size) |
| 339 |
{ |
| 340 |
debug("Empty id3 data!"); |
| 341 |
return; |
| 342 |
} |
| 343 |
|
| 344 |
/* We shall just copy the data. Client wants to decode itself. */ |
| 345 |
if(notranslate) |
| 346 |
{ |
| 347 |
/* Future: Add a path for ID3 errors. */ |
| 348 |
if(!mpg123_grow_string(sb, source_size+1)) |
| 349 |
{ |
| 350 |
if(noquiet) error("Cannot resize target string, out of memory?"); |
| 351 |
return; |
| 352 |
} |
| 353 |
sb->p[0] = (char)encoding; |
| 354 |
memcpy(sb->p+1, source, source_size); |
| 355 |
sb->fill = source_size+1; |
| 356 |
debug1("stored undecoded ID3 text of size %zu", source_size+1); |
| 357 |
return; |
| 358 |
} |
| 359 |
|
| 360 |
if(encoding > mpg123_id3_enc_max) |
| 361 |
{ |
| 362 |
if(noquiet) |
| 363 |
error1("Unknown text encoding %u, I take no chances, sorry!", encoding); |
| 364 |
return; |
| 365 |
} |
| 366 |
INT123_id3_to_utf8(sb, encoding, source, source_size, noquiet); |
| 367 |
|
| 368 |
if(sb->fill) debug1("UTF-8 string (the first one): %s", sb->p); |
| 369 |
else if(noquiet) error("unable to convert string to UTF-8 (out of memory, junk input?)!"); |
| 370 |
} |
| 371 |
|
| 372 |
/* On error, sb->size is 0. */ |
| 373 |
/* Also, encoding has been checked already! */ |
| 374 |
void INT123_id3_to_utf8(mpg123_string *sb, unsigned char encoding, const unsigned char *source, size_t source_size, int noquiet) |
| 375 |
{ |
| 376 |
unsigned int bwidth; |
| 377 |
if(sb) |
| 378 |
sb->fill = 0; |
| 379 |
debug1("encoding: %u", encoding); |
| 380 |
/* A note: ID3v2.3 uses UCS-2 non-variable 16bit encoding, v2.4 uses UTF16. |
| 381 |
UTF-16 uses a reserved/private range in UCS-2 to add the magic, so we just always treat it as UTF. */ |
| 382 |
bwidth = encoding_widths[encoding]; |
| 383 |
/* Hack! I've seen a stray zero byte before BOM. Is that supposed to happen? */ |
| 384 |
if(encoding != mpg123_id3_utf16be) /* UTF16be _can_ beging with a null byte! */ |
| 385 |
while(source_size > bwidth && source[0] == 0) |
| 386 |
{ |
| 387 |
--source_size; |
| 388 |
++source; |
| 389 |
debug("skipped leading zero"); |
| 390 |
} |
| 391 |
if(source_size % bwidth) |
| 392 |
{ |
| 393 |
/* When we need two bytes for a character, it's strange to have an uneven bytestream length. */ |
| 394 |
if(noquiet) warning2("Weird tag size %d for encoding %u - I will probably trim too early or something but I think the MP3 is broken.", (int)source_size, encoding); |
| 395 |
source_size -= source_size % bwidth; |
| 396 |
} |
| 397 |
text_converters[encoding](sb, source, source_size, noquiet); |
| 398 |
} |
| 399 |
|
| 400 |
/* You have checked encoding to be in the range already. */ |
| 401 |
static unsigned char *next_text(unsigned char* prev, unsigned char encoding, size_t limit) |
| 402 |
{ |
| 403 |
unsigned char *text = prev; |
| 404 |
size_t width = encoding_widths[encoding]; |
| 405 |
|
| 406 |
if(limit > PTRDIFF_MAX) |
| 407 |
return NULL; |
| 408 |
|
| 409 |
/* So I go lengths to find zero or double zero... |
| 410 |
Remember bug 2834636: Only check for aligned NULLs! */ |
| 411 |
while(text-prev < (ptrdiff_t)limit) |
| 412 |
{ |
| 413 |
if(text[0] == 0) |
| 414 |
{ |
| 415 |
if(width <= limit-(text-prev)) |
| 416 |
{ |
| 417 |
size_t i = 1; |
| 418 |
for(; i<width; ++i) if(text[i] != 0) break; |
| 419 |
|
| 420 |
if(i == width) /* found a null wide enough! */ |
| 421 |
{ |
| 422 |
text += width; |
| 423 |
break; |
| 424 |
} |
| 425 |
} |
| 426 |
else return NULL; /* No full character left? This text is broken */ |
| 427 |
} |
| 428 |
|
| 429 |
text += width; |
| 430 |
} |
| 431 |
if((size_t)(text-prev) >= limit) text = NULL; |
| 432 |
|
| 433 |
return text; |
| 434 |
} |
| 435 |
|
| 436 |
static const char *enc_name(unsigned char enc) |
| 437 |
{ |
| 438 |
switch(enc) |
| 439 |
{ |
| 440 |
case 0: return "Latin 1"; |
| 441 |
case 1: return "UTF-16 BOM"; |
| 442 |
case 2: return "UTF-16 BE"; |
| 443 |
case 3: return "UTF-8"; |
| 444 |
default: return "unknown!"; |
| 445 |
} |
| 446 |
} |
| 447 |
|
| 448 |
static void process_text(mpg123_handle *fr, unsigned char *realdata, size_t realsize, char *id) |
| 449 |
{ |
| 450 |
/* Text encoding $xx */ |
| 451 |
/* The text (encoded) ... */ |
| 452 |
if(realsize < 1) |
| 453 |
{ |
| 454 |
if(NOQUIET) error("Not even an encoding byte?"); |
| 455 |
return; |
| 456 |
} |
| 457 |
mpg123_text *t = add_text(fr, id); |
| 458 |
if(VERBOSE4) fprintf(stderr, "Note: Storing text from %s encoding\n", enc_name(realdata[0])); |
| 459 |
if(t == NULL) |
| 460 |
{ |
| 461 |
if(NOQUIET) error("Unable to attach new text!"); |
| 462 |
return; |
| 463 |
} |
| 464 |
mdebug("process_text: (over)writing entry with ID %s", t->id[0] |
| 465 |
? (char[5]) { t->id[0], t->id[1], t->id[2], t->id[3], 0 } |
| 466 |
: "(nil)" ); |
| 467 |
memcpy(t->id, id, 4); |
| 468 |
store_id3_text( &t->text, realdata[0], realdata+1, realsize-1 |
| 469 |
, NOQUIET, fr->p.flags & MPG123_PLAIN_ID3TEXT ); |
| 470 |
if(VERBOSE4) // Do not print unsanitized text to terminals! |
| 471 |
fprintf(stderr, "Note: ID3v2 %c%c%c%c text frame stored\n", id[0], id[1], id[2], id[3]); |
| 472 |
} |
| 473 |
|
| 474 |
static void process_picture(mpg123_handle *fr, unsigned char *realdata, size_t realsize) |
| 475 |
{ |
| 476 |
unsigned char encoding; |
| 477 |
mpg123_picture *i = NULL; |
| 478 |
unsigned char* workpoint = NULL; |
| 479 |
mpg123_string mime; mpg123_init_string(&mime); |
| 480 |
unsigned char image_type = 0; |
| 481 |
mpg123_string description; mpg123_init_string(&description); |
| 482 |
unsigned char *image_data = NULL; |
| 483 |
if(realsize < 1) |
| 484 |
{ |
| 485 |
debug("Empty id3 data!"); |
| 486 |
return; |
| 487 |
} |
| 488 |
encoding = realdata[0]; |
| 489 |
realdata++; realsize--; |
| 490 |
if(encoding > mpg123_id3_enc_max) |
| 491 |
{ |
| 492 |
if(NOQUIET) |
| 493 |
error1("Unknown text encoding %u, I take no chances, sorry!", encoding); |
| 494 |
return; |
| 495 |
} |
| 496 |
if(VERBOSE4) fprintf(stderr, "Note: Storing picture from APIC frame.\n"); |
| 497 |
|
| 498 |
/* get mime type (encoding is always latin-1) */ |
| 499 |
workpoint = next_text(realdata, 0, realsize); |
| 500 |
if(!workpoint) |
| 501 |
{ |
| 502 |
if(NOQUIET) |
| 503 |
error("Unable to get mime type for picture; skipping picture."); |
| 504 |
return; |
| 505 |
} |
| 506 |
INT123_id3_to_utf8(&mime, 0, realdata, workpoint - realdata, NOQUIET); |
| 507 |
realsize -= workpoint - realdata; |
| 508 |
realdata = workpoint; |
| 509 |
/* get picture type */ |
| 510 |
image_type = realdata[0]; |
| 511 |
realdata++; realsize--; |
| 512 |
/* get description (encoding is encoding) */ |
| 513 |
workpoint = next_text(realdata, encoding, realsize); |
| 514 |
if(!workpoint) |
| 515 |
{ |
| 516 |
if(NOQUIET) |
| 517 |
error("Unable to get description for picture; skipping picture."); |
| 518 |
mpg123_free_string(&mime); |
| 519 |
return; |
| 520 |
} |
| 521 |
INT123_id3_to_utf8(&description, encoding, realdata, workpoint - realdata, NOQUIET); |
| 522 |
realsize -= workpoint - realdata; |
| 523 |
if(realsize) |
| 524 |
image_data = (unsigned char*)malloc(realsize); |
| 525 |
if(!realsize || !image_data) { |
| 526 |
if(NOQUIET) |
| 527 |
error("No picture data or malloc failure; skipping picture."); |
| 528 |
mpg123_free_string(&description); |
| 529 |
mpg123_free_string(&mime); |
| 530 |
return; |
| 531 |
} |
| 532 |
memcpy(image_data, workpoint, realsize); |
| 533 |
|
| 534 |
// All data ready now, append to/replace in list. |
| 535 |
i = add_picture(fr, image_type, &description); |
| 536 |
if(!i) |
| 537 |
{ |
| 538 |
if(NOQUIET) |
| 539 |
error("Unable to attach new picture!"); |
| 540 |
free(image_data); |
| 541 |
mpg123_free_string(&description); |
| 542 |
mpg123_free_string(&mime); |
| 543 |
return; |
| 544 |
} |
| 545 |
|
| 546 |
// Either this is a fresh image, or one to be replaced. |
| 547 |
// We hand over memory, so old storage needs to be freed. |
| 548 |
free_mpg123_picture(i); |
| 549 |
i->type = image_type; |
| 550 |
i->size = realsize; |
| 551 |
i->data = image_data; |
| 552 |
mpg123_move_string(&mime, &i->mime_type); |
| 553 |
mpg123_move_string(&description, &i->description); |
| 554 |
if(VERBOSE4) |
| 555 |
fprintf(stderr, "Note: ID3v2 APIC picture frame of type: %d\n", i->type); |
| 556 |
} |
| 557 |
|
| 558 |
/* Store a new comment that perhaps is a RVA / RVA_ALBUM/AUDIOPHILE / RVA_MIX/RADIO one |
| 559 |
Special gimmik: It also stores USLT to the texts. Stucture is the same as for comments. */ |
| 560 |
static void process_comment(mpg123_handle *fr, enum frame_types tt, unsigned char *realdata, size_t realsize, int rva_level, char *id) |
| 561 |
{ |
| 562 |
/* Text encoding $xx */ |
| 563 |
/* Language $xx xx xx */ |
| 564 |
/* Short description (encoded!) <text> $00 (00) */ |
| 565 |
/* Then the comment text (encoded) ... */ |
| 566 |
unsigned char encoding = realdata[0]; |
| 567 |
char lang[3]; // realdata + 1 |
| 568 |
unsigned char *descr = realdata+4; |
| 569 |
unsigned char *text = NULL; |
| 570 |
mpg123_text *xcom = NULL; |
| 571 |
mpg123_text localcom; // UTF-8 variant for local processing, remember to clean up! |
| 572 |
init_mpg123_text(&localcom); |
| 573 |
|
| 574 |
if(realsize < (size_t)(descr-realdata)) |
| 575 |
{ |
| 576 |
if(NOQUIET) error1("Invalid frame size of %zu (too small for anything).", realsize); |
| 577 |
return; |
| 578 |
} |
| 579 |
if(encoding > mpg123_id3_enc_max) |
| 580 |
{ |
| 581 |
if(NOQUIET) |
| 582 |
error1("Unknown text encoding %u, I take no chances, sorry!", encoding); |
| 583 |
return; |
| 584 |
} |
| 585 |
memcpy(lang, realdata+1, 3); |
| 586 |
/* Be careful with finding the end of description, I have to honor encoding here. */ |
| 587 |
text = next_text(descr, encoding, realsize-(descr-realdata)); |
| 588 |
if(text == NULL) |
| 589 |
{ |
| 590 |
if(NOQUIET) |
| 591 |
error("No comment text / valid description?"); |
| 592 |
return; |
| 593 |
} |
| 594 |
{ // just vor variable scope |
| 595 |
mpg123_string description; |
| 596 |
mpg123_init_string(&description); |
| 597 |
// Store the text, with desired encoding, but for comments always a local copy in UTF-8. |
| 598 |
store_id3_text( &description, encoding, descr, text-descr |
| 599 |
, NOQUIET, fr->p.flags & MPG123_PLAIN_ID3TEXT ); |
| 600 |
if(tt == comment) |
| 601 |
store_id3_text( &localcom.description, encoding, descr, text-descr |
| 602 |
, NOQUIET, 0 ); |
| 603 |
if(VERBOSE4) |
| 604 |
fprintf( stderr, "Note: Storing comment from %s encoding\n" |
| 605 |
, enc_name(encoding) ); |
| 606 |
xcom = tt == uslt |
| 607 |
? add_uslt(fr, lang, &description) |
| 608 |
: add_comment(fr, lang, &description); |
| 609 |
if(xcom == NULL) |
| 610 |
{ |
| 611 |
if(NOQUIET) |
| 612 |
error("Unable to attach new comment!"); |
| 613 |
mpg123_free_string(&description); |
| 614 |
free_mpg123_text(&localcom); |
| 615 |
return; |
| 616 |
} |
| 617 |
memcpy(xcom->id, id, 4); |
| 618 |
memcpy(xcom->lang, lang, 3); |
| 619 |
// That takes over the description allocation. |
| 620 |
mpg123_move_string(&description, &xcom->description); |
| 621 |
} |
| 622 |
|
| 623 |
store_id3_text( &xcom->text, encoding, text, realsize-(text-realdata) |
| 624 |
, NOQUIET, fr->p.flags & MPG123_PLAIN_ID3TEXT ); |
| 625 |
/* Remember: I will probably decode the above (again) for rva comment checking. So no messing around, please. */ |
| 626 |
|
| 627 |
if(VERBOSE4) /* Do _not_ print the verbatim text: The encoding might be funny! */ |
| 628 |
{ |
| 629 |
fprintf(stderr, "Note: ID3 comm/uslt desc of length %zu.\n", xcom->description.fill); |
| 630 |
fprintf(stderr, "Note: ID3 comm/uslt text of length %zu.\n", xcom->text.fill); |
| 631 |
} |
| 632 |
/* Look out for RVA info only when we really deal with a straight comment. */ |
| 633 |
if(tt == comment && localcom.description.fill > 0) |
| 634 |
{ |
| 635 |
int rva_mode = -1; /* mix / album */ |
| 636 |
if( !strcasecmp(localcom.description.p, "rva") |
| 637 |
|| !strcasecmp(localcom.description.p, "rva_mix") |
| 638 |
|| !strcasecmp(localcom.description.p, "rva_track") |
| 639 |
|| !strcasecmp(localcom.description.p, "rva_radio") ) |
| 640 |
rva_mode = 0; |
| 641 |
else if( !strcasecmp(localcom.description.p, "rva_album") |
| 642 |
|| !strcasecmp(localcom.description.p, "rva_audiophile") |
| 643 |
|| !strcasecmp(localcom.description.p, "rva_user") ) |
| 644 |
rva_mode = 1; |
| 645 |
if((rva_mode > -1) && (fr->rva.level[rva_mode] <= rva_level)) |
| 646 |
{ |
| 647 |
/* Only translate the contents in here where we really need them. */ |
| 648 |
store_id3_text( &localcom.text, encoding, text, realsize-(text-realdata) |
| 649 |
, NOQUIET, 0 ); |
| 650 |
if(localcom.text.fill > 0) |
| 651 |
{ |
| 652 |
fr->rva.gain[rva_mode] = (float) atof(localcom.text.p); |
| 653 |
if(VERBOSE3) fprintf(stderr, "Note: RVA value %fdB\n", fr->rva.gain[rva_mode]); |
| 654 |
fr->rva.peak[rva_mode] = 0; |
| 655 |
fr->rva.level[rva_mode] = rva_level; |
| 656 |
} |
| 657 |
} |
| 658 |
} |
| 659 |
/* Make sure to free the local memory... */ |
| 660 |
free_mpg123_text(&localcom); |
| 661 |
} |
| 662 |
|
| 663 |
static void process_extra(mpg123_handle *fr, unsigned char* realdata, size_t realsize, int rva_level, char *id) |
| 664 |
{ |
| 665 |
/* Text encoding $xx */ |
| 666 |
/* Description ... $00 (00) */ |
| 667 |
/* Text ... */ |
| 668 |
unsigned char encoding = realdata[0]; |
| 669 |
unsigned char *descr = realdata+1; /* remember, the encoding is descr[-1] */ |
| 670 |
unsigned char *text; |
| 671 |
mpg123_text *xex; |
| 672 |
mpg123_text localex; |
| 673 |
|
| 674 |
if((int)realsize < descr-realdata) |
| 675 |
{ |
| 676 |
if(NOQUIET) error1("Invalid frame size of %lu (too small for anything).", (unsigned long)realsize); |
| 677 |
return; |
| 678 |
} |
| 679 |
if(encoding > mpg123_id3_enc_max) |
| 680 |
{ |
| 681 |
if(NOQUIET) |
| 682 |
error1("Unknown text encoding %u, I take no chances, sorry!", encoding); |
| 683 |
return; |
| 684 |
} |
| 685 |
text = next_text(descr, encoding, realsize-(descr-realdata)); |
| 686 |
if(VERBOSE4) fprintf(stderr, "Note: Storing extra from %s encoding\n", enc_name(realdata[0])); |
| 687 |
if(text == NULL) |
| 688 |
{ |
| 689 |
if(NOQUIET) error("No extra frame text / valid description?"); |
| 690 |
return; |
| 691 |
} |
| 692 |
{ // just vor variable scope |
| 693 |
mpg123_string description; |
| 694 |
mpg123_init_string(&description); |
| 695 |
/* The outside storage gets reencoded to UTF-8 only if not requested otherwise. |
| 696 |
Remember that we really need the -1 here to hand in the encoding byte!*/ |
| 697 |
store_id3_text( &description, encoding, descr, text-descr |
| 698 |
, NOQUIET, fr->p.flags & MPG123_PLAIN_ID3TEXT ); |
| 699 |
xex = add_extra(fr, &description); |
| 700 |
if(xex) |
| 701 |
mpg123_move_string(&description, &xex->description); |
| 702 |
else |
| 703 |
mpg123_free_string(&description); |
| 704 |
} |
| 705 |
if(xex == NULL) |
| 706 |
{ |
| 707 |
if(NOQUIET) error("Unable to attach new extra text!"); |
| 708 |
return; |
| 709 |
} |
| 710 |
memcpy(xex->id, id, 4); |
| 711 |
init_mpg123_text(&localex); /* For our local copy. */ |
| 712 |
|
| 713 |
/* Our local copy is always stored in UTF-8! */ |
| 714 |
store_id3_text(&localex.description, encoding, descr, text-descr, NOQUIET, 0); |
| 715 |
/* At first, only store the outside copy of the payload. We may not need the local copy. */ |
| 716 |
store_id3_text( &xex->text, encoding, text, realsize-(text-realdata) |
| 717 |
, NOQUIET, fr->p.flags & MPG123_PLAIN_ID3TEXT ); |
| 718 |
|
| 719 |
/* Now check if we would like to interpret this extra info for RVA. */ |
| 720 |
if(localex.description.fill > 0) |
| 721 |
{ |
| 722 |
int is_peak = 0; |
| 723 |
int rva_mode = -1; /* mix / album */ |
| 724 |
|
| 725 |
if(!strncasecmp(localex.description.p, "replaygain_track_",17)) |
| 726 |
{ |
| 727 |
if(VERBOSE3) fprintf(stderr, "Note: RVA ReplayGain track gain/peak\n"); |
| 728 |
|
| 729 |
rva_mode = 0; |
| 730 |
if(!strcasecmp(localex.description.p, "replaygain_track_peak")) is_peak = 1; |
| 731 |
else if(strcasecmp(localex.description.p, "replaygain_track_gain")) rva_mode = -1; |
| 732 |
} |
| 733 |
else |
| 734 |
if(!strncasecmp(localex.description.p, "replaygain_album_",17)) |
| 735 |
{ |
| 736 |
if(VERBOSE3) fprintf(stderr, "Note: RVA ReplayGain album gain/peak\n"); |
| 737 |
|
| 738 |
rva_mode = 1; |
| 739 |
if(!strcasecmp(localex.description.p, "replaygain_album_peak")) is_peak = 1; |
| 740 |
else if(strcasecmp(localex.description.p, "replaygain_album_gain")) rva_mode = -1; |
| 741 |
} |
| 742 |
if((rva_mode > -1) && (fr->rva.level[rva_mode] <= rva_level)) |
| 743 |
{ |
| 744 |
/* Now we need the translated copy of the data. */ |
| 745 |
store_id3_text( &localex.text, encoding, text, realsize-(text-realdata) |
| 746 |
, NOQUIET, 0 ); |
| 747 |
if(localex.text.fill > 0) |
| 748 |
{ |
| 749 |
if(is_peak) |
| 750 |
{ |
| 751 |
fr->rva.peak[rva_mode] = (float) atof(localex.text.p); |
| 752 |
if(VERBOSE3) fprintf(stderr, "Note: RVA peak %f\n", fr->rva.peak[rva_mode]); |
| 753 |
} |
| 754 |
else |
| 755 |
{ |
| 756 |
fr->rva.gain[rva_mode] = (float) atof(localex.text.p); |
| 757 |
if(VERBOSE3) fprintf(stderr, "Note: RVA gain %fdB\n", fr->rva.gain[rva_mode]); |
| 758 |
} |
| 759 |
fr->rva.level[rva_mode] = rva_level; |
| 760 |
} |
| 761 |
} |
| 762 |
} |
| 763 |
|
| 764 |
free_mpg123_text(&localex); |
| 765 |
} |
| 766 |
|
| 767 |
/* Make a ID3v2.3+ 4-byte ID from a ID3v2.2 3-byte ID |
| 768 |
Note that not all frames survived to 2.4; the mapping goes to 2.3 . |
| 769 |
A notable miss is the old RVA frame, which is very unspecific anyway. |
| 770 |
This function returns -1 when a not known 3 char ID was encountered, 0 otherwise. */ |
| 771 |
static int promote_framename(mpg123_handle *fr, char *id) /* fr because of VERBOSE macros */ |
| 772 |
{ |
| 773 |
size_t i; |
| 774 |
char *old[] = |
| 775 |
{ |
| 776 |
"COM", "TAL", "TBP", "TCM", "TCO", "TCR", "TDA", "TDY", "TEN", "TFT", |
| 777 |
"TIM", "TKE", "TLA", "TLE", "TMT", "TOA", "TOF", "TOL", "TOR", "TOT", |
| 778 |
"TP1", "TP2", "TP3", "TP4", "TPA", "TPB", "TRC", "TDA", "TRK", "TSI", |
| 779 |
"TSS", "TT1", "TT2", "TT3", "TXT", "TXX", "TYE" |
| 780 |
}; |
| 781 |
char *new[] = |
| 782 |
{ |
| 783 |
"COMM", "TALB", "TBPM", "TCOM", "TCON", "TCOP", "TDAT", "TDLY", "TENC", "TFLT", |
| 784 |
"TIME", "TKEY", "TLAN", "TLEN", "TMED", "TOPE", "TOFN", "TOLY", "TORY", "TOAL", |
| 785 |
"TPE1", "TPE2", "TPE3", "TPE4", "TPOS", "TPUB", "TSRC", "TRDA", "TRCK", "TSIZ", |
| 786 |
"TSSE", "TIT1", "TIT2", "TIT3", "TEXT", "TXXX", "TYER" |
| 787 |
}; |
| 788 |
for(i=0; i<sizeof(old)/sizeof(char*); ++i) |
| 789 |
{ |
| 790 |
if(!strncmp(id, old[i], 3)) |
| 791 |
{ |
| 792 |
memcpy(id, new[i], 4); |
| 793 |
if(VERBOSE3) fprintf(stderr, "Translated ID3v2.2 frame %s to %s\n", old[i], new[i]); |
| 794 |
return 0; |
| 795 |
} |
| 796 |
} |
| 797 |
if(VERBOSE3) fprintf(stderr, "Ignoring untranslated ID3v2.2 frame %c%c%c\n", id[0], id[1], id[2]); |
| 798 |
return -1; |
| 799 |
} |
| 800 |
|
| 801 |
#endif /* NO_ID3V2 */ |
| 802 |
|
| 803 |
static int store_id3v2( mpg123_handle *fr |
| 804 |
, unsigned long first4bytes, unsigned char buf[6], unsigned long length ) |
| 805 |
{ |
| 806 |
int ret = 1; |
| 807 |
int64_t ret2; |
| 808 |
unsigned long fullen = 10+length; |
| 809 |
if(fr->id3v2_raw) |
| 810 |
free(fr->id3v2_raw); |
| 811 |
fr->id3v2_size = 0; |
| 812 |
/* Allocate one byte more for a closing zero as safety catch for strlen(). */ |
| 813 |
fr->id3v2_raw = malloc(fullen+1); |
| 814 |
if(!fr->id3v2_raw) |
| 815 |
{ |
| 816 |
fr->err = MPG123_OUT_OF_MEM; |
| 817 |
if(NOQUIET) |
| 818 |
error1("ID3v2: Arrg! Unable to allocate %lu bytes" |
| 819 |
" for ID3v2 data - trying to skip instead.", length+1); |
| 820 |
if((ret2=fr->rd->skip_bytes(fr,length)) < 0) |
| 821 |
ret = ret2; |
| 822 |
else |
| 823 |
ret = 0; |
| 824 |
} |
| 825 |
else |
| 826 |
{ |
| 827 |
fr->id3v2_raw[0] = (first4bytes>>24) & 0xff; |
| 828 |
fr->id3v2_raw[1] = (first4bytes>>16) & 0xff; |
| 829 |
fr->id3v2_raw[2] = (first4bytes>>8) & 0xff; |
| 830 |
fr->id3v2_raw[3] = first4bytes & 0xff; |
| 831 |
memcpy(fr->id3v2_raw+4, buf, 6); |
| 832 |
if((ret2=fr->rd->read_frame_body(fr, fr->id3v2_raw+10, length)) < 0) |
| 833 |
{ |
| 834 |
ret=ret2; |
| 835 |
free(fr->id3v2_raw); |
| 836 |
fr->id3v2_raw = NULL; |
| 837 |
} |
| 838 |
else |
| 839 |
{ /* Closing with a zero for paranoia. */ |
| 840 |
fr->id3v2_raw[fullen] = 0; |
| 841 |
fr->id3v2_size = fullen; |
| 842 |
} |
| 843 |
} |
| 844 |
return ret; |
| 845 |
} |
| 846 |
|
| 847 |
/* |
| 848 |
trying to parse ID3v2.3 and ID3v2.4 tags... |
| 849 |
|
| 850 |
returns: 0: bad or just unparseable tag |
| 851 |
1: good, (possibly) new tag info |
| 852 |
<0: reader error (may need more data feed, try again) |
| 853 |
*/ |
| 854 |
int INT123_parse_new_id3(mpg123_handle *fr, unsigned long first4bytes) |
| 855 |
{ |
| 856 |
#define UNSYNC_FLAG 128 |
| 857 |
#define EXTHEAD_FLAG 64 /* ID3v2.3+ */ |
| 858 |
#define COMPRESS_FLAG 64 /* ID3v2.2 */ |
| 859 |
#define EXP_FLAG 32 |
| 860 |
#define FOOTER_FLAG 16 |
| 861 |
#define EXT_UPDATE_FLAG 64 /* ID3v2.4 only: extended header update flag */ |
| 862 |
#define UNKNOWN_FLAGS 15 /* 00001111*/ |
| 863 |
unsigned char buf[6]; |
| 864 |
unsigned long length=0; |
| 865 |
unsigned char flags = 0; |
| 866 |
int ret = 1; |
| 867 |
int64_t ret2; |
| 868 |
int storetag = 0; |
| 869 |
unsigned int footlen = 0; |
| 870 |
#ifndef NO_ID3V2 |
| 871 |
int skiptag = 0; |
| 872 |
#endif |
| 873 |
unsigned char major = first4bytes & 0xff; |
| 874 |
debug1("ID3v2: major tag version: %i", major); |
| 875 |
|
| 876 |
if(major == 0xff) return 0; /* Invalid... */ |
| 877 |
if((ret2 = fr->rd->read_frame_body(fr, buf, 6)) < 0) /* read more header information */ |
| 878 |
return ret2; |
| 879 |
|
| 880 |
if(buf[0] == 0xff) return 0; /* Revision, will never be 0xff. */ |
| 881 |
|
| 882 |
if(fr->p.flags & MPG123_STORE_RAW_ID3) |
| 883 |
storetag = 1; |
| 884 |
/* second new byte are some nice flags, if these are invalid skip the whole thing */ |
| 885 |
flags = buf[1]; |
| 886 |
debug1("ID3v2: flags 0x%08x", flags); |
| 887 |
/* use 4 bytes from buf to construct 28bit uint value and return 1; return 0 if bytes are not synchsafe */ |
| 888 |
#define synchsafe_to_long(buf,res) \ |
| 889 |
( \ |
| 890 |
(((buf)[0]|(buf)[1]|(buf)[2]|(buf)[3]) & 0x80) ? 0 : \ |
| 891 |
(res = (((unsigned long) (buf)[0]) << 21) \ |
| 892 |
| (((unsigned long) (buf)[1]) << 14) \ |
| 893 |
| (((unsigned long) (buf)[2]) << 7) \ |
| 894 |
| ((unsigned long) (buf)[3]) \ |
| 895 |
,1) \ |
| 896 |
) |
| 897 |
/* id3v2.3 does not store synchsafe frame sizes, but synchsafe tag size - doh! */ |
| 898 |
/* Remember: bytes_to_long() can yield ULONG_MAX on 32 bit platforms! */ |
| 899 |
#define bytes_to_long(buf,res) \ |
| 900 |
( \ |
| 901 |
major == 3 ? \ |
| 902 |
(res = (((unsigned long) (buf)[0]) << 24) \ |
| 903 |
| (((unsigned long) (buf)[1]) << 16) \ |
| 904 |
| (((unsigned long) (buf)[2]) << 8) \ |
| 905 |
| ((unsigned long) (buf)[3]) \ |
| 906 |
,1) : synchsafe_to_long(buf,res) \ |
| 907 |
) |
| 908 |
/* for id3v2.2 only */ |
| 909 |
#define threebytes_to_long(buf,res) \ |
| 910 |
( \ |
| 911 |
res = (((unsigned long) (buf)[0]) << 16) \ |
| 912 |
| (((unsigned long) (buf)[1]) << 8) \ |
| 913 |
| ((unsigned long) (buf)[2]) \ |
| 914 |
) |
| 915 |
|
| 916 |
/* length-10 or length-20 (footer present); 4 synchsafe integers == 28 bit number */ |
| 917 |
/* we have already read 10 bytes, so left are length or length+10 bytes belonging to tag */ |
| 918 |
/* Note: This is an 28 bit value in 32 bit storage, plenty of space for */ |
| 919 |
/* length+x for reasonable x. */ |
| 920 |
if(!synchsafe_to_long(buf+2,length)) |
| 921 |
{ |
| 922 |
if(NOQUIET) error4("Bad tag length (not synchsafe): 0x%02x%02x%02x%02x; You got a bad ID3 tag here.", buf[2],buf[3],buf[4],buf[5]); |
| 923 |
return 0; |
| 924 |
} |
| 925 |
if(flags & FOOTER_FLAG) |
| 926 |
footlen = 10; |
| 927 |
debug1("ID3v2: tag data length %lu", length); |
| 928 |
#ifndef NO_ID3V2 |
| 929 |
if(VERBOSE2) fprintf(stderr,"Note: ID3v2.%i rev %i tag of %lu bytes\n", major, buf[0], length); |
| 930 |
/* skip if unknown version/scary flags, parse otherwise */ |
| 931 |
if(fr->p.flags & MPG123_SKIP_ID3V2) |
| 932 |
{ |
| 933 |
if(VERBOSE3) |
| 934 |
fprintf(stderr, "Note: Skipping ID3v2 tag per user request.\n"); |
| 935 |
skiptag = 1; |
| 936 |
} |
| 937 |
if((flags & UNKNOWN_FLAGS) || (major > 4) || (major < 2)) |
| 938 |
{ |
| 939 |
if(NOQUIET) |
| 940 |
warning2( "ID3v2: Won't parse the ID3v2 tag with major version" |
| 941 |
" %u and flags 0x%xu - some extra code may be needed" |
| 942 |
, major, flags ); |
| 943 |
skiptag = 1; |
| 944 |
} |
| 945 |
// Standard says that compressed tags should be ignored as there isn't an agreed |
| 946 |
// compressoion scheme. |
| 947 |
if(major == 2 && flags & COMPRESS_FLAG) |
| 948 |
{ |
| 949 |
if(NOQUIET) |
| 950 |
warning("ID3v2: ignoring compressed ID3v2.2 tag"); |
| 951 |
skiptag = 1; |
| 952 |
} |
| 953 |
if(length < 10) |
| 954 |
{ |
| 955 |
if(NOQUIET) |
| 956 |
warning1("ID3v2: unrealistic small tag lengh %lu, skipping", length); |
| 957 |
skiptag = 1; |
| 958 |
} |
| 959 |
if(!skiptag) |
| 960 |
storetag = 1; |
| 961 |
#endif |
| 962 |
if(storetag) |
| 963 |
{ |
| 964 |
/* Stores whole tag with footer and an additonal trailing zero. */ |
| 965 |
if((ret2 = store_id3v2(fr, first4bytes, buf, length+footlen)) <= 0) |
| 966 |
return ret2; |
| 967 |
} |
| 968 |
#ifndef NO_ID3V2 |
| 969 |
if(skiptag) |
| 970 |
{ |
| 971 |
if(VERBOSE3) |
| 972 |
fprintf(stderr, "Note: skipped tag clearing possibly existing ID3v2 data"); |
| 973 |
INT123_reset_id3(fr); // Old data is invalid. |
| 974 |
#endif |
| 975 |
if(!storetag && (ret2=fr->rd->skip_bytes(fr,length+footlen))<0) |
| 976 |
ret=ret2; |
| 977 |
#ifndef NO_ID3V2 |
| 978 |
} |
| 979 |
else |
| 980 |
{ |
| 981 |
unsigned char* tagdata = fr->id3v2_raw+10; |
| 982 |
/* try to interpret that beast */ |
| 983 |
debug("ID3v2: analysing frames..."); |
| 984 |
if(length > 0) |
| 985 |
{ |
| 986 |
unsigned char extflags = 0; |
| 987 |
unsigned long tagpos = 0; |
| 988 |
/* bytes of frame title and of framesize value */ |
| 989 |
unsigned int head_part = major > 2 ? 4 : 3; |
| 990 |
unsigned int flag_part = major > 2 ? 2 : 0; |
| 991 |
/* The amount of bytes that are unconditionally read for each frame: */ |
| 992 |
/* ID, size, flags. */ |
| 993 |
unsigned int framebegin = head_part+head_part+flag_part; |
| 994 |
debug1("ID3v2: have read at all %lu bytes for the tag now", (unsigned long)length+6); |
| 995 |
if(flags & EXTHEAD_FLAG) |
| 996 |
{ |
| 997 |
debug("ID3v2: extended header"); |
| 998 |
if(!bytes_to_long(tagdata, tagpos) || tagpos >= length) |
| 999 |
{ |
| 1000 |
ret = 0; |
| 1001 |
if(NOQUIET) |
| 1002 |
error4( "Bad (non-synchsafe/too large) tag offset from extended header:" |
| 1003 |
"0x%02x%02x%02x%02x" |
| 1004 |
, tagdata[0], tagdata[1], tagdata[2], tagdata[3] ); |
| 1005 |
} else if(tagpos < 6) |
| 1006 |
{ |
| 1007 |
ret = 0; |
| 1008 |
if(NOQUIET) |
| 1009 |
merror("Extended header too small (%lu).", tagpos); |
| 1010 |
} |
| 1011 |
if(major == 3) |
| 1012 |
{ |
| 1013 |
tagpos += 4; // The size itself is not included. |
| 1014 |
if(tagpos >= length) |
| 1015 |
{ |
| 1016 |
ret = 0; |
| 1017 |
if(NOQUIET) |
| 1018 |
error("Too much extended v2.3 header."); |
| 1019 |
} |
| 1020 |
} else if(ret) // v2.4 and at least got my 6 bytes of ext header |
| 1021 |
{ |
| 1022 |
// Only v4 knows update frames, check for that. |
| 1023 |
// Need to step back. Header is 4 bytes length, one byte flag size, |
| 1024 |
// one byte flags. Flag size has to be 1! |
| 1025 |
if(tagdata[4] == 1 && tagdata[5] & EXT_UPDATE_FLAG) |
| 1026 |
{ |
| 1027 |
if(VERBOSE3) |
| 1028 |
fprintf(stderr, "Note: ID3v2.4 update tag\n"); |
| 1029 |
extflags |= EXT_UPDATE_FLAG; |
| 1030 |
} |
| 1031 |
} |
| 1032 |
} |
| 1033 |
if(!(extflags & EXT_UPDATE_FLAG)) |
| 1034 |
{ |
| 1035 |
if(VERBOSE3) |
| 1036 |
fprintf(stderr, "Note: non-update tag replacing existing ID3v2 data\n"); |
| 1037 |
INT123_reset_id3(fr); |
| 1038 |
} |
| 1039 |
if(ret > 0) |
| 1040 |
{ |
| 1041 |
char id[5]; |
| 1042 |
unsigned long framesize; |
| 1043 |
unsigned long fflags; /* need 16 bits, actually */ |
| 1044 |
id[4] = 0; |
| 1045 |
fr->id3v2.version = major; |
| 1046 |
/* Pos now advanced after ext head, now a frame has to follow. */ |
| 1047 |
/* Note: tagpos <= length, which is 28 bit integer, so both */ |
| 1048 |
/* far away from overflow for adding known small values. */ |
| 1049 |
/* I want to read at least one full header now. */ |
| 1050 |
while(length >= tagpos+framebegin) |
| 1051 |
{ |
| 1052 |
int i = 0; |
| 1053 |
unsigned long pos = tagpos; |
| 1054 |
/* level 1,2,3 - 0 is info from lame/info tag! */ |
| 1055 |
/* rva tags with ascending significance, then general frames */ |
| 1056 |
enum frame_types tt = unknown; |
| 1057 |
/* we may have entered the padding zone or any other strangeness: check if we have valid frame id characters */ |
| 1058 |
for(i=0; i< head_part; ++i) |
| 1059 |
if( !( ((tagdata[tagpos+i] > 47) && (tagdata[tagpos+i] < 58)) |
| 1060 |
|| ((tagdata[tagpos+i] > 64) && (tagdata[tagpos+i] < 91)) ) ) |
| 1061 |
{ |
| 1062 |
debug5("ID3v2: real tag data apparently ended after %lu bytes with 0x%02x%02x%02x%02x", tagpos, tagdata[tagpos], tagdata[tagpos+1], tagdata[tagpos+2], tagdata[tagpos+3]); |
| 1063 |
/* This is no hard error... let's just hope that we got something meaningful already (ret==1 in that case). */ |
| 1064 |
goto tagparse_cleanup; /* Need to escape two loops here. */ |
| 1065 |
} |
| 1066 |
if(ret > 0) |
| 1067 |
{ |
| 1068 |
/* 4 or 3 bytes id */ |
| 1069 |
strncpy(id, (char*) tagdata+pos, head_part); |
| 1070 |
id[head_part] = 0; /* terminate for 3 or 4 bytes */ |
| 1071 |
pos += head_part; |
| 1072 |
tagpos += head_part; |
| 1073 |
/* size as 32 bits or 28 bits */ |
| 1074 |
if(fr->id3v2.version == 2) threebytes_to_long(tagdata+pos, framesize); |
| 1075 |
else |
| 1076 |
if(!bytes_to_long(tagdata+pos, framesize)) |
| 1077 |
{ |
| 1078 |
/* Just assume that up to now there was some good data. */ |
| 1079 |
if(NOQUIET) error1("ID3v2: non-syncsafe size of %s frame, skipping the remainder of tag", id); |
| 1080 |
break; |
| 1081 |
} |
| 1082 |
if(VERBOSE3) fprintf(stderr, "Note: ID3v2 %s frame of size %lu\n", id, framesize); |
| 1083 |
tagpos += head_part; |
| 1084 |
pos += head_part; |
| 1085 |
if(fr->id3v2.version > 2) |
| 1086 |
{ |
| 1087 |
fflags = (((unsigned long) tagdata[pos]) << 8) | ((unsigned long) tagdata[pos+1]); |
| 1088 |
pos += 2; |
| 1089 |
tagpos += 2; |
| 1090 |
} |
| 1091 |
else fflags = 0; |
| 1092 |
|
| 1093 |
if(length - tagpos < framesize) |
| 1094 |
{ |
| 1095 |
if(NOQUIET) error("Whoa! ID3v2 frame claims to be larger than the whole rest of the tag."); |
| 1096 |
break; |
| 1097 |
} |
| 1098 |
tagpos += framesize; /* the important advancement in whole tag */ |
| 1099 |
/* for sanity, after full parsing tagpos should be == pos */ |
| 1100 |
/* debug4("ID3v2: found %s frame, size %lu (as bytes: 0x%08lx), flags 0x%016lx", id, framesize, framesize, fflags); */ |
| 1101 |
/* v2.4: %0abc0000 %0h00kmnp */ |
| 1102 |
/* v2.3: %abc00000 %ijk00000 */ |
| 1103 |
/* v2.2: just zero */ |
| 1104 |
#define V3 (major == 3) |
| 1105 |
#define BAD_FFLAGS (unsigned long) (V3 ? 7967 : 36784) |
| 1106 |
#define PRES_TAG_FFLAG (unsigned long) (V3 ? 32768 : 16384) |
| 1107 |
#define PRES_FILE_FFLAG (unsigned long) (V3 ? 16384 : 8192) |
| 1108 |
#define READ_ONLY_FFLAG (unsigned long) (V3 ? 8192 : 4096) |
| 1109 |
#define GROUP_FFLAG (unsigned long) (V3 ? 32 : 64) |
| 1110 |
#define COMPR_FFLAG (unsigned long) (V3 ? 128 : 8) |
| 1111 |
#define ENCR_FFLAG (unsigned long) (V3 ? 64 : 4) |
| 1112 |
#define UNSYNC_FFLAG (unsigned long) (V3 ? 0 : 2) |
| 1113 |
#define DATLEN_FFLAG (unsigned long) (V3 ? 0 : 1) |
| 1114 |
if(head_part < 4 && promote_framename(fr, id) != 0) continue; |
| 1115 |
|
| 1116 |
/* shall not or want not handle these */ |
| 1117 |
if(fflags & (BAD_FFLAGS | COMPR_FFLAG | ENCR_FFLAG)) |
| 1118 |
{ |
| 1119 |
if(NOQUIET) warning("ID3v2: skipping invalid/unsupported frame"); |
| 1120 |
continue; |
| 1121 |
} |
| 1122 |
|
| 1123 |
for(i = 0; i < KNOWN_FRAMES; ++i) |
| 1124 |
if(!strncmp(frame_type[i], id, 4)){ tt = i; break; } |
| 1125 |
|
| 1126 |
if(id[0] == 'T' && tt != extra) tt = text; |
| 1127 |
|
| 1128 |
if(tt != unknown) |
| 1129 |
{ |
| 1130 |
int rva_mode = -1; /* mix / album */ |
| 1131 |
unsigned long realsize = framesize; |
| 1132 |
unsigned char* realdata = tagdata+pos; |
| 1133 |
unsigned char* unsyncbuffer = NULL; |
| 1134 |
if(((flags & UNSYNC_FLAG) || (fflags & UNSYNC_FFLAG)) && framesize > 0) |
| 1135 |
{ |
| 1136 |
unsigned long ipos = 0; |
| 1137 |
unsigned long opos = 0; |
| 1138 |
debug("Id3v2: going to de-unsync the frame data"); |
| 1139 |
/* de-unsync: FF00 -> FF; real FF00 is simply represented as FF0000 ... */ |
| 1140 |
/* damn, that means I have to delete bytes from withing the data block... thus need temporal storage */ |
| 1141 |
/* standard mandates that de-unsync should always be safe if flag is set */ |
| 1142 |
realdata = unsyncbuffer = malloc(framesize+1); /* will need <= bytes, plus a safety zero */ |
| 1143 |
if(realdata == NULL) |
| 1144 |
{ |
| 1145 |
if(NOQUIET) error("ID3v2: unable to allocate working buffer for de-unsync"); |
| 1146 |
continue; |
| 1147 |
} |
| 1148 |
/* now going byte per byte through the data... */ |
| 1149 |
realdata[0] = tagdata[pos]; |
| 1150 |
opos = 1; |
| 1151 |
for(ipos = pos+1; ipos < pos+framesize; ++ipos) |
| 1152 |
{ |
| 1153 |
if(!((tagdata[ipos] == 0) && (tagdata[ipos-1] == 0xff))) |
| 1154 |
{ |
| 1155 |
realdata[opos++] = tagdata[ipos]; |
| 1156 |
} |
| 1157 |
} |
| 1158 |
realsize = opos; |
| 1159 |
/* Append a zero to keep strlen() safe. */ |
| 1160 |
realdata[realsize] = 0; |
| 1161 |
debug2("ID3v2: de-unsync made %lu out of %lu bytes", realsize, framesize); |
| 1162 |
} |
| 1163 |
// The spec says there is a group byte, without explicitly saying that it is |
| 1164 |
// the first thing following the header. I just assume so, because of the |
| 1165 |
// ordering of the flags. |
| 1166 |
if(fflags & GROUP_FFLAG) |
| 1167 |
{ // Just skip group byte. |
| 1168 |
if(realsize) |
| 1169 |
{ |
| 1170 |
if(VERBOSE3) |
| 1171 |
fprintf(stderr, "Note: frame of group %d\n", realdata[0]); |
| 1172 |
--realsize; |
| 1173 |
++realdata; |
| 1174 |
} else if(NOQUIET) |
| 1175 |
error("Grouped frame without group byte, even."); |
| 1176 |
} |
| 1177 |
if(fflags & DATLEN_FFLAG) |
| 1178 |
{ |
| 1179 |
// Spec says the original (without compression or unsync) data length follows, |
| 1180 |
// so it should match de-unsynced data now. |
| 1181 |
if(realsize >= 4) |
| 1182 |
{ |
| 1183 |
unsigned long datlen; |
| 1184 |
if(bytes_to_long(realdata, datlen) && datlen == realsize-4) |
| 1185 |
{ |
| 1186 |
realsize -= 4; |
| 1187 |
realdata += 4; |
| 1188 |
} else |
| 1189 |
{ |
| 1190 |
if(NOQUIET) |
| 1191 |
error("frame data length bad, skipping"); |
| 1192 |
realsize = 0; |
| 1193 |
} |
| 1194 |
} else |
| 1195 |
{ |
| 1196 |
realsize = 0; |
| 1197 |
if(NOQUIET) |
| 1198 |
error("frame truncated at frame data length, skipping"); |
| 1199 |
} |
| 1200 |
} |
| 1201 |
pos = 0; /* now at the beginning again... */ |
| 1202 |
/* Avoid reading over boundary, even if there is a */ |
| 1203 |
/* zero byte of padding for safety. */ |
| 1204 |
if(realsize) switch(tt) |
| 1205 |
{ |
| 1206 |
case comment: |
| 1207 |
case uslt: |
| 1208 |
process_comment(fr, tt, realdata, realsize, comment+1, id); |
| 1209 |
break; |
| 1210 |
case extra: /* perhaps foobar2000's work */ |
| 1211 |
process_extra(fr, realdata, realsize, extra+1, id); |
| 1212 |
break; |
| 1213 |
case rva2: /* "the" RVA tag */ |
| 1214 |
{ |
| 1215 |
/* starts with null-terminated identification */ |
| 1216 |
if(VERBOSE3) fprintf(stderr, "Note: RVA2 identification \"%s\"\n", realdata); |
| 1217 |
/* default: some individual value, mix mode */ |
| 1218 |
rva_mode = 0; |
| 1219 |
if( !strncasecmp((char*)realdata, "album", 5) |
| 1220 |
|| !strncasecmp((char*)realdata, "audiophile", 10) |
| 1221 |
|| !strncasecmp((char*)realdata, "user", 4)) |
| 1222 |
rva_mode = 1; |
| 1223 |
if(fr->rva.level[rva_mode] <= rva2+1) |
| 1224 |
{ |
| 1225 |
pos += strlen((char*) realdata) + 1; |
| 1226 |
debug2("got my pos: %lu - %lu", realsize, pos); |
| 1227 |
// channel and two bytes for RVA value |
| 1228 |
// pos possibly just past the safety zero, so one more than realsize |
| 1229 |
if(pos > realsize || realsize-pos < 3) |
| 1230 |
{ |
| 1231 |
if(NOQUIET) |
| 1232 |
error("bad RVA2 tag (truncated?)"); |
| 1233 |
} |
| 1234 |
else if(realdata[pos] == 1) |
| 1235 |
{ |
| 1236 |
++pos; |
| 1237 |
/* only handle master channel */ |
| 1238 |
debug("ID3v2: it is for the master channel"); |
| 1239 |
/* two bytes adjustment, one byte for bits representing peak - n bytes, eh bits, for peak */ |
| 1240 |
/* 16 bit signed integer = dB * 512. Do not shift signed integers! Multiply instead. |
| 1241 |
Also no implementation-defined casting. Reinterpret the pointer to signed char, then do |
| 1242 |
proper casting. */ |
| 1243 |
fr->rva.gain[rva_mode] = (float) ( |
| 1244 |
((short)((signed char*)realdata)[pos]) * 256 + (short)realdata[pos+1] ) / 512; |
| 1245 |
pos += 2; |
| 1246 |
if(VERBOSE3) fprintf(stderr, "Note: RVA value %fdB\n", fr->rva.gain[rva_mode]); |
| 1247 |
/* heh, the peak value is represented by a number of bits - but in what manner? Skipping that part */ |
| 1248 |
fr->rva.peak[rva_mode] = 0; |
| 1249 |
fr->rva.level[rva_mode] = rva2+1; |
| 1250 |
} |
| 1251 |
} |
| 1252 |
} |
| 1253 |
break; |
| 1254 |
/* non-rva metainfo, simply store... */ |
| 1255 |
case text: |
| 1256 |
process_text(fr, realdata, realsize, id); |
| 1257 |
break; |
| 1258 |
case picture: |
| 1259 |
if (fr->p.flags & MPG123_PICTURE) |
| 1260 |
process_picture(fr, realdata, realsize); |
| 1261 |
|
| 1262 |
break; |
| 1263 |
default: if(NOQUIET) error1("ID3v2: unknown frame type %i", tt); |
| 1264 |
} |
| 1265 |
if(unsyncbuffer) |
| 1266 |
free(unsyncbuffer); |
| 1267 |
} |
| 1268 |
#undef V3 |
| 1269 |
#undef BAD_FFLAGS |
| 1270 |
#undef PRES_TAG_FFLAG |
| 1271 |
#undef PRES_FILE_FFLAG |
| 1272 |
#undef READ_ONLY_FFLAG |
| 1273 |
#undef GROUP_FFLAG |
| 1274 |
#undef COMPR_FFLAG |
| 1275 |
#undef ENCR_FFLAG |
| 1276 |
#undef UNSYNC_FFLAG |
| 1277 |
#undef DATLEN_FFLAG |
| 1278 |
} |
| 1279 |
else break; |
| 1280 |
#undef KNOWN_FRAMES |
| 1281 |
} |
| 1282 |
} else |
| 1283 |
{ |
| 1284 |
if(VERBOSE3) |
| 1285 |
fprintf(stderr, "Note: faulty ID3v2 tag still clearing old data\n"); |
| 1286 |
INT123_reset_id3(fr); |
| 1287 |
} |
| 1288 |
} else // No new data, but still there was a tag that invalidates old data. |
| 1289 |
{ |
| 1290 |
if(VERBOSE3) |
| 1291 |
fprintf(stderr, "Note: empty ID3v2 clearing old data\n"); |
| 1292 |
INT123_reset_id3(fr); |
| 1293 |
} |
| 1294 |
tagparse_cleanup: |
| 1295 |
/* Get rid of stored raw data that should not be kept. */ |
| 1296 |
if(!(fr->p.flags & MPG123_STORE_RAW_ID3)) |
| 1297 |
{ |
| 1298 |
free(fr->id3v2_raw); |
| 1299 |
fr->id3v2_raw = NULL; |
| 1300 |
fr->id3v2_size = 0; |
| 1301 |
} |
| 1302 |
} |
| 1303 |
#endif /* NO_ID3V2 */ |
| 1304 |
return ret; |
| 1305 |
#undef UNSYNC_FLAG |
| 1306 |
#undef EXTHEAD_FLAG |
| 1307 |
#undef COMPRESS_FLAG |
| 1308 |
#undef EXP_FLAG |
| 1309 |
#undef FOOTER_FLAG |
| 1310 |
#undef EXT_UPDATE_FLAG |
| 1311 |
#undef UNKOWN_FLAGS |
| 1312 |
} |
| 1313 |
|
| 1314 |
#ifndef NO_ID3V2 /* Disabling all the rest... */ |
| 1315 |
|
| 1316 |
static void convert_latin1(mpg123_string *sb, const unsigned char* s, size_t l, const int noquiet) |
| 1317 |
{ |
| 1318 |
size_t length = l; |
| 1319 |
size_t i; |
| 1320 |
unsigned char *p; |
| 1321 |
/* determine real length, a latin1 character can at most take 2 in UTF8 */ |
| 1322 |
for(i=0; i<l; ++i) |
| 1323 |
if(s[i] >= 0x80) ++length; |
| 1324 |
|
| 1325 |
debug1("UTF-8 length: %lu", (unsigned long)length); |
| 1326 |
/* one extra zero byte for paranoia */ |
| 1327 |
if(!mpg123_grow_string(sb, length+1)) |
| 1328 |
return; |
| 1329 |
|
| 1330 |
p = (unsigned char*) sb->p; /* Signedness doesn't matter but it shows I thought about the non-issue */ |
| 1331 |
for(i=0; i<l; ++i) |
| 1332 |
if(s[i] < 0x80){ *p = s[i]; ++p; } |
| 1333 |
else /* two-byte encoding */ |
| 1334 |
{ |
| 1335 |
*p = 0xc0 | (s[i]>>6); |
| 1336 |
*(p+1) = 0x80 | (s[i] & 0x3f); |
| 1337 |
p+=2; |
| 1338 |
} |
| 1339 |
|
| 1340 |
sb->p[length] = 0; |
| 1341 |
sb->fill = length+1; |
| 1342 |
} |
| 1343 |
|
| 1344 |
/* |
| 1345 |
Check if we have a byte oder mark(s) there, return: |
| 1346 |
-1: little endian |
| 1347 |
0: no BOM |
| 1348 |
1: big endian |
| 1349 |
|
| 1350 |
This modifies source and len to indicate the data _after_ the BOM(s). |
| 1351 |
Note on nasty data: The last encountered BOM determines the endianness. |
| 1352 |
I have seen data with multiple BOMS, namely from "the" id3v2 program. |
| 1353 |
Not nice, but what should I do? |
| 1354 |
*/ |
| 1355 |
static int check_bom(const unsigned char** source, size_t *len) |
| 1356 |
{ |
| 1357 |
int last_bom = 0; |
| 1358 |
|
| 1359 |
while(*len >= 2) |
| 1360 |
{ |
| 1361 |
int this_bom = 0; |
| 1362 |
if((*source)[0] == 0xff && (*source)[1] == 0xfe) |
| 1363 |
this_bom = -1; |
| 1364 |
if((*source)[0] == 0xfe && (*source)[1] == 0xff) |
| 1365 |
this_bom = 1; |
| 1366 |
if(this_bom == 0) |
| 1367 |
break; |
| 1368 |
/* Skip the detected BOM. */ |
| 1369 |
last_bom = this_bom; |
| 1370 |
*source += 2; |
| 1371 |
*len -= 2; |
| 1372 |
} |
| 1373 |
return last_bom; |
| 1374 |
} |
| 1375 |
|
| 1376 |
#define FULLPOINT(f,s) ( (((f)&0x3ff)<<10) + ((s)&0x3ff) + 0x10000 ) |
| 1377 |
/* Remember: There's a limit at 0x1ffff. */ |
| 1378 |
#define UTF8LEN(x) ( (x)<0x80 ? 1 : ((x)<0x800 ? 2 : ((x)<0x10000 ? 3 : 4))) |
| 1379 |
static void convert_utf16bom(mpg123_string *sb, const unsigned char* s, size_t l, const int noquiet) |
| 1380 |
{ |
| 1381 |
size_t i; |
| 1382 |
size_t n; /* number bytes that make up full pairs */ |
| 1383 |
unsigned char *p; |
| 1384 |
size_t length = 0; /* the resulting UTF-8 length */ |
| 1385 |
/* Determine real length... extreme case can be more than utf-16 length. */ |
| 1386 |
size_t high = 0; |
| 1387 |
size_t low = 1; |
| 1388 |
int bom_endian; |
| 1389 |
|
| 1390 |
debug1("convert_utf16 with length %lu", (unsigned long)l); |
| 1391 |
|
| 1392 |
bom_endian = check_bom(&s, &l); |
| 1393 |
debug1("UTF16 endianness check: %i", bom_endian); |
| 1394 |
|
| 1395 |
if(bom_endian == -1) /* little-endian */ |
| 1396 |
{ |
| 1397 |
high = 1; /* The second byte is the high byte. */ |
| 1398 |
low = 0; /* The first byte is the low byte. */ |
| 1399 |
} |
| 1400 |
|
| 1401 |
n = (l/2)*2; /* number bytes that make up full pairs */ |
| 1402 |
|
| 1403 |
/* first: get length, check for errors -- stop at first one */ |
| 1404 |
for(i=0; i < n; i+=2) |
| 1405 |
{ |
| 1406 |
unsigned long point = ((unsigned long) s[i+high]<<8) + s[i+low]; |
| 1407 |
if((point & 0xfc00) == 0xd800) /* lead surrogate */ |
| 1408 |
{ |
| 1409 |
unsigned short second = (i+3 < l) ? (s[i+2+high]<<8) + s[i+2+low] : 0; |
| 1410 |
if((second & 0xfc00) == 0xdc00) /* good... */ |
| 1411 |
{ |
| 1412 |
point = FULLPOINT(point,second); |
| 1413 |
length += UTF8LEN(point); /* possibly 4 bytes */ |
| 1414 |
i+=2; /* We overstepped one word. */ |
| 1415 |
} |
| 1416 |
else /* if no valid pair, break here */ |
| 1417 |
{ |
| 1418 |
if(noquiet) error2("Invalid UTF16 surrogate pair at %li (0x%04lx).", (unsigned long)i, point); |
| 1419 |
n = i; /* Forget the half pair, END! */ |
| 1420 |
break; |
| 1421 |
} |
| 1422 |
} |
| 1423 |
else length += UTF8LEN(point); /* 1,2 or 3 bytes */ |
| 1424 |
} |
| 1425 |
|
| 1426 |
if(!mpg123_grow_string(sb, length+1)) |
| 1427 |
return; |
| 1428 |
|
| 1429 |
/* Now really convert, skip checks as these have been done just before. */ |
| 1430 |
p = (unsigned char*) sb->p; /* Signedness doesn't matter but it shows I thought about the non-issue */ |
| 1431 |
for(i=0; i < n; i+=2) |
| 1432 |
{ |
| 1433 |
unsigned long codepoint = ((unsigned long) s[i+high]<<8) + s[i+low]; |
| 1434 |
if((codepoint & 0xfc00) == 0xd800) /* lead surrogate */ |
| 1435 |
{ |
| 1436 |
unsigned short second = (s[i+2+high]<<8) + s[i+2+low]; |
| 1437 |
codepoint = FULLPOINT(codepoint,second); |
| 1438 |
i+=2; /* We overstepped one word. */ |
| 1439 |
} |
| 1440 |
if(codepoint < 0x80) *p++ = (unsigned char) codepoint; |
| 1441 |
else if(codepoint < 0x800) |
| 1442 |
{ |
| 1443 |
*p++ = (unsigned char) (0xc0 | (codepoint>>6)); |
| 1444 |
*p++ = (unsigned char) (0x80 | (codepoint & 0x3f)); |
| 1445 |
} |
| 1446 |
else if(codepoint < 0x10000) |
| 1447 |
{ |
| 1448 |
*p++ = (unsigned char) (0xe0 | (codepoint>>12)); |
| 1449 |
*p++ = 0x80 | ((codepoint>>6) & 0x3f); |
| 1450 |
*p++ = 0x80 | (codepoint & 0x3f); |
| 1451 |
} |
| 1452 |
else if (codepoint < 0x200000) |
| 1453 |
{ |
| 1454 |
*p++ = (unsigned char) (0xf0 | codepoint>>18); |
| 1455 |
*p++ = (unsigned char) (0x80 | ((codepoint>>12) & 0x3f)); |
| 1456 |
*p++ = (unsigned char) (0x80 | ((codepoint>>6) & 0x3f)); |
| 1457 |
*p++ = (unsigned char) (0x80 | (codepoint & 0x3f)); |
| 1458 |
} /* ignore bigger ones (that are not possible here anyway) */ |
| 1459 |
} |
| 1460 |
sb->p[sb->size-1] = 0; /* paranoia... */ |
| 1461 |
sb->fill = sb->size; |
| 1462 |
} |
| 1463 |
#undef UTF8LEN |
| 1464 |
#undef FULLPOINT |
| 1465 |
|
| 1466 |
static void convert_utf8(mpg123_string *sb, const unsigned char* source, size_t len, const int noquiet) |
| 1467 |
{ |
| 1468 |
if(mpg123_grow_string(sb, len+1)) |
| 1469 |
{ |
| 1470 |
memcpy(sb->p, source, len); |
| 1471 |
sb->p[len] = 0; |
| 1472 |
sb->fill = len+1; |
| 1473 |
} |
| 1474 |
} |
| 1475 |
|
| 1476 |
#endif |