id3_parser.cpp 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536
  1. /*
  2. * This file is part of the FreeStreamer project,
  3. * (C)Copyright 2011-2018 Matias Muhonen <mmu@iki.fi> 穆马帝
  4. * See the file ''LICENSE'' for using the code.
  5. *
  6. * https://github.com/muhku/FreeStreamer
  7. */
  8. #include "id3_parser.h"
  9. #include <vector>
  10. //#define ID3_DEBUG 1
  11. #if !defined ( ID3_DEBUG)
  12. #define ID3_TRACE(...) do {} while (0)
  13. #else
  14. #define ID3_TRACE(...) printf(__VA_ARGS__)
  15. #endif
  16. namespace astreamer {
  17. // Code from:
  18. // http://www.opensource.apple.com/source/libsecurity_manifest/libsecurity_manifest-29384/lib/SecureDownloadInternal.c
  19. // Returns a CFString containing the base64 representation of the data.
  20. // boolean argument for whether to line wrap at 64 columns or not.
  21. CFStringRef createBase64EncodedString(const UInt8* ptr, size_t len, int wrap) {
  22. const char* alphabet = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
  23. "abcdefghijklmnopqrstuvwxyz"
  24. "0123456789+/=";
  25. // base64 encoded data uses 4 ASCII characters to represent 3 octets.
  26. // There can be up to two == at the end of the base64 data for padding.
  27. // If we are line wrapping then we need space for one newline character
  28. // every 64 characters of output.
  29. // Rounded 4/3 up to 2 to avoid floating point math.
  30. //CFIndex max_len = (2*len) + 2;
  31. //if (wrap) len = len + ((2*len) / 64) + 1;
  32. CFMutableStringRef string = CFStringCreateMutable(NULL, 0);
  33. if (!string) return NULL;
  34. /*
  35. http://www.faqs.org/rfcs/rfc3548.html
  36. +--first octet--+-second octet--+--third octet--+
  37. |7 6 5 4 3 2 1 0|7 6 5 4 3 2 1 0|7 6 5 4 3 2 1 0|
  38. +-----------+---+-------+-------+---+-----------+
  39. |5 4 3 2 1 0|5 4 3 2 1 0|5 4 3 2 1 0|5 4 3 2 1 0|
  40. +--1.index--+--2.index--+--3.index--+--4.index--+
  41. */
  42. int i = 0; // octet offset into input data
  43. int column = 0; // output column number (used for line wrapping)
  44. for (;;) {
  45. UniChar c[16]; // buffer of characters to add to output
  46. int j = 0; // offset to place next character in buffer
  47. int index; // index into output alphabet
  48. #define ADDCHAR(_X_) do { c[j++] = _X_; if (wrap && (++column == 64)) { column = 0; c[j++] = '\n'; } } while (0);
  49. // 1.index
  50. index = (ptr[i] >> 2) & 0x3F;
  51. ADDCHAR(alphabet[index]);
  52. // 2.index
  53. index = (ptr[i] << 4) & 0x30;
  54. if ((i+1) < len) {
  55. index = index | ((ptr[i+1] >> 4) & 0x0F);
  56. ADDCHAR(alphabet[index]);
  57. } else { // end of input, pad as necessary
  58. ADDCHAR(alphabet[index]);
  59. ADDCHAR('=');
  60. ADDCHAR('=');
  61. }
  62. // 3.index
  63. if ((i+1) < len) {
  64. index = (ptr[i+1] << 2) & 0x3C;
  65. if ((i+2) < len) {
  66. index = index | ((ptr[i+2] >> 6) & 0x03);
  67. ADDCHAR(alphabet[index]);
  68. } else { // end of input, pad as necessary
  69. ADDCHAR(alphabet[index]);
  70. ADDCHAR('=');
  71. }
  72. }
  73. // 4.index
  74. if ((i+2) < len) {
  75. index = (ptr[i+2]) & 0x3F;
  76. ADDCHAR(alphabet[index]);
  77. }
  78. CFStringAppendCharacters(string, c, j);
  79. i += 3; // we processed 3 bytes of input
  80. if (i >= len) {
  81. // end of data, append newline if we haven't already
  82. if (wrap && c[j-1] != '\n') {
  83. c[0] = '\n';
  84. CFStringAppendCharacters(string, c, 1);
  85. }
  86. break;
  87. }
  88. }
  89. return string;
  90. }
  91. enum ID3_Parser_State {
  92. ID3_Parser_State_Initial = 0,
  93. ID3_Parser_State_Parse_Frames,
  94. ID3_Parser_State_Tag_Parsed,
  95. ID3_Parser_State_Not_Valid_Tag
  96. };
  97. /*
  98. * =======================================
  99. * Private class
  100. * =======================================
  101. */
  102. class ID3_Parser_Private {
  103. public:
  104. ID3_Parser_Private();
  105. ~ID3_Parser_Private();
  106. bool wantData();
  107. void feedData(UInt8 *data, UInt32 numBytes);
  108. void setState(ID3_Parser_State state);
  109. void reset();
  110. CFStringRef parseContent(UInt32 framesize, UInt32 pos, CFStringEncoding encoding, bool byteOrderMark);
  111. ID3_Parser *m_parser;
  112. ID3_Parser_State m_state;
  113. UInt32 m_bytesReceived;
  114. UInt32 m_tagSize;
  115. UInt8 m_majorVersion;
  116. bool m_hasFooter;
  117. bool m_usesUnsynchronisation;
  118. bool m_usesExtendedHeader;
  119. CFStringRef m_title;
  120. CFStringRef m_performer;
  121. CFStringRef m_coverArt;
  122. std::vector<UInt8> m_tagData;
  123. };
  124. /*
  125. * =======================================
  126. * Private class implementation
  127. * =======================================
  128. */
  129. ID3_Parser_Private::ID3_Parser_Private() :
  130. m_parser(0),
  131. m_state(ID3_Parser_State_Initial),
  132. m_bytesReceived(0),
  133. m_tagSize(0),
  134. m_majorVersion(0),
  135. m_hasFooter(false),
  136. m_usesUnsynchronisation(false),
  137. m_usesExtendedHeader(false),
  138. m_title(NULL),
  139. m_performer(NULL),
  140. m_coverArt(NULL)
  141. {
  142. }
  143. ID3_Parser_Private::~ID3_Parser_Private()
  144. {
  145. if (m_performer) {
  146. CFRelease(m_performer);
  147. m_performer = NULL;
  148. }
  149. if (m_title) {
  150. CFRelease(m_title);
  151. m_title = NULL;
  152. }
  153. if (m_coverArt) {
  154. CFRelease(m_coverArt);
  155. m_coverArt = NULL;
  156. }
  157. }
  158. bool ID3_Parser_Private::wantData()
  159. {
  160. if (m_state == ID3_Parser_State_Tag_Parsed) {
  161. return false;
  162. }
  163. if (m_state == ID3_Parser_State_Not_Valid_Tag) {
  164. return false;
  165. }
  166. return true;
  167. }
  168. void ID3_Parser_Private::feedData(UInt8 *data, UInt32 numBytes)
  169. {
  170. if (!wantData()) {
  171. return;
  172. }
  173. m_bytesReceived += numBytes;
  174. ID3_TRACE("received %i bytes, total bytes %i\n", numBytes, m_bytesReceived);
  175. for (CFIndex i=0; i < numBytes; i++) {
  176. m_tagData.push_back(data[i]);
  177. }
  178. bool enoughBytesToParse = true;
  179. while (enoughBytesToParse) {
  180. switch (m_state) {
  181. case ID3_Parser_State_Initial: {
  182. // Do we have enough bytes to determine if this is an ID3 tag or not?
  183. if (m_bytesReceived <= 9) {
  184. enoughBytesToParse = false;
  185. break;
  186. }
  187. if (!(m_tagData[0] == 'I' &&
  188. m_tagData[1] == 'D' &&
  189. m_tagData[2] == '3')) {
  190. ID3_TRACE("Not an ID3 tag, bailing out\n");
  191. // Does not begin with the tag header; not an ID3 tag
  192. setState(ID3_Parser_State_Not_Valid_Tag);
  193. enoughBytesToParse = false;
  194. break;
  195. }
  196. m_majorVersion = m_tagData[3];
  197. // Currently support only id3v2.2 and 2.3
  198. if (m_majorVersion != 2 && m_majorVersion != 3) {
  199. ID3_TRACE("ID3v2.%i not supported by the parser\n", m_majorVersion);
  200. setState(ID3_Parser_State_Not_Valid_Tag);
  201. enoughBytesToParse = false;
  202. break;
  203. }
  204. // Ignore the revision
  205. // Parse the flags
  206. if ((m_tagData[5] & 0x80) != 0) {
  207. m_usesUnsynchronisation = true;
  208. } else if ((m_tagData[5] & 0x40) != 0 && m_majorVersion >= 3) {
  209. m_usesExtendedHeader = true;
  210. } else if ((m_tagData[5] & 0x10) != 0 && m_majorVersion >= 3) {
  211. m_hasFooter = true;
  212. }
  213. m_tagSize = ((m_tagData[6] & 0x7F) << 21) | ((m_tagData[7] & 0x7F) << 14) |
  214. ((m_tagData[8] & 0x7F) << 7) | (m_tagData[9] & 0x7F);
  215. if (m_tagSize > 0) {
  216. if (m_hasFooter) {
  217. m_tagSize += 10;
  218. }
  219. m_tagSize += 10;
  220. ID3_TRACE("tag size: %i\n", m_tagSize);
  221. if (m_parser->m_delegate) {
  222. m_parser->m_delegate->id3tagSizeAvailable(m_tagSize);
  223. }
  224. setState(ID3_Parser_State_Parse_Frames);
  225. break;
  226. }
  227. setState(ID3_Parser_State_Not_Valid_Tag);
  228. enoughBytesToParse = false;
  229. break;
  230. }
  231. case ID3_Parser_State_Parse_Frames: {
  232. // Do we have enough data to parse the frames?
  233. if (m_tagData.size() < m_tagSize) {
  234. ID3_TRACE("Not enough data received for parsing, have %lu bytes, need %i bytes\n",
  235. m_tagData.size(),
  236. m_tagSize);
  237. enoughBytesToParse = false;
  238. break;
  239. }
  240. UInt32 pos = 10;
  241. // Do we have an extended header? If we do, skip it
  242. if (m_usesExtendedHeader) {
  243. UInt32 extendedHeaderSize = ((m_tagData[pos] << 21) |
  244. (m_tagData[pos+1] << 14) |
  245. (m_tagData[pos+2] << 7) |
  246. m_tagData[pos+3]);
  247. if (pos + extendedHeaderSize >= m_tagSize) {
  248. setState(ID3_Parser_State_Not_Valid_Tag);
  249. enoughBytesToParse = false;
  250. break;
  251. }
  252. ID3_TRACE("Skipping extended header, size %i\n", extendedHeaderSize);
  253. pos += extendedHeaderSize;
  254. }
  255. while (pos < m_tagSize) {
  256. char frameName[5];
  257. frameName[0] = m_tagData[pos];
  258. frameName[1] = m_tagData[pos+1];
  259. frameName[2] = m_tagData[pos+2];
  260. if (m_majorVersion >= 3) {
  261. frameName[3] = m_tagData[pos+3];
  262. } else {
  263. frameName[3] = 0;
  264. }
  265. frameName[4] = 0;
  266. UInt32 framesize = 0;
  267. if (m_majorVersion >= 3) {
  268. pos += 4;
  269. framesize = ((m_tagData[pos] << 21) |
  270. (m_tagData[pos+1] << 14) |
  271. (m_tagData[pos+2] << 7) |
  272. m_tagData[pos+3]);
  273. } else {
  274. pos += 3;
  275. framesize = ((m_tagData[pos] << 16) |
  276. (m_tagData[pos+1] << 8) |
  277. m_tagData[pos+2]);
  278. }
  279. if (framesize == 0) {
  280. setState(ID3_Parser_State_Not_Valid_Tag);
  281. enoughBytesToParse = false;
  282. // Break from the loop and then out of the case context
  283. goto ParseFramesExit;
  284. }
  285. if (m_majorVersion >= 3) {
  286. pos += 6;
  287. } else {
  288. pos += 3;
  289. }
  290. CFStringEncoding encoding;
  291. bool byteOrderMark = false;
  292. if (m_tagData[pos] == 3) {
  293. encoding = kCFStringEncodingUTF8;
  294. } else if (m_tagData[pos] == 2) {
  295. encoding = kCFStringEncodingUTF16BE;
  296. } else if (m_tagData[pos] == 1) {
  297. encoding = kCFStringEncodingUTF16;
  298. byteOrderMark = true;
  299. } else {
  300. // ISO-8859-1 is the default encoding
  301. encoding = kCFStringEncodingISOLatin1;
  302. }
  303. if (!strcmp(frameName, "TIT2") || !strcmp(frameName, "TT2")) {
  304. if (m_title) {
  305. CFRelease(m_title);
  306. }
  307. m_title = parseContent(framesize, pos + 1, encoding, byteOrderMark);
  308. ID3_TRACE("ID3 title parsed: '%s'\n", CFStringGetCStringPtr(m_title, CFStringGetSystemEncoding()));
  309. } else if (!strcmp(frameName, "TPE1") || !strcmp(frameName, "TP1")) {
  310. if (m_performer) {
  311. CFRelease(m_performer);
  312. }
  313. m_performer = parseContent(framesize, pos + 1, encoding, byteOrderMark);
  314. ID3_TRACE("ID3 performer parsed: '%s'\n", CFStringGetCStringPtr(m_performer, CFStringGetSystemEncoding()));
  315. } else if (!strcmp(frameName, "APIC")) {
  316. char imageType[65] = {0};
  317. size_t dataPos = pos+1;
  318. for (int i=0; m_tagData[dataPos]; i++,dataPos++) {
  319. imageType[i] = m_tagData[dataPos];
  320. }
  321. dataPos++;
  322. if (!strcmp(imageType, "image/jpeg") ||
  323. !strcmp(imageType, "image/png")) {
  324. ID3_TRACE("Image type %s, parsing, dataPos %zu\n", imageType, dataPos);
  325. // Skip the image description
  326. while (!m_tagData[++dataPos]);
  327. const size_t coverArtSize = framesize - ((dataPos - pos) + 5);
  328. UInt8 *bytes = new UInt8[coverArtSize];
  329. for (int i=0; i < coverArtSize; i++) {
  330. bytes[i] = m_tagData[dataPos+i];
  331. }
  332. if (m_coverArt) {
  333. CFRelease(m_coverArt);
  334. }
  335. m_coverArt = createBase64EncodedString(bytes, coverArtSize, 0);
  336. delete [] bytes;
  337. } else {
  338. ID3_TRACE("%s is an unknown type for image data, skipping\n", imageType);
  339. }
  340. } else {
  341. // Unknown/unhandled frame
  342. ID3_TRACE("Unknown/unhandled frame: %s, size %i\n", frameName, framesize);
  343. }
  344. pos += framesize;
  345. }
  346. // Push out the metadata
  347. if (m_parser->m_delegate) {
  348. std::map<CFStringRef,CFStringRef> metadataMap;
  349. if (m_performer && CFStringGetLength(m_performer) > 0) {
  350. metadataMap[CFSTR("MPMediaItemPropertyArtist")] =
  351. CFStringCreateCopy(kCFAllocatorDefault, m_performer);
  352. }
  353. if (m_title && CFStringGetLength(m_title) > 0) {
  354. metadataMap[CFSTR("MPMediaItemPropertyTitle")] =
  355. CFStringCreateCopy(kCFAllocatorDefault, m_title);
  356. }
  357. if (m_coverArt && CFStringGetLength(m_coverArt) > 0) {
  358. metadataMap[CFSTR("CoverArt")] =
  359. CFStringCreateCopy(kCFAllocatorDefault, m_coverArt);
  360. }
  361. m_parser->m_delegate->id3metaDataAvailable(metadataMap);
  362. }
  363. setState(ID3_Parser_State_Tag_Parsed);
  364. enoughBytesToParse = false;
  365. ParseFramesExit:
  366. break;
  367. }
  368. default:
  369. enoughBytesToParse = false;
  370. break;
  371. }
  372. }
  373. }
  374. void ID3_Parser_Private::setState(astreamer::ID3_Parser_State state)
  375. {
  376. m_state = state;
  377. }
  378. void ID3_Parser_Private::reset()
  379. {
  380. m_state = ID3_Parser_State_Initial;
  381. m_bytesReceived = 0;
  382. m_tagSize = 0;
  383. m_majorVersion = 0;
  384. m_hasFooter = false;
  385. m_usesUnsynchronisation = false;
  386. m_usesExtendedHeader = false;
  387. if (m_title) {
  388. CFRelease(m_title);
  389. m_title = NULL;
  390. }
  391. if (m_performer) {
  392. CFRelease(m_performer);
  393. m_performer = NULL;
  394. }
  395. if (m_coverArt) {
  396. CFRelease(m_coverArt);
  397. m_coverArt = NULL;
  398. }
  399. m_tagData.clear();
  400. }
  401. CFStringRef ID3_Parser_Private::parseContent(UInt32 framesize, UInt32 pos, CFStringEncoding encoding, bool byteOrderMark)
  402. {
  403. CFStringRef content = CFStringCreateWithBytes(kCFAllocatorDefault,
  404. &m_tagData[pos],
  405. framesize - 1,
  406. encoding,
  407. byteOrderMark);
  408. return content;
  409. }
  410. /*
  411. * =======================================
  412. * ID3_Parser implementation
  413. * =======================================
  414. */
  415. ID3_Parser::ID3_Parser() :
  416. m_delegate(0),
  417. m_private(new ID3_Parser_Private())
  418. {
  419. m_private->m_parser = this;
  420. }
  421. ID3_Parser::~ID3_Parser()
  422. {
  423. delete m_private;
  424. m_private = 0;
  425. }
  426. void ID3_Parser::reset()
  427. {
  428. m_private->reset();
  429. }
  430. bool ID3_Parser::wantData()
  431. {
  432. return m_private->wantData();
  433. }
  434. void ID3_Parser::feedData(UInt8 *data, UInt32 numBytes)
  435. {
  436. m_private->feedData(data, numBytes);
  437. }
  438. }