pickle.js 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562
  1. /* jshint esversion: 6 */
  2. var pickle = pickle || {};
  3. pickle.Unpickler = class {
  4. constructor(buffer) {
  5. this._reader = new pickle.Reader(buffer, 0);
  6. }
  7. load(function_call, persistent_load) {
  8. const reader = this._reader;
  9. const marker = [];
  10. let stack = [];
  11. const memo = new Map();
  12. while (reader.position < reader.length) {
  13. const opcode = reader.byte();
  14. switch (opcode) {
  15. case pickle.OpCode.PROTO: {
  16. const version = reader.byte();
  17. if (version > 5) {
  18. throw new pickle.Error("Unsupported protocol version '" + version + "'.");
  19. }
  20. break;
  21. }
  22. case pickle.OpCode.GLOBAL:
  23. stack.push([ reader.line(), reader.line() ].join('.'));
  24. break;
  25. case pickle.OpCode.STACK_GLOBAL:
  26. stack.push([ stack.pop(), stack.pop() ].reverse().join('.'));
  27. break;
  28. case pickle.OpCode.PUT: {
  29. const index = parseInt(reader.line(), 10);
  30. memo.set(index, stack[stack.length - 1]);
  31. break;
  32. }
  33. case pickle.OpCode.OBJ: {
  34. const items = stack;
  35. stack = marker.pop();
  36. stack.push(function_call(items.pop(), items));
  37. break;
  38. }
  39. case pickle.OpCode.GET: {
  40. const index = parseInt(reader.line(), 10);
  41. stack.push(memo.get(index));
  42. break;
  43. }
  44. case pickle.OpCode.POP:
  45. stack.pop();
  46. break;
  47. case pickle.OpCode.POP_MARK:
  48. stack = marker.pop();
  49. break;
  50. case pickle.OpCode.DUP:
  51. stack.push(stack[stack.length-1]);
  52. break;
  53. case pickle.OpCode.PERSID:
  54. stack.push(persistent_load(reader.line()));
  55. break;
  56. case pickle.OpCode.BINPERSID:
  57. stack.push(persistent_load(stack.pop()));
  58. break;
  59. case pickle.OpCode.REDUCE: {
  60. const items = stack.pop();
  61. const type = stack.pop();
  62. stack.push(function_call(type, items));
  63. break;
  64. }
  65. case pickle.OpCode.NEWOBJ: {
  66. const items = stack.pop();
  67. const type = stack.pop();
  68. stack.push(function_call(type, items));
  69. break;
  70. }
  71. case pickle.OpCode.BINGET:
  72. stack.push(memo.get(reader.byte()));
  73. break;
  74. case pickle.OpCode.LONG_BINGET:
  75. stack.push(memo.get(reader.uint32()));
  76. break;
  77. case pickle.OpCode.BINPUT:
  78. memo.set(reader.byte(), stack[stack.length - 1]);
  79. break;
  80. case pickle.OpCode.LONG_BINPUT:
  81. memo.set(reader.uint32(), stack[stack.length - 1]);
  82. break;
  83. case pickle.OpCode.BININT:
  84. stack.push(reader.int32());
  85. break;
  86. case pickle.OpCode.BININT1:
  87. stack.push(reader.byte());
  88. break;
  89. case pickle.OpCode.LONG:
  90. stack.push(parseInt(reader.line(), 10));
  91. break;
  92. case pickle.OpCode.BININT2:
  93. stack.push(reader.uint16());
  94. break;
  95. case pickle.OpCode.BINBYTES:
  96. stack.push(reader.bytes(reader.int32()));
  97. break;
  98. case pickle.OpCode.SHORT_BINBYTES:
  99. stack.push(reader.bytes(reader.byte()));
  100. break;
  101. case pickle.OpCode.FLOAT:
  102. stack.push(parseFloat(reader.line()));
  103. break;
  104. case pickle.OpCode.BINFLOAT:
  105. stack.push(reader.float64());
  106. break;
  107. case pickle.OpCode.INT: {
  108. const value = reader.line();
  109. if (value == '01') {
  110. stack.push(true);
  111. }
  112. else if (value == '00') {
  113. stack.push(false);
  114. }
  115. else {
  116. stack.push(parseInt(value, 10));
  117. }
  118. break;
  119. }
  120. case pickle.OpCode.EMPTY_LIST:
  121. stack.push([]);
  122. break;
  123. case pickle.OpCode.EMPTY_TUPLE:
  124. stack.push([]);
  125. break;
  126. case pickle.OpCode.EMPTY_SET:
  127. stack.push([]);
  128. break;
  129. case pickle.OpCode.ADDITEMS: {
  130. const items = stack;
  131. stack = marker.pop();
  132. const obj = stack[stack.length - 1];
  133. for (let i = 0; i < items.length; i++) {
  134. obj.push(items[i]);
  135. }
  136. break;
  137. }
  138. case pickle.OpCode.DICT: {
  139. const items = stack;
  140. stack = marker.pop();
  141. const dict = {};
  142. for (let i = 0; i < items.length; i += 2) {
  143. dict[items[i]] = items[i + 1];
  144. }
  145. stack.push(dict);
  146. break;
  147. }
  148. case pickle.OpCode.LIST: {
  149. const items = stack;
  150. stack = marker.pop();
  151. stack.push(items);
  152. break;
  153. }
  154. case pickle.OpCode.TUPLE: {
  155. const items = stack;
  156. stack = marker.pop();
  157. stack.push(items);
  158. break;
  159. }
  160. case pickle.OpCode.SETITEM: {
  161. const value = stack.pop();
  162. const key = stack.pop();
  163. const obj = stack[stack.length - 1];
  164. if (obj.__setitem__) {
  165. obj.__setitem__(key, value);
  166. }
  167. else {
  168. obj[key] = value;
  169. }
  170. break;
  171. }
  172. case pickle.OpCode.SETITEMS: {
  173. const items = stack;
  174. stack = marker.pop();
  175. const obj = stack[stack.length - 1];
  176. for (let i = 0; i < items.length; i += 2) {
  177. if (obj.__setitem__) {
  178. obj.__setitem__(items[i], items[i + 1]);
  179. }
  180. else {
  181. obj[items[i]] = items[i + 1];
  182. }
  183. }
  184. break;
  185. }
  186. case pickle.OpCode.EMPTY_DICT:
  187. stack.push({});
  188. break;
  189. case pickle.OpCode.APPEND: {
  190. const append = stack.pop();
  191. stack[stack.length-1].push(append);
  192. break;
  193. }
  194. case pickle.OpCode.APPENDS: {
  195. const appends = stack;
  196. stack = marker.pop();
  197. const list = stack[stack.length - 1];
  198. list.push.apply(list, appends);
  199. break;
  200. }
  201. case pickle.OpCode.STRING: {
  202. const str = reader.line();
  203. stack.push(str.substr(1, str.length - 2));
  204. break;
  205. }
  206. case pickle.OpCode.BINSTRING:
  207. stack.push(reader.string(reader.uint32()));
  208. break;
  209. case pickle.OpCode.SHORT_BINSTRING:
  210. stack.push(reader.string(reader.byte()));
  211. break;
  212. case pickle.OpCode.UNICODE:
  213. stack.push(reader.line());
  214. break;
  215. case pickle.OpCode.BINUNICODE:
  216. stack.push(reader.string(reader.uint32(), 'utf-8'));
  217. break;
  218. case pickle.OpCode.SHORT_BINUNICODE:
  219. stack.push(reader.string(reader.byte(), 'utf-8'));
  220. break;
  221. case pickle.OpCode.BUILD: {
  222. const state = stack.pop();
  223. let obj = stack.pop();
  224. if (obj.__setstate__) {
  225. if (obj.__setstate__.__call__) {
  226. obj.__setstate__.__call__([ obj, state ]);
  227. }
  228. else {
  229. obj.__setstate__(state);
  230. }
  231. }
  232. else {
  233. for (const p in state) {
  234. obj[p] = state[p];
  235. }
  236. }
  237. if (obj.__read__) {
  238. obj = obj.__read__(this);
  239. }
  240. stack.push(obj);
  241. break;
  242. }
  243. case pickle.OpCode.MARK:
  244. marker.push(stack);
  245. stack = [];
  246. break;
  247. case pickle.OpCode.NEWTRUE:
  248. stack.push(true);
  249. break;
  250. case pickle.OpCode.NEWFALSE:
  251. stack.push(false);
  252. break;
  253. case pickle.OpCode.LONG1: {
  254. const data = reader.bytes(reader.byte());
  255. let number = 0;
  256. switch (data.length) {
  257. case 0: number = 0; break;
  258. case 1: number = data[0]; break;
  259. case 2: number = data[1] << 8 | data[0]; break;
  260. case 3: number = data[2] << 16 | data[1] << 8 | data[0]; break;
  261. case 4: number = data[3] << 24 | data[2] << 16 | data[1] << 8 | data[0]; break;
  262. default: number = Array.prototype.slice.call(data, 0); break;
  263. }
  264. stack.push(number);
  265. break;
  266. }
  267. case pickle.OpCode.LONG4:
  268. // TODO decode LONG4
  269. stack.push(reader.bytes(reader.uint32()));
  270. break;
  271. case pickle.OpCode.TUPLE1:
  272. stack.push([ stack.pop() ]);
  273. break;
  274. case pickle.OpCode.TUPLE2: {
  275. const b = stack.pop();
  276. const a = stack.pop();
  277. stack.push([ a, b ]);
  278. break;
  279. }
  280. case pickle.OpCode.TUPLE3: {
  281. const c = stack.pop();
  282. const b = stack.pop();
  283. const a = stack.pop();
  284. stack.push([ a, b, c ]);
  285. break;
  286. }
  287. case pickle.OpCode.MEMOIZE:
  288. memo.set(memo.size, stack[stack.length - 1]);
  289. break;
  290. case pickle.OpCode.FRAME:
  291. reader.bytes(8);
  292. break;
  293. case pickle.OpCode.NONE:
  294. stack.push(null);
  295. break;
  296. case pickle.OpCode.STOP:
  297. return stack.pop();
  298. default:
  299. throw new pickle.Error("Unknown opcode '" + opcode + "'.");
  300. }
  301. }
  302. throw new pickle.Error('Unexpected end of file.');
  303. }
  304. read(size) {
  305. return this._reader.bytes(size);
  306. }
  307. unescape(token, size) {
  308. const length = token.length;
  309. const a = new Uint8Array(length);
  310. if (size && size == length) {
  311. for (let p = 0; p < size; p++) {
  312. a[p] = token.charCodeAt(p);
  313. }
  314. return a;
  315. }
  316. let i = 0;
  317. let o = 0;
  318. while (i < length) {
  319. let c = token.charCodeAt(i++);
  320. if (c !== 0x5C || i >= length) {
  321. a[o++] = c;
  322. }
  323. else {
  324. c = token.charCodeAt(i++);
  325. switch (c) {
  326. case 0x27: a[o++] = 0x27; break; // '
  327. case 0x5C: a[o++] = 0x5C; break; // \\
  328. case 0x22: a[o++] = 0x22; break; // "
  329. case 0x72: a[o++] = 0x0D; break; // \r
  330. case 0x6E: a[o++] = 0x0A; break; // \n
  331. case 0x74: a[o++] = 0x09; break; // \t
  332. case 0x62: a[o++] = 0x08; break; // \b
  333. case 0x58: // x
  334. case 0x78: { // X
  335. const xsi = i - 1;
  336. const xso = o;
  337. for (let xi = 0; xi < 2; xi++) {
  338. if (i >= length) {
  339. i = xsi;
  340. o = xso;
  341. a[o] = 0x5c;
  342. break;
  343. }
  344. let xd = token.charCodeAt(i++);
  345. xd = xd >= 65 && xd <= 70 ? xd - 55 : xd >= 97 && xd <= 102 ? xd - 87 : xd >= 48 && xd <= 57 ? xd - 48 : -1;
  346. if (xd === -1) {
  347. i = xsi;
  348. o = xso;
  349. a[o] = 0x5c;
  350. break;
  351. }
  352. a[o] = a[o] << 4 | xd;
  353. }
  354. o++;
  355. break;
  356. }
  357. default:
  358. if (c < 48 || c > 57) { // 0-9
  359. a[o++] = 0x5c;
  360. a[o++] = c;
  361. }
  362. else {
  363. i--;
  364. const osi = i;
  365. const oso = o;
  366. for (let oi = 0; oi < 3; oi++) {
  367. if (i >= length) {
  368. i = osi;
  369. o = oso;
  370. a[o] = 0x5c;
  371. break;
  372. }
  373. const od = token.charCodeAt(i++);
  374. if (od < 48 || od > 57) {
  375. i = osi;
  376. o = oso;
  377. a[o] = 0x5c;
  378. break;
  379. }
  380. a[o] = a[o] << 3 | od - 48;
  381. }
  382. o++;
  383. }
  384. break;
  385. }
  386. }
  387. }
  388. return a.slice(0, o);
  389. }
  390. };
  391. // https://svn.python.org/projects/python/trunk/Lib/pickletools.py
  392. // https://github.com/python/cpython/blob/master/Lib/pickle.py
  393. pickle.OpCode = {
  394. MARK: 40, // '('
  395. EMPTY_TUPLE: 41, // ')'
  396. STOP: 46, // '.'
  397. POP: 48, // '0'
  398. POP_MARK: 49, // '1'
  399. DUP: 50, // '2'
  400. BINBYTES: 66, // 'B' (Protocol 3)
  401. SHORT_BINBYTES: 67, // 'C' (Protocol 3)
  402. FLOAT: 70, // 'F'
  403. BINFLOAT: 71, // 'G'
  404. INT: 73, // 'I'
  405. BININT: 74, // 'J'
  406. BININT1: 75, // 'K'
  407. LONG: 76, // 'L'
  408. BININT2: 77, // 'M'
  409. NONE: 78, // 'N'
  410. PERSID: 80, // 'P'
  411. BINPERSID: 81, // 'Q'
  412. REDUCE: 82, // 'R'
  413. STRING: 83, // 'S'
  414. BINSTRING: 84, // 'T'
  415. SHORT_BINSTRING: 85, // 'U'
  416. UNICODE: 86, // 'V'
  417. BINUNICODE: 88, // 'X'
  418. EMPTY_LIST: 93, // ']'
  419. APPEND: 97, // 'a'
  420. BUILD: 98, // 'b'
  421. GLOBAL: 99, // 'c'
  422. DICT: 100, // 'd'
  423. APPENDS: 101, // 'e'
  424. GET: 103, // 'g'
  425. BINGET: 104, // 'h'
  426. LONG_BINGET: 106, // 'j'
  427. LIST: 108, // 'l'
  428. OBJ: 111, // 'o'
  429. PUT: 112, // 'p'
  430. BINPUT: 113, // 'q'
  431. LONG_BINPUT: 114, // 'r'
  432. SETITEM: 115, // 's'
  433. TUPLE: 116, // 't'
  434. SETITEMS: 117, // 'u'
  435. EMPTY_DICT: 125, // '}'
  436. PROTO: 128,
  437. NEWOBJ: 129,
  438. TUPLE1: 133, // '\x85'
  439. TUPLE2: 134, // '\x86'
  440. TUPLE3: 135, // '\x87'
  441. NEWTRUE: 136, // '\x88'
  442. NEWFALSE: 137, // '\x89'
  443. LONG1: 138, // '\x8a'
  444. LONG4: 139, // '\x8b'
  445. SHORT_BINUNICODE: 140, // '\x8c' (Protocol 4)
  446. BINUNICODE8: 141, // '\x8d' (Protocol 4)
  447. BINBYTES8: 142, // '\x8e' (Protocol 4)
  448. EMPTY_SET: 143, // '\x8f' (Protocol 4)
  449. ADDITEMS: 144, // '\x90' (Protocol 4)
  450. FROZENSET: 145, // '\x91' (Protocol 4)
  451. NEWOBJ_EX: 146, // '\x92' (Protocol 4)
  452. STACK_GLOBAL: 147, // '\x93' (Protocol 4)
  453. MEMOIZE: 148, // '\x94' (Protocol 4)
  454. FRAME: 149 // '\x95' (Protocol 4)
  455. };
  456. pickle.Reader = class {
  457. constructor(buffer) {
  458. if (buffer) {
  459. this._buffer = buffer;
  460. this._dataView = new DataView(buffer.buffer, buffer.byteOffset, buffer.byteLength);
  461. this._position = 0;
  462. }
  463. pickle.Reader._utf8Decoder = pickle.Reader._utf8Decoder || new TextDecoder('utf-8');
  464. pickle.Reader._asciiDecoder = pickle.Reader._asciiDecoder || new TextDecoder('ascii');
  465. }
  466. get length() {
  467. return this._buffer.byteLength;
  468. }
  469. get position() {
  470. return this._position;
  471. }
  472. byte() {
  473. const position = this._position;
  474. this.skip(1);
  475. return this._dataView.getUint8(position);
  476. }
  477. bytes(length) {
  478. const position = this._position;
  479. this.skip(length);
  480. return this._buffer.subarray(position, this._position);
  481. }
  482. uint16() {
  483. const position = this.position;
  484. this.skip(2);
  485. return this._dataView.getUint16(position, true);
  486. }
  487. int32() {
  488. const position = this.position;
  489. this.skip(4);
  490. return this._dataView.getInt32(position, true);
  491. }
  492. uint32() {
  493. const position = this.position;
  494. this.skip(4);
  495. return this._dataView.getUint32(position, true);
  496. }
  497. float32() {
  498. const position = this.position;
  499. this.skip(4);
  500. return this._dataView.getFloat32(position, true);
  501. }
  502. float64() {
  503. const position = this.position;
  504. this.skip(8);
  505. return this._dataView.getFloat64(position, true);
  506. }
  507. skip(offset) {
  508. this._position += offset;
  509. if (this._position > this._buffer.length) {
  510. throw new pickle.Error('Expected ' + (this._position - this._buffer.length) + ' more bytes. The file might be corrupted. Unexpected end of file.');
  511. }
  512. }
  513. string(size, encoding) {
  514. const data = this.bytes(size);
  515. return (encoding == 'utf-8') ?
  516. pickle.Reader._utf8Decoder.decode(data) :
  517. pickle.Reader._asciiDecoder.decode(data);
  518. }
  519. line() {
  520. const index = this._buffer.indexOf(0x0A, this._position);
  521. if (index == -1) {
  522. throw new pickle.Error("Could not find end of line.");
  523. }
  524. const size = index - this._position;
  525. const text = this.string(size, 'ascii');
  526. this.skip(1);
  527. return text;
  528. }
  529. };
  530. pickle.Error = class extends Error {
  531. constructor(message) {
  532. super(message);
  533. this.name = 'Unpickle Error';
  534. }
  535. };
  536. if (typeof module !== 'undefined' && typeof module.exports === 'object') {
  537. module.exports.Unpickler = pickle.Unpickler;
  538. }