Multipart.js 9.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356
  1. /* eslint-disable no-fallthrough */
  2. /* eslint-disable no-bitwise */
  3. /* eslint-disable no-plusplus */
  4. /* eslint-disable no-underscore-dangle */
  5. import { Transform } from 'node:stream';
  6. import * as errors from '../FormidableError.js';
  7. import FormidableError from '../FormidableError.js';
  8. let s = 0;
  9. const STATE = {
  10. PARSER_UNINITIALIZED: s++,
  11. START: s++,
  12. START_BOUNDARY: s++,
  13. HEADER_FIELD_START: s++,
  14. HEADER_FIELD: s++,
  15. HEADER_VALUE_START: s++,
  16. HEADER_VALUE: s++,
  17. HEADER_VALUE_ALMOST_DONE: s++,
  18. HEADERS_ALMOST_DONE: s++,
  19. PART_DATA_START: s++,
  20. PART_DATA: s++,
  21. PART_END: s++,
  22. END: s++,
  23. };
  24. let f = 1;
  25. const FBOUNDARY = { PART_BOUNDARY: f, LAST_BOUNDARY: (f *= 2) };
  26. const LF = 10;
  27. const CR = 13;
  28. const SPACE = 32;
  29. const HYPHEN = 45;
  30. const COLON = 58;
  31. const A = 97;
  32. const Z = 122;
  33. function lower(c) {
  34. return c | 0x20;
  35. }
  36. export const STATES = {};
  37. Object.keys(STATE).forEach((stateName) => {
  38. STATES[stateName] = STATE[stateName];
  39. });
  40. class MultipartParser extends Transform {
  41. constructor(options = {}) {
  42. super({ readableObjectMode: true });
  43. this.boundary = null;
  44. this.boundaryChars = null;
  45. this.lookbehind = null;
  46. this.bufferLength = 0;
  47. this.state = STATE.PARSER_UNINITIALIZED;
  48. this.globalOptions = { ...options };
  49. this.index = null;
  50. this.flags = 0;
  51. }
  52. _endUnexpected() {
  53. return new FormidableError(
  54. `MultipartParser.end(): stream ended unexpectedly: ${this.explain()}`,
  55. errors.malformedMultipart,
  56. 400,
  57. );
  58. }
  59. _flush(done) {
  60. if (
  61. (this.state === STATE.HEADER_FIELD_START && this.index === 0) ||
  62. (this.state === STATE.PART_DATA && this.index === this.boundary.length)
  63. ) {
  64. this._handleCallback('partEnd');
  65. this._handleCallback('end');
  66. done();
  67. } else if (this.state !== STATE.END) {
  68. done(this._endUnexpected());
  69. } else {
  70. done();
  71. }
  72. }
  73. initWithBoundary(str) {
  74. this.boundary = Buffer.from(`\r\n--${str}`);
  75. this.lookbehind = Buffer.alloc(this.boundary.length + 8);
  76. this.state = STATE.START;
  77. this.boundaryChars = {};
  78. for (let i = 0; i < this.boundary.length; i++) {
  79. this.boundaryChars[this.boundary[i]] = true;
  80. }
  81. }
  82. // eslint-disable-next-line max-params
  83. _handleCallback(name, buf, start, end) {
  84. if (start !== undefined && start === end) {
  85. return;
  86. }
  87. this.push({ name, buffer: buf, start, end });
  88. }
  89. // eslint-disable-next-line max-statements
  90. _transform(buffer, _, done) {
  91. let i = 0;
  92. let prevIndex = this.index;
  93. let { index, state, flags } = this;
  94. const { lookbehind, boundary, boundaryChars } = this;
  95. const boundaryLength = boundary.length;
  96. const boundaryEnd = boundaryLength - 1;
  97. this.bufferLength = buffer.length;
  98. let c = null;
  99. let cl = null;
  100. const setMark = (name, idx) => {
  101. this[`${name}Mark`] = typeof idx === 'number' ? idx : i;
  102. };
  103. const clearMarkSymbol = (name) => {
  104. delete this[`${name}Mark`];
  105. };
  106. const dataCallback = (name, shouldClear) => {
  107. const markSymbol = `${name}Mark`;
  108. if (!(markSymbol in this)) {
  109. return;
  110. }
  111. if (!shouldClear) {
  112. this._handleCallback(name, buffer, this[markSymbol], buffer.length);
  113. setMark(name, 0);
  114. } else {
  115. this._handleCallback(name, buffer, this[markSymbol], i);
  116. clearMarkSymbol(name);
  117. }
  118. };
  119. for (i = 0; i < this.bufferLength; i++) {
  120. c = buffer[i];
  121. switch (state) {
  122. case STATE.PARSER_UNINITIALIZED:
  123. done(this._endUnexpected());
  124. return;
  125. case STATE.START:
  126. index = 0;
  127. state = STATE.START_BOUNDARY;
  128. case STATE.START_BOUNDARY:
  129. if (index === boundary.length - 2) {
  130. if (c === HYPHEN) {
  131. flags |= FBOUNDARY.LAST_BOUNDARY;
  132. } else if (c !== CR) {
  133. done(this._endUnexpected());
  134. return;
  135. }
  136. index++;
  137. break;
  138. } else if (index - 1 === boundary.length - 2) {
  139. if (flags & FBOUNDARY.LAST_BOUNDARY && c === HYPHEN) {
  140. this._handleCallback('end');
  141. state = STATE.END;
  142. flags = 0;
  143. } else if (!(flags & FBOUNDARY.LAST_BOUNDARY) && c === LF) {
  144. index = 0;
  145. this._handleCallback('partBegin');
  146. state = STATE.HEADER_FIELD_START;
  147. } else {
  148. done(this._endUnexpected());
  149. return;
  150. }
  151. break;
  152. }
  153. if (c !== boundary[index + 2]) {
  154. index = -2;
  155. }
  156. if (c === boundary[index + 2]) {
  157. index++;
  158. }
  159. break;
  160. case STATE.HEADER_FIELD_START:
  161. state = STATE.HEADER_FIELD;
  162. setMark('headerField');
  163. index = 0;
  164. case STATE.HEADER_FIELD:
  165. if (c === CR) {
  166. clearMarkSymbol('headerField');
  167. state = STATE.HEADERS_ALMOST_DONE;
  168. break;
  169. }
  170. index++;
  171. if (c === HYPHEN) {
  172. break;
  173. }
  174. if (c === COLON) {
  175. if (index === 1) {
  176. // empty header field
  177. done(this._endUnexpected());
  178. return;
  179. }
  180. dataCallback('headerField', true);
  181. state = STATE.HEADER_VALUE_START;
  182. break;
  183. }
  184. cl = lower(c);
  185. if (cl < A || cl > Z) {
  186. done(this._endUnexpected());
  187. return;
  188. }
  189. break;
  190. case STATE.HEADER_VALUE_START:
  191. if (c === SPACE) {
  192. break;
  193. }
  194. setMark('headerValue');
  195. state = STATE.HEADER_VALUE;
  196. case STATE.HEADER_VALUE:
  197. if (c === CR) {
  198. dataCallback('headerValue', true);
  199. this._handleCallback('headerEnd');
  200. state = STATE.HEADER_VALUE_ALMOST_DONE;
  201. }
  202. break;
  203. case STATE.HEADER_VALUE_ALMOST_DONE:
  204. if (c !== LF) {
  205. done(this._endUnexpected());
  206. return;
  207. }
  208. state = STATE.HEADER_FIELD_START;
  209. break;
  210. case STATE.HEADERS_ALMOST_DONE:
  211. if (c !== LF) {
  212. done(this._endUnexpected());
  213. return;
  214. }
  215. this._handleCallback('headersEnd');
  216. state = STATE.PART_DATA_START;
  217. break;
  218. case STATE.PART_DATA_START:
  219. state = STATE.PART_DATA;
  220. setMark('partData');
  221. case STATE.PART_DATA:
  222. prevIndex = index;
  223. if (index === 0) {
  224. // boyer-moore derived algorithm to safely skip non-boundary data
  225. i += boundaryEnd;
  226. while (i < this.bufferLength && !(buffer[i] in boundaryChars)) {
  227. i += boundaryLength;
  228. }
  229. i -= boundaryEnd;
  230. c = buffer[i];
  231. }
  232. if (index < boundary.length) {
  233. if (boundary[index] === c) {
  234. if (index === 0) {
  235. dataCallback('partData', true);
  236. }
  237. index++;
  238. } else {
  239. index = 0;
  240. }
  241. } else if (index === boundary.length) {
  242. index++;
  243. if (c === CR) {
  244. // CR = part boundary
  245. flags |= FBOUNDARY.PART_BOUNDARY;
  246. } else if (c === HYPHEN) {
  247. // HYPHEN = end boundary
  248. flags |= FBOUNDARY.LAST_BOUNDARY;
  249. } else {
  250. index = 0;
  251. }
  252. } else if (index - 1 === boundary.length) {
  253. if (flags & FBOUNDARY.PART_BOUNDARY) {
  254. index = 0;
  255. if (c === LF) {
  256. // unset the PART_BOUNDARY flag
  257. flags &= ~FBOUNDARY.PART_BOUNDARY;
  258. this._handleCallback('partEnd');
  259. this._handleCallback('partBegin');
  260. state = STATE.HEADER_FIELD_START;
  261. break;
  262. }
  263. } else if (flags & FBOUNDARY.LAST_BOUNDARY) {
  264. if (c === HYPHEN) {
  265. this._handleCallback('partEnd');
  266. this._handleCallback('end');
  267. state = STATE.END;
  268. flags = 0;
  269. } else {
  270. index = 0;
  271. }
  272. } else {
  273. index = 0;
  274. }
  275. }
  276. if (index > 0) {
  277. // when matching a possible boundary, keep a lookbehind reference
  278. // in case it turns out to be a false lead
  279. lookbehind[index - 1] = c;
  280. } else if (prevIndex > 0) {
  281. // if our boundary turned out to be rubbish, the captured lookbehind
  282. // belongs to partData
  283. this._handleCallback('partData', lookbehind, 0, prevIndex);
  284. prevIndex = 0;
  285. setMark('partData');
  286. // reconsider the current character even so it interrupted the sequence
  287. // it could be the beginning of a new sequence
  288. i--;
  289. }
  290. break;
  291. case STATE.END:
  292. break;
  293. default:
  294. done(this._endUnexpected());
  295. return;
  296. }
  297. }
  298. dataCallback('headerField');
  299. dataCallback('headerValue');
  300. dataCallback('partData');
  301. this.index = index;
  302. this.state = state;
  303. this.flags = flags;
  304. done();
  305. return this.bufferLength;
  306. }
  307. explain() {
  308. return `state = ${MultipartParser.stateToString(this.state)}`;
  309. }
  310. }
  311. // eslint-disable-next-line consistent-return
  312. MultipartParser.stateToString = (stateNumber) => {
  313. // eslint-disable-next-line no-restricted-syntax, guard-for-in
  314. for (const stateName in STATE) {
  315. const number = STATE[stateName];
  316. if (number === stateNumber) return stateName;
  317. }
  318. };
  319. export default Object.assign(MultipartParser, { STATES });