sax.js 10.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359
  1. function parseSync(xstr, options) {
  2. const dummy = () => {};
  3. let {onStartNode: _onStartNode = dummy,
  4. onEndNode: _onEndNode = dummy,
  5. onTextNode: _onTextNode = dummy,
  6. onCdata: _onCdata = dummy,
  7. onComment: _onComment = dummy,
  8. onProgress: _onProgress = dummy,
  9. innerCut = new Set()
  10. } = options;
  11. let i = 0;
  12. const len = xstr.length;
  13. const progStep = len/20;
  14. let nextProg = 0;
  15. let cutCounter = 0;
  16. let cutTag = '';
  17. let inCdata;
  18. let inComment;
  19. let leftData = 0;
  20. while (i < len) {
  21. inCdata = false;
  22. inComment = false;
  23. let singleTag = false;
  24. let left = xstr.indexOf('<', i);
  25. if (left < 0)
  26. break;
  27. leftData = left;
  28. if (left < len - 2 && xstr[left + 1] == '!') {
  29. if (xstr[left + 2] == '-') {
  30. const leftComment = xstr.indexOf('<!--', left);
  31. if (leftComment == left) {
  32. inComment = true;
  33. leftData = left + 3;
  34. }
  35. }
  36. if (!inComment && xstr[left + 2] == '[') {
  37. const leftCdata = xstr.indexOf('<![CDATA[', left);
  38. if (leftCdata == left) {
  39. inCdata = true;
  40. leftData = left + 8;
  41. }
  42. }
  43. }
  44. if (left != i) {
  45. const text = xstr.substr(i, left - i);
  46. _onTextNode(text, cutCounter, cutTag);
  47. }
  48. let right = null;
  49. let rightData = null;
  50. if (inCdata) {
  51. rightData = xstr.indexOf(']]>', leftData + 1);
  52. if (rightData < 0)
  53. break;
  54. right = rightData + 2;
  55. } else if (inComment) {
  56. rightData = xstr.indexOf('-->', leftData + 1);
  57. if (rightData < 0)
  58. break;
  59. right = rightData + 2;
  60. } else {
  61. rightData = xstr.indexOf('>', leftData + 1);
  62. if (rightData < 0)
  63. break;
  64. right = rightData;
  65. if (xstr[right - 1] === '/') {
  66. singleTag = true;
  67. rightData--;
  68. }
  69. }
  70. let tagData = xstr.substr(leftData + 1, rightData - leftData - 1);
  71. if (inCdata) {
  72. _onCdata(tagData, cutCounter, cutTag);
  73. } else if (inComment) {
  74. _onComment(tagData, cutCounter, cutTag);
  75. } else {
  76. let tag = '';
  77. let tail = '';
  78. const firstSpace = tagData.indexOf(' ');
  79. if (firstSpace >= 0) {
  80. tail = tagData.substr(firstSpace);
  81. tag = tagData.substr(0, firstSpace);
  82. } else {
  83. tag = tagData;
  84. }
  85. tag = tag.toLowerCase();
  86. let endTag = (singleTag ? tag : '');
  87. if (tag === '' || tag[0] !== '/') {
  88. _onStartNode(tag, tail, singleTag, cutCounter, cutTag);
  89. } else {
  90. endTag = tag.substr(1);
  91. }
  92. if (endTag)
  93. _onEndNode(endTag, tail, singleTag, cutCounter, cutTag);
  94. if (innerCut.has(tag) && (!cutCounter || cutTag === tag)) {
  95. if (!cutCounter)
  96. cutTag = tag;
  97. cutCounter++;
  98. }
  99. if (cutTag === endTag) {
  100. cutCounter = (cutCounter > 0 ? cutCounter - 1 : 0);
  101. if (!cutCounter)
  102. cutTag = '';
  103. }
  104. }
  105. if (right >= nextProg) {
  106. _onProgress(Math.round(right/(len + 1)*100));
  107. nextProg += progStep;
  108. }
  109. i = right + 1;
  110. }
  111. if (i < len) {
  112. if (inCdata) {
  113. _onCdata(xstr.substr(leftData, len - leftData), cutCounter, cutTag);
  114. } else if (inComment) {
  115. _onComment(xstr.substr(leftData, len - leftData), cutCounter, cutTag);
  116. } else {
  117. _onTextNode(xstr.substr(i, len - i), cutCounter, cutTag);
  118. }
  119. }
  120. _onProgress(100);
  121. }
  122. //асинхронная копия parseSync
  123. //делается заменой "_on" => "await _on" после while
  124. async function parse(xstr, options) {
  125. const dummy = () => {};
  126. let {onStartNode: _onStartNode = dummy,
  127. onEndNode: _onEndNode = dummy,
  128. onTextNode: _onTextNode = dummy,
  129. onCdata: _onCdata = dummy,
  130. onComment: _onComment = dummy,
  131. onProgress: _onProgress = dummy,
  132. innerCut = new Set()
  133. } = options;
  134. let i = 0;
  135. const len = xstr.length;
  136. const progStep = len/20;
  137. let nextProg = 0;
  138. let cutCounter = 0;
  139. let cutTag = '';
  140. let inCdata;
  141. let inComment;
  142. let leftData = 0;
  143. while (i < len) {
  144. inCdata = false;
  145. inComment = false;
  146. let singleTag = false;
  147. let left = xstr.indexOf('<', i);
  148. if (left < 0)
  149. break;
  150. leftData = left;
  151. if (left < len - 2 && xstr[left + 1] == '!') {
  152. if (xstr[left + 2] == '-') {
  153. const leftComment = xstr.indexOf('<!--', left);
  154. if (leftComment == left) {
  155. inComment = true;
  156. leftData = left + 3;
  157. }
  158. }
  159. if (!inComment && xstr[left + 2] == '[') {
  160. const leftCdata = xstr.indexOf('<![CDATA[', left);
  161. if (leftCdata == left) {
  162. inCdata = true;
  163. leftData = left + 8;
  164. }
  165. }
  166. }
  167. if (left != i) {
  168. const text = xstr.substr(i, left - i);
  169. await _onTextNode(text, cutCounter, cutTag);
  170. }
  171. let right = null;
  172. let rightData = null;
  173. if (inCdata) {
  174. rightData = xstr.indexOf(']]>', leftData + 1);
  175. if (rightData < 0)
  176. break;
  177. right = rightData + 2;
  178. } else if (inComment) {
  179. rightData = xstr.indexOf('-->', leftData + 1);
  180. if (rightData < 0)
  181. break;
  182. right = rightData + 2;
  183. } else {
  184. rightData = xstr.indexOf('>', leftData + 1);
  185. if (rightData < 0)
  186. break;
  187. right = rightData;
  188. if (xstr[right - 1] === '/') {
  189. singleTag = true;
  190. rightData--;
  191. }
  192. }
  193. let tagData = xstr.substr(leftData + 1, rightData - leftData - 1);
  194. if (inCdata) {
  195. await _onCdata(tagData, cutCounter, cutTag);
  196. } else if (inComment) {
  197. await _onComment(tagData, cutCounter, cutTag);
  198. } else {
  199. let tag = '';
  200. let tail = '';
  201. const firstSpace = tagData.indexOf(' ');
  202. if (firstSpace >= 0) {
  203. tail = tagData.substr(firstSpace);
  204. tag = tagData.substr(0, firstSpace);
  205. } else {
  206. tag = tagData;
  207. }
  208. tag = tag.toLowerCase();
  209. let endTag = (singleTag ? tag : '');
  210. if (tag === '' || tag[0] !== '/') {
  211. await _onStartNode(tag, tail, singleTag, cutCounter, cutTag);
  212. } else {
  213. endTag = tag.substr(1);
  214. }
  215. if (endTag)
  216. await _onEndNode(endTag, tail, singleTag, cutCounter, cutTag);
  217. if (innerCut.has(tag) && (!cutCounter || cutTag === tag)) {
  218. if (!cutCounter)
  219. cutTag = tag;
  220. cutCounter++;
  221. }
  222. if (cutTag === endTag) {
  223. cutCounter = (cutCounter > 0 ? cutCounter - 1 : 0);
  224. if (!cutCounter)
  225. cutTag = '';
  226. }
  227. }
  228. if (right >= nextProg) {
  229. await _onProgress(Math.round(right/(len + 1)*100));
  230. nextProg += progStep;
  231. }
  232. i = right + 1;
  233. }
  234. if (i < len) {
  235. if (inCdata) {
  236. await _onCdata(xstr.substr(leftData, len - leftData), cutCounter, cutTag);
  237. } else if (inComment) {
  238. await _onComment(xstr.substr(leftData, len - leftData), cutCounter, cutTag);
  239. } else {
  240. await _onTextNode(xstr.substr(i, len - i), cutCounter, cutTag);
  241. }
  242. }
  243. await _onProgress(100);
  244. }
  245. function getAttrsSync(tail) {
  246. let result = {};
  247. let name = '';
  248. let value = '';
  249. let vOpen = '';
  250. let inName = false;
  251. let inValue = false;
  252. let waitValue = false;
  253. let waitEq = false;
  254. const pushResult = () => {
  255. if (name != '') {
  256. let ns = '';
  257. if (name.indexOf(':') >= 0) {
  258. [ns, name] = name.split(':');
  259. }
  260. result[name] = {value, ns};
  261. }
  262. name = '';
  263. value = '';
  264. vOpen = '';
  265. inName = false;
  266. inValue = false;
  267. waitValue = false;
  268. waitEq = false;
  269. };
  270. tail = tail.replace(/[\t\n\r]/g, ' ');
  271. for (let i = 0; i < tail.length; i++) {
  272. const c = tail.charAt(i);
  273. if (c == ' ') {
  274. if (inValue) {
  275. if (vOpen == '"')
  276. value += c;
  277. else
  278. pushResult();
  279. } else if (inName) {
  280. waitEq = true;
  281. inName = false;
  282. }
  283. } else if (!inValue && c == '=') {
  284. waitEq = false;
  285. waitValue = true;
  286. inName = false;
  287. } else if (c == '"') {
  288. if (inValue) {
  289. pushResult();
  290. } else if (waitValue) {
  291. inValue = true;
  292. vOpen = '"';
  293. }
  294. } else if (inValue) {
  295. value += c;
  296. } else if (inName) {
  297. name += c;
  298. } else if (waitEq) {
  299. pushResult();
  300. inName = true;
  301. name = c;
  302. } else if (waitValue) {
  303. waitValue = false;
  304. inValue = true;
  305. vOpen = ' ';
  306. value = c;
  307. } else {
  308. inName = true;
  309. name = c;
  310. }
  311. }
  312. if (name != '')
  313. pushResult();
  314. return result;
  315. }
  316. module.exports = {
  317. parseSync,
  318. getAttrsSync,
  319. parse
  320. }