sax.js 8.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278
  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/10;
  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. const text = xstr.substr(i, left - i);
  45. _onTextNode(text, cutCounter, cutTag);
  46. let right = null;
  47. let rightData = null;
  48. if (inCdata) {
  49. rightData = xstr.indexOf(']]>', leftData + 1);
  50. if (rightData < 0)
  51. break;
  52. right = rightData + 2;
  53. } else if (inComment) {
  54. rightData = xstr.indexOf('-->', leftData + 1);
  55. if (rightData < 0)
  56. break;
  57. right = rightData + 2;
  58. } else {
  59. rightData = xstr.indexOf('>', leftData + 1);
  60. if (rightData < 0)
  61. break;
  62. right = rightData;
  63. if (xstr[right - 1] === '/') {
  64. singleTag = true;
  65. rightData--;
  66. }
  67. }
  68. let tagData = xstr.substr(leftData + 1, rightData - leftData - 1);
  69. if (inCdata) {
  70. _onCdata(tagData, cutCounter, cutTag);
  71. } else if (inComment) {
  72. _onComment(tagData, cutCounter, cutTag);
  73. } else {
  74. let tag = '';
  75. let tail = '';
  76. const firstSpace = tagData.indexOf(' ');
  77. if (firstSpace >= 0) {
  78. tail = tagData.substr(firstSpace);
  79. tag = tagData.substr(0, firstSpace);
  80. } else {
  81. tag = tagData;
  82. }
  83. tag = tag.toLowerCase();
  84. let endTag = (singleTag ? tag : '');
  85. if (tag === '' || tag[0] !== '/') {
  86. _onStartNode(tag, tail, singleTag, cutCounter, cutTag);
  87. } else {
  88. endTag = tag.substr(1);
  89. }
  90. if (endTag)
  91. _onEndNode(endTag, tail, singleTag, cutCounter, cutTag);
  92. if (innerCut.has(tag) && (!cutCounter || cutTag === tag)) {
  93. if (!cutCounter)
  94. cutTag = tag;
  95. cutCounter++;
  96. }
  97. if (cutTag === endTag) {
  98. cutCounter = (cutCounter > 0 ? cutCounter - 1 : 0);
  99. if (!cutCounter)
  100. cutTag = '';
  101. }
  102. }
  103. if (right >= nextProg) {
  104. _onProgress(Math.round(right/(len + 1)*100));
  105. nextProg += progStep;
  106. }
  107. i = right + 1;
  108. }
  109. if (i < len) {
  110. if (inCdata) {
  111. _onCdata(xstr.substr(leftData, len - leftData), cutCounter, cutTag);
  112. } else if (inComment) {
  113. _onComment(xstr.substr(leftData, len - leftData), cutCounter, cutTag);
  114. } else {
  115. _onTextNode(xstr.substr(i, len - i), cutCounter, cutTag);
  116. }
  117. }
  118. _onProgress(100);
  119. }
  120. //асинхронная копия parseSync
  121. //делается заменой "_on" => "await _on" после while
  122. async function parse(xstr, options) {
  123. const dummy = () => {};
  124. let {onStartNode: _onStartNode = dummy,
  125. onEndNode: _onEndNode = dummy,
  126. onTextNode: _onTextNode = dummy,
  127. onCdata: _onCdata = dummy,
  128. onComment: _onComment = dummy,
  129. onProgress: _onProgress = dummy,
  130. innerCut = new Set()
  131. } = options;
  132. let i = 0;
  133. const len = xstr.length;
  134. const progStep = len/10;
  135. let nextProg = 0;
  136. let cutCounter = 0;
  137. let cutTag = '';
  138. let inCdata;
  139. let inComment;
  140. let leftData = 0;
  141. while (i < len) {
  142. inCdata = false;
  143. inComment = false;
  144. let singleTag = false;
  145. let left = xstr.indexOf('<', i);
  146. if (left < 0)
  147. break;
  148. leftData = left;
  149. if (left < len - 2 && xstr[left + 1] == '!') {
  150. if (xstr[left + 2] == '-') {
  151. const leftComment = xstr.indexOf('<!--', left);
  152. if (leftComment == left) {
  153. inComment = true;
  154. leftData = left + 3;
  155. }
  156. }
  157. if (!inComment && xstr[left + 2] == '[') {
  158. const leftCdata = xstr.indexOf('<![CDATA[', left);
  159. if (leftCdata == left) {
  160. inCdata = true;
  161. leftData = left + 8;
  162. }
  163. }
  164. }
  165. const text = xstr.substr(i, left - i);
  166. await _onTextNode(text, cutCounter, cutTag);
  167. let right = null;
  168. let rightData = null;
  169. if (inCdata) {
  170. rightData = xstr.indexOf(']]>', leftData + 1);
  171. if (rightData < 0)
  172. break;
  173. right = rightData + 2;
  174. } else if (inComment) {
  175. rightData = xstr.indexOf('-->', leftData + 1);
  176. if (rightData < 0)
  177. break;
  178. right = rightData + 2;
  179. } else {
  180. rightData = xstr.indexOf('>', leftData + 1);
  181. if (rightData < 0)
  182. break;
  183. right = rightData;
  184. if (xstr[right - 1] === '/') {
  185. singleTag = true;
  186. rightData--;
  187. }
  188. }
  189. let tagData = xstr.substr(leftData + 1, rightData - leftData - 1);
  190. if (inCdata) {
  191. await _onCdata(tagData, cutCounter, cutTag);
  192. } else if (inComment) {
  193. await _onComment(tagData, cutCounter, cutTag);
  194. } else {
  195. let tag = '';
  196. let tail = '';
  197. const firstSpace = tagData.indexOf(' ');
  198. if (firstSpace >= 0) {
  199. tail = tagData.substr(firstSpace);
  200. tag = tagData.substr(0, firstSpace);
  201. } else {
  202. tag = tagData;
  203. }
  204. tag = tag.toLowerCase();
  205. let endTag = (singleTag ? tag : '');
  206. if (tag === '' || tag[0] !== '/') {
  207. await _onStartNode(tag, tail, singleTag, cutCounter, cutTag);
  208. } else {
  209. endTag = tag.substr(1);
  210. }
  211. if (endTag)
  212. await _onEndNode(endTag, tail, singleTag, cutCounter, cutTag);
  213. if (innerCut.has(tag) && (!cutCounter || cutTag === tag)) {
  214. if (!cutCounter)
  215. cutTag = tag;
  216. cutCounter++;
  217. }
  218. if (cutTag === endTag) {
  219. cutCounter = (cutCounter > 0 ? cutCounter - 1 : 0);
  220. if (!cutCounter)
  221. cutTag = '';
  222. }
  223. }
  224. if (right >= nextProg) {
  225. await _onProgress(Math.round(right/(len + 1)*100));
  226. nextProg += progStep;
  227. }
  228. i = right + 1;
  229. }
  230. if (i < len) {
  231. if (inCdata) {
  232. await _onCdata(xstr.substr(leftData, len - leftData), cutCounter, cutTag);
  233. } else if (inComment) {
  234. await _onComment(xstr.substr(leftData, len - leftData), cutCounter, cutTag);
  235. } else {
  236. await _onTextNode(xstr.substr(i, len - i), cutCounter, cutTag);
  237. }
  238. }
  239. await _onProgress(100);
  240. }
  241. module.exports = {
  242. parseSync,
  243. parse
  244. }