1 | const Tokenizer = require('./Tokenizer.js');
|
2 | const { defaults } = require('./defaults.js');
|
3 | const { block, inline } = require('./rules.js');
|
4 |
|
5 |
|
6 |
|
7 |
|
8 | function smartypants(text) {
|
9 | return text
|
10 |
|
11 | .replace(/---/g, '\u2014')
|
12 |
|
13 | .replace(/--/g, '\u2013')
|
14 |
|
15 | .replace(/(^|[-\u2014/(\[{"\s])'/g, '$1\u2018')
|
16 |
|
17 | .replace(/'/g, '\u2019')
|
18 |
|
19 | .replace(/(^|[-\u2014/(\[{\u2018\s])"/g, '$1\u201c')
|
20 |
|
21 | .replace(/"/g, '\u201d')
|
22 |
|
23 | .replace(/\.{3}/g, '\u2026');
|
24 | }
|
25 |
|
26 |
|
27 |
|
28 |
|
29 | function mangle(text) {
|
30 | let out = '',
|
31 | i,
|
32 | ch;
|
33 |
|
34 | const l = text.length;
|
35 | for (i = 0; i < l; i++) {
|
36 | ch = text.charCodeAt(i);
|
37 | if (Math.random() > 0.5) {
|
38 | ch = 'x' + ch.toString(16);
|
39 | }
|
40 | out += '&#' + ch + ';';
|
41 | }
|
42 |
|
43 | return out;
|
44 | }
|
45 |
|
46 |
|
47 |
|
48 |
|
49 | module.exports = class Lexer {
|
50 | constructor(options) {
|
51 | this.tokens = [];
|
52 | this.tokens.links = Object.create(null);
|
53 | this.options = options || defaults;
|
54 | this.options.tokenizer = this.options.tokenizer || new Tokenizer();
|
55 | this.tokenizer = this.options.tokenizer;
|
56 | this.tokenizer.options = this.options;
|
57 |
|
58 | const rules = {
|
59 | block: block.normal,
|
60 | inline: inline.normal
|
61 | };
|
62 |
|
63 | if (this.options.pedantic) {
|
64 | rules.block = block.pedantic;
|
65 | rules.inline = inline.pedantic;
|
66 | } else if (this.options.gfm) {
|
67 | rules.block = block.gfm;
|
68 | if (this.options.breaks) {
|
69 | rules.inline = inline.breaks;
|
70 | } else {
|
71 | rules.inline = inline.gfm;
|
72 | }
|
73 | }
|
74 | this.tokenizer.rules = rules;
|
75 | }
|
76 |
|
77 | |
78 |
|
79 |
|
80 | static get rules() {
|
81 | return {
|
82 | block,
|
83 | inline
|
84 | };
|
85 | }
|
86 |
|
87 | |
88 |
|
89 |
|
90 | static lex(src, options) {
|
91 | const lexer = new Lexer(options);
|
92 | return lexer.lex(src);
|
93 | }
|
94 |
|
95 | |
96 |
|
97 |
|
98 | lex(src) {
|
99 | src = src
|
100 | .replace(/\r\n|\r/g, '\n')
|
101 | .replace(/\t/g, ' ');
|
102 |
|
103 | this.blockTokens(src, this.tokens, true);
|
104 |
|
105 | this.inline(this.tokens);
|
106 |
|
107 | return this.tokens;
|
108 | }
|
109 |
|
110 | |
111 |
|
112 |
|
113 | blockTokens(src, tokens = [], top = true) {
|
114 | src = src.replace(/^ +$/gm, '');
|
115 | let token, i, l, lastToken;
|
116 |
|
117 | while (src) {
|
118 |
|
119 | if (token = this.tokenizer.space(src)) {
|
120 | src = src.substring(token.raw.length);
|
121 | if (token.type) {
|
122 | tokens.push(token);
|
123 | }
|
124 | continue;
|
125 | }
|
126 |
|
127 |
|
128 | if (token = this.tokenizer.code(src, tokens)) {
|
129 | src = src.substring(token.raw.length);
|
130 | if (token.type) {
|
131 | tokens.push(token);
|
132 | } else {
|
133 | lastToken = tokens[tokens.length - 1];
|
134 | lastToken.raw += '\n' + token.raw;
|
135 | lastToken.text += '\n' + token.text;
|
136 | }
|
137 | continue;
|
138 | }
|
139 |
|
140 |
|
141 | if (token = this.tokenizer.fences(src)) {
|
142 | src = src.substring(token.raw.length);
|
143 | tokens.push(token);
|
144 | continue;
|
145 | }
|
146 |
|
147 |
|
148 | if (token = this.tokenizer.heading(src)) {
|
149 | src = src.substring(token.raw.length);
|
150 | tokens.push(token);
|
151 | continue;
|
152 | }
|
153 |
|
154 |
|
155 | if (token = this.tokenizer.nptable(src)) {
|
156 | src = src.substring(token.raw.length);
|
157 | tokens.push(token);
|
158 | continue;
|
159 | }
|
160 |
|
161 |
|
162 | if (token = this.tokenizer.hr(src)) {
|
163 | src = src.substring(token.raw.length);
|
164 | tokens.push(token);
|
165 | continue;
|
166 | }
|
167 |
|
168 |
|
169 | if (token = this.tokenizer.blockquote(src)) {
|
170 | src = src.substring(token.raw.length);
|
171 | token.tokens = this.blockTokens(token.text, [], top);
|
172 | tokens.push(token);
|
173 | continue;
|
174 | }
|
175 |
|
176 |
|
177 | if (token = this.tokenizer.list(src)) {
|
178 | src = src.substring(token.raw.length);
|
179 | l = token.items.length;
|
180 | for (i = 0; i < l; i++) {
|
181 | token.items[i].tokens = this.blockTokens(token.items[i].text, [], false);
|
182 | }
|
183 | tokens.push(token);
|
184 | continue;
|
185 | }
|
186 |
|
187 |
|
188 | if (token = this.tokenizer.html(src)) {
|
189 | src = src.substring(token.raw.length);
|
190 | tokens.push(token);
|
191 | continue;
|
192 | }
|
193 |
|
194 |
|
195 | if (top && (token = this.tokenizer.def(src))) {
|
196 | src = src.substring(token.raw.length);
|
197 | if (!this.tokens.links[token.tag]) {
|
198 | this.tokens.links[token.tag] = {
|
199 | href: token.href,
|
200 | title: token.title
|
201 | };
|
202 | }
|
203 | continue;
|
204 | }
|
205 |
|
206 |
|
207 | if (token = this.tokenizer.table(src)) {
|
208 | src = src.substring(token.raw.length);
|
209 | tokens.push(token);
|
210 | continue;
|
211 | }
|
212 |
|
213 |
|
214 | if (token = this.tokenizer.lheading(src)) {
|
215 | src = src.substring(token.raw.length);
|
216 | tokens.push(token);
|
217 | continue;
|
218 | }
|
219 |
|
220 |
|
221 | if (top && (token = this.tokenizer.paragraph(src))) {
|
222 | src = src.substring(token.raw.length);
|
223 | tokens.push(token);
|
224 | continue;
|
225 | }
|
226 |
|
227 |
|
228 | if (token = this.tokenizer.text(src, tokens)) {
|
229 | src = src.substring(token.raw.length);
|
230 | if (token.type) {
|
231 | tokens.push(token);
|
232 | } else {
|
233 | lastToken = tokens[tokens.length - 1];
|
234 | lastToken.raw += '\n' + token.raw;
|
235 | lastToken.text += '\n' + token.text;
|
236 | }
|
237 | continue;
|
238 | }
|
239 |
|
240 | if (src) {
|
241 | const errMsg = 'Infinite loop on byte: ' + src.charCodeAt(0);
|
242 | if (this.options.silent) {
|
243 | console.error(errMsg);
|
244 | break;
|
245 | } else {
|
246 | throw new Error(errMsg);
|
247 | }
|
248 | }
|
249 | }
|
250 |
|
251 | return tokens;
|
252 | }
|
253 |
|
254 | inline(tokens) {
|
255 | let i,
|
256 | j,
|
257 | k,
|
258 | l2,
|
259 | row,
|
260 | token;
|
261 |
|
262 | const l = tokens.length;
|
263 | for (i = 0; i < l; i++) {
|
264 | token = tokens[i];
|
265 | switch (token.type) {
|
266 | case 'paragraph':
|
267 | case 'text':
|
268 | case 'heading': {
|
269 | token.tokens = [];
|
270 | this.inlineTokens(token.text, token.tokens);
|
271 | break;
|
272 | }
|
273 | case 'table': {
|
274 | token.tokens = {
|
275 | header: [],
|
276 | cells: []
|
277 | };
|
278 |
|
279 |
|
280 | l2 = token.header.length;
|
281 | for (j = 0; j < l2; j++) {
|
282 | token.tokens.header[j] = [];
|
283 | this.inlineTokens(token.header[j], token.tokens.header[j]);
|
284 | }
|
285 |
|
286 |
|
287 | l2 = token.cells.length;
|
288 | for (j = 0; j < l2; j++) {
|
289 | row = token.cells[j];
|
290 | token.tokens.cells[j] = [];
|
291 | for (k = 0; k < row.length; k++) {
|
292 | token.tokens.cells[j][k] = [];
|
293 | this.inlineTokens(row[k], token.tokens.cells[j][k]);
|
294 | }
|
295 | }
|
296 |
|
297 | break;
|
298 | }
|
299 | case 'blockquote': {
|
300 | this.inline(token.tokens);
|
301 | break;
|
302 | }
|
303 | case 'list': {
|
304 | l2 = token.items.length;
|
305 | for (j = 0; j < l2; j++) {
|
306 | this.inline(token.items[j].tokens);
|
307 | }
|
308 | break;
|
309 | }
|
310 | default: {
|
311 |
|
312 | }
|
313 | }
|
314 | }
|
315 |
|
316 | return tokens;
|
317 | }
|
318 |
|
319 | |
320 |
|
321 |
|
322 | inlineTokens(src, tokens = [], inLink = false, inRawBlock = false) {
|
323 | let token;
|
324 |
|
325 | while (src) {
|
326 |
|
327 | if (token = this.tokenizer.escape(src)) {
|
328 | src = src.substring(token.raw.length);
|
329 | tokens.push(token);
|
330 | continue;
|
331 | }
|
332 |
|
333 |
|
334 | if (token = this.tokenizer.tag(src, inLink, inRawBlock)) {
|
335 | src = src.substring(token.raw.length);
|
336 | inLink = token.inLink;
|
337 | inRawBlock = token.inRawBlock;
|
338 | tokens.push(token);
|
339 | continue;
|
340 | }
|
341 |
|
342 |
|
343 | if (token = this.tokenizer.link(src)) {
|
344 | src = src.substring(token.raw.length);
|
345 | if (token.type === 'link') {
|
346 | token.tokens = this.inlineTokens(token.text, [], true, inRawBlock);
|
347 | }
|
348 | tokens.push(token);
|
349 | continue;
|
350 | }
|
351 |
|
352 |
|
353 | if (token = this.tokenizer.reflink(src, this.tokens.links)) {
|
354 | src = src.substring(token.raw.length);
|
355 | if (token.type === 'link') {
|
356 | token.tokens = this.inlineTokens(token.text, [], true, inRawBlock);
|
357 | }
|
358 | tokens.push(token);
|
359 | continue;
|
360 | }
|
361 |
|
362 |
|
363 | if (token = this.tokenizer.strong(src)) {
|
364 | src = src.substring(token.raw.length);
|
365 | token.tokens = this.inlineTokens(token.text, [], inLink, inRawBlock);
|
366 | tokens.push(token);
|
367 | continue;
|
368 | }
|
369 |
|
370 |
|
371 | if (token = this.tokenizer.em(src)) {
|
372 | src = src.substring(token.raw.length);
|
373 | token.tokens = this.inlineTokens(token.text, [], inLink, inRawBlock);
|
374 | tokens.push(token);
|
375 | continue;
|
376 | }
|
377 |
|
378 |
|
379 | if (token = this.tokenizer.codespan(src)) {
|
380 | src = src.substring(token.raw.length);
|
381 | tokens.push(token);
|
382 | continue;
|
383 | }
|
384 |
|
385 |
|
386 | if (token = this.tokenizer.br(src)) {
|
387 | src = src.substring(token.raw.length);
|
388 | tokens.push(token);
|
389 | continue;
|
390 | }
|
391 |
|
392 |
|
393 | if (token = this.tokenizer.del(src)) {
|
394 | src = src.substring(token.raw.length);
|
395 | token.tokens = this.inlineTokens(token.text, [], inLink, inRawBlock);
|
396 | tokens.push(token);
|
397 | continue;
|
398 | }
|
399 |
|
400 |
|
401 | if (token = this.tokenizer.autolink(src, mangle)) {
|
402 | src = src.substring(token.raw.length);
|
403 | tokens.push(token);
|
404 | continue;
|
405 | }
|
406 |
|
407 |
|
408 | if (!inLink && (token = this.tokenizer.url(src, mangle))) {
|
409 | src = src.substring(token.raw.length);
|
410 | tokens.push(token);
|
411 | continue;
|
412 | }
|
413 |
|
414 |
|
415 | if (token = this.tokenizer.inlineText(src, inRawBlock, smartypants)) {
|
416 | src = src.substring(token.raw.length);
|
417 | tokens.push(token);
|
418 | continue;
|
419 | }
|
420 |
|
421 | if (src) {
|
422 | const errMsg = 'Infinite loop on byte: ' + src.charCodeAt(0);
|
423 | if (this.options.silent) {
|
424 | console.error(errMsg);
|
425 | break;
|
426 | } else {
|
427 | throw new Error(errMsg);
|
428 | }
|
429 | }
|
430 | }
|
431 |
|
432 | return tokens;
|
433 | }
|
434 | };
|