mdream 0.1.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE.md +9 -0
- package/README.md +185 -0
- package/bin/mdream.mjs +2 -0
- package/dist/cli.d.mts +2 -0
- package/dist/cli.d.ts +2 -0
- package/dist/cli.mjs +25 -0
- package/dist/index.d.mts +15 -0
- package/dist/index.d.ts +15 -0
- package/dist/index.mjs +13 -0
- package/dist/plugins.d.mts +88 -0
- package/dist/plugins.d.ts +88 -0
- package/dist/plugins.mjs +4 -0
- package/dist/preset/minimal.d.mts +11 -0
- package/dist/preset/minimal.d.ts +11 -0
- package/dist/preset/minimal.mjs +39 -0
- package/dist/shared/mdream.-hdaPj9a.mjs +280 -0
- package/dist/shared/mdream.5zaIXVJz.mjs +508 -0
- package/dist/shared/mdream.C8ruysN5.mjs +291 -0
- package/dist/shared/mdream.DUeWbUFG.mjs +1432 -0
- package/dist/shared/mdream.a2AvjJLp.d.mts +218 -0
- package/dist/shared/mdream.a2AvjJLp.d.ts +218 -0
- package/dist/shared/mdream.cpEmpxyh.mjs +105 -0
- package/package.json +62 -0
|
@@ -0,0 +1,1432 @@
|
|
|
1
|
+
import { ae as TEXT_NODE, af as NodeEventEnter, ab as collectNodeContent, aa as ELEMENT_NODE, a3 as TAG_PRE, M as TAG_LI, a4 as TAG_BLOCKQUOTE, ag as NO_SPACING, ah as DEFAULT_BLOCK_SPACING, ai as TABLE_ROW_SPACING, aj as LIST_ITEM_SPACING, ak as BLOCKQUOTE_SPACING, al as MARKDOWN_STRIKETHROUGH, am as MARKDOWN_HORIZONTAL_RULE, J as TAG_DD, K as TAG_DT, L as TAG_DL, e as TAG_ADDRESS, an as TAG_RP, ao as TAG_RT, ap as TAG_RUBY, aq as TAG_BDO, ar as TAG_TIME, as as TAG_VAR, at as TAG_DFN, au as TAG_CITE, av as TAG_U, z as TAG_ASIDE, aw as TAG_PLAINTEXT, ax as TAG_XMP, ay as TAG_NOFRAMES, az as TAG_NOSCRIPT, aA as TAG_SMALL, aB as TAG_SAMP, aC as TAG_Q, aD as TAG_MARK, aE as TAG_ABBR, aF as TAG_TEMPLATE, aG as TAG_PROGRESS, aH as TAG_METER, aI as TAG_DIALOG, aJ as TAG_MAP, s as TAG_IFRAME, aK as TAG_CANVAS, _ as TAG_VIDEO, Z as TAG_AUDIO, aL as TAG_LEGEND, t as TAG_FIELDSET, aM as TAG_OPTION, v as TAG_TEXTAREA, u as TAG_SELECT, Y as TAG_SVG, aN as TAG_WBR, aO as TAG_TRACK, aP as TAG_SOURCE, aQ as TAG_PARAM, aR as TAG_KEYGEN, w as TAG_INPUT, r as TAG_EMBED, aS as TAG_COL, aT as TAG_BASE, aU as TAG_AREA, aV as TAG_LINK, y as TAG_FORM, B as TAG_FOOTER, aW as TAG_KBD, S as TAG_TFOOT, U as TAG_TBODY, aX as TAG_CENTER, X as TAG_TABLE, T as TAG_BODY, x as TAG_BUTTON, aY as TAG_LABEL, A as TAG_NAV, j as TAG_SPAN, a5 as TAG_DIV, a6 as TAG_P, P as TAG_TD, Q as TAG_TH, R as TAG_TR, V as TAG_THEAD, a1 as TAG_IMG, d as TAG_A, O as TAG_UL, N as TAG_OL, a2 as TAG_CODE, aZ as MARKDOWN_CODE_BLOCK, a_ as MARKDOWN_INLINE_CODE, a$ as TAG_INS, b0 as TAG_SUP, b1 as TAG_SUB, b2 as TAG_DEL, m as TAG_I, n as TAG_EM, o as TAG_B, p as TAG_STRONG, l as TAG_HR, D as TAG_H6, E as TAG_H5, F as TAG_H4, G as TAG_H3, H as TAG_H2, I as TAG_H1, k as TAG_BR, ad as TAG_META, h as TAG_STYLE, i as TAG_SCRIPT, ac as TAG_TITLE, f as TAG_SUMMARY, g as TAG_DETAILS, b as TAG_HEAD, b3 as MARKDOWN_EMPHASIS, b4 as MARKDOWN_STRONG, b5 as HTML_ENTITIES, b6 as MAX_TAG_ID, b7 as assembleBufferedContent, b8 as TagIdMap, b9 as NodeEventExit } from './mdream.-hdaPj9a.mjs';
|
|
2
|
+
|
|
3
|
+
function needsSpacing(lastChar, firstChar) {
|
|
4
|
+
const noSpaceLastChars = /* @__PURE__ */ new Set(["\n", " ", "[", ">", "_", "*", "`", "|", "#", "<", "("]);
|
|
5
|
+
const noSpaceFirstChars = /* @__PURE__ */ new Set([" ", "\n", " ", "_", "*", "`", "|", ">", "#"]);
|
|
6
|
+
return !noSpaceLastChars.has(lastChar) && !noSpaceFirstChars.has(firstChar);
|
|
7
|
+
}
|
|
8
|
+
function shouldAddSpacingBeforeText(lastChar, lastNode, textNode) {
|
|
9
|
+
return lastChar && lastChar !== "\n" && lastChar !== " " && lastChar !== "[" && lastChar !== ">" && !lastNode?.tagHandler?.isInline && textNode.value[0] !== " ";
|
|
10
|
+
}
|
|
11
|
+
function processTextNodeWithPlugins(node, state) {
|
|
12
|
+
if (!state.plugins?.length)
|
|
13
|
+
return void 0;
|
|
14
|
+
for (const plugin of state.plugins) {
|
|
15
|
+
if (!plugin.processTextNode)
|
|
16
|
+
continue;
|
|
17
|
+
const result = plugin.processTextNode(node, state);
|
|
18
|
+
if (result) {
|
|
19
|
+
if (result.skip)
|
|
20
|
+
return result;
|
|
21
|
+
return { content: result.content, skip: false };
|
|
22
|
+
}
|
|
23
|
+
}
|
|
24
|
+
return void 0;
|
|
25
|
+
}
|
|
26
|
+
function processHtmlEventToMarkdown(event, state) {
|
|
27
|
+
const { type: eventType, node } = event;
|
|
28
|
+
const lastNode = state.lastNode;
|
|
29
|
+
state.lastNode = event.node;
|
|
30
|
+
const buff = state.regionContentBuffers.get(node.regionId || 0) || [];
|
|
31
|
+
const lastBuffEntry = buff[buff.length - 1];
|
|
32
|
+
const lastChar = lastBuffEntry?.charAt(lastBuffEntry.length - 1) || "";
|
|
33
|
+
let secondLastChar;
|
|
34
|
+
if (lastBuffEntry?.length > 1) {
|
|
35
|
+
secondLastChar = lastBuffEntry.charAt(lastBuffEntry.length - 2);
|
|
36
|
+
} else {
|
|
37
|
+
secondLastChar = buff[buff.length - 2]?.charAt(buff[buff.length - 2].length - 1);
|
|
38
|
+
}
|
|
39
|
+
if (node.type === TEXT_NODE && eventType === NodeEventEnter) {
|
|
40
|
+
const textNode = node;
|
|
41
|
+
if (textNode.value) {
|
|
42
|
+
if (state.plugins?.length) {
|
|
43
|
+
const pluginResult = processTextNodeWithPlugins(textNode, state);
|
|
44
|
+
if (pluginResult) {
|
|
45
|
+
if (pluginResult.skip) {
|
|
46
|
+
return;
|
|
47
|
+
}
|
|
48
|
+
textNode.value = pluginResult.content;
|
|
49
|
+
}
|
|
50
|
+
}
|
|
51
|
+
if (textNode.value === " " && lastChar === "\n") {
|
|
52
|
+
return;
|
|
53
|
+
}
|
|
54
|
+
if (shouldAddSpacingBeforeText(lastChar, lastNode, textNode)) {
|
|
55
|
+
textNode.value = ` ${textNode.value}`;
|
|
56
|
+
}
|
|
57
|
+
collectNodeContent(textNode, textNode.value, state);
|
|
58
|
+
}
|
|
59
|
+
state.lastTextNode = textNode;
|
|
60
|
+
return;
|
|
61
|
+
}
|
|
62
|
+
if (node.type !== ELEMENT_NODE) {
|
|
63
|
+
return;
|
|
64
|
+
}
|
|
65
|
+
const context = { node, state };
|
|
66
|
+
const output = [];
|
|
67
|
+
const lastFragment = state.lastContentCache;
|
|
68
|
+
if (state.plugins?.length) {
|
|
69
|
+
const results = [];
|
|
70
|
+
const fn = eventType === NodeEventEnter ? "onNodeEnter" : "onNodeExit";
|
|
71
|
+
for (const plugin of state.plugins) {
|
|
72
|
+
if (!plugin[fn])
|
|
73
|
+
continue;
|
|
74
|
+
const result = plugin[fn](event.node, state);
|
|
75
|
+
if (result) {
|
|
76
|
+
results.push(result);
|
|
77
|
+
}
|
|
78
|
+
}
|
|
79
|
+
output.push(...results);
|
|
80
|
+
}
|
|
81
|
+
let lastNewLines = 0;
|
|
82
|
+
if (lastChar === "\n") {
|
|
83
|
+
lastNewLines++;
|
|
84
|
+
}
|
|
85
|
+
if (secondLastChar === "\n") {
|
|
86
|
+
lastNewLines++;
|
|
87
|
+
}
|
|
88
|
+
const eventFn = eventType === NodeEventEnter ? "enter" : "exit";
|
|
89
|
+
const handler = node.tagHandler;
|
|
90
|
+
if (!output.length && handler?.[eventFn]) {
|
|
91
|
+
const res = handler[eventFn](context);
|
|
92
|
+
if (res) {
|
|
93
|
+
output.push(res);
|
|
94
|
+
}
|
|
95
|
+
}
|
|
96
|
+
const newLineConfig = calculateNewLineConfig(node);
|
|
97
|
+
const newLines = Math.max(0, (newLineConfig[eventType] || 0) - lastNewLines);
|
|
98
|
+
if (newLines > 0) {
|
|
99
|
+
if (!buff.length) {
|
|
100
|
+
for (const fragment of output) {
|
|
101
|
+
collectNodeContent(node, fragment, state);
|
|
102
|
+
}
|
|
103
|
+
return;
|
|
104
|
+
}
|
|
105
|
+
const newlinesStr = "\n".repeat(newLines);
|
|
106
|
+
if (lastChar === " " && buff?.length) {
|
|
107
|
+
buff[buff.length - 1] = buff[buff.length - 1].substring(0, buff[buff.length - 1].length - 1);
|
|
108
|
+
}
|
|
109
|
+
if (eventType === NodeEventEnter) {
|
|
110
|
+
output.unshift(newlinesStr);
|
|
111
|
+
} else {
|
|
112
|
+
output.push(newlinesStr);
|
|
113
|
+
}
|
|
114
|
+
} else {
|
|
115
|
+
if (lastFragment && state.lastTextNode?.containsWhitespace && !!node.parent && "value" in state.lastTextNode && typeof state.lastTextNode.value === "string") {
|
|
116
|
+
if (!node.parent.depthMap[TAG_PRE] || node.parent.tagId === TAG_PRE) {
|
|
117
|
+
const originalLength = lastFragment.length;
|
|
118
|
+
const trimmed = lastFragment.trimEnd();
|
|
119
|
+
const trimmedChars = originalLength - trimmed.length;
|
|
120
|
+
if (trimmedChars > 0) {
|
|
121
|
+
if (buff?.length && buff[buff.length - 1] === lastFragment) {
|
|
122
|
+
buff[buff.length - 1] = trimmed;
|
|
123
|
+
}
|
|
124
|
+
}
|
|
125
|
+
state.lastTextNode = void 0;
|
|
126
|
+
}
|
|
127
|
+
}
|
|
128
|
+
}
|
|
129
|
+
if (output[0]?.[0] && eventType === NodeEventEnter && lastChar && needsSpacing(lastChar, output[0][0])) {
|
|
130
|
+
collectNodeContent(node, " ", state);
|
|
131
|
+
}
|
|
132
|
+
for (const fragment of output) {
|
|
133
|
+
collectNodeContent(node, fragment, state);
|
|
134
|
+
}
|
|
135
|
+
}
|
|
136
|
+
function calculateNewLineConfig(node) {
|
|
137
|
+
const tagId = node.tagId;
|
|
138
|
+
const depthMap = node.depthMap;
|
|
139
|
+
if (tagId !== TAG_LI && depthMap[TAG_LI] > 0 || tagId !== TAG_BLOCKQUOTE && depthMap[TAG_BLOCKQUOTE] > 0) {
|
|
140
|
+
return NO_SPACING;
|
|
141
|
+
}
|
|
142
|
+
let currParent = node.parent;
|
|
143
|
+
while (currParent) {
|
|
144
|
+
if (currParent.tagHandler?.collapsesInnerWhiteSpace) {
|
|
145
|
+
return NO_SPACING;
|
|
146
|
+
}
|
|
147
|
+
currParent = currParent.parent;
|
|
148
|
+
}
|
|
149
|
+
if (node.tagHandler?.spacing) {
|
|
150
|
+
return node.tagHandler?.spacing;
|
|
151
|
+
}
|
|
152
|
+
return DEFAULT_BLOCK_SPACING;
|
|
153
|
+
}
|
|
154
|
+
|
|
155
|
+
function resolveUrl(url, origin) {
|
|
156
|
+
if (!url)
|
|
157
|
+
return url;
|
|
158
|
+
if (url.startsWith("//")) {
|
|
159
|
+
return `https:${url}`;
|
|
160
|
+
}
|
|
161
|
+
if (origin) {
|
|
162
|
+
if (url.startsWith("/") && origin) {
|
|
163
|
+
const cleanOrigin = origin.endsWith("/") ? origin.slice(0, -1) : origin;
|
|
164
|
+
return `${cleanOrigin}${url}`;
|
|
165
|
+
}
|
|
166
|
+
if (url.startsWith("./")) {
|
|
167
|
+
return `${origin}/${url.slice(2)}`;
|
|
168
|
+
}
|
|
169
|
+
if (!url.startsWith("http")) {
|
|
170
|
+
const cleanUrl = url.startsWith("/") ? url.slice(1) : url;
|
|
171
|
+
return `${origin}/${cleanUrl}`;
|
|
172
|
+
}
|
|
173
|
+
}
|
|
174
|
+
return url;
|
|
175
|
+
}
|
|
176
|
+
function isInsideTableCell(node) {
|
|
177
|
+
return node.depthMap[TAG_TD] > 0;
|
|
178
|
+
}
|
|
179
|
+
function getLanguageFromClass(className) {
|
|
180
|
+
if (!className)
|
|
181
|
+
return "";
|
|
182
|
+
const langParts = className.split(" ").map((c) => c.split("language-")[1]).filter(Boolean);
|
|
183
|
+
return langParts.length > 0 ? langParts[0].trim() : "";
|
|
184
|
+
}
|
|
185
|
+
function handleHeading(depth) {
|
|
186
|
+
return {
|
|
187
|
+
enter: ({ node }) => {
|
|
188
|
+
if (node.depthMap[TAG_A]) {
|
|
189
|
+
return `<h${depth}>`;
|
|
190
|
+
}
|
|
191
|
+
return `${"#".repeat(depth)} `;
|
|
192
|
+
},
|
|
193
|
+
exit: ({ node }) => {
|
|
194
|
+
if (node.depthMap[TAG_A]) {
|
|
195
|
+
return `</h${depth}>`;
|
|
196
|
+
}
|
|
197
|
+
},
|
|
198
|
+
collapsesInnerWhiteSpace: true
|
|
199
|
+
};
|
|
200
|
+
}
|
|
201
|
+
const Strong = {
|
|
202
|
+
enter: ({ node }) => {
|
|
203
|
+
if (node.depthMap[TAG_B] > 1) {
|
|
204
|
+
return "";
|
|
205
|
+
}
|
|
206
|
+
return MARKDOWN_STRONG;
|
|
207
|
+
},
|
|
208
|
+
exit: ({ node }) => {
|
|
209
|
+
if (node.depthMap[TAG_B] > 1) {
|
|
210
|
+
return "";
|
|
211
|
+
}
|
|
212
|
+
return MARKDOWN_STRONG;
|
|
213
|
+
},
|
|
214
|
+
collapsesInnerWhiteSpace: true,
|
|
215
|
+
spacing: NO_SPACING,
|
|
216
|
+
isInline: true
|
|
217
|
+
};
|
|
218
|
+
const Emphasis = {
|
|
219
|
+
enter: ({ node }) => {
|
|
220
|
+
if (node.depthMap[TAG_I] > 1) {
|
|
221
|
+
return "";
|
|
222
|
+
}
|
|
223
|
+
return MARKDOWN_EMPHASIS;
|
|
224
|
+
},
|
|
225
|
+
exit: ({ node }) => {
|
|
226
|
+
if (node.depthMap[TAG_I] > 1) {
|
|
227
|
+
return "";
|
|
228
|
+
}
|
|
229
|
+
return MARKDOWN_EMPHASIS;
|
|
230
|
+
},
|
|
231
|
+
collapsesInnerWhiteSpace: true,
|
|
232
|
+
spacing: NO_SPACING,
|
|
233
|
+
isInline: true
|
|
234
|
+
};
|
|
235
|
+
const tagHandlers = {
|
|
236
|
+
// Numeric tag constants
|
|
237
|
+
[TAG_HEAD]: {
|
|
238
|
+
// No special handling for head - plugins will handle frontmatter
|
|
239
|
+
spacing: NO_SPACING,
|
|
240
|
+
collapsesInnerWhiteSpace: true
|
|
241
|
+
},
|
|
242
|
+
[TAG_DETAILS]: {
|
|
243
|
+
enter: () => "<details>",
|
|
244
|
+
exit: () => "</details>\n\n"
|
|
245
|
+
},
|
|
246
|
+
[TAG_SUMMARY]: {
|
|
247
|
+
enter: () => "<summary>",
|
|
248
|
+
exit: () => "</summary>\n\n"
|
|
249
|
+
},
|
|
250
|
+
[TAG_TITLE]: {
|
|
251
|
+
// No special handling for title - plugins will handle frontmatter
|
|
252
|
+
collapsesInnerWhiteSpace: true,
|
|
253
|
+
isNonNesting: true,
|
|
254
|
+
spacing: NO_SPACING
|
|
255
|
+
},
|
|
256
|
+
[TAG_SCRIPT]: {
|
|
257
|
+
excludesTextNodes: true,
|
|
258
|
+
isNonNesting: true
|
|
259
|
+
},
|
|
260
|
+
[TAG_STYLE]: {
|
|
261
|
+
isNonNesting: true,
|
|
262
|
+
excludesTextNodes: true
|
|
263
|
+
},
|
|
264
|
+
[TAG_META]: {
|
|
265
|
+
// No special handling for meta - plugins will handle frontmatter
|
|
266
|
+
collapsesInnerWhiteSpace: true,
|
|
267
|
+
isSelfClosing: true,
|
|
268
|
+
spacing: NO_SPACING
|
|
269
|
+
},
|
|
270
|
+
[TAG_BR]: {
|
|
271
|
+
enter: ({ node }) => {
|
|
272
|
+
return isInsideTableCell(node) ? "<br>" : void 0;
|
|
273
|
+
},
|
|
274
|
+
isSelfClosing: true,
|
|
275
|
+
spacing: NO_SPACING,
|
|
276
|
+
collapsesInnerWhiteSpace: true,
|
|
277
|
+
isInline: true
|
|
278
|
+
},
|
|
279
|
+
[TAG_H1]: handleHeading(1),
|
|
280
|
+
[TAG_H2]: handleHeading(2),
|
|
281
|
+
[TAG_H3]: handleHeading(3),
|
|
282
|
+
[TAG_H4]: handleHeading(4),
|
|
283
|
+
[TAG_H5]: handleHeading(5),
|
|
284
|
+
[TAG_H6]: handleHeading(6),
|
|
285
|
+
[TAG_HR]: {
|
|
286
|
+
enter: () => MARKDOWN_HORIZONTAL_RULE,
|
|
287
|
+
isSelfClosing: true
|
|
288
|
+
},
|
|
289
|
+
[TAG_STRONG]: Strong,
|
|
290
|
+
[TAG_B]: Strong,
|
|
291
|
+
[TAG_EM]: Emphasis,
|
|
292
|
+
[TAG_I]: Emphasis,
|
|
293
|
+
[TAG_DEL]: {
|
|
294
|
+
enter: () => MARKDOWN_STRIKETHROUGH,
|
|
295
|
+
exit: () => MARKDOWN_STRIKETHROUGH,
|
|
296
|
+
collapsesInnerWhiteSpace: true,
|
|
297
|
+
spacing: NO_SPACING,
|
|
298
|
+
isInline: true
|
|
299
|
+
},
|
|
300
|
+
[TAG_SUB]: {
|
|
301
|
+
enter: () => "<sub>",
|
|
302
|
+
exit: () => "</sub>",
|
|
303
|
+
collapsesInnerWhiteSpace: true,
|
|
304
|
+
spacing: NO_SPACING,
|
|
305
|
+
isInline: true
|
|
306
|
+
},
|
|
307
|
+
[TAG_SUP]: {
|
|
308
|
+
enter: () => "<sup>",
|
|
309
|
+
exit: () => "</sup>",
|
|
310
|
+
collapsesInnerWhiteSpace: true,
|
|
311
|
+
spacing: NO_SPACING,
|
|
312
|
+
isInline: true
|
|
313
|
+
},
|
|
314
|
+
[TAG_INS]: {
|
|
315
|
+
enter: () => "<ins>",
|
|
316
|
+
exit: () => "</ins>",
|
|
317
|
+
collapsesInnerWhiteSpace: true,
|
|
318
|
+
spacing: NO_SPACING,
|
|
319
|
+
isInline: true
|
|
320
|
+
},
|
|
321
|
+
[TAG_BLOCKQUOTE]: {
|
|
322
|
+
enter: ({ node }) => {
|
|
323
|
+
const depth = node.depthMap[TAG_BLOCKQUOTE] || 1;
|
|
324
|
+
let prefix = "> ".repeat(depth);
|
|
325
|
+
if (node.depthMap[TAG_LI] > 0) {
|
|
326
|
+
prefix = `
|
|
327
|
+
${" ".repeat(node.depthMap[TAG_LI])}${prefix}`;
|
|
328
|
+
}
|
|
329
|
+
return prefix;
|
|
330
|
+
},
|
|
331
|
+
spacing: BLOCKQUOTE_SPACING
|
|
332
|
+
},
|
|
333
|
+
[TAG_CODE]: {
|
|
334
|
+
enter: ({ node }) => {
|
|
335
|
+
if ((node.depthMap[TAG_PRE] || 0) > 0) {
|
|
336
|
+
const language = getLanguageFromClass(node.attributes?.class);
|
|
337
|
+
return `${MARKDOWN_CODE_BLOCK}${language}
|
|
338
|
+
`;
|
|
339
|
+
}
|
|
340
|
+
return MARKDOWN_INLINE_CODE;
|
|
341
|
+
},
|
|
342
|
+
exit: ({ node }) => {
|
|
343
|
+
return node.depthMap[TAG_PRE] > 0 ? `
|
|
344
|
+
${MARKDOWN_CODE_BLOCK}` : MARKDOWN_INLINE_CODE;
|
|
345
|
+
},
|
|
346
|
+
collapsesInnerWhiteSpace: true,
|
|
347
|
+
spacing: NO_SPACING,
|
|
348
|
+
isInline: true
|
|
349
|
+
},
|
|
350
|
+
[TAG_UL]: {
|
|
351
|
+
enter: ({ node }) => isInsideTableCell(node) ? "<ul>" : void 0,
|
|
352
|
+
exit: ({ node }) => isInsideTableCell(node) ? "</ul>" : void 0
|
|
353
|
+
},
|
|
354
|
+
[TAG_LI]: {
|
|
355
|
+
enter: ({ node }) => {
|
|
356
|
+
if (isInsideTableCell(node)) {
|
|
357
|
+
return "<li>";
|
|
358
|
+
}
|
|
359
|
+
const depth = (node.depthMap[TAG_UL] || 0) + (node.depthMap[TAG_OL] || 0) - 1;
|
|
360
|
+
const isOrdered = node.parent?.tagId === TAG_OL;
|
|
361
|
+
const indent = " ".repeat(Math.max(0, depth));
|
|
362
|
+
const marker = isOrdered ? `${node.index + 1}. ` : "- ";
|
|
363
|
+
return `${indent}${marker}`;
|
|
364
|
+
},
|
|
365
|
+
exit: ({ node }) => isInsideTableCell(node) ? "</li>" : void 0,
|
|
366
|
+
spacing: LIST_ITEM_SPACING
|
|
367
|
+
},
|
|
368
|
+
[TAG_A]: {
|
|
369
|
+
enter: ({ node }) => {
|
|
370
|
+
if (node.attributes?.href) {
|
|
371
|
+
return "[";
|
|
372
|
+
}
|
|
373
|
+
},
|
|
374
|
+
exit: ({ node, state }) => {
|
|
375
|
+
if (!node.attributes?.href) {
|
|
376
|
+
return "";
|
|
377
|
+
}
|
|
378
|
+
const href = resolveUrl(node.attributes?.href || "", state.options?.origin);
|
|
379
|
+
let title = node.attributes?.title;
|
|
380
|
+
const lastContent = state.lastContentCache;
|
|
381
|
+
if (lastContent === title) {
|
|
382
|
+
title = "";
|
|
383
|
+
}
|
|
384
|
+
return title ? `](${href} "${title}")` : `](${href})`;
|
|
385
|
+
},
|
|
386
|
+
collapsesInnerWhiteSpace: true,
|
|
387
|
+
spacing: NO_SPACING,
|
|
388
|
+
isInline: true
|
|
389
|
+
},
|
|
390
|
+
[TAG_IMG]: {
|
|
391
|
+
enter: ({ node, state }) => {
|
|
392
|
+
const alt = node.attributes?.alt || "";
|
|
393
|
+
const src = resolveUrl(node.attributes?.src || "", state.options?.origin);
|
|
394
|
+
return ``;
|
|
395
|
+
},
|
|
396
|
+
collapsesInnerWhiteSpace: true,
|
|
397
|
+
isSelfClosing: true,
|
|
398
|
+
spacing: NO_SPACING,
|
|
399
|
+
isInline: true
|
|
400
|
+
},
|
|
401
|
+
[TAG_TABLE]: {
|
|
402
|
+
enter: ({ node, state }) => {
|
|
403
|
+
if (isInsideTableCell(node)) {
|
|
404
|
+
return "<table>";
|
|
405
|
+
}
|
|
406
|
+
if (node.depthMap[TAG_TABLE] <= 1) {
|
|
407
|
+
state.tableRenderedTable = false;
|
|
408
|
+
}
|
|
409
|
+
state.tableColumnAlignments = [];
|
|
410
|
+
},
|
|
411
|
+
exit: ({ node }) => isInsideTableCell(node) ? "</table>" : void 0
|
|
412
|
+
},
|
|
413
|
+
[TAG_THEAD]: {
|
|
414
|
+
enter: ({ node }) => {
|
|
415
|
+
if (isInsideTableCell(node)) {
|
|
416
|
+
return "<thead>";
|
|
417
|
+
}
|
|
418
|
+
},
|
|
419
|
+
exit: ({ node }) => isInsideTableCell(node) ? "</thead>" : void 0,
|
|
420
|
+
spacing: TABLE_ROW_SPACING,
|
|
421
|
+
excludesTextNodes: true
|
|
422
|
+
},
|
|
423
|
+
[TAG_TR]: {
|
|
424
|
+
enter: ({ node, state }) => {
|
|
425
|
+
if (isInsideTableCell(node)) {
|
|
426
|
+
return "<tr>";
|
|
427
|
+
}
|
|
428
|
+
state.tableCurrentRowCells = 0;
|
|
429
|
+
return "| ";
|
|
430
|
+
},
|
|
431
|
+
exit: ({ node, state }) => {
|
|
432
|
+
if (isInsideTableCell(node) || node.depthMap[TAG_TABLE] > 1) {
|
|
433
|
+
return "</tr>";
|
|
434
|
+
}
|
|
435
|
+
if (!state.tableRenderedTable) {
|
|
436
|
+
state.tableRenderedTable = true;
|
|
437
|
+
const alignments = state.tableColumnAlignments;
|
|
438
|
+
while (alignments.length < state.tableCurrentRowCells) {
|
|
439
|
+
alignments.push("");
|
|
440
|
+
}
|
|
441
|
+
const alignmentMarkers = alignments.map((align) => {
|
|
442
|
+
switch (align) {
|
|
443
|
+
case "left":
|
|
444
|
+
return ":---";
|
|
445
|
+
case "center":
|
|
446
|
+
return ":---:";
|
|
447
|
+
case "right":
|
|
448
|
+
return "---:";
|
|
449
|
+
default:
|
|
450
|
+
return "---";
|
|
451
|
+
}
|
|
452
|
+
});
|
|
453
|
+
return ` |
|
|
454
|
+
| ${alignmentMarkers.join(" | ")} |`;
|
|
455
|
+
}
|
|
456
|
+
return " |";
|
|
457
|
+
},
|
|
458
|
+
excludesTextNodes: true,
|
|
459
|
+
spacing: TABLE_ROW_SPACING
|
|
460
|
+
},
|
|
461
|
+
[TAG_TH]: {
|
|
462
|
+
enter: ({ node, state }) => {
|
|
463
|
+
if (node.depthMap[TAG_TABLE] > 1) {
|
|
464
|
+
return "<th>";
|
|
465
|
+
}
|
|
466
|
+
const align = node.attributes?.align?.toLowerCase();
|
|
467
|
+
if (align) {
|
|
468
|
+
state.tableColumnAlignments.push(align);
|
|
469
|
+
} else if (state.tableColumnAlignments.length <= state.tableCurrentRowCells) {
|
|
470
|
+
state.tableColumnAlignments.push("");
|
|
471
|
+
}
|
|
472
|
+
return node.index === 0 ? "" : " | ";
|
|
473
|
+
},
|
|
474
|
+
exit: ({ node, state }) => {
|
|
475
|
+
if (node.depthMap[TAG_TABLE] > 1) {
|
|
476
|
+
return "</th>";
|
|
477
|
+
}
|
|
478
|
+
state.tableCurrentRowCells++;
|
|
479
|
+
},
|
|
480
|
+
collapsesInnerWhiteSpace: true,
|
|
481
|
+
spacing: NO_SPACING
|
|
482
|
+
},
|
|
483
|
+
[TAG_TD]: {
|
|
484
|
+
enter: ({ node }) => {
|
|
485
|
+
if (node.depthMap[TAG_TABLE] > 1) {
|
|
486
|
+
return "<td>";
|
|
487
|
+
}
|
|
488
|
+
return node.index === 0 ? "" : " | ";
|
|
489
|
+
},
|
|
490
|
+
exit: ({ node, state }) => {
|
|
491
|
+
if (node.depthMap[TAG_TABLE] > 1) {
|
|
492
|
+
return "</td>";
|
|
493
|
+
}
|
|
494
|
+
state.tableCurrentRowCells++;
|
|
495
|
+
},
|
|
496
|
+
collapsesInnerWhiteSpace: true,
|
|
497
|
+
spacing: NO_SPACING
|
|
498
|
+
},
|
|
499
|
+
[TAG_P]: {},
|
|
500
|
+
[TAG_DIV]: {},
|
|
501
|
+
[TAG_SPAN]: {
|
|
502
|
+
collapsesInnerWhiteSpace: true,
|
|
503
|
+
spacing: NO_SPACING,
|
|
504
|
+
isInline: true
|
|
505
|
+
},
|
|
506
|
+
[TAG_NAV]: {},
|
|
507
|
+
[TAG_LABEL]: {
|
|
508
|
+
collapsesInnerWhiteSpace: true,
|
|
509
|
+
spacing: NO_SPACING,
|
|
510
|
+
isInline: true
|
|
511
|
+
},
|
|
512
|
+
[TAG_BUTTON]: {
|
|
513
|
+
collapsesInnerWhiteSpace: true,
|
|
514
|
+
isInline: true
|
|
515
|
+
},
|
|
516
|
+
[TAG_BODY]: { spacing: NO_SPACING },
|
|
517
|
+
[TAG_CENTER]: {
|
|
518
|
+
// if in table cell we preserve
|
|
519
|
+
enter: ({ node }) => {
|
|
520
|
+
if (node.depthMap[TAG_TABLE] > 1) {
|
|
521
|
+
return "<center>";
|
|
522
|
+
}
|
|
523
|
+
},
|
|
524
|
+
exit: ({ node }) => {
|
|
525
|
+
if (node.depthMap[TAG_TABLE] > 1) {
|
|
526
|
+
return "</center>";
|
|
527
|
+
}
|
|
528
|
+
},
|
|
529
|
+
spacing: NO_SPACING
|
|
530
|
+
},
|
|
531
|
+
[TAG_TBODY]: {
|
|
532
|
+
spacing: NO_SPACING,
|
|
533
|
+
excludesTextNodes: true
|
|
534
|
+
},
|
|
535
|
+
[TAG_TFOOT]: {
|
|
536
|
+
spacing: TABLE_ROW_SPACING,
|
|
537
|
+
excludesTextNodes: true
|
|
538
|
+
},
|
|
539
|
+
[TAG_KBD]: {
|
|
540
|
+
enter: () => "`",
|
|
541
|
+
exit: () => "`",
|
|
542
|
+
collapsesInnerWhiteSpace: true,
|
|
543
|
+
spacing: NO_SPACING,
|
|
544
|
+
isInline: true
|
|
545
|
+
},
|
|
546
|
+
[TAG_FOOTER]: {
|
|
547
|
+
spacing: NO_SPACING
|
|
548
|
+
},
|
|
549
|
+
[TAG_FORM]: {
|
|
550
|
+
spacing: NO_SPACING
|
|
551
|
+
},
|
|
552
|
+
[TAG_LINK]: {
|
|
553
|
+
isSelfClosing: true,
|
|
554
|
+
spacing: NO_SPACING,
|
|
555
|
+
collapsesInnerWhiteSpace: true,
|
|
556
|
+
isInline: true
|
|
557
|
+
},
|
|
558
|
+
[TAG_AREA]: {
|
|
559
|
+
isSelfClosing: true,
|
|
560
|
+
spacing: NO_SPACING,
|
|
561
|
+
isInline: true
|
|
562
|
+
},
|
|
563
|
+
[TAG_BASE]: {
|
|
564
|
+
isSelfClosing: true,
|
|
565
|
+
spacing: NO_SPACING,
|
|
566
|
+
isInline: true
|
|
567
|
+
},
|
|
568
|
+
[TAG_COL]: {
|
|
569
|
+
isSelfClosing: true,
|
|
570
|
+
spacing: NO_SPACING
|
|
571
|
+
},
|
|
572
|
+
[TAG_EMBED]: {
|
|
573
|
+
isSelfClosing: true,
|
|
574
|
+
spacing: NO_SPACING
|
|
575
|
+
},
|
|
576
|
+
[TAG_INPUT]: {
|
|
577
|
+
isSelfClosing: true,
|
|
578
|
+
spacing: NO_SPACING,
|
|
579
|
+
isInline: true
|
|
580
|
+
},
|
|
581
|
+
[TAG_KEYGEN]: {
|
|
582
|
+
isSelfClosing: true,
|
|
583
|
+
spacing: NO_SPACING,
|
|
584
|
+
isInline: true
|
|
585
|
+
},
|
|
586
|
+
[TAG_PARAM]: {
|
|
587
|
+
isSelfClosing: true,
|
|
588
|
+
spacing: NO_SPACING
|
|
589
|
+
},
|
|
590
|
+
[TAG_SOURCE]: {
|
|
591
|
+
isSelfClosing: true,
|
|
592
|
+
spacing: NO_SPACING
|
|
593
|
+
},
|
|
594
|
+
[TAG_TRACK]: {
|
|
595
|
+
isSelfClosing: true,
|
|
596
|
+
spacing: NO_SPACING
|
|
597
|
+
},
|
|
598
|
+
[TAG_WBR]: {
|
|
599
|
+
isSelfClosing: true,
|
|
600
|
+
spacing: NO_SPACING,
|
|
601
|
+
isInline: true
|
|
602
|
+
},
|
|
603
|
+
[TAG_SVG]: {
|
|
604
|
+
spacing: NO_SPACING
|
|
605
|
+
},
|
|
606
|
+
[TAG_SELECT]: {
|
|
607
|
+
spacing: NO_SPACING
|
|
608
|
+
},
|
|
609
|
+
[TAG_TEXTAREA]: {
|
|
610
|
+
isNonNesting: true,
|
|
611
|
+
spacing: NO_SPACING
|
|
612
|
+
},
|
|
613
|
+
[TAG_OPTION]: {
|
|
614
|
+
isNonNesting: true,
|
|
615
|
+
spacing: NO_SPACING
|
|
616
|
+
},
|
|
617
|
+
[TAG_FIELDSET]: {
|
|
618
|
+
spacing: NO_SPACING
|
|
619
|
+
},
|
|
620
|
+
[TAG_LEGEND]: {
|
|
621
|
+
spacing: NO_SPACING
|
|
622
|
+
},
|
|
623
|
+
[TAG_AUDIO]: {
|
|
624
|
+
spacing: NO_SPACING
|
|
625
|
+
},
|
|
626
|
+
[TAG_VIDEO]: {
|
|
627
|
+
spacing: NO_SPACING
|
|
628
|
+
},
|
|
629
|
+
[TAG_CANVAS]: {
|
|
630
|
+
spacing: NO_SPACING
|
|
631
|
+
},
|
|
632
|
+
[TAG_IFRAME]: {
|
|
633
|
+
isNonNesting: true,
|
|
634
|
+
spacing: NO_SPACING
|
|
635
|
+
},
|
|
636
|
+
[TAG_MAP]: {
|
|
637
|
+
spacing: NO_SPACING
|
|
638
|
+
},
|
|
639
|
+
[TAG_DIALOG]: {
|
|
640
|
+
spacing: NO_SPACING
|
|
641
|
+
},
|
|
642
|
+
[TAG_METER]: {
|
|
643
|
+
spacing: NO_SPACING
|
|
644
|
+
},
|
|
645
|
+
[TAG_PROGRESS]: {
|
|
646
|
+
spacing: NO_SPACING
|
|
647
|
+
},
|
|
648
|
+
[TAG_TEMPLATE]: {
|
|
649
|
+
spacing: NO_SPACING
|
|
650
|
+
},
|
|
651
|
+
[TAG_ABBR]: {
|
|
652
|
+
enter: () => "",
|
|
653
|
+
exit: () => "",
|
|
654
|
+
collapsesInnerWhiteSpace: true,
|
|
655
|
+
spacing: NO_SPACING,
|
|
656
|
+
isInline: true
|
|
657
|
+
},
|
|
658
|
+
[TAG_MARK]: {
|
|
659
|
+
enter: () => "<mark>",
|
|
660
|
+
exit: () => "</mark>",
|
|
661
|
+
collapsesInnerWhiteSpace: true,
|
|
662
|
+
spacing: NO_SPACING,
|
|
663
|
+
isInline: true
|
|
664
|
+
},
|
|
665
|
+
[TAG_Q]: {
|
|
666
|
+
enter: () => '"',
|
|
667
|
+
exit: () => '"',
|
|
668
|
+
collapsesInnerWhiteSpace: true,
|
|
669
|
+
spacing: NO_SPACING,
|
|
670
|
+
isInline: true
|
|
671
|
+
},
|
|
672
|
+
[TAG_SAMP]: {
|
|
673
|
+
enter: () => "`",
|
|
674
|
+
exit: () => "`",
|
|
675
|
+
collapsesInnerWhiteSpace: true,
|
|
676
|
+
spacing: NO_SPACING,
|
|
677
|
+
isInline: true
|
|
678
|
+
},
|
|
679
|
+
[TAG_SMALL]: {
|
|
680
|
+
enter: () => "",
|
|
681
|
+
exit: () => "",
|
|
682
|
+
collapsesInnerWhiteSpace: true,
|
|
683
|
+
spacing: NO_SPACING,
|
|
684
|
+
isInline: true
|
|
685
|
+
},
|
|
686
|
+
[TAG_NOSCRIPT]: {
|
|
687
|
+
excludesTextNodes: true,
|
|
688
|
+
spacing: NO_SPACING
|
|
689
|
+
},
|
|
690
|
+
[TAG_NOFRAMES]: {
|
|
691
|
+
isNonNesting: true,
|
|
692
|
+
spacing: NO_SPACING
|
|
693
|
+
},
|
|
694
|
+
[TAG_XMP]: {
|
|
695
|
+
isNonNesting: true,
|
|
696
|
+
spacing: NO_SPACING
|
|
697
|
+
},
|
|
698
|
+
[TAG_PLAINTEXT]: {
|
|
699
|
+
isNonNesting: true,
|
|
700
|
+
spacing: NO_SPACING
|
|
701
|
+
},
|
|
702
|
+
[TAG_ASIDE]: {
|
|
703
|
+
spacing: NO_SPACING
|
|
704
|
+
},
|
|
705
|
+
[TAG_U]: {
|
|
706
|
+
enter: () => {
|
|
707
|
+
return "<u>";
|
|
708
|
+
},
|
|
709
|
+
exit: () => {
|
|
710
|
+
return "</u>";
|
|
711
|
+
},
|
|
712
|
+
collapsesInnerWhiteSpace: true,
|
|
713
|
+
spacing: NO_SPACING,
|
|
714
|
+
isInline: true
|
|
715
|
+
},
|
|
716
|
+
[TAG_CITE]: {
|
|
717
|
+
enter: () => "*",
|
|
718
|
+
exit: () => "*",
|
|
719
|
+
collapsesInnerWhiteSpace: true,
|
|
720
|
+
spacing: NO_SPACING,
|
|
721
|
+
isInline: true
|
|
722
|
+
},
|
|
723
|
+
[TAG_DFN]: {
|
|
724
|
+
enter: () => "**",
|
|
725
|
+
exit: () => "**",
|
|
726
|
+
collapsesInnerWhiteSpace: true,
|
|
727
|
+
spacing: NO_SPACING,
|
|
728
|
+
isInline: true
|
|
729
|
+
},
|
|
730
|
+
[TAG_VAR]: {
|
|
731
|
+
enter: () => "`",
|
|
732
|
+
exit: () => "`",
|
|
733
|
+
collapsesInnerWhiteSpace: true,
|
|
734
|
+
spacing: NO_SPACING,
|
|
735
|
+
isInline: true
|
|
736
|
+
},
|
|
737
|
+
[TAG_TIME]: {
|
|
738
|
+
enter: () => "",
|
|
739
|
+
exit: () => "",
|
|
740
|
+
collapsesInnerWhiteSpace: true,
|
|
741
|
+
spacing: NO_SPACING,
|
|
742
|
+
isInline: true
|
|
743
|
+
},
|
|
744
|
+
[TAG_BDO]: {
|
|
745
|
+
enter: () => "",
|
|
746
|
+
exit: () => "",
|
|
747
|
+
collapsesInnerWhiteSpace: true,
|
|
748
|
+
spacing: NO_SPACING,
|
|
749
|
+
isInline: true
|
|
750
|
+
},
|
|
751
|
+
[TAG_RUBY]: {
|
|
752
|
+
enter: () => "",
|
|
753
|
+
exit: () => "",
|
|
754
|
+
collapsesInnerWhiteSpace: true,
|
|
755
|
+
spacing: NO_SPACING,
|
|
756
|
+
isInline: true
|
|
757
|
+
},
|
|
758
|
+
[TAG_RT]: {
|
|
759
|
+
enter: () => "",
|
|
760
|
+
exit: () => "",
|
|
761
|
+
collapsesInnerWhiteSpace: true,
|
|
762
|
+
spacing: NO_SPACING,
|
|
763
|
+
isInline: true
|
|
764
|
+
},
|
|
765
|
+
[TAG_RP]: {
|
|
766
|
+
enter: () => "",
|
|
767
|
+
exit: () => "",
|
|
768
|
+
collapsesInnerWhiteSpace: true,
|
|
769
|
+
spacing: NO_SPACING,
|
|
770
|
+
isInline: true
|
|
771
|
+
},
|
|
772
|
+
[TAG_ADDRESS]: {
|
|
773
|
+
enter: () => "<address>",
|
|
774
|
+
exit: () => "</address>",
|
|
775
|
+
spacing: NO_SPACING,
|
|
776
|
+
collapsesInnerWhiteSpace: true
|
|
777
|
+
},
|
|
778
|
+
[TAG_DL]: {
|
|
779
|
+
spacing: NO_SPACING,
|
|
780
|
+
enter: () => "<dl>",
|
|
781
|
+
exit: () => "</dl>"
|
|
782
|
+
},
|
|
783
|
+
[TAG_DT]: {
|
|
784
|
+
// Definition term
|
|
785
|
+
enter: () => "<dt>",
|
|
786
|
+
exit: () => "</dt>",
|
|
787
|
+
collapsesInnerWhiteSpace: true,
|
|
788
|
+
spacing: [0, 1]
|
|
789
|
+
},
|
|
790
|
+
[TAG_DD]: {
|
|
791
|
+
// Definition term
|
|
792
|
+
enter: () => "<dd>",
|
|
793
|
+
exit: () => "</dd>",
|
|
794
|
+
spacing: [0, 1]
|
|
795
|
+
}
|
|
796
|
+
};
|
|
797
|
+
|
|
798
|
+
function decodeHTMLEntities(text) {
|
|
799
|
+
let result = "";
|
|
800
|
+
let i = 0;
|
|
801
|
+
while (i < text.length) {
|
|
802
|
+
if (text[i] === "&") {
|
|
803
|
+
let match = false;
|
|
804
|
+
for (const [entity, replacement] of Object.entries(HTML_ENTITIES)) {
|
|
805
|
+
if (text.startsWith(entity, i)) {
|
|
806
|
+
result += replacement;
|
|
807
|
+
i += entity.length;
|
|
808
|
+
match = true;
|
|
809
|
+
break;
|
|
810
|
+
}
|
|
811
|
+
}
|
|
812
|
+
if (match)
|
|
813
|
+
continue;
|
|
814
|
+
if (i + 2 < text.length && text[i + 1] === "#") {
|
|
815
|
+
const start = i;
|
|
816
|
+
i += 2;
|
|
817
|
+
const isHex = text[i] === "x" || text[i] === "X";
|
|
818
|
+
if (isHex)
|
|
819
|
+
i++;
|
|
820
|
+
const numStart = i;
|
|
821
|
+
while (i < text.length && text[i] !== ";") {
|
|
822
|
+
i++;
|
|
823
|
+
}
|
|
824
|
+
if (i < text.length && text[i] === ";") {
|
|
825
|
+
const numStr = text.substring(numStart, i);
|
|
826
|
+
const base = isHex ? 16 : 10;
|
|
827
|
+
try {
|
|
828
|
+
const codePoint = Number.parseInt(numStr, base);
|
|
829
|
+
if (!Number.isNaN(codePoint)) {
|
|
830
|
+
result += String.fromCodePoint(codePoint);
|
|
831
|
+
i++;
|
|
832
|
+
continue;
|
|
833
|
+
}
|
|
834
|
+
} catch {
|
|
835
|
+
}
|
|
836
|
+
}
|
|
837
|
+
i = start;
|
|
838
|
+
}
|
|
839
|
+
}
|
|
840
|
+
result += text[i];
|
|
841
|
+
i++;
|
|
842
|
+
}
|
|
843
|
+
return result;
|
|
844
|
+
}
|
|
845
|
+
function traverseUpToFirstBlockNode(node) {
|
|
846
|
+
let firstBlockParent = node;
|
|
847
|
+
const parentsToIncrement = [firstBlockParent];
|
|
848
|
+
while (firstBlockParent.tagHandler?.isInline) {
|
|
849
|
+
if (!firstBlockParent.parent) {
|
|
850
|
+
break;
|
|
851
|
+
}
|
|
852
|
+
firstBlockParent = firstBlockParent.parent;
|
|
853
|
+
parentsToIncrement.push(firstBlockParent);
|
|
854
|
+
}
|
|
855
|
+
return parentsToIncrement;
|
|
856
|
+
}
|
|
857
|
+
|
|
858
|
+
const LT_CHAR = 60;
|
|
859
|
+
const GT_CHAR = 62;
|
|
860
|
+
const SLASH_CHAR = 47;
|
|
861
|
+
const EQUALS_CHAR = 61;
|
|
862
|
+
const QUOTE_CHAR = 34;
|
|
863
|
+
const APOS_CHAR = 39;
|
|
864
|
+
const EXCLAMATION_CHAR = 33;
|
|
865
|
+
const AMPERSAND_CHAR = 38;
|
|
866
|
+
const BACKSLASH_CHAR = 92;
|
|
867
|
+
const DASH_CHAR = 45;
|
|
868
|
+
const SPACE_CHAR = 32;
|
|
869
|
+
const TAB_CHAR = 9;
|
|
870
|
+
const NEWLINE_CHAR = 10;
|
|
871
|
+
const CARRIAGE_RETURN_CHAR = 13;
|
|
872
|
+
const EMPTY_ATTRIBUTES = Object.freeze({});
|
|
873
|
+
function copyDepthMap(depthMap) {
|
|
874
|
+
return new Uint8Array(depthMap);
|
|
875
|
+
}
|
|
876
|
+
function isWhitespace(charCode) {
|
|
877
|
+
return charCode === SPACE_CHAR || charCode === TAB_CHAR || charCode === NEWLINE_CHAR || charCode === CARRIAGE_RETURN_CHAR;
|
|
878
|
+
}
|
|
879
|
+
function runProcessAttributesHooks(node, state) {
|
|
880
|
+
if (!state.plugins?.length)
|
|
881
|
+
return;
|
|
882
|
+
for (const plugin of state.plugins) {
|
|
883
|
+
if (plugin.processAttributes) {
|
|
884
|
+
plugin.processAttributes(node, state);
|
|
885
|
+
}
|
|
886
|
+
}
|
|
887
|
+
}
|
|
888
|
+
function parseHTML(htmlChunk, state, handleEvent) {
|
|
889
|
+
let textBuffer = "";
|
|
890
|
+
state.depthMap ??= new Uint8Array(MAX_TAG_ID);
|
|
891
|
+
state.depth ??= 0;
|
|
892
|
+
state.lastCharWasWhitespace ??= true;
|
|
893
|
+
state.justClosedTag ??= false;
|
|
894
|
+
state.isFirstTextInElement ??= false;
|
|
895
|
+
let i = 0;
|
|
896
|
+
const chunkLength = htmlChunk.length;
|
|
897
|
+
while (i < chunkLength) {
|
|
898
|
+
const currentCharCode = htmlChunk.charCodeAt(i);
|
|
899
|
+
if (currentCharCode !== LT_CHAR) {
|
|
900
|
+
if (currentCharCode === AMPERSAND_CHAR) {
|
|
901
|
+
state.hasEncodedHtmlEntity = true;
|
|
902
|
+
}
|
|
903
|
+
if (isWhitespace(currentCharCode)) {
|
|
904
|
+
const inPreTag = state.depthMap[TAG_PRE] > 0;
|
|
905
|
+
if (state.justClosedTag) {
|
|
906
|
+
state.justClosedTag = false;
|
|
907
|
+
state.lastCharWasWhitespace = false;
|
|
908
|
+
}
|
|
909
|
+
if (!inPreTag && state.lastCharWasWhitespace) {
|
|
910
|
+
i++;
|
|
911
|
+
continue;
|
|
912
|
+
}
|
|
913
|
+
if (inPreTag) {
|
|
914
|
+
textBuffer += htmlChunk[i];
|
|
915
|
+
} else {
|
|
916
|
+
if (currentCharCode === SPACE_CHAR || !state.lastCharWasWhitespace) {
|
|
917
|
+
textBuffer += " ";
|
|
918
|
+
}
|
|
919
|
+
}
|
|
920
|
+
state.lastCharWasWhitespace = true;
|
|
921
|
+
state.textBufferContainsWhitespace = true;
|
|
922
|
+
} else {
|
|
923
|
+
state.textBufferContainsNonWhitespace = true;
|
|
924
|
+
state.lastCharWasWhitespace = false;
|
|
925
|
+
state.justClosedTag = false;
|
|
926
|
+
if (currentCharCode === 124 && state.depthMap[TAG_TABLE]) {
|
|
927
|
+
textBuffer += "\\|";
|
|
928
|
+
} else if (currentCharCode === 96 && (state.depthMap[TAG_CODE] || state.depthMap[TAG_PRE])) {
|
|
929
|
+
textBuffer += "\\`";
|
|
930
|
+
} else if (currentCharCode === 91 && state.depthMap[TAG_A]) {
|
|
931
|
+
textBuffer += "\\[";
|
|
932
|
+
} else if (currentCharCode === 93 && state.depthMap[TAG_A]) {
|
|
933
|
+
textBuffer += "\\]";
|
|
934
|
+
} else if (currentCharCode === 62 && state.depthMap[TAG_BLOCKQUOTE]) {
|
|
935
|
+
textBuffer += "\\>";
|
|
936
|
+
} else {
|
|
937
|
+
textBuffer += htmlChunk[i];
|
|
938
|
+
}
|
|
939
|
+
}
|
|
940
|
+
i++;
|
|
941
|
+
continue;
|
|
942
|
+
}
|
|
943
|
+
if (i + 1 >= chunkLength) {
|
|
944
|
+
textBuffer += htmlChunk[i];
|
|
945
|
+
break;
|
|
946
|
+
}
|
|
947
|
+
const nextCharCode = htmlChunk.charCodeAt(i + 1);
|
|
948
|
+
if (nextCharCode === EXCLAMATION_CHAR) {
|
|
949
|
+
if (textBuffer.length > 0) {
|
|
950
|
+
processTextBuffer(textBuffer, state, handleEvent);
|
|
951
|
+
textBuffer = "";
|
|
952
|
+
}
|
|
953
|
+
const result = processCommentOrDoctype(htmlChunk, i);
|
|
954
|
+
if (result.complete) {
|
|
955
|
+
i = result.newPosition;
|
|
956
|
+
} else {
|
|
957
|
+
textBuffer += result.remainingText;
|
|
958
|
+
break;
|
|
959
|
+
}
|
|
960
|
+
} else if (nextCharCode === SLASH_CHAR) {
|
|
961
|
+
if (textBuffer.length > 0) {
|
|
962
|
+
processTextBuffer(textBuffer, state, handleEvent);
|
|
963
|
+
textBuffer = "";
|
|
964
|
+
}
|
|
965
|
+
const result = processClosingTag(htmlChunk, i, state, handleEvent);
|
|
966
|
+
if (result.complete) {
|
|
967
|
+
i = result.newPosition;
|
|
968
|
+
} else {
|
|
969
|
+
textBuffer += result.remainingText;
|
|
970
|
+
break;
|
|
971
|
+
}
|
|
972
|
+
} else {
|
|
973
|
+
let i2 = i + 1;
|
|
974
|
+
const tagNameStart = i2;
|
|
975
|
+
let tagNameEnd = -1;
|
|
976
|
+
const chunkLength2 = htmlChunk.length;
|
|
977
|
+
while (i2 < chunkLength2) {
|
|
978
|
+
const c = htmlChunk.charCodeAt(i2);
|
|
979
|
+
if (isWhitespace(c) || c === SLASH_CHAR || c === GT_CHAR) {
|
|
980
|
+
tagNameEnd = i2;
|
|
981
|
+
break;
|
|
982
|
+
}
|
|
983
|
+
i2++;
|
|
984
|
+
}
|
|
985
|
+
if (tagNameEnd === -1) {
|
|
986
|
+
textBuffer += htmlChunk.substring(i);
|
|
987
|
+
break;
|
|
988
|
+
}
|
|
989
|
+
const tagName = htmlChunk.substring(tagNameStart, tagNameEnd).toLowerCase();
|
|
990
|
+
if (!tagName) {
|
|
991
|
+
i = tagNameEnd;
|
|
992
|
+
break;
|
|
993
|
+
}
|
|
994
|
+
const tagId = TagIdMap[tagName] ?? -1;
|
|
995
|
+
i2 = tagNameEnd;
|
|
996
|
+
if (state.currentNode?.tagHandler?.isNonNesting) {
|
|
997
|
+
if (tagId !== state.currentNode?.tagId) {
|
|
998
|
+
textBuffer += htmlChunk[i++];
|
|
999
|
+
continue;
|
|
1000
|
+
}
|
|
1001
|
+
}
|
|
1002
|
+
if (textBuffer.length > 0) {
|
|
1003
|
+
processTextBuffer(textBuffer, state, handleEvent);
|
|
1004
|
+
textBuffer = "";
|
|
1005
|
+
}
|
|
1006
|
+
const result = processOpeningTag(tagName, tagId, htmlChunk, i2, state, handleEvent);
|
|
1007
|
+
if (result.skip) {
|
|
1008
|
+
textBuffer += htmlChunk[i++];
|
|
1009
|
+
} else if (result.complete) {
|
|
1010
|
+
i = result.newPosition;
|
|
1011
|
+
if (!result.selfClosing) {
|
|
1012
|
+
state.isFirstTextInElement = true;
|
|
1013
|
+
}
|
|
1014
|
+
} else {
|
|
1015
|
+
textBuffer += result.remainingText;
|
|
1016
|
+
break;
|
|
1017
|
+
}
|
|
1018
|
+
}
|
|
1019
|
+
}
|
|
1020
|
+
return textBuffer;
|
|
1021
|
+
}
|
|
1022
|
+
function processTextBuffer(textBuffer, state, handleEvent) {
|
|
1023
|
+
const containsNonWhitespace = state.textBufferContainsNonWhitespace;
|
|
1024
|
+
const containsWhitespace = state.textBufferContainsWhitespace;
|
|
1025
|
+
state.textBufferContainsNonWhitespace = false;
|
|
1026
|
+
state.textBufferContainsWhitespace = false;
|
|
1027
|
+
if (!state.currentNode || state.currentNode?.tagHandler?.excludesTextNodes) {
|
|
1028
|
+
return;
|
|
1029
|
+
}
|
|
1030
|
+
const inPreTag = state.depthMap[TAG_PRE] > 0;
|
|
1031
|
+
if (!inPreTag && !containsNonWhitespace && !state.currentNode.childTextNodeIndex) {
|
|
1032
|
+
return;
|
|
1033
|
+
}
|
|
1034
|
+
let text = textBuffer;
|
|
1035
|
+
if (text.length === 0) {
|
|
1036
|
+
return;
|
|
1037
|
+
}
|
|
1038
|
+
const parentsToIncrement = traverseUpToFirstBlockNode(state.currentNode);
|
|
1039
|
+
const firstBlockParent = parentsToIncrement[parentsToIncrement.length - 1];
|
|
1040
|
+
if (containsWhitespace && !firstBlockParent?.childTextNodeIndex) {
|
|
1041
|
+
let start = 0;
|
|
1042
|
+
while (start < text.length && (inPreTag ? text.charCodeAt(start) === 10 || text.charCodeAt(start) === 13 : isWhitespace(text.charCodeAt(start)))) {
|
|
1043
|
+
start++;
|
|
1044
|
+
}
|
|
1045
|
+
if (start > 0) {
|
|
1046
|
+
text = text.substring(start);
|
|
1047
|
+
}
|
|
1048
|
+
}
|
|
1049
|
+
if (state.hasEncodedHtmlEntity) {
|
|
1050
|
+
text = decodeHTMLEntities(String(text));
|
|
1051
|
+
state.hasEncodedHtmlEntity = false;
|
|
1052
|
+
}
|
|
1053
|
+
const textNode = {
|
|
1054
|
+
type: TEXT_NODE,
|
|
1055
|
+
value: text,
|
|
1056
|
+
parent: state.currentNode,
|
|
1057
|
+
regionId: state.currentNode?.regionId,
|
|
1058
|
+
index: state.currentNode.currentWalkIndex++,
|
|
1059
|
+
depth: state.depth,
|
|
1060
|
+
containsWhitespace
|
|
1061
|
+
};
|
|
1062
|
+
for (const parent of parentsToIncrement) {
|
|
1063
|
+
parent.childTextNodeIndex = (parent.childTextNodeIndex || 0) + 1;
|
|
1064
|
+
}
|
|
1065
|
+
handleEvent({ type: NodeEventEnter, node: textNode });
|
|
1066
|
+
state.lastTextNode = textNode;
|
|
1067
|
+
}
|
|
1068
|
+
function processClosingTag(htmlChunk, position, state, handleEvent) {
|
|
1069
|
+
let i = position + 2;
|
|
1070
|
+
const tagNameStart = i;
|
|
1071
|
+
const chunkLength = htmlChunk.length;
|
|
1072
|
+
let foundClose = false;
|
|
1073
|
+
while (i < chunkLength) {
|
|
1074
|
+
const charCode = htmlChunk.charCodeAt(i);
|
|
1075
|
+
if (charCode === GT_CHAR) {
|
|
1076
|
+
foundClose = true;
|
|
1077
|
+
break;
|
|
1078
|
+
}
|
|
1079
|
+
i++;
|
|
1080
|
+
}
|
|
1081
|
+
if (!foundClose) {
|
|
1082
|
+
return {
|
|
1083
|
+
complete: false,
|
|
1084
|
+
newPosition: position,
|
|
1085
|
+
remainingText: htmlChunk.substring(position)
|
|
1086
|
+
};
|
|
1087
|
+
}
|
|
1088
|
+
const tagName = htmlChunk.substring(tagNameStart, i).toLowerCase();
|
|
1089
|
+
const tagId = TagIdMap[tagName] ?? -1;
|
|
1090
|
+
if (state.currentNode?.tagHandler?.isNonNesting && tagId !== state.currentNode.tagId) {
|
|
1091
|
+
return {
|
|
1092
|
+
complete: false,
|
|
1093
|
+
newPosition: position,
|
|
1094
|
+
remainingText: htmlChunk.substring(position)
|
|
1095
|
+
};
|
|
1096
|
+
}
|
|
1097
|
+
let curr = state.currentNode;
|
|
1098
|
+
if (curr) {
|
|
1099
|
+
let match = curr.tagId !== tagId;
|
|
1100
|
+
while (curr && match) {
|
|
1101
|
+
closeNode(curr, state, handleEvent);
|
|
1102
|
+
curr = curr.parent;
|
|
1103
|
+
match = curr?.tagId !== tagId;
|
|
1104
|
+
}
|
|
1105
|
+
}
|
|
1106
|
+
if (curr) {
|
|
1107
|
+
closeNode(state.currentNode, state, handleEvent);
|
|
1108
|
+
}
|
|
1109
|
+
state.justClosedTag = true;
|
|
1110
|
+
return {
|
|
1111
|
+
complete: true,
|
|
1112
|
+
newPosition: i + 1,
|
|
1113
|
+
// Skip past '>'
|
|
1114
|
+
remainingText: ""
|
|
1115
|
+
};
|
|
1116
|
+
}
|
|
1117
|
+
function closeNode(node, state, handleEvent) {
|
|
1118
|
+
if (!node) {
|
|
1119
|
+
return;
|
|
1120
|
+
}
|
|
1121
|
+
if (node.tagId === TAG_A && !node.childTextNodeIndex) {
|
|
1122
|
+
const prefix = node.attributes?.title || node.attributes?.["aria-label"] || "";
|
|
1123
|
+
if (prefix) {
|
|
1124
|
+
node.childTextNodeIndex = 1;
|
|
1125
|
+
const textNode = {
|
|
1126
|
+
type: TEXT_NODE,
|
|
1127
|
+
value: prefix,
|
|
1128
|
+
parent: node,
|
|
1129
|
+
index: 0,
|
|
1130
|
+
depth: node.depth + 1
|
|
1131
|
+
};
|
|
1132
|
+
handleEvent({ type: NodeEventEnter, node: textNode });
|
|
1133
|
+
for (const parent of traverseUpToFirstBlockNode(node)) {
|
|
1134
|
+
parent.childTextNodeIndex = (parent.childTextNodeIndex || 0) + 1;
|
|
1135
|
+
}
|
|
1136
|
+
}
|
|
1137
|
+
}
|
|
1138
|
+
if (node.tagId) {
|
|
1139
|
+
state.depthMap[node.tagId] = Math.max(0, state.depthMap[node.tagId] - 1);
|
|
1140
|
+
}
|
|
1141
|
+
state.depth--;
|
|
1142
|
+
handleEvent({ type: NodeEventExit, node });
|
|
1143
|
+
state.currentNode = state.currentNode.parent;
|
|
1144
|
+
state.hasEncodedHtmlEntity = false;
|
|
1145
|
+
state.justClosedTag = true;
|
|
1146
|
+
}
|
|
1147
|
+
function processCommentOrDoctype(htmlChunk, position) {
|
|
1148
|
+
let i = position;
|
|
1149
|
+
const chunkLength = htmlChunk.length;
|
|
1150
|
+
if (i + 3 < chunkLength && htmlChunk.charCodeAt(i + 2) === DASH_CHAR && htmlChunk.charCodeAt(i + 3) === DASH_CHAR) {
|
|
1151
|
+
i += 4;
|
|
1152
|
+
while (i < chunkLength - 2) {
|
|
1153
|
+
if (htmlChunk.charCodeAt(i) === DASH_CHAR && htmlChunk.charCodeAt(i + 1) === DASH_CHAR && htmlChunk.charCodeAt(i + 2) === GT_CHAR) {
|
|
1154
|
+
i += 3;
|
|
1155
|
+
return {
|
|
1156
|
+
complete: true,
|
|
1157
|
+
newPosition: i,
|
|
1158
|
+
remainingText: ""
|
|
1159
|
+
};
|
|
1160
|
+
}
|
|
1161
|
+
i++;
|
|
1162
|
+
}
|
|
1163
|
+
return {
|
|
1164
|
+
complete: false,
|
|
1165
|
+
newPosition: position,
|
|
1166
|
+
remainingText: htmlChunk.substring(position)
|
|
1167
|
+
};
|
|
1168
|
+
} else {
|
|
1169
|
+
i += 2;
|
|
1170
|
+
while (i < chunkLength) {
|
|
1171
|
+
if (htmlChunk.charCodeAt(i) === GT_CHAR) {
|
|
1172
|
+
i++;
|
|
1173
|
+
return {
|
|
1174
|
+
complete: true,
|
|
1175
|
+
newPosition: i,
|
|
1176
|
+
remainingText: ""
|
|
1177
|
+
};
|
|
1178
|
+
}
|
|
1179
|
+
i++;
|
|
1180
|
+
}
|
|
1181
|
+
return {
|
|
1182
|
+
complete: false,
|
|
1183
|
+
newPosition: i,
|
|
1184
|
+
remainingText: htmlChunk.substring(position, i)
|
|
1185
|
+
};
|
|
1186
|
+
}
|
|
1187
|
+
}
|
|
1188
|
+
function processOpeningTag(tagName, tagId, htmlChunk, i, state, handleEvent) {
|
|
1189
|
+
if (state.currentNode?.tagHandler?.isNonNesting) {
|
|
1190
|
+
closeNode(state.currentNode, state, handleEvent);
|
|
1191
|
+
}
|
|
1192
|
+
const tagHandler = tagHandlers[tagId];
|
|
1193
|
+
const result = processTagAttributes(htmlChunk, i, tagHandler);
|
|
1194
|
+
if (!result.complete) {
|
|
1195
|
+
return {
|
|
1196
|
+
complete: false,
|
|
1197
|
+
newPosition: i,
|
|
1198
|
+
remainingText: `<${tagName}${result.attrBuffer}`,
|
|
1199
|
+
selfClosing: false
|
|
1200
|
+
};
|
|
1201
|
+
}
|
|
1202
|
+
const currentTagCount = state.depthMap[tagId];
|
|
1203
|
+
state.depthMap[tagId] = currentTagCount + 1;
|
|
1204
|
+
state.depth++;
|
|
1205
|
+
i = result.newPosition;
|
|
1206
|
+
if (state.currentNode) {
|
|
1207
|
+
state.currentNode.currentWalkIndex = state.currentNode.currentWalkIndex || 0;
|
|
1208
|
+
}
|
|
1209
|
+
const currentWalkIndex = state.currentNode ? state.currentNode.currentWalkIndex++ : 0;
|
|
1210
|
+
const tag = {
|
|
1211
|
+
type: ELEMENT_NODE,
|
|
1212
|
+
name: tagName,
|
|
1213
|
+
attributes: result.attributes,
|
|
1214
|
+
parent: state.currentNode,
|
|
1215
|
+
depthMap: copyDepthMap(state.depthMap),
|
|
1216
|
+
depth: state.depth,
|
|
1217
|
+
index: currentWalkIndex,
|
|
1218
|
+
regionId: state.currentNode?.regionId,
|
|
1219
|
+
tagId,
|
|
1220
|
+
tagHandler
|
|
1221
|
+
};
|
|
1222
|
+
state.lastTextNode = tag;
|
|
1223
|
+
if (state.options?.plugins) {
|
|
1224
|
+
runProcessAttributesHooks(tag, state);
|
|
1225
|
+
}
|
|
1226
|
+
handleEvent({ type: NodeEventEnter, node: tag });
|
|
1227
|
+
const parentNode = tag;
|
|
1228
|
+
parentNode.currentWalkIndex = 0;
|
|
1229
|
+
state.currentNode = parentNode;
|
|
1230
|
+
state.hasEncodedHtmlEntity = false;
|
|
1231
|
+
if (result.selfClosing) {
|
|
1232
|
+
closeNode(tag, state, handleEvent);
|
|
1233
|
+
state.justClosedTag = true;
|
|
1234
|
+
} else {
|
|
1235
|
+
state.justClosedTag = false;
|
|
1236
|
+
}
|
|
1237
|
+
return {
|
|
1238
|
+
complete: true,
|
|
1239
|
+
newPosition: i,
|
|
1240
|
+
remainingText: "",
|
|
1241
|
+
selfClosing: result.selfClosing
|
|
1242
|
+
};
|
|
1243
|
+
}
|
|
1244
|
+
function processTagAttributes(htmlChunk, position, tagHandler) {
|
|
1245
|
+
let i = position;
|
|
1246
|
+
const chunkLength = htmlChunk.length;
|
|
1247
|
+
const selfClosing = tagHandler?.isSelfClosing || false;
|
|
1248
|
+
const attrStartPos = i;
|
|
1249
|
+
let insideQuote = false;
|
|
1250
|
+
let quoteChar = 0;
|
|
1251
|
+
let prevChar = 0;
|
|
1252
|
+
while (i < chunkLength) {
|
|
1253
|
+
const c = htmlChunk.charCodeAt(i);
|
|
1254
|
+
if (insideQuote) {
|
|
1255
|
+
if (c === quoteChar && prevChar !== BACKSLASH_CHAR) {
|
|
1256
|
+
insideQuote = false;
|
|
1257
|
+
}
|
|
1258
|
+
i++;
|
|
1259
|
+
continue;
|
|
1260
|
+
} else if (c === QUOTE_CHAR || c === APOS_CHAR) {
|
|
1261
|
+
insideQuote = true;
|
|
1262
|
+
quoteChar = c;
|
|
1263
|
+
} else if (c === SLASH_CHAR && i + 1 < chunkLength && htmlChunk.charCodeAt(i + 1) === GT_CHAR) {
|
|
1264
|
+
const attrStr = htmlChunk.substring(attrStartPos, i).trim();
|
|
1265
|
+
return {
|
|
1266
|
+
complete: true,
|
|
1267
|
+
newPosition: i + 2,
|
|
1268
|
+
attributes: parseAttributes(attrStr),
|
|
1269
|
+
selfClosing: true,
|
|
1270
|
+
attrBuffer: attrStr
|
|
1271
|
+
};
|
|
1272
|
+
} else if (c === GT_CHAR) {
|
|
1273
|
+
const attrStr = htmlChunk.substring(attrStartPos, i).trim();
|
|
1274
|
+
return {
|
|
1275
|
+
complete: true,
|
|
1276
|
+
newPosition: i + 1,
|
|
1277
|
+
attributes: parseAttributes(attrStr),
|
|
1278
|
+
selfClosing,
|
|
1279
|
+
attrBuffer: attrStr
|
|
1280
|
+
};
|
|
1281
|
+
}
|
|
1282
|
+
i++;
|
|
1283
|
+
prevChar = c;
|
|
1284
|
+
}
|
|
1285
|
+
return {
|
|
1286
|
+
complete: false,
|
|
1287
|
+
newPosition: i,
|
|
1288
|
+
attributes: EMPTY_ATTRIBUTES,
|
|
1289
|
+
selfClosing: false,
|
|
1290
|
+
attrBuffer: htmlChunk.substring(attrStartPos, i)
|
|
1291
|
+
};
|
|
1292
|
+
}
|
|
1293
|
+
function parseAttributes(attrStr) {
|
|
1294
|
+
if (!attrStr)
|
|
1295
|
+
return EMPTY_ATTRIBUTES;
|
|
1296
|
+
const result = {};
|
|
1297
|
+
const len = attrStr.length;
|
|
1298
|
+
let i = 0;
|
|
1299
|
+
const WHITESPACE = 0;
|
|
1300
|
+
const NAME = 1;
|
|
1301
|
+
const AFTER_NAME = 2;
|
|
1302
|
+
const BEFORE_VALUE = 3;
|
|
1303
|
+
const QUOTED_VALUE = 4;
|
|
1304
|
+
const UNQUOTED_VALUE = 5;
|
|
1305
|
+
let state = WHITESPACE;
|
|
1306
|
+
let nameStart = 0;
|
|
1307
|
+
let nameEnd = 0;
|
|
1308
|
+
let valueStart = 0;
|
|
1309
|
+
let quoteChar = 0;
|
|
1310
|
+
let name = "";
|
|
1311
|
+
while (i < len) {
|
|
1312
|
+
const charCode = attrStr.charCodeAt(i);
|
|
1313
|
+
const isSpace = isWhitespace(charCode);
|
|
1314
|
+
switch (state) {
|
|
1315
|
+
case WHITESPACE:
|
|
1316
|
+
if (!isSpace) {
|
|
1317
|
+
state = NAME;
|
|
1318
|
+
nameStart = i;
|
|
1319
|
+
}
|
|
1320
|
+
break;
|
|
1321
|
+
case NAME:
|
|
1322
|
+
if (charCode === EQUALS_CHAR || isSpace) {
|
|
1323
|
+
nameEnd = i;
|
|
1324
|
+
name = attrStr.substring(nameStart, nameEnd).toLowerCase();
|
|
1325
|
+
state = charCode === EQUALS_CHAR ? BEFORE_VALUE : AFTER_NAME;
|
|
1326
|
+
}
|
|
1327
|
+
break;
|
|
1328
|
+
case AFTER_NAME:
|
|
1329
|
+
if (charCode === EQUALS_CHAR) {
|
|
1330
|
+
state = BEFORE_VALUE;
|
|
1331
|
+
} else if (!isSpace) {
|
|
1332
|
+
result[name] = "";
|
|
1333
|
+
state = NAME;
|
|
1334
|
+
nameStart = i;
|
|
1335
|
+
}
|
|
1336
|
+
break;
|
|
1337
|
+
case BEFORE_VALUE:
|
|
1338
|
+
if (charCode === QUOTE_CHAR || charCode === APOS_CHAR) {
|
|
1339
|
+
quoteChar = charCode;
|
|
1340
|
+
state = QUOTED_VALUE;
|
|
1341
|
+
valueStart = i + 1;
|
|
1342
|
+
} else if (!isSpace) {
|
|
1343
|
+
state = UNQUOTED_VALUE;
|
|
1344
|
+
valueStart = i;
|
|
1345
|
+
}
|
|
1346
|
+
break;
|
|
1347
|
+
case QUOTED_VALUE:
|
|
1348
|
+
if (charCode === BACKSLASH_CHAR && i + 1 < len) {
|
|
1349
|
+
i++;
|
|
1350
|
+
} else if (charCode === quoteChar) {
|
|
1351
|
+
result[name] = attrStr.substring(valueStart, i);
|
|
1352
|
+
state = WHITESPACE;
|
|
1353
|
+
}
|
|
1354
|
+
break;
|
|
1355
|
+
case UNQUOTED_VALUE:
|
|
1356
|
+
if (isSpace || charCode === GT_CHAR) {
|
|
1357
|
+
result[name] = attrStr.substring(valueStart, i);
|
|
1358
|
+
state = WHITESPACE;
|
|
1359
|
+
}
|
|
1360
|
+
break;
|
|
1361
|
+
}
|
|
1362
|
+
i++;
|
|
1363
|
+
}
|
|
1364
|
+
if (name) {
|
|
1365
|
+
if (state === QUOTED_VALUE || state === UNQUOTED_VALUE) {
|
|
1366
|
+
result[name] = attrStr.substring(valueStart, i);
|
|
1367
|
+
} else if (state === NAME || state === AFTER_NAME || state === BEFORE_VALUE) {
|
|
1368
|
+
nameEnd = nameEnd || i;
|
|
1369
|
+
name = name || attrStr.substring(nameStart, nameEnd).toLowerCase();
|
|
1370
|
+
result[name] = "";
|
|
1371
|
+
}
|
|
1372
|
+
}
|
|
1373
|
+
return result;
|
|
1374
|
+
}
|
|
1375
|
+
function processPartialHTMLToMarkdown(partialHtml, state = {}) {
|
|
1376
|
+
state.depthMap ??= new Uint8Array(MAX_TAG_ID);
|
|
1377
|
+
state.plugins = [...state.options?.plugins || []];
|
|
1378
|
+
state.regionToggles ??= /* @__PURE__ */ new Map();
|
|
1379
|
+
state.regionContentBuffers ??= /* @__PURE__ */ new Map();
|
|
1380
|
+
state.regionToggles.set(0, true);
|
|
1381
|
+
state.regionContentBuffers.set(0, []);
|
|
1382
|
+
const fullState = state;
|
|
1383
|
+
function handleEvent(event) {
|
|
1384
|
+
for (const plugin of fullState.plugins || []) {
|
|
1385
|
+
const res = plugin.beforeNodeProcess?.(event, fullState);
|
|
1386
|
+
if (typeof res === "object" && res.skip) {
|
|
1387
|
+
return;
|
|
1388
|
+
}
|
|
1389
|
+
}
|
|
1390
|
+
if (event.node.type === TEXT_NODE) {
|
|
1391
|
+
processHtmlEventToMarkdown(event, fullState);
|
|
1392
|
+
return;
|
|
1393
|
+
}
|
|
1394
|
+
processHtmlEventToMarkdown(event, fullState);
|
|
1395
|
+
}
|
|
1396
|
+
const unprocessedHtml = parseHTML(partialHtml, fullState, handleEvent);
|
|
1397
|
+
const assembledContent = assembleBufferedContent(fullState);
|
|
1398
|
+
return { chunk: assembledContent, remainingHTML: unprocessedHtml };
|
|
1399
|
+
}
|
|
1400
|
+
|
|
1401
|
+
async function* streamHtmlToMarkdown(htmlStream, options = {}) {
|
|
1402
|
+
if (!htmlStream) {
|
|
1403
|
+
throw new Error("Invalid HTML stream provided");
|
|
1404
|
+
}
|
|
1405
|
+
const decoder = new TextDecoder();
|
|
1406
|
+
const reader = htmlStream.getReader();
|
|
1407
|
+
const state = {
|
|
1408
|
+
options
|
|
1409
|
+
};
|
|
1410
|
+
let remainingHtml = "";
|
|
1411
|
+
try {
|
|
1412
|
+
while (true) {
|
|
1413
|
+
const { done, value } = await reader.read();
|
|
1414
|
+
if (done) {
|
|
1415
|
+
break;
|
|
1416
|
+
}
|
|
1417
|
+
const htmlContent = `${remainingHtml}${typeof value === "string" ? value : decoder.decode(value, { stream: true })}`;
|
|
1418
|
+
const result = processPartialHTMLToMarkdown(htmlContent, state);
|
|
1419
|
+
if (result.chunk) {
|
|
1420
|
+
yield result.chunk;
|
|
1421
|
+
}
|
|
1422
|
+
remainingHtml = result.remainingHTML;
|
|
1423
|
+
}
|
|
1424
|
+
} finally {
|
|
1425
|
+
if (remainingHtml) {
|
|
1426
|
+
decoder.decode(new Uint8Array(0), { stream: false });
|
|
1427
|
+
}
|
|
1428
|
+
reader.releaseLock();
|
|
1429
|
+
}
|
|
1430
|
+
}
|
|
1431
|
+
|
|
1432
|
+
export { processPartialHTMLToMarkdown as p, streamHtmlToMarkdown as s };
|