@oh-my-pi/pi-utils 18.6.3 → 18.8.0
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/CHANGELOG.md +16 -0
- package/dist/types/ar/index.d.ts +1 -0
- package/dist/types/ar/zip-package.d.ts +21 -0
- package/dist/types/docx/converter.d.ts +2 -0
- package/dist/types/dom/core.d.ts +15 -4
- package/dist/types/dom/parser.d.ts +5 -3
- package/dist/types/dom/selector.d.ts +7 -1
- package/dist/types/logger/rotating-file.d.ts +4 -2
- package/dist/types/marked/core.d.ts +7 -6
- package/dist/types/marked-list.d.ts +38 -0
- package/dist/types/math-delimiters.d.ts +10 -0
- package/dist/types/readability/types.d.ts +2 -0
- package/dist/types/turndown/types.d.ts +7 -0
- package/dist/types/vterm/buffer.d.ts +16 -5
- package/package.json +2 -2
- package/src/ar/index.ts +1 -0
- package/src/ar/zip-package.ts +57 -0
- package/src/chalk.ts +66 -23
- package/src/docx/converter.ts +26 -12
- package/src/dom/core.ts +354 -111
- package/src/dom/parser.ts +42 -19
- package/src/dom/selector.ts +328 -95
- package/src/fetch-retry.ts +2 -0
- package/src/json-parse.ts +21 -6
- package/src/logger/rotating-file.ts +5 -3
- package/src/logger.ts +39 -23
- package/src/marked/core.ts +42 -18
- package/src/marked-list.ts +73 -0
- package/src/math-delimiters.ts +16 -0
- package/src/prompt.ts +3 -1
- package/src/ptree.ts +28 -35
- package/src/readability/readability.ts +349 -39
- package/src/readability/types.ts +2 -0
- package/src/stream.ts +34 -13
- package/src/temp.ts +1 -10
- package/src/template.ts +7 -4
- package/src/turndown/gfm.ts +16 -5
- package/src/turndown/html.ts +28 -8
- package/src/turndown/service.ts +43 -15
- package/src/turndown/types.ts +7 -0
- package/src/vterm/buffer.ts +31 -11
- package/src/vterm/terminal.ts +185 -92
|
@@ -34,6 +34,9 @@ const UNLIKELY_ROLES = new Set(["menu", "menubar", "complementary", "navigation"
|
|
|
34
34
|
const ARTICLE_TYPES =
|
|
35
35
|
/^(?:Article|AdvertiserContentArticle|NewsArticle|AnalysisNewsArticle|OpinionNewsArticle|ReportageNewsArticle|ReviewNewsArticle|Report|ScholarlyArticle|MedicalScholarlyArticle|SocialMediaPosting|BlogPosting|LiveBlogPosting|DiscussionForumPosting|TechArticle|APIReference)$/;
|
|
36
36
|
const NORMALIZE = /\s{2,}/g;
|
|
37
|
+
const ELEMENT_NODE = 1;
|
|
38
|
+
const TEXT_NODE = 3;
|
|
39
|
+
const CDATA_SECTION_NODE = 4;
|
|
37
40
|
|
|
38
41
|
type Metadata = {
|
|
39
42
|
title?: string;
|
|
@@ -119,6 +122,273 @@ function linkDensity(node: ReadabilityElement): number {
|
|
|
119
122
|
return linked / total;
|
|
120
123
|
}
|
|
121
124
|
|
|
125
|
+
/** Whether a UTF-16 code unit is whitespace to `String#trim` and `\s`. */
|
|
126
|
+
function isSpace(code: number): boolean {
|
|
127
|
+
if (code <= 0x20) return code === 0x20 || (code >= 0x09 && code <= 0x0d);
|
|
128
|
+
if (code < 0xa0) return false;
|
|
129
|
+
return (
|
|
130
|
+
code === 0xa0 ||
|
|
131
|
+
code === 0x1680 ||
|
|
132
|
+
(code >= 0x2000 && code <= 0x200a) ||
|
|
133
|
+
code === 0x2028 ||
|
|
134
|
+
code === 0x2029 ||
|
|
135
|
+
code === 0x202f ||
|
|
136
|
+
code === 0x205f ||
|
|
137
|
+
code === 0x3000 ||
|
|
138
|
+
code === 0xfeff
|
|
139
|
+
);
|
|
140
|
+
}
|
|
141
|
+
|
|
142
|
+
/** Separators `#scoreParagraph` counts clauses by. */
|
|
143
|
+
function isSeparator(code: number): boolean {
|
|
144
|
+
return (
|
|
145
|
+
code === 0x2c ||
|
|
146
|
+
code === 0x060c ||
|
|
147
|
+
code === 0xfe50 ||
|
|
148
|
+
code === 0xfe10 ||
|
|
149
|
+
code === 0xfe11 ||
|
|
150
|
+
code === 0x2e41 ||
|
|
151
|
+
code === 0x2e34 ||
|
|
152
|
+
code === 0x2e32 ||
|
|
153
|
+
code === 0xff0c
|
|
154
|
+
);
|
|
155
|
+
}
|
|
156
|
+
|
|
157
|
+
const SEPARATORS = /[\u002c\u060c\ufe50\ufe10\ufe11\u2e41\u2e34\u2e32\uff0c]/;
|
|
158
|
+
|
|
159
|
+
/** {@link TextRun} flags: no non-whitespace text; starts with whitespace; ends with whitespace. */
|
|
160
|
+
const BLANK = 1;
|
|
161
|
+
const LEADING_SPACE = 2;
|
|
162
|
+
const TRAILING_SPACE = 4;
|
|
163
|
+
|
|
164
|
+
/** Measures `text()` of concatenated pieces without building the string. */
|
|
165
|
+
class TextRun {
|
|
166
|
+
length = 0;
|
|
167
|
+
/** A blank run that contains whitespace carries both space flags. */
|
|
168
|
+
flags = BLANK;
|
|
169
|
+
commas = 0;
|
|
170
|
+
separators = 0;
|
|
171
|
+
|
|
172
|
+
reset(): void {
|
|
173
|
+
this.length = 0;
|
|
174
|
+
this.flags = BLANK;
|
|
175
|
+
this.commas = 0;
|
|
176
|
+
this.separators = 0;
|
|
177
|
+
}
|
|
178
|
+
|
|
179
|
+
/** Append a run measured earlier. */
|
|
180
|
+
append(length: number, flags: number, commas: number, separators: number): void {
|
|
181
|
+
this.commas += commas;
|
|
182
|
+
this.separators += separators;
|
|
183
|
+
if (flags & BLANK) {
|
|
184
|
+
if (flags & LEADING_SPACE) this.flags |= this.flags & BLANK ? LEADING_SPACE | TRAILING_SPACE : TRAILING_SPACE;
|
|
185
|
+
return;
|
|
186
|
+
}
|
|
187
|
+
if (this.flags & BLANK) {
|
|
188
|
+
this.length = length;
|
|
189
|
+
this.flags = (this.flags & LEADING_SPACE) | flags;
|
|
190
|
+
return;
|
|
191
|
+
}
|
|
192
|
+
// Whitespace between two non-blank runs collapses to a single character.
|
|
193
|
+
if (this.flags & TRAILING_SPACE || flags & LEADING_SPACE) this.length++;
|
|
194
|
+
this.length += length;
|
|
195
|
+
this.flags = (this.flags & LEADING_SPACE) | (flags & TRAILING_SPACE);
|
|
196
|
+
}
|
|
197
|
+
|
|
198
|
+
appendText(value: string): void {
|
|
199
|
+
if (!value) return;
|
|
200
|
+
let length = 0;
|
|
201
|
+
let commas = 0;
|
|
202
|
+
let separators = 0;
|
|
203
|
+
let seen = false;
|
|
204
|
+
let gap = false;
|
|
205
|
+
for (let index = 0; index < value.length; index++) {
|
|
206
|
+
const code = value.charCodeAt(index);
|
|
207
|
+
if (isSpace(code)) {
|
|
208
|
+
if (seen) gap = true;
|
|
209
|
+
continue;
|
|
210
|
+
}
|
|
211
|
+
if (gap) {
|
|
212
|
+
length++;
|
|
213
|
+
gap = false;
|
|
214
|
+
}
|
|
215
|
+
length++;
|
|
216
|
+
seen = true;
|
|
217
|
+
if (isSeparator(code)) {
|
|
218
|
+
separators++;
|
|
219
|
+
if (code === 0x2c) commas++;
|
|
220
|
+
}
|
|
221
|
+
}
|
|
222
|
+
const flags = seen
|
|
223
|
+
? (isSpace(value.charCodeAt(0)) ? LEADING_SPACE : 0) | (gap ? TRAILING_SPACE : 0)
|
|
224
|
+
: BLANK | LEADING_SPACE | TRAILING_SPACE;
|
|
225
|
+
this.append(length, flags, commas, separators);
|
|
226
|
+
}
|
|
227
|
+
}
|
|
228
|
+
|
|
229
|
+
function isTag(node: ReadabilityElement, upper: string, lower: string): boolean {
|
|
230
|
+
return node.tagName === upper || node.tagName === lower;
|
|
231
|
+
}
|
|
232
|
+
|
|
233
|
+
/**
|
|
234
|
+
* `text()` lengths, comma counts, descendant tag counts and link density of every element in a tree, from one
|
|
235
|
+
* bottom-up pass over the whole tree holding the first element asked about. Answers describe the tree as
|
|
236
|
+
* measured, so only ask about elements whose subtree has not changed since.
|
|
237
|
+
*/
|
|
238
|
+
class SubtreeStats {
|
|
239
|
+
readonly #countsAsText: (node: ReadabilityNode) => boolean;
|
|
240
|
+
readonly #index = new Map<ReadabilityNode, number>();
|
|
241
|
+
readonly #length: number[] = [];
|
|
242
|
+
readonly #flags: number[] = [];
|
|
243
|
+
readonly #commas: number[] = [];
|
|
244
|
+
readonly #separators: number[] = [];
|
|
245
|
+
readonly #paragraphs: number[] = [];
|
|
246
|
+
readonly #images: number[] = [];
|
|
247
|
+
readonly #inputs: number[] = [];
|
|
248
|
+
/** Range of an element's descendant anchors in {@link #links}. */
|
|
249
|
+
readonly #linkStart: number[] = [];
|
|
250
|
+
readonly #linkEnd: number[] = [];
|
|
251
|
+
/** `linkDensity` term of each anchor, in document order. */
|
|
252
|
+
readonly #links: number[] = [];
|
|
253
|
+
|
|
254
|
+
/** `countsAsText` tells whether a non-element child's `textContent` is part of its parent's. */
|
|
255
|
+
constructor(countsAsText: (node: ReadabilityNode) => boolean) {
|
|
256
|
+
this.#countsAsText = countsAsText;
|
|
257
|
+
}
|
|
258
|
+
|
|
259
|
+
/** `text(node).length`. */
|
|
260
|
+
length(node: ReadabilityElement): number {
|
|
261
|
+
const index = this.#at(node);
|
|
262
|
+
return index < 0 ? text(node).length : this.#length[index];
|
|
263
|
+
}
|
|
264
|
+
|
|
265
|
+
/** `!text(node)`. */
|
|
266
|
+
blank(node: ReadabilityElement): boolean {
|
|
267
|
+
const index = this.#at(node);
|
|
268
|
+
return index < 0 ? !text(node) : (this.#flags[index] & BLANK) !== 0;
|
|
269
|
+
}
|
|
270
|
+
|
|
271
|
+
/** `text(node).split(",").length - 1`. */
|
|
272
|
+
commas(node: ReadabilityElement): number {
|
|
273
|
+
const index = this.#at(node);
|
|
274
|
+
return index < 0 ? text(node).split(",").length - 1 : this.#commas[index];
|
|
275
|
+
}
|
|
276
|
+
|
|
277
|
+
/** `text(node).split(SEPARATORS).length - 1`. */
|
|
278
|
+
separators(node: ReadabilityElement): number {
|
|
279
|
+
const index = this.#at(node);
|
|
280
|
+
return index < 0 ? text(node).split(SEPARATORS).length - 1 : this.#separators[index];
|
|
281
|
+
}
|
|
282
|
+
|
|
283
|
+
/** `node.getElementsByTagName("p").length`. */
|
|
284
|
+
paragraphs(node: ReadabilityElement): number {
|
|
285
|
+
const index = this.#at(node);
|
|
286
|
+
return index < 0 ? node.getElementsByTagName("p").length : this.#paragraphs[index];
|
|
287
|
+
}
|
|
288
|
+
|
|
289
|
+
/** `node.getElementsByTagName("img").length`. */
|
|
290
|
+
images(node: ReadabilityElement): number {
|
|
291
|
+
const index = this.#at(node);
|
|
292
|
+
return index < 0 ? node.getElementsByTagName("img").length : this.#images[index];
|
|
293
|
+
}
|
|
294
|
+
|
|
295
|
+
/** `node.getElementsByTagName("input").length`. */
|
|
296
|
+
inputs(node: ReadabilityElement): number {
|
|
297
|
+
const index = this.#at(node);
|
|
298
|
+
return index < 0 ? node.getElementsByTagName("input").length : this.#inputs[index];
|
|
299
|
+
}
|
|
300
|
+
|
|
301
|
+
/** `linkDensity(node)`, summed in the same order so the result is bit-identical. */
|
|
302
|
+
linkDensity(node: ReadabilityElement): number {
|
|
303
|
+
const index = this.#at(node);
|
|
304
|
+
if (index < 0) return linkDensity(node);
|
|
305
|
+
const total = this.#length[index];
|
|
306
|
+
if (!total) return 0;
|
|
307
|
+
let linked = 0;
|
|
308
|
+
for (let link = this.#linkStart[index]; link < this.#linkEnd[index]; link++) linked += this.#links[link];
|
|
309
|
+
return linked / total;
|
|
310
|
+
}
|
|
311
|
+
|
|
312
|
+
#at(node: ReadabilityElement): number {
|
|
313
|
+
const index = this.#index.get(node);
|
|
314
|
+
if (index !== undefined) return index;
|
|
315
|
+
if (node.nodeType !== ELEMENT_NODE) return -1;
|
|
316
|
+
let root: ReadabilityNode = node;
|
|
317
|
+
while (root.parentNode?.nodeType === ELEMENT_NODE) root = root.parentNode;
|
|
318
|
+
this.#measure(root as ReadabilityElement);
|
|
319
|
+
return this.#index.get(node) ?? -1;
|
|
320
|
+
}
|
|
321
|
+
|
|
322
|
+
#measure(root: ReadabilityElement): void {
|
|
323
|
+
const first = this.#length.length;
|
|
324
|
+
const order: ReadabilityElement[] = [];
|
|
325
|
+
const anchors: ReadabilityElement[] = [];
|
|
326
|
+
const pending = [root];
|
|
327
|
+
while (pending.length) {
|
|
328
|
+
const node = pending.pop();
|
|
329
|
+
if (!node) continue;
|
|
330
|
+
this.#index.set(node, first + order.length);
|
|
331
|
+
order.push(node);
|
|
332
|
+
// An anchor counts for its ancestors only, so its own range starts after it.
|
|
333
|
+
if (isTag(node, "A", "a")) anchors.push(node);
|
|
334
|
+
this.#linkStart.push(this.#links.length + anchors.length);
|
|
335
|
+
this.#linkEnd.push(0);
|
|
336
|
+
this.#length.push(0);
|
|
337
|
+
this.#flags.push(0);
|
|
338
|
+
this.#commas.push(0);
|
|
339
|
+
this.#separators.push(0);
|
|
340
|
+
this.#paragraphs.push(0);
|
|
341
|
+
this.#images.push(0);
|
|
342
|
+
this.#inputs.push(0);
|
|
343
|
+
const children = node.childNodes;
|
|
344
|
+
for (let child = children.length - 1; child >= 0; child--) {
|
|
345
|
+
if (children[child].nodeType === ELEMENT_NODE) pending.push(children[child] as ReadabilityElement);
|
|
346
|
+
}
|
|
347
|
+
}
|
|
348
|
+
const run = new TextRun();
|
|
349
|
+
// Reverse document order reaches every element after all of its descendants.
|
|
350
|
+
for (let offset = order.length - 1; offset >= 0; offset--) {
|
|
351
|
+
const node = order[offset];
|
|
352
|
+
const index = first + offset;
|
|
353
|
+
const children = node.childNodes;
|
|
354
|
+
// A template's parsed children are not part of its text in every DOM; ask it directly.
|
|
355
|
+
const leaf = !children.length || isTag(node, "TEMPLATE", "template");
|
|
356
|
+
run.reset();
|
|
357
|
+
if (leaf) run.appendText(node.textContent ?? "");
|
|
358
|
+
let paragraphs = 0;
|
|
359
|
+
let images = 0;
|
|
360
|
+
let inputs = 0;
|
|
361
|
+
let linkEnd = this.#linkStart[index];
|
|
362
|
+
for (let child = 0; child < children.length; child++) {
|
|
363
|
+
const childNode = children[child];
|
|
364
|
+
if (childNode.nodeType !== ELEMENT_NODE) {
|
|
365
|
+
if (!leaf && this.#countsAsText(childNode)) run.appendText(childNode.textContent ?? "");
|
|
366
|
+
continue;
|
|
367
|
+
}
|
|
368
|
+
const element = childNode as ReadabilityElement;
|
|
369
|
+
const at = this.#index.get(element) ?? -1;
|
|
370
|
+
if (!leaf) run.append(this.#length[at], this.#flags[at], this.#commas[at], this.#separators[at]);
|
|
371
|
+
paragraphs += this.#paragraphs[at] + (isTag(element, "P", "p") ? 1 : 0);
|
|
372
|
+
images += this.#images[at] + (isTag(element, "IMG", "img") ? 1 : 0);
|
|
373
|
+
inputs += this.#inputs[at] + (isTag(element, "INPUT", "input") ? 1 : 0);
|
|
374
|
+
linkEnd = this.#linkEnd[at];
|
|
375
|
+
}
|
|
376
|
+
this.#length[index] = run.length;
|
|
377
|
+
this.#flags[index] = run.flags;
|
|
378
|
+
this.#commas[index] = run.commas;
|
|
379
|
+
this.#separators[index] = run.separators;
|
|
380
|
+
this.#paragraphs[index] = paragraphs;
|
|
381
|
+
this.#images[index] = images;
|
|
382
|
+
this.#inputs[index] = inputs;
|
|
383
|
+
this.#linkEnd[index] = linkEnd;
|
|
384
|
+
}
|
|
385
|
+
for (const anchor of anchors) {
|
|
386
|
+
const length = this.#length[this.#index.get(anchor) ?? -1];
|
|
387
|
+
this.#links.push(length * ((anchor.getAttribute("href") ?? "").startsWith("#") ? 0.3 : 1));
|
|
388
|
+
}
|
|
389
|
+
}
|
|
390
|
+
}
|
|
391
|
+
|
|
122
392
|
function visible(node: ReadabilityElement): boolean {
|
|
123
393
|
const style = node.getAttribute("style")?.toLowerCase() ?? "";
|
|
124
394
|
return (
|
|
@@ -135,6 +405,24 @@ function removeAll(root: ReadabilityNode, tags: readonly string[]): void {
|
|
|
135
405
|
}
|
|
136
406
|
}
|
|
137
407
|
|
|
408
|
+
/** Replace `parent`'s children, detaching the old ones in one step where the DOM supports it. */
|
|
409
|
+
function replaceChildren(parent: ReadabilityNode, children: readonly ReadabilityNode[]): void {
|
|
410
|
+
if (parent.replaceChildren) parent.replaceChildren();
|
|
411
|
+
else while (parent.firstChild) parent.firstChild.remove();
|
|
412
|
+
for (const child of children) parent.appendChild(child);
|
|
413
|
+
}
|
|
414
|
+
|
|
415
|
+
/** Move all of `from`'s children to the end of `to` without detaching them one by one. */
|
|
416
|
+
function moveChildren(from: ReadabilityNode, to: ReadabilityNode): void {
|
|
417
|
+
if (!from.replaceChildren) {
|
|
418
|
+
while (from.firstChild) to.appendChild(from.firstChild);
|
|
419
|
+
return;
|
|
420
|
+
}
|
|
421
|
+
const children = Array.from(from.childNodes);
|
|
422
|
+
from.replaceChildren();
|
|
423
|
+
for (const child of children) to.appendChild(child);
|
|
424
|
+
}
|
|
425
|
+
|
|
138
426
|
function entityDecode(value: string | undefined | null): string | undefined | null {
|
|
139
427
|
if (!value) return value;
|
|
140
428
|
const named: Record<string, string> = { quot: '"', amp: "&", apos: "'", lt: "<", gt: ">" };
|
|
@@ -283,6 +571,8 @@ export class Readability<T = string> {
|
|
|
283
571
|
readonly #scores = new Map<ReadabilityElement, number>();
|
|
284
572
|
#byline: string | undefined;
|
|
285
573
|
#lang: string | null = null;
|
|
574
|
+
/** Per non-text node type, whether this DOM counts its `textContent` in a parent's. */
|
|
575
|
+
readonly #textByNodeType = new Map<number, boolean>();
|
|
286
576
|
|
|
287
577
|
constructor(document: ReadabilityDocument, options: ReadabilityOptions<T> = {}) {
|
|
288
578
|
this.#document = document;
|
|
@@ -303,10 +593,11 @@ export class Readability<T = string> {
|
|
|
303
593
|
removeAll(this.#document, ["script", "style"]);
|
|
304
594
|
const body = this.#document.body;
|
|
305
595
|
if (!body) return null;
|
|
306
|
-
|
|
596
|
+
// Retries start from the body as it was before the first attempt rearranged it.
|
|
597
|
+
const pristine = Array.from(body.childNodes, node => node.cloneNode(true));
|
|
307
598
|
const attempts: Attempt[] = [];
|
|
308
599
|
for (const mode of [0, 1, 2, 3]) {
|
|
309
|
-
if (mode) body.
|
|
600
|
+
if (mode) replaceChildren(body, mode === 3 ? pristine : pristine.map(node => node.cloneNode(true)));
|
|
310
601
|
this.#scores.clear();
|
|
311
602
|
this.#byline = undefined;
|
|
312
603
|
const attempt = this.#extract(body, documentElement, metadata.title ?? "", mode);
|
|
@@ -356,10 +647,13 @@ export class Readability<T = string> {
|
|
|
356
647
|
node.remove();
|
|
357
648
|
continue;
|
|
358
649
|
}
|
|
359
|
-
if (!this.#byline
|
|
360
|
-
|
|
361
|
-
|
|
362
|
-
|
|
650
|
+
if (!this.#byline) {
|
|
651
|
+
const byline = this.#bylineText(node, label);
|
|
652
|
+
if (byline) {
|
|
653
|
+
this.#byline = byline;
|
|
654
|
+
node.remove();
|
|
655
|
+
continue;
|
|
656
|
+
}
|
|
363
657
|
}
|
|
364
658
|
if (!titleRemoved && /^(?:H1|H2)$/.test(node.tagName) && this.#similar(articleTitle, text(node)) > 0.75) {
|
|
365
659
|
titleRemoved = true;
|
|
@@ -375,12 +669,14 @@ export class Readability<T = string> {
|
|
|
375
669
|
}
|
|
376
670
|
if (SCORE_TAGS.has(node.tagName)) scored.push(node);
|
|
377
671
|
}
|
|
378
|
-
|
|
672
|
+
// Scoring and sibling selection below only read the tree, so one measurement serves them all.
|
|
673
|
+
const stats = new SubtreeStats(this.#countsAsText);
|
|
674
|
+
for (const paragraph of scored) this.#scoreParagraph(paragraph, weightClasses, stats);
|
|
379
675
|
let top: ReadabilityElement | undefined;
|
|
380
676
|
let topScore = Number.NEGATIVE_INFINITY;
|
|
381
677
|
for (const [candidate, raw] of this.#scores) {
|
|
382
678
|
if (candidate.tagName === "BODY" || candidate.tagName === "HTML") continue;
|
|
383
|
-
const score = raw * (1 - linkDensity(candidate));
|
|
679
|
+
const score = raw * (1 - stats.linkDensity(candidate));
|
|
384
680
|
this.#scores.set(candidate, score);
|
|
385
681
|
if (score > topScore) {
|
|
386
682
|
top = candidate;
|
|
@@ -399,18 +695,19 @@ export class Readability<T = string> {
|
|
|
399
695
|
const siblings = parent ? elements(parent.children) : [top];
|
|
400
696
|
const threshold = Math.max(10, (this.#scores.get(top) ?? topScore) * 0.2);
|
|
401
697
|
for (const sibling of siblings) {
|
|
402
|
-
|
|
698
|
+
// Moving earlier siblings into the article leaves this sibling's subtree, and so its stats, intact.
|
|
699
|
+
const siblingLength = stats.length(sibling);
|
|
403
700
|
const sameClassBonus =
|
|
404
701
|
sibling.className && sibling.className === top.className ? (this.#scores.get(top) ?? 0) * 0.2 : 0;
|
|
405
702
|
const include =
|
|
406
703
|
sibling === top ||
|
|
407
704
|
(this.#scores.get(sibling) ?? 0) + sameClassBonus >= threshold ||
|
|
408
705
|
(sibling.tagName === "P" &&
|
|
409
|
-
((
|
|
410
|
-
(
|
|
411
|
-
|
|
412
|
-
linkDensity(sibling) === 0 &&
|
|
413
|
-
/\.(?: |$)/.test(
|
|
706
|
+
((siblingLength > 80 && stats.linkDensity(sibling) < 0.25) ||
|
|
707
|
+
(siblingLength > 0 &&
|
|
708
|
+
siblingLength < 80 &&
|
|
709
|
+
stats.linkDensity(sibling) === 0 &&
|
|
710
|
+
/\.(?: |$)/.test(text(sibling)))));
|
|
414
711
|
if (!include) continue;
|
|
415
712
|
if (["DIV", "ARTICLE", "SECTION", "P", "OL", "UL"].includes(sibling.tagName)) {
|
|
416
713
|
article.appendChild(sibling);
|
|
@@ -419,14 +716,14 @@ export class Readability<T = string> {
|
|
|
419
716
|
const replacement = this.#document.createElement("DIV");
|
|
420
717
|
for (const attribute of Array.from(sibling.attributes))
|
|
421
718
|
replacement.setAttribute(attribute.name, attribute.value);
|
|
422
|
-
|
|
719
|
+
moveChildren(sibling, replacement);
|
|
423
720
|
article.appendChild(replacement);
|
|
424
721
|
}
|
|
425
722
|
this.#clean(article, mode < 3);
|
|
426
723
|
const page = this.#document.createElement("DIV");
|
|
427
724
|
page.id = "readability-page-1";
|
|
428
725
|
page.className = "page";
|
|
429
|
-
|
|
726
|
+
moveChildren(article, page);
|
|
430
727
|
article.appendChild(page);
|
|
431
728
|
const content = text(article);
|
|
432
729
|
let dir: string | null | undefined;
|
|
@@ -441,13 +738,10 @@ export class Readability<T = string> {
|
|
|
441
738
|
return { element: article, length: content.length, dir };
|
|
442
739
|
}
|
|
443
740
|
|
|
444
|
-
#scoreParagraph(node: ReadabilityElement, weightClasses: boolean): void {
|
|
445
|
-
const
|
|
446
|
-
if (
|
|
447
|
-
const score =
|
|
448
|
-
1 +
|
|
449
|
-
content.split(/[\u002c\u060c\ufe50\ufe10\ufe11\u2e41\u2e34\u2e32\uff0c]/).length +
|
|
450
|
-
Math.min(Math.floor(content.length / 100), 3);
|
|
741
|
+
#scoreParagraph(node: ReadabilityElement, weightClasses: boolean, stats: SubtreeStats): void {
|
|
742
|
+
const length = stats.length(node);
|
|
743
|
+
if (length < 25) return;
|
|
744
|
+
const score = 1 + (stats.separators(node) + 1) + Math.min(Math.floor(length / 100), 3);
|
|
451
745
|
let ancestor = node.parentNode;
|
|
452
746
|
for (let level = 0; ancestor && level < 5; level++, ancestor = ancestor.parentNode) {
|
|
453
747
|
const element = ancestor as ReadabilityElement;
|
|
@@ -465,19 +759,19 @@ export class Readability<T = string> {
|
|
|
465
759
|
if (classWeight(heading) < 0 || linkDensity(heading) > 0.33) heading.remove();
|
|
466
760
|
}
|
|
467
761
|
if (conditional) {
|
|
762
|
+
// Nodes come in document order, so removing one only changes ancestors that were already judged.
|
|
763
|
+
const stats = new SubtreeStats(this.#countsAsText);
|
|
468
764
|
for (const node of elements(root.querySelectorAll("table, ul, div"))) {
|
|
469
765
|
if (node === root) continue;
|
|
470
|
-
const
|
|
471
|
-
const
|
|
472
|
-
const images = node.getElementsByTagName("img").length;
|
|
473
|
-
const inputs = node.getElementsByTagName("input").length;
|
|
766
|
+
const paragraphs = stats.paragraphs(node);
|
|
767
|
+
const images = stats.images(node);
|
|
474
768
|
if (
|
|
475
769
|
classWeight(node) < 0 ||
|
|
476
|
-
(
|
|
477
|
-
(
|
|
770
|
+
(stats.blank(node) && !images) ||
|
|
771
|
+
(stats.commas(node) + 1 < 10 &&
|
|
478
772
|
((images > paragraphs && paragraphs > 0) ||
|
|
479
|
-
inputs > Math.floor(paragraphs / 3) ||
|
|
480
|
-
linkDensity(node) > 0.5))
|
|
773
|
+
stats.inputs(node) > Math.floor(paragraphs / 3) ||
|
|
774
|
+
stats.linkDensity(node) > 0.5))
|
|
481
775
|
)
|
|
482
776
|
node.remove();
|
|
483
777
|
}
|
|
@@ -512,17 +806,33 @@ export class Readability<T = string> {
|
|
|
512
806
|
}
|
|
513
807
|
}
|
|
514
808
|
|
|
515
|
-
|
|
809
|
+
/** Byline text when `node` is marked as one; the attribute checks run first so most nodes never build text. */
|
|
810
|
+
#bylineText(node: ReadabilityElement, label: string): string | undefined {
|
|
811
|
+
if (
|
|
812
|
+
node.getAttribute("rel") !== "author" &&
|
|
813
|
+
!(node.getAttribute("itemprop") ?? "").includes("author") &&
|
|
814
|
+
!BYLINE.test(label)
|
|
815
|
+
)
|
|
816
|
+
return undefined;
|
|
516
817
|
const value = text(node);
|
|
517
|
-
return
|
|
518
|
-
value.length > 0 &&
|
|
519
|
-
value.length < 100 &&
|
|
520
|
-
(node.getAttribute("rel") === "author" ||
|
|
521
|
-
(node.getAttribute("itemprop") ?? "").includes("author") ||
|
|
522
|
-
BYLINE.test(label))
|
|
523
|
-
);
|
|
818
|
+
return value.length > 0 && value.length < 100 ? value : undefined;
|
|
524
819
|
}
|
|
525
820
|
|
|
821
|
+
/** Whether a non-element child's `textContent` is part of its parent's (text is; comments are in some DOMs). */
|
|
822
|
+
readonly #countsAsText = (node: ReadabilityNode): boolean => {
|
|
823
|
+
if (node.nodeType === TEXT_NODE || node.nodeType === CDATA_SECTION_NODE) return true;
|
|
824
|
+
const value = node.textContent;
|
|
825
|
+
if (!value) return false;
|
|
826
|
+
let counts = this.#textByNodeType.get(node.nodeType);
|
|
827
|
+
if (counts === undefined) {
|
|
828
|
+
const probe = this.#document.createElement("DIV");
|
|
829
|
+
probe.appendChild(node.cloneNode(false));
|
|
830
|
+
counts = probe.textContent === value;
|
|
831
|
+
this.#textByNodeType.set(node.nodeType, counts);
|
|
832
|
+
}
|
|
833
|
+
return counts;
|
|
834
|
+
};
|
|
835
|
+
|
|
526
836
|
#similar(left: string, right: string): number {
|
|
527
837
|
const leftTokens = left.toLowerCase().split(/\W+/).filter(Boolean);
|
|
528
838
|
const rightTokens = right.toLowerCase().split(/\W+/).filter(Boolean);
|
package/src/readability/types.ts
CHANGED
|
@@ -14,6 +14,8 @@ export interface ReadabilityNode {
|
|
|
14
14
|
appendChild(node: ReadabilityNode): ReadabilityNode;
|
|
15
15
|
cloneNode(deep?: boolean): ReadabilityNode;
|
|
16
16
|
remove(): void;
|
|
17
|
+
/** Detach all children; used when present to move or reset children in one step. */
|
|
18
|
+
replaceChildren?(): void;
|
|
17
19
|
}
|
|
18
20
|
|
|
19
21
|
/** Minimal standards-shaped DOM element accepted by the readability extractor. */
|
package/src/stream.ts
CHANGED
|
@@ -6,6 +6,10 @@ import { parseStreamingJson } from "./json-parse";
|
|
|
6
6
|
const LF = 0x0a;
|
|
7
7
|
const CR = 0x0d;
|
|
8
8
|
|
|
9
|
+
// Capacity above this is released once the buffered bytes no longer need it,
|
|
10
|
+
// so one oversized frame does not pin its peak buffer for the stream's life.
|
|
11
|
+
const CONCAT_SINK_RETAIN_BYTES = 1024 * 1024;
|
|
12
|
+
|
|
9
13
|
/**
|
|
10
14
|
* Split a byte stream on LF boundaries.
|
|
11
15
|
*
|
|
@@ -89,6 +93,21 @@ export class ConcatSink {
|
|
|
89
93
|
return next;
|
|
90
94
|
}
|
|
91
95
|
|
|
96
|
+
/** Drop or right-size an oversized buffer after the buffered length shrank. */
|
|
97
|
+
#releaseCapacity(): void {
|
|
98
|
+
const space = this.#space;
|
|
99
|
+
if (!space || space.length <= CONCAT_SINK_RETAIN_BYTES) return;
|
|
100
|
+
const length = this.#length;
|
|
101
|
+
if (length === 0) {
|
|
102
|
+
this.#space = undefined;
|
|
103
|
+
return;
|
|
104
|
+
}
|
|
105
|
+
if (length * 4 > space.length) return;
|
|
106
|
+
const next = Buffer.allocUnsafe(length);
|
|
107
|
+
space.copy(next, 0, 0, length);
|
|
108
|
+
this.#space = next;
|
|
109
|
+
}
|
|
110
|
+
|
|
92
111
|
append(chunk: Uint8Array) {
|
|
93
112
|
const n = chunk.length;
|
|
94
113
|
if (!n) return;
|
|
@@ -99,11 +118,10 @@ export class ConcatSink {
|
|
|
99
118
|
}
|
|
100
119
|
|
|
101
120
|
reset(chunk: Uint8Array) {
|
|
121
|
+
this.#length = 0;
|
|
122
|
+
this.#releaseCapacity();
|
|
102
123
|
const n = chunk.length;
|
|
103
|
-
if (!n)
|
|
104
|
-
this.#length = 0;
|
|
105
|
-
return;
|
|
106
|
-
}
|
|
124
|
+
if (!n) return;
|
|
107
125
|
const space = this.#ensureCapacity(n);
|
|
108
126
|
space.set(chunk, 0);
|
|
109
127
|
this.#length = n;
|
|
@@ -127,14 +145,16 @@ export class ConcatSink {
|
|
|
127
145
|
if (count <= 0) return;
|
|
128
146
|
if (count >= this.#length) {
|
|
129
147
|
this.#length = 0;
|
|
130
|
-
|
|
148
|
+
} else {
|
|
149
|
+
this.#space!.copyWithin(0, count, this.#length);
|
|
150
|
+
this.#length -= count;
|
|
131
151
|
}
|
|
132
|
-
this.#
|
|
133
|
-
this.#length -= count;
|
|
152
|
+
this.#releaseCapacity();
|
|
134
153
|
}
|
|
135
154
|
|
|
136
155
|
clear() {
|
|
137
156
|
this.#length = 0;
|
|
157
|
+
this.#releaseCapacity();
|
|
138
158
|
}
|
|
139
159
|
|
|
140
160
|
/**
|
|
@@ -229,13 +249,14 @@ export class ConcatSink {
|
|
|
229
249
|
if (error) throw error;
|
|
230
250
|
if (done) {
|
|
231
251
|
this.#length = 0;
|
|
232
|
-
|
|
233
|
-
|
|
234
|
-
|
|
235
|
-
|
|
236
|
-
|
|
252
|
+
} else {
|
|
253
|
+
const rem = total - read;
|
|
254
|
+
if (rem < total) {
|
|
255
|
+
space.copyWithin(0, read, total);
|
|
256
|
+
}
|
|
257
|
+
this.#length = rem;
|
|
237
258
|
}
|
|
238
|
-
this.#
|
|
259
|
+
this.#releaseCapacity();
|
|
239
260
|
}
|
|
240
261
|
}
|
|
241
262
|
|
package/src/temp.ts
CHANGED
|
@@ -84,7 +84,6 @@ const kRemoveRetries = 40;
|
|
|
84
84
|
// was too short for some test cleanup scenarios.
|
|
85
85
|
const kRemoveRetryDelayMs = 50;
|
|
86
86
|
const kRetryableRemoveErrorCodes = new Set(["EBUSY", "EPERM", "ENOTEMPTY"]);
|
|
87
|
-
const kSleepBuffer = new Int32Array(new SharedArrayBuffer(4));
|
|
88
87
|
|
|
89
88
|
/** Removes a path recursively, retrying transient Windows deletion failures. */
|
|
90
89
|
export async function removeWithRetries(target: string): Promise<void> {
|
|
@@ -112,7 +111,7 @@ export function removeSyncWithRetries(target: string): void {
|
|
|
112
111
|
} catch (err) {
|
|
113
112
|
if (!shouldRetryRemove(err, attempt)) throw err;
|
|
114
113
|
if (attempt === 0) Bun.gc(true);
|
|
115
|
-
sleepSync(kRemoveRetryDelayMs);
|
|
114
|
+
Bun.sleepSync(kRemoveRetryDelayMs);
|
|
116
115
|
}
|
|
117
116
|
}
|
|
118
117
|
}
|
|
@@ -130,11 +129,3 @@ function isRetryableRemoveError(err: unknown): boolean {
|
|
|
130
129
|
kRetryableRemoveErrorCodes.has(err.code)
|
|
131
130
|
);
|
|
132
131
|
}
|
|
133
|
-
|
|
134
|
-
function sleepSync(ms: number): void {
|
|
135
|
-
if ("sleepSync" in Bun && typeof Bun.sleepSync === "function") {
|
|
136
|
-
Bun.sleepSync(ms);
|
|
137
|
-
return;
|
|
138
|
-
}
|
|
139
|
-
Atomics.wait(kSleepBuffer, 0, 0, ms);
|
|
140
|
-
}
|
package/src/template.ts
CHANGED
|
@@ -226,7 +226,9 @@ function findTagEnd(source: string, start: number, triple: boolean): number {
|
|
|
226
226
|
|
|
227
227
|
function parseTemplate(source: string): Node[] {
|
|
228
228
|
const root: Node[] = [];
|
|
229
|
-
|
|
229
|
+
// `chained` marks a block opened by `{{else name …}}`: it has no closing tag of
|
|
230
|
+
// its own and closes with the block it chains from, as in Handlebars.
|
|
231
|
+
const stack: { node: BlockNode; target: Node[]; inverted: boolean; chained: boolean }[] = [];
|
|
230
232
|
let target = root;
|
|
231
233
|
let cursor = 0;
|
|
232
234
|
while (cursor < source.length) {
|
|
@@ -263,7 +265,7 @@ function parseTemplate(source: string): Node[] {
|
|
|
263
265
|
};
|
|
264
266
|
target.push(nested);
|
|
265
267
|
target = nested.body;
|
|
266
|
-
stack.push({ node: nested, target: parentTarget, inverted: false });
|
|
268
|
+
stack.push({ node: nested, target: parentTarget, inverted: false, chained: true });
|
|
267
269
|
}
|
|
268
270
|
continue;
|
|
269
271
|
}
|
|
@@ -277,12 +279,13 @@ function parseTemplate(source: string): Node[] {
|
|
|
277
279
|
inverted,
|
|
278
280
|
};
|
|
279
281
|
target.push(node);
|
|
280
|
-
stack.push({ node, target, inverted });
|
|
282
|
+
stack.push({ node, target, inverted, chained: false });
|
|
281
283
|
target = node.body;
|
|
282
284
|
continue;
|
|
283
285
|
}
|
|
284
286
|
if (raw.startsWith("/")) {
|
|
285
|
-
|
|
287
|
+
let current = stack.pop();
|
|
288
|
+
while (current?.chained) current = stack.pop();
|
|
286
289
|
if (!current || current.node.expression.name !== raw.slice(1).trim())
|
|
287
290
|
throw new Error(`Parse error: mismatched ${raw}`);
|
|
288
291
|
if (current.inverted) [current.node.body, current.node.inverse] = [current.node.inverse, current.node.body];
|