@steipete/summarize-core 0.7.0
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/README.md +7 -0
- package/dist/esm/content/index.js +5 -0
- package/dist/esm/content/index.js.map +1 -0
- package/dist/esm/content/link-preview/client.js +28 -0
- package/dist/esm/content/link-preview/client.js.map +1 -0
- package/dist/esm/content/link-preview/content/article.js +155 -0
- package/dist/esm/content/link-preview/content/article.js.map +1 -0
- package/dist/esm/content/link-preview/content/cleaner.js +55 -0
- package/dist/esm/content/link-preview/content/cleaner.js.map +1 -0
- package/dist/esm/content/link-preview/content/constants.js +7 -0
- package/dist/esm/content/link-preview/content/constants.js.map +1 -0
- package/dist/esm/content/link-preview/content/fetcher.js +124 -0
- package/dist/esm/content/link-preview/content/fetcher.js.map +1 -0
- package/dist/esm/content/link-preview/content/firecrawl.js +86 -0
- package/dist/esm/content/link-preview/content/firecrawl.js.map +1 -0
- package/dist/esm/content/link-preview/content/html.js +162 -0
- package/dist/esm/content/link-preview/content/html.js.map +1 -0
- package/dist/esm/content/link-preview/content/index.js +345 -0
- package/dist/esm/content/link-preview/content/index.js.map +1 -0
- package/dist/esm/content/link-preview/content/jsonld.js +77 -0
- package/dist/esm/content/link-preview/content/jsonld.js.map +1 -0
- package/dist/esm/content/link-preview/content/parsers.js +77 -0
- package/dist/esm/content/link-preview/content/parsers.js.map +1 -0
- package/dist/esm/content/link-preview/content/podcast-utils.js +79 -0
- package/dist/esm/content/link-preview/content/podcast-utils.js.map +1 -0
- package/dist/esm/content/link-preview/content/readability.js +53 -0
- package/dist/esm/content/link-preview/content/readability.js.map +1 -0
- package/dist/esm/content/link-preview/content/twitter-utils.js +68 -0
- package/dist/esm/content/link-preview/content/twitter-utils.js.map +1 -0
- package/dist/esm/content/link-preview/content/types.js +4 -0
- package/dist/esm/content/link-preview/content/types.js.map +1 -0
- package/dist/esm/content/link-preview/content/utils.js +164 -0
- package/dist/esm/content/link-preview/content/utils.js.map +1 -0
- package/dist/esm/content/link-preview/content/video.js +96 -0
- package/dist/esm/content/link-preview/content/video.js.map +1 -0
- package/dist/esm/content/link-preview/content/youtube.js +82 -0
- package/dist/esm/content/link-preview/content/youtube.js.map +1 -0
- package/dist/esm/content/link-preview/deps.js +20 -0
- package/dist/esm/content/link-preview/deps.js.map +1 -0
- package/dist/esm/content/link-preview/fetch-with-timeout.js +35 -0
- package/dist/esm/content/link-preview/fetch-with-timeout.js.map +1 -0
- package/dist/esm/content/link-preview/types.js +2 -0
- package/dist/esm/content/link-preview/types.js.map +1 -0
- package/dist/esm/content/transcript/cache.js +79 -0
- package/dist/esm/content/transcript/cache.js.map +1 -0
- package/dist/esm/content/transcript/index.js +130 -0
- package/dist/esm/content/transcript/index.js.map +1 -0
- package/dist/esm/content/transcript/normalize.js +43 -0
- package/dist/esm/content/transcript/normalize.js.map +1 -0
- package/dist/esm/content/transcript/providers/generic.js +11 -0
- package/dist/esm/content/transcript/providers/generic.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/apple-flow.js +222 -0
- package/dist/esm/content/transcript/providers/podcast/apple-flow.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/apple.js +38 -0
- package/dist/esm/content/transcript/providers/podcast/apple.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/constants.js +8 -0
- package/dist/esm/content/transcript/providers/podcast/constants.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/flow-context.js +2 -0
- package/dist/esm/content/transcript/providers/podcast/flow-context.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/itunes.js +134 -0
- package/dist/esm/content/transcript/providers/podcast/itunes.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/json.js +34 -0
- package/dist/esm/content/transcript/providers/podcast/json.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/media.js +345 -0
- package/dist/esm/content/transcript/providers/podcast/media.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/results.js +28 -0
- package/dist/esm/content/transcript/providers/podcast/results.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/rss.js +253 -0
- package/dist/esm/content/transcript/providers/podcast/rss.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/spotify-flow.js +218 -0
- package/dist/esm/content/transcript/providers/podcast/spotify-flow.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast/spotify.js +113 -0
- package/dist/esm/content/transcript/providers/podcast/spotify.js.map +1 -0
- package/dist/esm/content/transcript/providers/podcast.js +222 -0
- package/dist/esm/content/transcript/providers/podcast.js.map +1 -0
- package/dist/esm/content/transcript/providers/youtube/api.js +257 -0
- package/dist/esm/content/transcript/providers/youtube/api.js.map +1 -0
- package/dist/esm/content/transcript/providers/youtube/apify.js +55 -0
- package/dist/esm/content/transcript/providers/youtube/apify.js.map +1 -0
- package/dist/esm/content/transcript/providers/youtube/captions.js +409 -0
- package/dist/esm/content/transcript/providers/youtube/captions.js.map +1 -0
- package/dist/esm/content/transcript/providers/youtube/yt-dlp.js +166 -0
- package/dist/esm/content/transcript/providers/youtube/yt-dlp.js.map +1 -0
- package/dist/esm/content/transcript/providers/youtube.js +173 -0
- package/dist/esm/content/transcript/providers/youtube.js.map +1 -0
- package/dist/esm/content/transcript/types.js +2 -0
- package/dist/esm/content/transcript/types.js.map +1 -0
- package/dist/esm/content/transcript/utils.js +259 -0
- package/dist/esm/content/transcript/utils.js.map +1 -0
- package/dist/esm/index.js +4 -0
- package/dist/esm/index.js.map +1 -0
- package/dist/esm/language.js +126 -0
- package/dist/esm/language.js.map +1 -0
- package/dist/esm/prompts/cli.js +20 -0
- package/dist/esm/prompts/cli.js.map +1 -0
- package/dist/esm/prompts/file.js +48 -0
- package/dist/esm/prompts/file.js.map +1 -0
- package/dist/esm/prompts/index.js +4 -0
- package/dist/esm/prompts/index.js.map +1 -0
- package/dist/esm/prompts/link-summary.js +116 -0
- package/dist/esm/prompts/link-summary.js.map +1 -0
- package/dist/esm/shared/contracts.js +2 -0
- package/dist/esm/shared/contracts.js.map +1 -0
- package/dist/esm/transcription/whisper/constants.js +8 -0
- package/dist/esm/transcription/whisper/constants.js.map +1 -0
- package/dist/esm/transcription/whisper/core.js +303 -0
- package/dist/esm/transcription/whisper/core.js.map +1 -0
- package/dist/esm/transcription/whisper/fal.js +41 -0
- package/dist/esm/transcription/whisper/fal.js.map +1 -0
- package/dist/esm/transcription/whisper/ffmpeg.js +179 -0
- package/dist/esm/transcription/whisper/ffmpeg.js.map +1 -0
- package/dist/esm/transcription/whisper/openai.js +47 -0
- package/dist/esm/transcription/whisper/openai.js.map +1 -0
- package/dist/esm/transcription/whisper/types.js +2 -0
- package/dist/esm/transcription/whisper/types.js.map +1 -0
- package/dist/esm/transcription/whisper/utils.js +63 -0
- package/dist/esm/transcription/whisper/utils.js.map +1 -0
- package/dist/esm/transcription/whisper/whisper-cpp.js +227 -0
- package/dist/esm/transcription/whisper/whisper-cpp.js.map +1 -0
- package/dist/esm/transcription/whisper.js +5 -0
- package/dist/esm/transcription/whisper.js.map +1 -0
- package/dist/types/content/index.d.ts +5 -0
- package/dist/types/content/link-preview/client.d.ts +18 -0
- package/dist/types/content/link-preview/content/article.d.ts +4 -0
- package/dist/types/content/link-preview/content/cleaner.d.ts +12 -0
- package/dist/types/content/link-preview/content/constants.d.ts +6 -0
- package/dist/types/content/link-preview/content/fetcher.d.ts +16 -0
- package/dist/types/content/link-preview/content/firecrawl.d.ts +14 -0
- package/dist/types/content/link-preview/content/html.d.ts +17 -0
- package/dist/types/content/link-preview/content/index.d.ts +4 -0
- package/dist/types/content/link-preview/content/jsonld.d.ts +6 -0
- package/dist/types/content/link-preview/content/parsers.d.ts +7 -0
- package/dist/types/content/link-preview/content/podcast-utils.d.ts +7 -0
- package/dist/types/content/link-preview/content/readability.d.ts +8 -0
- package/dist/types/content/link-preview/content/twitter-utils.d.ts +4 -0
- package/dist/types/content/link-preview/content/types.d.ts +61 -0
- package/dist/types/content/link-preview/content/utils.d.ts +17 -0
- package/dist/types/content/link-preview/content/video.d.ts +5 -0
- package/dist/types/content/link-preview/content/youtube.d.ts +1 -0
- package/dist/types/content/link-preview/deps.d.ts +167 -0
- package/dist/types/content/link-preview/fetch-with-timeout.d.ts +4 -0
- package/dist/types/content/link-preview/types.d.ts +37 -0
- package/dist/types/content/transcript/cache.d.ts +29 -0
- package/dist/types/content/transcript/index.d.ts +9 -0
- package/dist/types/content/transcript/normalize.d.ts +3 -0
- package/dist/types/content/transcript/providers/generic.d.ts +3 -0
- package/dist/types/content/transcript/providers/podcast/apple-flow.d.ts +4 -0
- package/dist/types/content/transcript/providers/podcast/apple.d.ts +6 -0
- package/dist/types/content/transcript/providers/podcast/constants.d.ts +7 -0
- package/dist/types/content/transcript/providers/podcast/flow-context.d.ts +11 -0
- package/dist/types/content/transcript/providers/podcast/itunes.d.ts +17 -0
- package/dist/types/content/transcript/providers/podcast/json.d.ts +8 -0
- package/dist/types/content/transcript/providers/podcast/media.d.ts +42 -0
- package/dist/types/content/transcript/providers/podcast/results.d.ts +10 -0
- package/dist/types/content/transcript/providers/podcast/rss.d.ts +22 -0
- package/dist/types/content/transcript/providers/podcast/spotify-flow.d.ts +3 -0
- package/dist/types/content/transcript/providers/podcast/spotify.d.ts +24 -0
- package/dist/types/content/transcript/providers/podcast.d.ts +20 -0
- package/dist/types/content/transcript/providers/youtube/api.d.ts +26 -0
- package/dist/types/content/transcript/providers/youtube/apify.d.ts +1 -0
- package/dist/types/content/transcript/providers/youtube/captions.d.ts +7 -0
- package/dist/types/content/transcript/providers/youtube/yt-dlp.d.ts +17 -0
- package/dist/types/content/transcript/providers/youtube.d.ts +3 -0
- package/dist/types/content/transcript/types.d.ts +30 -0
- package/dist/types/content/transcript/utils.d.ts +8 -0
- package/dist/types/index.d.ts +4 -0
- package/dist/types/language.d.ts +25 -0
- package/dist/types/prompts/cli.d.ts +10 -0
- package/dist/types/prompts/file.d.ts +17 -0
- package/dist/types/prompts/index.d.ts +4 -0
- package/dist/types/prompts/link-summary.d.ts +29 -0
- package/dist/types/shared/contracts.d.ts +2 -0
- package/dist/types/transcription/whisper/constants.d.ts +7 -0
- package/dist/types/transcription/whisper/core.d.ts +20 -0
- package/dist/types/transcription/whisper/fal.d.ts +1 -0
- package/dist/types/transcription/whisper/ffmpeg.d.ts +16 -0
- package/dist/types/transcription/whisper/openai.d.ts +2 -0
- package/dist/types/transcription/whisper/types.d.ts +17 -0
- package/dist/types/transcription/whisper/utils.d.ts +5 -0
- package/dist/types/transcription/whisper/whisper-cpp.d.ts +9 -0
- package/dist/types/transcription/whisper.d.ts +5 -0
- package/package.json +54 -0
package/README.md
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
1
|
+
export { createLinkPreviewClient, } from './link-preview/client.js';
|
|
2
|
+
export { DEFAULT_CACHE_MODE, DEFAULT_MAX_CONTENT_CHARACTERS, DEFAULT_TIMEOUT_MS, } from './link-preview/content/types.js';
|
|
3
|
+
export { ProgressKind } from './link-preview/deps.js';
|
|
4
|
+
export { CACHE_MODES, } from './link-preview/types.js';
|
|
5
|
+
//# sourceMappingURL=index.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../src/content/index.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,uBAAuB,GAGxB,MAAM,0BAA0B,CAAA;AACjC,OAAO,EACL,kBAAkB,EAClB,8BAA8B,EAC9B,kBAAkB,GAGnB,MAAM,iCAAiC,CAAA;AAUxC,OAAO,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAA;AACrD,OAAO,EACL,WAAW,GAIZ,MAAM,yBAAyB,CAAA"}
|
|
@@ -0,0 +1,28 @@
|
|
|
1
|
+
import { fetchLinkContent } from './content/index.js';
|
|
2
|
+
export function createLinkPreviewClient(options = {}) {
|
|
3
|
+
const fetchImpl = options.fetch ?? ((...args) => globalThis.fetch(...args));
|
|
4
|
+
const scrape = options.scrapeWithFirecrawl ?? null;
|
|
5
|
+
const apifyApiToken = typeof options.apifyApiToken === 'string' ? options.apifyApiToken : null;
|
|
6
|
+
const ytDlpPath = typeof options.ytDlpPath === 'string' ? options.ytDlpPath : null;
|
|
7
|
+
const falApiKey = typeof options.falApiKey === 'string' ? options.falApiKey : null;
|
|
8
|
+
const openaiApiKey = typeof options.openaiApiKey === 'string' ? options.openaiApiKey : null;
|
|
9
|
+
const convertHtmlToMarkdown = options.convertHtmlToMarkdown ?? null;
|
|
10
|
+
const transcriptCache = options.transcriptCache ?? null;
|
|
11
|
+
const readTweetWithBird = typeof options.readTweetWithBird === 'function' ? options.readTweetWithBird : null;
|
|
12
|
+
const onProgress = typeof options.onProgress === 'function' ? options.onProgress : null;
|
|
13
|
+
return {
|
|
14
|
+
fetchLinkContent: (url, contentOptions) => fetchLinkContent(url, contentOptions, {
|
|
15
|
+
fetch: fetchImpl,
|
|
16
|
+
scrapeWithFirecrawl: scrape,
|
|
17
|
+
apifyApiToken,
|
|
18
|
+
ytDlpPath,
|
|
19
|
+
falApiKey,
|
|
20
|
+
openaiApiKey,
|
|
21
|
+
convertHtmlToMarkdown,
|
|
22
|
+
transcriptCache,
|
|
23
|
+
readTweetWithBird,
|
|
24
|
+
onProgress,
|
|
25
|
+
}),
|
|
26
|
+
};
|
|
27
|
+
}
|
|
28
|
+
//# sourceMappingURL=client.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"client.js","sourceRoot":"","sources":["../../../../src/content/link-preview/client.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAA;AA2BrD,MAAM,UAAU,uBAAuB,CAAC,UAAoC,EAAE;IAC5E,MAAM,SAAS,GACb,OAAO,CAAC,KAAK,IAAI,CAAC,CAAC,GAAG,IAA8B,EAAE,EAAE,CAAC,UAAU,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC,CAAC,CAAA;IACrF,MAAM,MAAM,GAA+B,OAAO,CAAC,mBAAmB,IAAI,IAAI,CAAA;IAC9E,MAAM,aAAa,GAAG,OAAO,OAAO,CAAC,aAAa,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC,CAAC,IAAI,CAAA;IAC9F,MAAM,SAAS,GAAG,OAAO,OAAO,CAAC,SAAS,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAA;IAClF,MAAM,SAAS,GAAG,OAAO,OAAO,CAAC,SAAS,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAA;IAClF,MAAM,YAAY,GAAG,OAAO,OAAO,CAAC,YAAY,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,YAAY,CAAC,CAAC,CAAC,IAAI,CAAA;IAC3F,MAAM,qBAAqB,GAAiC,OAAO,CAAC,qBAAqB,IAAI,IAAI,CAAA;IACjG,MAAM,eAAe,GAA2B,OAAO,CAAC,eAAe,IAAI,IAAI,CAAA;IAC/E,MAAM,iBAAiB,GACrB,OAAO,OAAO,CAAC,iBAAiB,KAAK,UAAU,CAAC,CAAC,CAAC,OAAO,CAAC,iBAAiB,CAAC,CAAC,CAAC,IAAI,CAAA;IACpF,MAAM,UAAU,GAAG,OAAO,OAAO,CAAC,UAAU,KAAK,UAAU,CAAC,CAAC,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI,CAAA;IAEvF,OAAO;QACL,gBAAgB,EAAE,CAAC,GAAW,EAAE,cAAwC,EAAE,EAAE,CAC1E,gBAAgB,CAAC,GAAG,EAAE,cAAc,EAAE;YACpC,KAAK,EAAE,SAAS;YAChB,mBAAmB,EAAE,MAAM;YAC3B,aAAa;YACb,SAAS;YACT,SAAS;YACT,YAAY;YACZ,qBAAqB;YACrB,eAAe;YACf,iBAAiB;YACjB,UAAU;SACX,CAAC;KACL,CAAA;AACH,CAAC"}
|
|
@@ -0,0 +1,155 @@
|
|
|
1
|
+
import { load } from 'cheerio';
|
|
2
|
+
import sanitizeHtml from 'sanitize-html';
|
|
3
|
+
import { decodeHtmlEntities, normalizeWhitespace } from './cleaner.js';
|
|
4
|
+
const MIN_SEGMENT_LENGTH = 30;
|
|
5
|
+
export function sanitizeHtmlForMarkdownConversion(html) {
|
|
6
|
+
return sanitizeHtml(html, {
|
|
7
|
+
allowedTags: [
|
|
8
|
+
'article',
|
|
9
|
+
'section',
|
|
10
|
+
'div',
|
|
11
|
+
'p',
|
|
12
|
+
'h1',
|
|
13
|
+
'h2',
|
|
14
|
+
'h3',
|
|
15
|
+
'h4',
|
|
16
|
+
'h5',
|
|
17
|
+
'h6',
|
|
18
|
+
'ol',
|
|
19
|
+
'ul',
|
|
20
|
+
'li',
|
|
21
|
+
'blockquote',
|
|
22
|
+
'pre',
|
|
23
|
+
'code',
|
|
24
|
+
'span',
|
|
25
|
+
'strong',
|
|
26
|
+
'em',
|
|
27
|
+
'br',
|
|
28
|
+
'a',
|
|
29
|
+
],
|
|
30
|
+
allowedAttributes: {
|
|
31
|
+
a: ['href'],
|
|
32
|
+
},
|
|
33
|
+
nonTextTags: [
|
|
34
|
+
'style',
|
|
35
|
+
'script',
|
|
36
|
+
'noscript',
|
|
37
|
+
'template',
|
|
38
|
+
'svg',
|
|
39
|
+
'canvas',
|
|
40
|
+
'iframe',
|
|
41
|
+
'object',
|
|
42
|
+
'embed',
|
|
43
|
+
],
|
|
44
|
+
textFilter(text) {
|
|
45
|
+
return decodeHtmlEntities(text);
|
|
46
|
+
},
|
|
47
|
+
});
|
|
48
|
+
}
|
|
49
|
+
export function extractArticleContent(html) {
|
|
50
|
+
const segments = collectSegmentsFromHtml(html);
|
|
51
|
+
if (segments.length > 0) {
|
|
52
|
+
return segments.join('\n');
|
|
53
|
+
}
|
|
54
|
+
const fallback = normalizeWhitespace(extractPlainText(html));
|
|
55
|
+
return fallback ?? '';
|
|
56
|
+
}
|
|
57
|
+
export function collectSegmentsFromHtml(html) {
|
|
58
|
+
const sanitized = sanitizeHtml(html, {
|
|
59
|
+
allowedTags: [
|
|
60
|
+
'article',
|
|
61
|
+
'section',
|
|
62
|
+
'div',
|
|
63
|
+
'p',
|
|
64
|
+
'h1',
|
|
65
|
+
'h2',
|
|
66
|
+
'h3',
|
|
67
|
+
'h4',
|
|
68
|
+
'h5',
|
|
69
|
+
'h6',
|
|
70
|
+
'ol',
|
|
71
|
+
'ul',
|
|
72
|
+
'li',
|
|
73
|
+
'blockquote',
|
|
74
|
+
'pre',
|
|
75
|
+
'code',
|
|
76
|
+
'span',
|
|
77
|
+
'strong',
|
|
78
|
+
'em',
|
|
79
|
+
'br',
|
|
80
|
+
],
|
|
81
|
+
allowedAttributes: {},
|
|
82
|
+
nonTextTags: [
|
|
83
|
+
'style',
|
|
84
|
+
'script',
|
|
85
|
+
'noscript',
|
|
86
|
+
'template',
|
|
87
|
+
'svg',
|
|
88
|
+
'canvas',
|
|
89
|
+
'iframe',
|
|
90
|
+
'object',
|
|
91
|
+
'embed',
|
|
92
|
+
],
|
|
93
|
+
textFilter(text) {
|
|
94
|
+
return decodeHtmlEntities(text);
|
|
95
|
+
},
|
|
96
|
+
});
|
|
97
|
+
const $ = load(sanitized);
|
|
98
|
+
const segments = [];
|
|
99
|
+
$('h1,h2,h3,h4,h5,h6,li,p,blockquote,pre').each((_, element) => {
|
|
100
|
+
if (!('tagName' in element) || typeof element.tagName !== 'string') {
|
|
101
|
+
return;
|
|
102
|
+
}
|
|
103
|
+
const tag = element.tagName.toLowerCase();
|
|
104
|
+
const raw = $(element).text();
|
|
105
|
+
const text = normalizeWhitespace(raw).replaceAll(/\n+/g, ' ');
|
|
106
|
+
if (!text || text.length === 0) {
|
|
107
|
+
return;
|
|
108
|
+
}
|
|
109
|
+
if (tag.startsWith('h')) {
|
|
110
|
+
if (text.length >= 10) {
|
|
111
|
+
segments.push(text);
|
|
112
|
+
}
|
|
113
|
+
return;
|
|
114
|
+
}
|
|
115
|
+
if (tag === 'li') {
|
|
116
|
+
if (text.length >= 20) {
|
|
117
|
+
segments.push(`• ${text}`);
|
|
118
|
+
}
|
|
119
|
+
return;
|
|
120
|
+
}
|
|
121
|
+
if (text.length < MIN_SEGMENT_LENGTH) {
|
|
122
|
+
return;
|
|
123
|
+
}
|
|
124
|
+
segments.push(text);
|
|
125
|
+
});
|
|
126
|
+
if (segments.length === 0) {
|
|
127
|
+
const fallback = normalizeWhitespace($('body').text() || sanitized);
|
|
128
|
+
return fallback ? [fallback] : [];
|
|
129
|
+
}
|
|
130
|
+
return mergeConsecutiveSegments(segments);
|
|
131
|
+
}
|
|
132
|
+
export function extractPlainText(html) {
|
|
133
|
+
const stripped = sanitizeHtml(html, {
|
|
134
|
+
allowedTags: [],
|
|
135
|
+
allowedAttributes: {},
|
|
136
|
+
nonTextTags: [
|
|
137
|
+
'style',
|
|
138
|
+
'script',
|
|
139
|
+
'noscript',
|
|
140
|
+
'template',
|
|
141
|
+
'svg',
|
|
142
|
+
'canvas',
|
|
143
|
+
'iframe',
|
|
144
|
+
'object',
|
|
145
|
+
'embed',
|
|
146
|
+
],
|
|
147
|
+
});
|
|
148
|
+
return decodeHtmlEntities(stripped);
|
|
149
|
+
}
|
|
150
|
+
function mergeConsecutiveSegments(segments) {
|
|
151
|
+
// Keep headings as separate segments; merging short segments mostly collapses headings into the
|
|
152
|
+
// previous paragraph ("... Conclusion"), which reads worse than a standalone heading line.
|
|
153
|
+
return segments.filter(Boolean);
|
|
154
|
+
}
|
|
155
|
+
//# sourceMappingURL=article.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"article.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/article.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,IAAI,EAAE,MAAM,SAAS,CAAA;AAC9B,OAAO,YAAY,MAAM,eAAe,CAAA;AAExC,OAAO,EAAE,kBAAkB,EAAE,mBAAmB,EAAE,MAAM,cAAc,CAAA;AAEtE,MAAM,kBAAkB,GAAG,EAAE,CAAA;AAE7B,MAAM,UAAU,iCAAiC,CAAC,IAAY;IAC5D,OAAO,YAAY,CAAC,IAAI,EAAE;QACxB,WAAW,EAAE;YACX,SAAS;YACT,SAAS;YACT,KAAK;YACL,GAAG;YACH,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,YAAY;YACZ,KAAK;YACL,MAAM;YACN,MAAM;YACN,QAAQ;YACR,IAAI;YACJ,IAAI;YACJ,GAAG;SACJ;QACD,iBAAiB,EAAE;YACjB,CAAC,EAAE,CAAC,MAAM,CAAC;SACZ;QACD,WAAW,EAAE;YACX,OAAO;YACP,QAAQ;YACR,UAAU;YACV,UAAU;YACV,KAAK;YACL,QAAQ;YACR,QAAQ;YACR,QAAQ;YACR,OAAO;SACR;QACD,UAAU,CAAC,IAAY;YACrB,OAAO,kBAAkB,CAAC,IAAI,CAAC,CAAA;QACjC,CAAC;KACF,CAAC,CAAA;AACJ,CAAC;AAED,MAAM,UAAU,qBAAqB,CAAC,IAAY;IAChD,MAAM,QAAQ,GAAG,uBAAuB,CAAC,IAAI,CAAC,CAAA;IAC9C,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACxB,OAAO,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAC5B,CAAC;IACD,MAAM,QAAQ,GAAG,mBAAmB,CAAC,gBAAgB,CAAC,IAAI,CAAC,CAAC,CAAA;IAC5D,OAAO,QAAQ,IAAI,EAAE,CAAA;AACvB,CAAC;AAED,MAAM,UAAU,uBAAuB,CAAC,IAAY;IAClD,MAAM,SAAS,GAAG,YAAY,CAAC,IAAI,EAAE;QACnC,WAAW,EAAE;YACX,SAAS;YACT,SAAS;YACT,KAAK;YACL,GAAG;YACH,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,IAAI;YACJ,YAAY;YACZ,KAAK;YACL,MAAM;YACN,MAAM;YACN,QAAQ;YACR,IAAI;YACJ,IAAI;SACL;QACD,iBAAiB,EAAE,EAAE;QACrB,WAAW,EAAE;YACX,OAAO;YACP,QAAQ;YACR,UAAU;YACV,UAAU;YACV,KAAK;YACL,QAAQ;YACR,QAAQ;YACR,QAAQ;YACR,OAAO;SACR;QACD,UAAU,CAAC,IAAY;YACrB,OAAO,kBAAkB,CAAC,IAAI,CAAC,CAAA;QACjC,CAAC;KACF,CAAC,CAAA;IAEF,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,CAAA;IACzB,MAAM,QAAQ,GAAa,EAAE,CAAA;IAE7B,CAAC,CAAC,uCAAuC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,EAAE;QAC7D,IAAI,CAAC,CAAC,SAAS,IAAI,OAAO,CAAC,IAAI,OAAO,OAAO,CAAC,OAAO,KAAK,QAAQ,EAAE,CAAC;YACnE,OAAM;QACR,CAAC;QAED,MAAM,GAAG,GAAG,OAAO,CAAC,OAAO,CAAC,WAAW,EAAE,CAAA;QAEzC,MAAM,GAAG,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,CAAA;QAC7B,MAAM,IAAI,GAAG,mBAAmB,CAAC,GAAG,CAAC,CAAC,UAAU,CAAC,MAAM,EAAE,GAAG,CAAC,CAAA;QAC7D,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC/B,OAAM;QACR,CAAC;QAED,IAAI,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YACxB,IAAI,IAAI,CAAC,MAAM,IAAI,EAAE,EAAE,CAAC;gBACtB,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YACrB,CAAC;YACD,OAAM;QACR,CAAC;QAED,IAAI,GAAG,KAAK,IAAI,EAAE,CAAC;YACjB,IAAI,IAAI,CAAC,MAAM,IAAI,EAAE,EAAE,CAAC;gBACtB,QAAQ,CAAC,IAAI,CAAC,KAAK,IAAI,EAAE,CAAC,CAAA;YAC5B,CAAC;YACD,OAAM;QACR,CAAC;QAED,IAAI,IAAI,CAAC,MAAM,GAAG,kBAAkB,EAAE,CAAC;YACrC,OAAM;QACR,CAAC;QAED,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IACrB,CAAC,CAAC,CAAA;IAEF,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC1B,MAAM,QAAQ,GAAG,mBAAmB,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,IAAI,SAAS,CAAC,CAAA;QACnE,OAAO,QAAQ,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;IACnC,CAAC;IAED,OAAO,wBAAwB,CAAC,QAAQ,CAAC,CAAA;AAC3C,CAAC;AAED,MAAM,UAAU,gBAAgB,CAAC,IAAY;IAC3C,MAAM,QAAQ,GAAG,YAAY,CAAC,IAAI,EAAE;QAClC,WAAW,EAAE,EAAE;QACf,iBAAiB,EAAE,EAAE;QACrB,WAAW,EAAE;YACX,OAAO;YACP,QAAQ;YACR,UAAU;YACV,UAAU;YACV,KAAK;YACL,QAAQ;YACR,QAAQ;YACR,QAAQ;YACR,OAAO;SACR;KACF,CAAC,CAAA;IACF,OAAO,kBAAkB,CAAC,QAAQ,CAAC,CAAA;AACrC,CAAC;AAED,SAAS,wBAAwB,CAAC,QAAkB;IAClD,gGAAgG;IAChG,2FAA2F;IAC3F,OAAO,QAAQ,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;AACjC,CAAC"}
|
|
@@ -0,0 +1,55 @@
|
|
|
1
|
+
import { compact } from 'es-toolkit';
|
|
2
|
+
const WORD_SPLIT_PATTERN = /\s+/g;
|
|
3
|
+
export function normalizeForPrompt(input) {
|
|
4
|
+
return input
|
|
5
|
+
.replaceAll('\u00A0', ' ')
|
|
6
|
+
.replaceAll(/[\t ]+/g, ' ')
|
|
7
|
+
.replaceAll(/\s*\n\s*/g, '\n')
|
|
8
|
+
.replaceAll(/\n{3,}/g, '\n\n')
|
|
9
|
+
.trim();
|
|
10
|
+
}
|
|
11
|
+
export function normalizeWhitespace(input) {
|
|
12
|
+
return input
|
|
13
|
+
.replaceAll('\u00A0', ' ')
|
|
14
|
+
.replaceAll(/[\t ]+/g, ' ')
|
|
15
|
+
.replaceAll(/\s*\n\s*/g, '\n')
|
|
16
|
+
.trim();
|
|
17
|
+
}
|
|
18
|
+
export function decodeHtmlEntities(input) {
|
|
19
|
+
return input
|
|
20
|
+
.replaceAll('&', '&')
|
|
21
|
+
.replaceAll('<', '<')
|
|
22
|
+
.replaceAll('>', '>')
|
|
23
|
+
.replaceAll('"', '"')
|
|
24
|
+
.replaceAll(''', "'")
|
|
25
|
+
.replaceAll(''', "'")
|
|
26
|
+
.replaceAll('/', '/')
|
|
27
|
+
.replaceAll(' ', ' ');
|
|
28
|
+
}
|
|
29
|
+
export function normalizeCandidate(value) {
|
|
30
|
+
if (!value) {
|
|
31
|
+
return null;
|
|
32
|
+
}
|
|
33
|
+
const trimmed = value.replaceAll(/\s+/g, ' ').trim();
|
|
34
|
+
return trimmed.length > 0 ? trimmed : null;
|
|
35
|
+
}
|
|
36
|
+
export function clipAtSentenceBoundary(input, maxLength) {
|
|
37
|
+
if (input.length <= maxLength) {
|
|
38
|
+
return input;
|
|
39
|
+
}
|
|
40
|
+
const slice = input.slice(0, maxLength);
|
|
41
|
+
const lastSentenceBreak = Math.max(slice.lastIndexOf('. '), slice.lastIndexOf('! '), slice.lastIndexOf('? '), slice.lastIndexOf('\n\n'));
|
|
42
|
+
if (lastSentenceBreak > maxLength * 0.5) {
|
|
43
|
+
return slice.slice(0, lastSentenceBreak + 1);
|
|
44
|
+
}
|
|
45
|
+
return slice;
|
|
46
|
+
}
|
|
47
|
+
export function applyContentBudget(baseContent, maxCharacters) {
|
|
48
|
+
const totalCharacters = baseContent.length;
|
|
49
|
+
const truncated = totalCharacters > maxCharacters;
|
|
50
|
+
const clipped = truncated ? clipAtSentenceBoundary(baseContent, maxCharacters) : baseContent;
|
|
51
|
+
const content = clipped.trim();
|
|
52
|
+
const wordCount = content.length > 0 ? compact(content.split(WORD_SPLIT_PATTERN)).length : 0;
|
|
53
|
+
return { content, truncated, totalCharacters, wordCount };
|
|
54
|
+
}
|
|
55
|
+
//# sourceMappingURL=cleaner.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"cleaner.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/cleaner.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,OAAO,EAAE,MAAM,YAAY,CAAA;AAEpC,MAAM,kBAAkB,GAAG,MAAM,CAAA;AASjC,MAAM,UAAU,kBAAkB,CAAC,KAAa;IAC9C,OAAO,KAAK;SACT,UAAU,CAAC,QAAQ,EAAE,GAAG,CAAC;SACzB,UAAU,CAAC,SAAS,EAAE,GAAG,CAAC;SAC1B,UAAU,CAAC,WAAW,EAAE,IAAI,CAAC;SAC7B,UAAU,CAAC,SAAS,EAAE,MAAM,CAAC;SAC7B,IAAI,EAAE,CAAA;AACX,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,KAAa;IAC/C,OAAO,KAAK;SACT,UAAU,CAAC,QAAQ,EAAE,GAAG,CAAC;SACzB,UAAU,CAAC,SAAS,EAAE,GAAG,CAAC;SAC1B,UAAU,CAAC,WAAW,EAAE,IAAI,CAAC;SAC7B,IAAI,EAAE,CAAA;AACX,CAAC;AAED,MAAM,UAAU,kBAAkB,CAAC,KAAa;IAC9C,OAAO,KAAK;SACT,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC;SACxB,UAAU,CAAC,MAAM,EAAE,GAAG,CAAC;SACvB,UAAU,CAAC,MAAM,EAAE,GAAG,CAAC;SACvB,UAAU,CAAC,QAAQ,EAAE,GAAG,CAAC;SACzB,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC;SACxB,UAAU,CAAC,QAAQ,EAAE,GAAG,CAAC;SACzB,UAAU,CAAC,QAAQ,EAAE,GAAG,CAAC;SACzB,UAAU,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAA;AAC9B,CAAC;AAED,MAAM,UAAU,kBAAkB,CAAC,KAAgC;IACjE,IAAI,CAAC,KAAK,EAAE,CAAC;QACX,OAAO,IAAI,CAAA;IACb,CAAC;IACD,MAAM,OAAO,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IACpD,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAA;AAC5C,CAAC;AAED,MAAM,UAAU,sBAAsB,CAAC,KAAa,EAAE,SAAiB;IACrE,IAAI,KAAK,CAAC,MAAM,IAAI,SAAS,EAAE,CAAC;QAC9B,OAAO,KAAK,CAAA;IACd,CAAC;IACD,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,SAAS,CAAC,CAAA;IACvC,MAAM,iBAAiB,GAAG,IAAI,CAAC,GAAG,CAChC,KAAK,CAAC,WAAW,CAAC,IAAI,CAAC,EACvB,KAAK,CAAC,WAAW,CAAC,IAAI,CAAC,EACvB,KAAK,CAAC,WAAW,CAAC,IAAI,CAAC,EACvB,KAAK,CAAC,WAAW,CAAC,MAAM,CAAC,CAC1B,CAAA;IACD,IAAI,iBAAiB,GAAG,SAAS,GAAG,GAAG,EAAE,CAAC;QACxC,OAAO,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,iBAAiB,GAAG,CAAC,CAAC,CAAA;IAC9C,CAAC;IACD,OAAO,KAAK,CAAA;AACd,CAAC;AAED,MAAM,UAAU,kBAAkB,CAChC,WAAmB,EACnB,aAAqB;IAErB,MAAM,eAAe,GAAG,WAAW,CAAC,MAAM,CAAA;IAC1C,MAAM,SAAS,GAAG,eAAe,GAAG,aAAa,CAAA;IACjD,MAAM,OAAO,GAAG,SAAS,CAAC,CAAC,CAAC,sBAAsB,CAAC,WAAW,EAAE,aAAa,CAAC,CAAC,CAAC,CAAC,WAAW,CAAA;IAC5F,MAAM,OAAO,GAAG,OAAO,CAAC,IAAI,EAAE,CAAA;IAC9B,MAAM,SAAS,GAAG,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,OAAO,CAAC,KAAK,CAAC,kBAAkB,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAA;IAC5F,OAAO,EAAE,OAAO,EAAE,SAAS,EAAE,eAAe,EAAE,SAAS,EAAE,CAAA;AAC3D,CAAC"}
|
|
@@ -0,0 +1,7 @@
|
|
|
1
|
+
export const BLOCKED_HTML_HINT_PATTERN = /access denied|attention required|captcha|cloudflare|enable javascript|forbidden|please turn javascript on|verify you are human/i;
|
|
2
|
+
export const MIN_HTML_CONTENT_CHARACTERS = 200;
|
|
3
|
+
export const MIN_READABILITY_CONTENT_CHARACTERS = 200;
|
|
4
|
+
export const MIN_METADATA_DESCRIPTION_CHARACTERS = 120;
|
|
5
|
+
export const READABILITY_RELATIVE_THRESHOLD = 0.6;
|
|
6
|
+
export const MIN_HTML_DOCUMENT_CHARACTERS_FOR_FALLBACK = 5000;
|
|
7
|
+
//# sourceMappingURL=constants.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"constants.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/constants.ts"],"names":[],"mappings":"AAAA,MAAM,CAAC,MAAM,yBAAyB,GACpC,iIAAiI,CAAA;AACnI,MAAM,CAAC,MAAM,2BAA2B,GAAG,GAAG,CAAA;AAC9C,MAAM,CAAC,MAAM,kCAAkC,GAAG,GAAG,CAAA;AACrD,MAAM,CAAC,MAAM,mCAAmC,GAAG,GAAG,CAAA;AACtD,MAAM,CAAC,MAAM,8BAA8B,GAAG,GAAG,CAAA;AACjD,MAAM,CAAC,MAAM,yCAAyC,GAAG,IAAI,CAAA"}
|
|
@@ -0,0 +1,124 @@
|
|
|
1
|
+
import { isYouTubeUrl } from '../../transcript/utils.js';
|
|
2
|
+
import { appendNote } from './utils.js';
|
|
3
|
+
const REQUEST_HEADERS = {
|
|
4
|
+
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0 Safari/537.36',
|
|
5
|
+
Accept: 'text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8',
|
|
6
|
+
'Accept-Language': 'en-US,en;q=0.9',
|
|
7
|
+
'Cache-Control': 'no-cache',
|
|
8
|
+
Pragma: 'no-cache',
|
|
9
|
+
};
|
|
10
|
+
const DEFAULT_REQUEST_TIMEOUT_MS = 5000;
|
|
11
|
+
export async function fetchHtmlDocument(fetchImpl, url, { timeoutMs, onProgress, } = {}) {
|
|
12
|
+
onProgress?.({ kind: 'fetch-html-start', url });
|
|
13
|
+
const controller = new AbortController();
|
|
14
|
+
const effectiveTimeoutMs = typeof timeoutMs === 'number' && Number.isFinite(timeoutMs)
|
|
15
|
+
? timeoutMs
|
|
16
|
+
: DEFAULT_REQUEST_TIMEOUT_MS;
|
|
17
|
+
const timeout = setTimeout(() => {
|
|
18
|
+
controller.abort();
|
|
19
|
+
}, effectiveTimeoutMs);
|
|
20
|
+
try {
|
|
21
|
+
const response = await fetchImpl(url, {
|
|
22
|
+
headers: REQUEST_HEADERS,
|
|
23
|
+
redirect: 'follow',
|
|
24
|
+
signal: controller.signal,
|
|
25
|
+
});
|
|
26
|
+
if (!response.ok) {
|
|
27
|
+
throw new Error(`Failed to fetch HTML document (status ${response.status})`);
|
|
28
|
+
}
|
|
29
|
+
const contentType = response.headers.get('content-type')?.toLowerCase() ?? null;
|
|
30
|
+
if (contentType &&
|
|
31
|
+
!contentType.includes('text/html') &&
|
|
32
|
+
!contentType.includes('application/xhtml+xml') &&
|
|
33
|
+
!contentType.includes('application/xml') &&
|
|
34
|
+
!contentType.includes('text/xml') &&
|
|
35
|
+
!contentType.includes('application/rss+xml') &&
|
|
36
|
+
!contentType.includes('application/atom+xml') &&
|
|
37
|
+
!contentType.startsWith('text/')) {
|
|
38
|
+
throw new Error(`Unsupported content-type for HTML document fetch: ${contentType}`);
|
|
39
|
+
}
|
|
40
|
+
const totalBytes = (() => {
|
|
41
|
+
const raw = response.headers.get('content-length');
|
|
42
|
+
if (!raw)
|
|
43
|
+
return null;
|
|
44
|
+
const parsed = Number(raw);
|
|
45
|
+
return Number.isFinite(parsed) && parsed > 0 ? Math.floor(parsed) : null;
|
|
46
|
+
})();
|
|
47
|
+
const body = response.body;
|
|
48
|
+
if (!body) {
|
|
49
|
+
const text = await response.text();
|
|
50
|
+
const bytes = new TextEncoder().encode(text).byteLength;
|
|
51
|
+
onProgress?.({ kind: 'fetch-html-done', url, downloadedBytes: bytes, totalBytes });
|
|
52
|
+
return text;
|
|
53
|
+
}
|
|
54
|
+
const reader = body.getReader();
|
|
55
|
+
const decoder = new TextDecoder();
|
|
56
|
+
let downloadedBytes = 0;
|
|
57
|
+
let text = '';
|
|
58
|
+
onProgress?.({ kind: 'fetch-html-progress', url, downloadedBytes: 0, totalBytes });
|
|
59
|
+
while (true) {
|
|
60
|
+
const { value, done } = await reader.read();
|
|
61
|
+
if (done)
|
|
62
|
+
break;
|
|
63
|
+
if (!value)
|
|
64
|
+
continue;
|
|
65
|
+
downloadedBytes += value.byteLength;
|
|
66
|
+
text += decoder.decode(value, { stream: true });
|
|
67
|
+
onProgress?.({ kind: 'fetch-html-progress', url, downloadedBytes, totalBytes });
|
|
68
|
+
}
|
|
69
|
+
text += decoder.decode();
|
|
70
|
+
onProgress?.({ kind: 'fetch-html-done', url, downloadedBytes, totalBytes });
|
|
71
|
+
return text;
|
|
72
|
+
}
|
|
73
|
+
catch (error) {
|
|
74
|
+
if (error instanceof DOMException && error.name === 'AbortError') {
|
|
75
|
+
throw new Error('Fetching HTML document timed out');
|
|
76
|
+
}
|
|
77
|
+
throw error;
|
|
78
|
+
}
|
|
79
|
+
finally {
|
|
80
|
+
clearTimeout(timeout);
|
|
81
|
+
}
|
|
82
|
+
}
|
|
83
|
+
export async function fetchWithFirecrawl(url, scrapeWithFirecrawl, options = {}) {
|
|
84
|
+
const timeoutMs = options.timeoutMs;
|
|
85
|
+
const cacheMode = options.cacheMode ?? 'default';
|
|
86
|
+
const onProgress = typeof options.onProgress === 'function' ? options.onProgress : null;
|
|
87
|
+
const reason = typeof options.reason === 'string' ? options.reason : null;
|
|
88
|
+
const diagnostics = {
|
|
89
|
+
attempted: false,
|
|
90
|
+
used: false,
|
|
91
|
+
cacheMode,
|
|
92
|
+
cacheStatus: cacheMode === 'bypass' ? 'bypassed' : 'unknown',
|
|
93
|
+
notes: null,
|
|
94
|
+
};
|
|
95
|
+
if (isYouTubeUrl(url)) {
|
|
96
|
+
diagnostics.notes = appendNote(diagnostics.notes, 'Skipped Firecrawl for YouTube URL');
|
|
97
|
+
return { payload: null, diagnostics };
|
|
98
|
+
}
|
|
99
|
+
if (!scrapeWithFirecrawl) {
|
|
100
|
+
diagnostics.notes = appendNote(diagnostics.notes, 'Firecrawl is not configured');
|
|
101
|
+
return { payload: null, diagnostics };
|
|
102
|
+
}
|
|
103
|
+
diagnostics.attempted = true;
|
|
104
|
+
onProgress?.({ kind: 'firecrawl-start', url, reason: reason ?? 'firecrawl' });
|
|
105
|
+
try {
|
|
106
|
+
const payload = await scrapeWithFirecrawl(url, { timeoutMs, cacheMode });
|
|
107
|
+
if (!payload) {
|
|
108
|
+
diagnostics.notes = appendNote(diagnostics.notes, 'Firecrawl returned no content payload');
|
|
109
|
+
onProgress?.({ kind: 'firecrawl-done', url, ok: false, markdownBytes: null, htmlBytes: null });
|
|
110
|
+
return { payload: null, diagnostics };
|
|
111
|
+
}
|
|
112
|
+
const encoder = new TextEncoder();
|
|
113
|
+
const markdownBytes = typeof payload.markdown === 'string' ? encoder.encode(payload.markdown).byteLength : null;
|
|
114
|
+
const htmlBytes = typeof payload.html === 'string' ? encoder.encode(payload.html).byteLength : null;
|
|
115
|
+
onProgress?.({ kind: 'firecrawl-done', url, ok: true, markdownBytes, htmlBytes });
|
|
116
|
+
return { payload, diagnostics };
|
|
117
|
+
}
|
|
118
|
+
catch (error) {
|
|
119
|
+
diagnostics.notes = appendNote(diagnostics.notes, `Firecrawl error: ${error instanceof Error ? error.message : 'unknown error'}`);
|
|
120
|
+
onProgress?.({ kind: 'firecrawl-done', url, ok: false, markdownBytes: null, htmlBytes: null });
|
|
121
|
+
return { payload: null, diagnostics };
|
|
122
|
+
}
|
|
123
|
+
}
|
|
124
|
+
//# sourceMappingURL=fetcher.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"fetcher.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/fetcher.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AAQxD,OAAO,EAAE,UAAU,EAAE,MAAM,YAAY,CAAA;AAEvC,MAAM,eAAe,GAA2B;IAC9C,YAAY,EACV,iHAAiH;IACnH,MAAM,EACJ,kGAAkG;IACpG,iBAAiB,EAAE,gBAAgB;IACnC,eAAe,EAAE,UAAU;IAC3B,MAAM,EAAE,UAAU;CACnB,CAAA;AAED,MAAM,0BAA0B,GAAG,IAAI,CAAA;AAOvC,MAAM,CAAC,KAAK,UAAU,iBAAiB,CACrC,SAAuB,EACvB,GAAW,EACX,EACE,SAAS,EACT,UAAU,MACiF,EAAE;IAE/F,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,kBAAkB,EAAE,GAAG,EAAE,CAAC,CAAA;IAE/C,MAAM,UAAU,GAAG,IAAI,eAAe,EAAE,CAAA;IACxC,MAAM,kBAAkB,GACtB,OAAO,SAAS,KAAK,QAAQ,IAAI,MAAM,CAAC,QAAQ,CAAC,SAAS,CAAC;QACzD,CAAC,CAAC,SAAS;QACX,CAAC,CAAC,0BAA0B,CAAA;IAChC,MAAM,OAAO,GAAG,UAAU,CAAC,GAAG,EAAE;QAC9B,UAAU,CAAC,KAAK,EAAE,CAAA;IACpB,CAAC,EAAE,kBAAkB,CAAC,CAAA;IAEtB,IAAI,CAAC;QACH,MAAM,QAAQ,GAAG,MAAM,SAAS,CAAC,GAAG,EAAE;YACpC,OAAO,EAAE,eAAe;YACxB,QAAQ,EAAE,QAAQ;YAClB,MAAM,EAAE,UAAU,CAAC,MAAM;SAC1B,CAAC,CAAA;QAEF,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;YACjB,MAAM,IAAI,KAAK,CAAC,yCAAyC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAA;QAC9E,CAAC;QAED,MAAM,WAAW,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,cAAc,CAAC,EAAE,WAAW,EAAE,IAAI,IAAI,CAAA;QAC/E,IACE,WAAW;YACX,CAAC,WAAW,CAAC,QAAQ,CAAC,WAAW,CAAC;YAClC,CAAC,WAAW,CAAC,QAAQ,CAAC,uBAAuB,CAAC;YAC9C,CAAC,WAAW,CAAC,QAAQ,CAAC,iBAAiB,CAAC;YACxC,CAAC,WAAW,CAAC,QAAQ,CAAC,UAAU,CAAC;YACjC,CAAC,WAAW,CAAC,QAAQ,CAAC,qBAAqB,CAAC;YAC5C,CAAC,WAAW,CAAC,QAAQ,CAAC,sBAAsB,CAAC;YAC7C,CAAC,WAAW,CAAC,UAAU,CAAC,OAAO,CAAC,EAChC,CAAC;YACD,MAAM,IAAI,KAAK,CAAC,qDAAqD,WAAW,EAAE,CAAC,CAAA;QACrF,CAAC;QAED,MAAM,UAAU,GAAG,CAAC,GAAG,EAAE;YACvB,MAAM,GAAG,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,gBAAgB,CAAC,CAAA;YAClD,IAAI,CAAC,GAAG;gBAAE,OAAO,IAAI,CAAA;YACrB,MAAM,MAAM,GAAG,MAAM,CAAC,GAAG,CAAC,CAAA;YAC1B,OAAO,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;QAC1E,CAAC,CAAC,EAAE,CAAA;QAEJ,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAA;QAC1B,IAAI,CAAC,IAAI,EAAE,CAAC;YACV,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAC,IAAI,EAAE,CAAA;YAClC,MAAM,KAAK,GAAG,IAAI,WAAW,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,UAAU,CAAA;YACvD,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,iBAAiB,EAAE,GAAG,EAAE,eAAe,EAAE,KAAK,EAAE,UAAU,EAAE,CAAC,CAAA;YAClF,OAAO,IAAI,CAAA;QACb,CAAC;QAED,MAAM,MAAM,GAAG,IAAI,CAAC,SAAS,EAAE,CAAA;QAC/B,MAAM,OAAO,GAAG,IAAI,WAAW,EAAE,CAAA;QACjC,IAAI,eAAe,GAAG,CAAC,CAAA;QACvB,IAAI,IAAI,GAAG,EAAE,CAAA;QAEb,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,qBAAqB,EAAE,GAAG,EAAE,eAAe,EAAE,CAAC,EAAE,UAAU,EAAE,CAAC,CAAA;QAElF,OAAO,IAAI,EAAE,CAAC;YACZ,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAA;YAC3C,IAAI,IAAI;gBAAE,MAAK;YACf,IAAI,CAAC,KAAK;gBAAE,SAAQ;YACpB,eAAe,IAAI,KAAK,CAAC,UAAU,CAAA;YACnC,IAAI,IAAI,OAAO,CAAC,MAAM,CAAC,KAAK,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC,CAAA;YAC/C,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,qBAAqB,EAAE,GAAG,EAAE,eAAe,EAAE,UAAU,EAAE,CAAC,CAAA;QACjF,CAAC;QAED,IAAI,IAAI,OAAO,CAAC,MAAM,EAAE,CAAA;QACxB,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,iBAAiB,EAAE,GAAG,EAAE,eAAe,EAAE,UAAU,EAAE,CAAC,CAAA;QAC3E,OAAO,IAAI,CAAA;IACb,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAI,KAAK,YAAY,YAAY,IAAI,KAAK,CAAC,IAAI,KAAK,YAAY,EAAE,CAAC;YACjE,MAAM,IAAI,KAAK,CAAC,kCAAkC,CAAC,CAAA;QACrD,CAAC;QACD,MAAM,KAAK,CAAA;IACb,CAAC;YAAS,CAAC;QACT,YAAY,CAAC,OAAO,CAAC,CAAA;IACvB,CAAC;AACH,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,kBAAkB,CACtC,GAAW,EACX,mBAA+C,EAC/C,UAKI,EAAE;IAEN,MAAM,SAAS,GAAG,OAAO,CAAC,SAAS,CAAA;IACnC,MAAM,SAAS,GAAc,OAAO,CAAC,SAAS,IAAI,SAAS,CAAA;IAC3D,MAAM,UAAU,GAAG,OAAO,OAAO,CAAC,UAAU,KAAK,UAAU,CAAC,CAAC,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI,CAAA;IACvF,MAAM,MAAM,GAAG,OAAO,OAAO,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAA;IACzE,MAAM,WAAW,GAAyB;QACxC,SAAS,EAAE,KAAK;QAChB,IAAI,EAAE,KAAK;QACX,SAAS;QACT,WAAW,EAAE,SAAS,KAAK,QAAQ,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,SAAS;QAC5D,KAAK,EAAE,IAAI;KACZ,CAAA;IAED,IAAI,YAAY,CAAC,GAAG,CAAC,EAAE,CAAC;QACtB,WAAW,CAAC,KAAK,GAAG,UAAU,CAAC,WAAW,CAAC,KAAK,EAAE,mCAAmC,CAAC,CAAA;QACtF,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,WAAW,EAAE,CAAA;IACvC,CAAC;IAED,IAAI,CAAC,mBAAmB,EAAE,CAAC;QACzB,WAAW,CAAC,KAAK,GAAG,UAAU,CAAC,WAAW,CAAC,KAAK,EAAE,6BAA6B,CAAC,CAAA;QAChF,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,WAAW,EAAE,CAAA;IACvC,CAAC;IAED,WAAW,CAAC,SAAS,GAAG,IAAI,CAAA;IAC5B,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,iBAAiB,EAAE,GAAG,EAAE,MAAM,EAAE,MAAM,IAAI,WAAW,EAAE,CAAC,CAAA;IAE7E,IAAI,CAAC;QACH,MAAM,OAAO,GAAG,MAAM,mBAAmB,CAAC,GAAG,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,CAAC,CAAA;QACxE,IAAI,CAAC,OAAO,EAAE,CAAC;YACb,WAAW,CAAC,KAAK,GAAG,UAAU,CAAC,WAAW,CAAC,KAAK,EAAE,uCAAuC,CAAC,CAAA;YAC1F,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,GAAG,EAAE,EAAE,EAAE,KAAK,EAAE,aAAa,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;YAC9F,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,WAAW,EAAE,CAAA;QACvC,CAAC;QAED,MAAM,OAAO,GAAG,IAAI,WAAW,EAAE,CAAA;QACjC,MAAM,aAAa,GACjB,OAAO,OAAO,CAAC,QAAQ,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI,CAAA;QAC3F,MAAM,SAAS,GACb,OAAO,OAAO,CAAC,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI,CAAA;QACnF,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,GAAG,EAAE,EAAE,EAAE,IAAI,EAAE,aAAa,EAAE,SAAS,EAAE,CAAC,CAAA;QAEjF,OAAO,EAAE,OAAO,EAAE,WAAW,EAAE,CAAA;IACjC,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,WAAW,CAAC,KAAK,GAAG,UAAU,CAC5B,WAAW,CAAC,KAAK,EACjB,oBAAoB,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,EAAE,CAC/E,CAAA;QACD,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,GAAG,EAAE,EAAE,EAAE,KAAK,EAAE,aAAa,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;QAC9F,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,WAAW,EAAE,CAAA;IACvC,CAAC;AACH,CAAC"}
|
|
@@ -0,0 +1,86 @@
|
|
|
1
|
+
import { resolveTranscriptForLink } from '../../transcript/index.js';
|
|
2
|
+
import { extractArticleContent, extractPlainText } from './article.js';
|
|
3
|
+
import { normalizeForPrompt } from './cleaner.js';
|
|
4
|
+
import { BLOCKED_HTML_HINT_PATTERN, MIN_HTML_CONTENT_CHARACTERS, MIN_HTML_DOCUMENT_CHARACTERS_FOR_FALLBACK, MIN_METADATA_DESCRIPTION_CHARACTERS, READABILITY_RELATIVE_THRESHOLD, } from './constants.js';
|
|
5
|
+
import { extractJsonLdContent } from './jsonld.js';
|
|
6
|
+
import { extractMetadataFromFirecrawl, extractMetadataFromHtml } from './parsers.js';
|
|
7
|
+
import { isPodcastHost, isPodcastLikeJsonLdType } from './podcast-utils.js';
|
|
8
|
+
import { appendNote, ensureTranscriptDiagnostics, finalizeExtractedLinkContent, pickFirstText, safeHostname, selectBaseContent, } from './utils.js';
|
|
9
|
+
import { detectPrimaryVideoFromHtml } from './video.js';
|
|
10
|
+
export function shouldFallbackToFirecrawl(html) {
|
|
11
|
+
const plainText = normalizeForPrompt(extractPlainText(html));
|
|
12
|
+
if (BLOCKED_HTML_HINT_PATTERN.test(plainText))
|
|
13
|
+
return true;
|
|
14
|
+
const normalized = normalizeForPrompt(extractArticleContent(html));
|
|
15
|
+
if (normalized.length >= MIN_HTML_CONTENT_CHARACTERS) {
|
|
16
|
+
return false;
|
|
17
|
+
}
|
|
18
|
+
// Avoid spending Firecrawl on truly small/simple pages where the extracted HTML content is short but
|
|
19
|
+
// likely complete (e.g. https://example.com). Only treat "thin" content as a Firecrawl signal when
|
|
20
|
+
// the HTML document itself is large (SSR/app-shell pages, blocked pages without a match, etc.).
|
|
21
|
+
return html.length >= MIN_HTML_DOCUMENT_CHARACTERS_FOR_FALLBACK;
|
|
22
|
+
}
|
|
23
|
+
export async function buildResultFromFirecrawl({ url, payload, cacheMode, maxCharacters, youtubeTranscriptMode, firecrawlDiagnostics, markdownRequested, deps, }) {
|
|
24
|
+
const normalizedMarkdown = normalizeForPrompt(payload.markdown ?? '');
|
|
25
|
+
if (normalizedMarkdown.length === 0) {
|
|
26
|
+
firecrawlDiagnostics.notes = appendNote(firecrawlDiagnostics.notes, 'Firecrawl markdown normalization yielded empty text');
|
|
27
|
+
return null;
|
|
28
|
+
}
|
|
29
|
+
const jsonLd = payload.html ? extractJsonLdContent(payload.html) : null;
|
|
30
|
+
const isPodcastJsonLd = isPodcastLikeJsonLdType(jsonLd?.type);
|
|
31
|
+
const transcriptResolution = await resolveTranscriptForLink(url, payload.html ?? null, deps, {
|
|
32
|
+
youtubeTranscriptMode,
|
|
33
|
+
cacheMode,
|
|
34
|
+
});
|
|
35
|
+
const htmlMetadata = payload.html
|
|
36
|
+
? extractMetadataFromHtml(payload.html, url)
|
|
37
|
+
: { title: null, description: null, siteName: null };
|
|
38
|
+
const metadata = extractMetadataFromFirecrawl(payload.metadata ?? null);
|
|
39
|
+
const title = pickFirstText([jsonLd?.title, metadata.title, htmlMetadata.title]);
|
|
40
|
+
const description = pickFirstText([
|
|
41
|
+
jsonLd?.description,
|
|
42
|
+
metadata.description,
|
|
43
|
+
htmlMetadata.description,
|
|
44
|
+
]);
|
|
45
|
+
const siteName = pickFirstText([metadata.siteName, htmlMetadata.siteName, safeHostname(url)]);
|
|
46
|
+
const descriptionCandidate = description ? normalizeForPrompt(description) : '';
|
|
47
|
+
const preferDescription = descriptionCandidate.length >= MIN_METADATA_DESCRIPTION_CHARACTERS &&
|
|
48
|
+
(isPodcastJsonLd ||
|
|
49
|
+
isPodcastHost(url) ||
|
|
50
|
+
normalizedMarkdown.length < MIN_HTML_CONTENT_CHARACTERS ||
|
|
51
|
+
descriptionCandidate.length >= normalizedMarkdown.length * READABILITY_RELATIVE_THRESHOLD);
|
|
52
|
+
const baseCandidate = preferDescription ? descriptionCandidate : normalizedMarkdown;
|
|
53
|
+
const baseContent = selectBaseContent(baseCandidate, transcriptResolution.text);
|
|
54
|
+
if (baseContent.length === 0) {
|
|
55
|
+
firecrawlDiagnostics.notes = appendNote(firecrawlDiagnostics.notes, 'Firecrawl produced content that normalized to an empty string');
|
|
56
|
+
return null;
|
|
57
|
+
}
|
|
58
|
+
firecrawlDiagnostics.used = true;
|
|
59
|
+
const transcriptDiagnostics = ensureTranscriptDiagnostics(transcriptResolution, cacheMode ?? 'default');
|
|
60
|
+
const video = payload.html ? detectPrimaryVideoFromHtml(payload.html, url) : null;
|
|
61
|
+
const isVideoOnly = !transcriptResolution.text &&
|
|
62
|
+
normalizedMarkdown.length < MIN_HTML_CONTENT_CHARACTERS &&
|
|
63
|
+
video !== null;
|
|
64
|
+
return finalizeExtractedLinkContent({
|
|
65
|
+
url,
|
|
66
|
+
baseContent,
|
|
67
|
+
maxCharacters,
|
|
68
|
+
title,
|
|
69
|
+
description,
|
|
70
|
+
siteName,
|
|
71
|
+
transcriptResolution,
|
|
72
|
+
video,
|
|
73
|
+
isVideoOnly,
|
|
74
|
+
diagnostics: {
|
|
75
|
+
strategy: 'firecrawl',
|
|
76
|
+
firecrawl: firecrawlDiagnostics,
|
|
77
|
+
markdown: {
|
|
78
|
+
requested: markdownRequested,
|
|
79
|
+
used: true,
|
|
80
|
+
provider: 'firecrawl',
|
|
81
|
+
},
|
|
82
|
+
transcript: transcriptDiagnostics,
|
|
83
|
+
},
|
|
84
|
+
});
|
|
85
|
+
}
|
|
86
|
+
//# sourceMappingURL=firecrawl.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"firecrawl.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/firecrawl.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,wBAAwB,EAAE,MAAM,2BAA2B,CAAA;AAGpE,OAAO,EAAE,qBAAqB,EAAE,gBAAgB,EAAE,MAAM,cAAc,CAAA;AACtE,OAAO,EAAE,kBAAkB,EAAE,MAAM,cAAc,CAAA;AACjD,OAAO,EACL,yBAAyB,EACzB,2BAA2B,EAC3B,yCAAyC,EACzC,mCAAmC,EACnC,8BAA8B,GAC/B,MAAM,gBAAgB,CAAA;AACvB,OAAO,EAAE,oBAAoB,EAAE,MAAM,aAAa,CAAA;AAClD,OAAO,EAAE,4BAA4B,EAAE,uBAAuB,EAAE,MAAM,cAAc,CAAA;AACpF,OAAO,EAAE,aAAa,EAAE,uBAAuB,EAAE,MAAM,oBAAoB,CAAA;AAE3E,OAAO,EACL,UAAU,EACV,2BAA2B,EAC3B,4BAA4B,EAC5B,aAAa,EACb,YAAY,EACZ,iBAAiB,GAClB,MAAM,YAAY,CAAA;AACnB,OAAO,EAAE,0BAA0B,EAAE,MAAM,YAAY,CAAA;AAEvD,MAAM,UAAU,yBAAyB,CAAC,IAAY;IACpD,MAAM,SAAS,GAAG,kBAAkB,CAAC,gBAAgB,CAAC,IAAI,CAAC,CAAC,CAAA;IAC5D,IAAI,yBAAyB,CAAC,IAAI,CAAC,SAAS,CAAC;QAAE,OAAO,IAAI,CAAA;IAC1D,MAAM,UAAU,GAAG,kBAAkB,CAAC,qBAAqB,CAAC,IAAI,CAAC,CAAC,CAAA;IAClE,IAAI,UAAU,CAAC,MAAM,IAAI,2BAA2B,EAAE,CAAC;QACrD,OAAO,KAAK,CAAA;IACd,CAAC;IAED,qGAAqG;IACrG,mGAAmG;IACnG,gGAAgG;IAChG,OAAO,IAAI,CAAC,MAAM,IAAI,yCAAyC,CAAA;AACjE,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,wBAAwB,CAAC,EAC7C,GAAG,EACH,OAAO,EACP,SAAS,EACT,aAAa,EACb,qBAAqB,EACrB,oBAAoB,EACpB,iBAAiB,EACjB,IAAI,GAUL;IACC,MAAM,kBAAkB,GAAG,kBAAkB,CAAC,OAAO,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IACrE,IAAI,kBAAkB,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACpC,oBAAoB,CAAC,KAAK,GAAG,UAAU,CACrC,oBAAoB,CAAC,KAAK,EAC1B,qDAAqD,CACtD,CAAA;QACD,OAAO,IAAI,CAAA;IACb,CAAC;IAED,MAAM,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,oBAAoB,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;IACvE,MAAM,eAAe,GAAG,uBAAuB,CAAC,MAAM,EAAE,IAAI,CAAC,CAAA;IAE7D,MAAM,oBAAoB,GAAG,MAAM,wBAAwB,CAAC,GAAG,EAAE,OAAO,CAAC,IAAI,IAAI,IAAI,EAAE,IAAI,EAAE;QAC3F,qBAAqB;QACrB,SAAS;KACV,CAAC,CAAA;IACF,MAAM,YAAY,GAAG,OAAO,CAAC,IAAI;QAC/B,CAAC,CAAC,uBAAuB,CAAC,OAAO,CAAC,IAAI,EAAE,GAAG,CAAC;QAC5C,CAAC,CAAC,EAAE,KAAK,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAA;IACtD,MAAM,QAAQ,GAAG,4BAA4B,CAAC,OAAO,CAAC,QAAQ,IAAI,IAAI,CAAC,CAAA;IAEvE,MAAM,KAAK,GAAG,aAAa,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,QAAQ,CAAC,KAAK,EAAE,YAAY,CAAC,KAAK,CAAC,CAAC,CAAA;IAChF,MAAM,WAAW,GAAG,aAAa,CAAC;QAChC,MAAM,EAAE,WAAW;QACnB,QAAQ,CAAC,WAAW;QACpB,YAAY,CAAC,WAAW;KACzB,CAAC,CAAA;IACF,MAAM,QAAQ,GAAG,aAAa,CAAC,CAAC,QAAQ,CAAC,QAAQ,EAAE,YAAY,CAAC,QAAQ,EAAE,YAAY,CAAC,GAAG,CAAC,CAAC,CAAC,CAAA;IAE7F,MAAM,oBAAoB,GAAG,WAAW,CAAC,CAAC,CAAC,kBAAkB,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;IAC/E,MAAM,iBAAiB,GACrB,oBAAoB,CAAC,MAAM,IAAI,mCAAmC;QAClE,CAAC,eAAe;YACd,aAAa,CAAC,GAAG,CAAC;YAClB,kBAAkB,CAAC,MAAM,GAAG,2BAA2B;YACvD,oBAAoB,CAAC,MAAM,IAAI,kBAAkB,CAAC,MAAM,GAAG,8BAA8B,CAAC,CAAA;IAC9F,MAAM,aAAa,GAAG,iBAAiB,CAAC,CAAC,CAAC,oBAAoB,CAAC,CAAC,CAAC,kBAAkB,CAAA;IACnF,MAAM,WAAW,GAAG,iBAAiB,CAAC,aAAa,EAAE,oBAAoB,CAAC,IAAI,CAAC,CAAA;IAC/E,IAAI,WAAW,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC7B,oBAAoB,CAAC,KAAK,GAAG,UAAU,CACrC,oBAAoB,CAAC,KAAK,EAC1B,+DAA+D,CAChE,CAAA;QACD,OAAO,IAAI,CAAA;IACb,CAAC;IAED,oBAAoB,CAAC,IAAI,GAAG,IAAI,CAAA;IAEhC,MAAM,qBAAqB,GAAG,2BAA2B,CACvD,oBAAoB,EACpB,SAAS,IAAI,SAAS,CACvB,CAAA;IAED,MAAM,KAAK,GAAG,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,0BAA0B,CAAC,OAAO,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;IACjF,MAAM,WAAW,GACf,CAAC,oBAAoB,CAAC,IAAI;QAC1B,kBAAkB,CAAC,MAAM,GAAG,2BAA2B;QACvD,KAAK,KAAK,IAAI,CAAA;IAEhB,OAAO,4BAA4B,CAAC;QAClC,GAAG;QACH,WAAW;QACX,aAAa;QACb,KAAK;QACL,WAAW;QACX,QAAQ;QACR,oBAAoB;QACpB,KAAK;QACL,WAAW;QACX,WAAW,EAAE;YACX,QAAQ,EAAE,WAAW;YACrB,SAAS,EAAE,oBAAoB;YAC/B,QAAQ,EAAE;gBACR,SAAS,EAAE,iBAAiB;gBAC5B,IAAI,EAAE,IAAI;gBACV,QAAQ,EAAE,WAAW;aACtB;YACD,UAAU,EAAE,qBAAqB;SAClC;KACF,CAAC,CAAA;AACJ,CAAC"}
|