@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
|
@@ -0,0 +1,162 @@
|
|
|
1
|
+
import { resolveTranscriptForLink } from '../../transcript/index.js';
|
|
2
|
+
import { extractYouTubeVideoId, isYouTubeUrl, isYouTubeVideoUrl } from '../../transcript/utils.js';
|
|
3
|
+
import { extractArticleContent, sanitizeHtmlForMarkdownConversion } from './article.js';
|
|
4
|
+
import { normalizeForPrompt } from './cleaner.js';
|
|
5
|
+
import { MIN_HTML_CONTENT_CHARACTERS, MIN_METADATA_DESCRIPTION_CHARACTERS, MIN_READABILITY_CONTENT_CHARACTERS, READABILITY_RELATIVE_THRESHOLD, } from './constants.js';
|
|
6
|
+
import { extractJsonLdContent } from './jsonld.js';
|
|
7
|
+
import { extractMetadataFromHtml } from './parsers.js';
|
|
8
|
+
import { isPodcastHost, isPodcastLikeJsonLdType } from './podcast-utils.js';
|
|
9
|
+
import { extractReadabilityFromHtml, toReadabilityHtml } from './readability.js';
|
|
10
|
+
import { ensureTranscriptDiagnostics, finalizeExtractedLinkContent, pickFirstText, selectBaseContent, } from './utils.js';
|
|
11
|
+
import { detectPrimaryVideoFromHtml } from './video.js';
|
|
12
|
+
import { extractYouTubeShortDescription } from './youtube.js';
|
|
13
|
+
const LEADING_CONTROL_PATTERN = /^[\s\p{Cc}]+/u;
|
|
14
|
+
function stripLeadingTitle(content, title) {
|
|
15
|
+
if (!(content && title)) {
|
|
16
|
+
return content;
|
|
17
|
+
}
|
|
18
|
+
const normalizedTitle = title.trim();
|
|
19
|
+
if (normalizedTitle.length === 0) {
|
|
20
|
+
return content;
|
|
21
|
+
}
|
|
22
|
+
const trimmedContent = content.trimStart();
|
|
23
|
+
if (!trimmedContent.toLowerCase().startsWith(normalizedTitle.toLowerCase())) {
|
|
24
|
+
return content;
|
|
25
|
+
}
|
|
26
|
+
const remainderOriginal = trimmedContent.slice(normalizedTitle.length);
|
|
27
|
+
const remainder = remainderOriginal.replace(LEADING_CONTROL_PATTERN, '');
|
|
28
|
+
return remainder;
|
|
29
|
+
}
|
|
30
|
+
export async function buildResultFromHtmlDocument({ url, html, cacheMode, maxCharacters, youtubeTranscriptMode, firecrawlDiagnostics, markdownRequested, markdownMode, timeoutMs, deps, readabilityCandidate, }) {
|
|
31
|
+
if (isYouTubeVideoUrl(url) && !extractYouTubeVideoId(url)) {
|
|
32
|
+
throw new Error('Invalid YouTube video id in URL');
|
|
33
|
+
}
|
|
34
|
+
const { title, description, siteName } = extractMetadataFromHtml(html, url);
|
|
35
|
+
const jsonLd = extractJsonLdContent(html);
|
|
36
|
+
const mergedTitle = pickFirstText([jsonLd?.title, title]);
|
|
37
|
+
const mergedDescription = pickFirstText([jsonLd?.description, description]);
|
|
38
|
+
const isPodcastJsonLd = isPodcastLikeJsonLdType(jsonLd?.type);
|
|
39
|
+
const readability = readabilityCandidate ?? (await extractReadabilityFromHtml(html, url));
|
|
40
|
+
const readabilityText = readability?.text ? normalizeForPrompt(readability.text) : '';
|
|
41
|
+
const readabilityHtml = toReadabilityHtml(readability);
|
|
42
|
+
const normalizedSegmentsFromHtml = normalizeForPrompt(extractArticleContent(html));
|
|
43
|
+
const normalizedSegmentsFromReadabilityHtml = readabilityHtml
|
|
44
|
+
? normalizeForPrompt(extractArticleContent(readabilityHtml))
|
|
45
|
+
: '';
|
|
46
|
+
const preferReadabilityHtml = normalizedSegmentsFromReadabilityHtml.length >= MIN_READABILITY_CONTENT_CHARACTERS &&
|
|
47
|
+
(normalizedSegmentsFromHtml.length < MIN_HTML_CONTENT_CHARACTERS ||
|
|
48
|
+
normalizedSegmentsFromReadabilityHtml.length >=
|
|
49
|
+
normalizedSegmentsFromHtml.length * READABILITY_RELATIVE_THRESHOLD);
|
|
50
|
+
const normalizedSegments = preferReadabilityHtml
|
|
51
|
+
? normalizedSegmentsFromReadabilityHtml
|
|
52
|
+
: normalizedSegmentsFromHtml;
|
|
53
|
+
const preferReadabilityText = !preferReadabilityHtml &&
|
|
54
|
+
readabilityText.length >= MIN_READABILITY_CONTENT_CHARACTERS &&
|
|
55
|
+
(normalizedSegmentsFromHtml.length < MIN_HTML_CONTENT_CHARACTERS ||
|
|
56
|
+
readabilityText.length >= normalizedSegmentsFromHtml.length * READABILITY_RELATIVE_THRESHOLD);
|
|
57
|
+
const preferReadability = preferReadabilityHtml || preferReadabilityText;
|
|
58
|
+
const effectiveNormalized = preferReadabilityText ? readabilityText : normalizedSegments;
|
|
59
|
+
const descriptionCandidate = mergedDescription ? normalizeForPrompt(mergedDescription) : '';
|
|
60
|
+
const preferDescription = descriptionCandidate.length >= MIN_METADATA_DESCRIPTION_CHARACTERS &&
|
|
61
|
+
(isPodcastJsonLd ||
|
|
62
|
+
isPodcastHost(url) ||
|
|
63
|
+
(!preferReadability &&
|
|
64
|
+
(effectiveNormalized.length < MIN_HTML_CONTENT_CHARACTERS ||
|
|
65
|
+
descriptionCandidate.length >=
|
|
66
|
+
effectiveNormalized.length * READABILITY_RELATIVE_THRESHOLD)));
|
|
67
|
+
const effectiveNormalizedWithDescription = preferDescription
|
|
68
|
+
? descriptionCandidate
|
|
69
|
+
: effectiveNormalized;
|
|
70
|
+
const transcriptResolution = await resolveTranscriptForLink(url, html, deps, {
|
|
71
|
+
youtubeTranscriptMode,
|
|
72
|
+
cacheMode,
|
|
73
|
+
});
|
|
74
|
+
const youtubeDescription = transcriptResolution.text === null ? extractYouTubeShortDescription(html) : null;
|
|
75
|
+
const baseCandidate = youtubeDescription
|
|
76
|
+
? normalizeForPrompt(youtubeDescription)
|
|
77
|
+
: effectiveNormalizedWithDescription;
|
|
78
|
+
let baseContent = selectBaseContent(baseCandidate, transcriptResolution.text);
|
|
79
|
+
if (baseContent === normalizedSegments) {
|
|
80
|
+
baseContent = stripLeadingTitle(baseContent, mergedTitle ?? title);
|
|
81
|
+
}
|
|
82
|
+
const transcriptDiagnostics = ensureTranscriptDiagnostics(transcriptResolution, cacheMode ?? 'default');
|
|
83
|
+
const markdownDiagnostics = await (async () => {
|
|
84
|
+
if (!markdownRequested) {
|
|
85
|
+
return { requested: false, used: false, provider: null, notes: null };
|
|
86
|
+
}
|
|
87
|
+
if (isYouTubeUrl(url)) {
|
|
88
|
+
return {
|
|
89
|
+
requested: true,
|
|
90
|
+
used: false,
|
|
91
|
+
provider: null,
|
|
92
|
+
notes: 'Skipping Markdown conversion for YouTube URLs',
|
|
93
|
+
};
|
|
94
|
+
}
|
|
95
|
+
if (!deps.convertHtmlToMarkdown) {
|
|
96
|
+
return {
|
|
97
|
+
requested: true,
|
|
98
|
+
used: false,
|
|
99
|
+
provider: null,
|
|
100
|
+
notes: 'No HTML→Markdown converter configured',
|
|
101
|
+
};
|
|
102
|
+
}
|
|
103
|
+
try {
|
|
104
|
+
const htmlForMarkdown = markdownMode === 'readability' && readabilityHtml ? readabilityHtml : html;
|
|
105
|
+
const sanitizedHtml = sanitizeHtmlForMarkdownConversion(htmlForMarkdown);
|
|
106
|
+
const markdown = await deps.convertHtmlToMarkdown({
|
|
107
|
+
url,
|
|
108
|
+
html: sanitizedHtml,
|
|
109
|
+
title: mergedTitle ?? title,
|
|
110
|
+
siteName,
|
|
111
|
+
timeoutMs,
|
|
112
|
+
});
|
|
113
|
+
const normalizedMarkdown = normalizeForPrompt(markdown);
|
|
114
|
+
if (normalizedMarkdown.length === 0) {
|
|
115
|
+
return {
|
|
116
|
+
requested: true,
|
|
117
|
+
used: false,
|
|
118
|
+
provider: null,
|
|
119
|
+
notes: 'HTML→Markdown conversion returned empty content',
|
|
120
|
+
};
|
|
121
|
+
}
|
|
122
|
+
baseContent = normalizedMarkdown;
|
|
123
|
+
return {
|
|
124
|
+
requested: true,
|
|
125
|
+
used: true,
|
|
126
|
+
provider: 'llm',
|
|
127
|
+
notes: markdownMode === 'readability' && readabilityHtml
|
|
128
|
+
? 'Readability HTML used for markdown input'
|
|
129
|
+
: null,
|
|
130
|
+
};
|
|
131
|
+
}
|
|
132
|
+
catch (error) {
|
|
133
|
+
const message = error instanceof Error ? error.message : String(error);
|
|
134
|
+
return {
|
|
135
|
+
requested: true,
|
|
136
|
+
used: false,
|
|
137
|
+
provider: null,
|
|
138
|
+
notes: `HTML→Markdown conversion failed: ${message}`,
|
|
139
|
+
};
|
|
140
|
+
}
|
|
141
|
+
})();
|
|
142
|
+
const video = detectPrimaryVideoFromHtml(html, url);
|
|
143
|
+
const isVideoOnly = !transcriptResolution.text && baseContent.length < MIN_HTML_CONTENT_CHARACTERS && video !== null;
|
|
144
|
+
return finalizeExtractedLinkContent({
|
|
145
|
+
url,
|
|
146
|
+
baseContent,
|
|
147
|
+
maxCharacters,
|
|
148
|
+
title: mergedTitle ?? title,
|
|
149
|
+
description: mergedDescription ?? description,
|
|
150
|
+
siteName,
|
|
151
|
+
transcriptResolution,
|
|
152
|
+
video,
|
|
153
|
+
isVideoOnly,
|
|
154
|
+
diagnostics: {
|
|
155
|
+
strategy: 'html',
|
|
156
|
+
firecrawl: firecrawlDiagnostics,
|
|
157
|
+
markdown: markdownDiagnostics,
|
|
158
|
+
transcript: transcriptDiagnostics,
|
|
159
|
+
},
|
|
160
|
+
});
|
|
161
|
+
}
|
|
162
|
+
//# sourceMappingURL=html.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"html.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/html.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,wBAAwB,EAAE,MAAM,2BAA2B,CAAA;AACpE,OAAO,EAAE,qBAAqB,EAAE,YAAY,EAAE,iBAAiB,EAAE,MAAM,2BAA2B,CAAA;AAGlG,OAAO,EAAE,qBAAqB,EAAE,iCAAiC,EAAE,MAAM,cAAc,CAAA;AACvF,OAAO,EAAE,kBAAkB,EAAE,MAAM,cAAc,CAAA;AACjD,OAAO,EACL,2BAA2B,EAC3B,mCAAmC,EACnC,kCAAkC,EAClC,8BAA8B,GAC/B,MAAM,gBAAgB,CAAA;AACvB,OAAO,EAAE,oBAAoB,EAAE,MAAM,aAAa,CAAA;AAClD,OAAO,EAAE,uBAAuB,EAAE,MAAM,cAAc,CAAA;AACtD,OAAO,EAAE,aAAa,EAAE,uBAAuB,EAAE,MAAM,oBAAoB,CAAA;AAC3E,OAAO,EAAE,0BAA0B,EAAE,iBAAiB,EAAE,MAAM,kBAAkB,CAAA;AAEhF,OAAO,EACL,2BAA2B,EAC3B,4BAA4B,EAC5B,aAAa,EACb,iBAAiB,GAClB,MAAM,YAAY,CAAA;AACnB,OAAO,EAAE,0BAA0B,EAAE,MAAM,YAAY,CAAA;AACvD,OAAO,EAAE,8BAA8B,EAAE,MAAM,cAAc,CAAA;AAE7D,MAAM,uBAAuB,GAAG,eAAe,CAAA;AAE/C,SAAS,iBAAiB,CAAC,OAAe,EAAE,KAAgC;IAC1E,IAAI,CAAC,CAAC,OAAO,IAAI,KAAK,CAAC,EAAE,CAAC;QACxB,OAAO,OAAO,CAAA;IAChB,CAAC;IAED,MAAM,eAAe,GAAG,KAAK,CAAC,IAAI,EAAE,CAAA;IACpC,IAAI,eAAe,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACjC,OAAO,OAAO,CAAA;IAChB,CAAC;IAED,MAAM,cAAc,GAAG,OAAO,CAAC,SAAS,EAAE,CAAA;IAC1C,IAAI,CAAC,cAAc,CAAC,WAAW,EAAE,CAAC,UAAU,CAAC,eAAe,CAAC,WAAW,EAAE,CAAC,EAAE,CAAC;QAC5E,OAAO,OAAO,CAAA;IAChB,CAAC;IAED,MAAM,iBAAiB,GAAG,cAAc,CAAC,KAAK,CAAC,eAAe,CAAC,MAAM,CAAC,CAAA;IACtE,MAAM,SAAS,GAAG,iBAAiB,CAAC,OAAO,CAAC,uBAAuB,EAAE,EAAE,CAAC,CAAA;IACxE,OAAO,SAAS,CAAA;AAClB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,2BAA2B,CAAC,EAChD,GAAG,EACH,IAAI,EACJ,SAAS,EACT,aAAa,EACb,qBAAqB,EACrB,oBAAoB,EACpB,iBAAiB,EACjB,YAAY,EACZ,SAAS,EACT,IAAI,EACJ,oBAAoB,GAarB;IACC,IAAI,iBAAiB,CAAC,GAAG,CAAC,IAAI,CAAC,qBAAqB,CAAC,GAAG,CAAC,EAAE,CAAC;QAC1D,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAA;IACpD,CAAC;IAED,MAAM,EAAE,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,GAAG,uBAAuB,CAAC,IAAI,EAAE,GAAG,CAAC,CAAA;IAC3E,MAAM,MAAM,GAAG,oBAAoB,CAAC,IAAI,CAAC,CAAA;IACzC,MAAM,WAAW,GAAG,aAAa,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAA;IACzD,MAAM,iBAAiB,GAAG,aAAa,CAAC,CAAC,MAAM,EAAE,WAAW,EAAE,WAAW,CAAC,CAAC,CAAA;IAC3E,MAAM,eAAe,GAAG,uBAAuB,CAAC,MAAM,EAAE,IAAI,CAAC,CAAA;IAC7D,MAAM,WAAW,GAAG,oBAAoB,IAAI,CAAC,MAAM,0BAA0B,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC,CAAA;IACzF,MAAM,eAAe,GAAG,WAAW,EAAE,IAAI,CAAC,CAAC,CAAC,kBAAkB,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;IACrF,MAAM,eAAe,GAAG,iBAAiB,CAAC,WAAW,CAAC,CAAA;IAEtD,MAAM,0BAA0B,GAAG,kBAAkB,CAAC,qBAAqB,CAAC,IAAI,CAAC,CAAC,CAAA;IAClF,MAAM,qCAAqC,GAAG,eAAe;QAC3D,CAAC,CAAC,kBAAkB,CAAC,qBAAqB,CAAC,eAAe,CAAC,CAAC;QAC5D,CAAC,CAAC,EAAE,CAAA;IACN,MAAM,qBAAqB,GACzB,qCAAqC,CAAC,MAAM,IAAI,kCAAkC;QAClF,CAAC,0BAA0B,CAAC,MAAM,GAAG,2BAA2B;YAC9D,qCAAqC,CAAC,MAAM;gBAC1C,0BAA0B,CAAC,MAAM,GAAG,8BAA8B,CAAC,CAAA;IACzE,MAAM,kBAAkB,GAAG,qBAAqB;QAC9C,CAAC,CAAC,qCAAqC;QACvC,CAAC,CAAC,0BAA0B,CAAA;IAE9B,MAAM,qBAAqB,GACzB,CAAC,qBAAqB;QACtB,eAAe,CAAC,MAAM,IAAI,kCAAkC;QAC5D,CAAC,0BAA0B,CAAC,MAAM,GAAG,2BAA2B;YAC9D,eAAe,CAAC,MAAM,IAAI,0BAA0B,CAAC,MAAM,GAAG,8BAA8B,CAAC,CAAA;IACjG,MAAM,iBAAiB,GAAG,qBAAqB,IAAI,qBAAqB,CAAA;IACxE,MAAM,mBAAmB,GAAG,qBAAqB,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,kBAAkB,CAAA;IACxF,MAAM,oBAAoB,GAAG,iBAAiB,CAAC,CAAC,CAAC,kBAAkB,CAAC,iBAAiB,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;IAC3F,MAAM,iBAAiB,GACrB,oBAAoB,CAAC,MAAM,IAAI,mCAAmC;QAClE,CAAC,eAAe;YACd,aAAa,CAAC,GAAG,CAAC;YAClB,CAAC,CAAC,iBAAiB;gBACjB,CAAC,mBAAmB,CAAC,MAAM,GAAG,2BAA2B;oBACvD,oBAAoB,CAAC,MAAM;wBACzB,mBAAmB,CAAC,MAAM,GAAG,8BAA8B,CAAC,CAAC,CAAC,CAAA;IACxE,MAAM,kCAAkC,GAAG,iBAAiB;QAC1D,CAAC,CAAC,oBAAoB;QACtB,CAAC,CAAC,mBAAmB,CAAA;IACvB,MAAM,oBAAoB,GAAG,MAAM,wBAAwB,CAAC,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE;QAC3E,qBAAqB;QACrB,SAAS;KACV,CAAC,CAAA;IAEF,MAAM,kBAAkB,GACtB,oBAAoB,CAAC,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,8BAA8B,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;IAClF,MAAM,aAAa,GAAG,kBAAkB;QACtC,CAAC,CAAC,kBAAkB,CAAC,kBAAkB,CAAC;QACxC,CAAC,CAAC,kCAAkC,CAAA;IAEtC,IAAI,WAAW,GAAG,iBAAiB,CAAC,aAAa,EAAE,oBAAoB,CAAC,IAAI,CAAC,CAAA;IAC7E,IAAI,WAAW,KAAK,kBAAkB,EAAE,CAAC;QACvC,WAAW,GAAG,iBAAiB,CAAC,WAAW,EAAE,WAAW,IAAI,KAAK,CAAC,CAAA;IACpE,CAAC;IAED,MAAM,qBAAqB,GAAG,2BAA2B,CACvD,oBAAoB,EACpB,SAAS,IAAI,SAAS,CACvB,CAAA;IAED,MAAM,mBAAmB,GAAwB,MAAM,CAAC,KAAK,IAAI,EAAE;QACjE,IAAI,CAAC,iBAAiB,EAAE,CAAC;YACvB,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,CAAA;QACvE,CAAC;QAED,IAAI,YAAY,CAAC,GAAG,CAAC,EAAE,CAAC;YACtB,OAAO;gBACL,SAAS,EAAE,IAAI;gBACf,IAAI,EAAE,KAAK;gBACX,QAAQ,EAAE,IAAI;gBACd,KAAK,EAAE,+CAA+C;aACvD,CAAA;QACH,CAAC;QAED,IAAI,CAAC,IAAI,CAAC,qBAAqB,EAAE,CAAC;YAChC,OAAO;gBACL,SAAS,EAAE,IAAI;gBACf,IAAI,EAAE,KAAK;gBACX,QAAQ,EAAE,IAAI;gBACd,KAAK,EAAE,uCAAuC;aAC/C,CAAA;QACH,CAAC;QAED,IAAI,CAAC;YACH,MAAM,eAAe,GACnB,YAAY,KAAK,aAAa,IAAI,eAAe,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,IAAI,CAAA;YAC5E,MAAM,aAAa,GAAG,iCAAiC,CAAC,eAAe,CAAC,CAAA;YACxE,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,qBAAqB,CAAC;gBAChD,GAAG;gBACH,IAAI,EAAE,aAAa;gBACnB,KAAK,EAAE,WAAW,IAAI,KAAK;gBAC3B,QAAQ;gBACR,SAAS;aACV,CAAC,CAAA;YACF,MAAM,kBAAkB,GAAG,kBAAkB,CAAC,QAAQ,CAAC,CAAA;YACvD,IAAI,kBAAkB,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBACpC,OAAO;oBACL,SAAS,EAAE,IAAI;oBACf,IAAI,EAAE,KAAK;oBACX,QAAQ,EAAE,IAAI;oBACd,KAAK,EAAE,iDAAiD;iBACzD,CAAA;YACH,CAAC;YAED,WAAW,GAAG,kBAAkB,CAAA;YAChC,OAAO;gBACL,SAAS,EAAE,IAAI;gBACf,IAAI,EAAE,IAAI;gBACV,QAAQ,EAAE,KAAK;gBACf,KAAK,EACH,YAAY,KAAK,aAAa,IAAI,eAAe;oBAC/C,CAAC,CAAC,0CAA0C;oBAC5C,CAAC,CAAC,IAAI;aACX,CAAA;QACH,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,MAAM,OAAO,GAAG,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAA;YACtE,OAAO;gBACL,SAAS,EAAE,IAAI;gBACf,IAAI,EAAE,KAAK;gBACX,QAAQ,EAAE,IAAI;gBACd,KAAK,EAAE,oCAAoC,OAAO,EAAE;aACrD,CAAA;QACH,CAAC;IACH,CAAC,CAAC,EAAE,CAAA;IAEJ,MAAM,KAAK,GAAG,0BAA0B,CAAC,IAAI,EAAE,GAAG,CAAC,CAAA;IACnD,MAAM,WAAW,GACf,CAAC,oBAAoB,CAAC,IAAI,IAAI,WAAW,CAAC,MAAM,GAAG,2BAA2B,IAAI,KAAK,KAAK,IAAI,CAAA;IAElG,OAAO,4BAA4B,CAAC;QAClC,GAAG;QACH,WAAW;QACX,aAAa;QACb,KAAK,EAAE,WAAW,IAAI,KAAK;QAC3B,WAAW,EAAE,iBAAiB,IAAI,WAAW;QAC7C,QAAQ;QACR,oBAAoB;QACpB,KAAK;QACL,WAAW;QACX,WAAW,EAAE;YACX,QAAQ,EAAE,MAAM;YAChB,SAAS,EAAE,oBAAoB;YAC/B,QAAQ,EAAE,mBAAmB;YAC7B,UAAU,EAAE,qBAAqB;SAClC;KACF,CAAC,CAAA;AACJ,CAAC"}
|
|
@@ -0,0 +1,345 @@
|
|
|
1
|
+
import { resolveTranscriptForLink } from '../../transcript/index.js';
|
|
2
|
+
import { isYouTubeUrl } from '../../transcript/utils.js';
|
|
3
|
+
import { normalizeForPrompt } from './cleaner.js';
|
|
4
|
+
import { MIN_READABILITY_CONTENT_CHARACTERS } from './constants.js';
|
|
5
|
+
import { fetchHtmlDocument, fetchWithFirecrawl } from './fetcher.js';
|
|
6
|
+
import { buildResultFromFirecrawl, shouldFallbackToFirecrawl } from './firecrawl.js';
|
|
7
|
+
import { buildResultFromHtmlDocument } from './html.js';
|
|
8
|
+
import { extractApplePodcastIds, extractSpotifyEpisodeId } from './podcast-utils.js';
|
|
9
|
+
import { extractReadabilityFromHtml } from './readability.js';
|
|
10
|
+
import { isAnubisHtml, isBlockedTwitterContent, isTwitterStatusUrl, toNitterUrls, } from './twitter-utils.js';
|
|
11
|
+
import { appendNote, ensureTranscriptDiagnostics, finalizeExtractedLinkContent, resolveCacheMode, resolveFirecrawlMode, resolveMaxCharacters, resolveTimeoutMs, selectBaseContent, } from './utils.js';
|
|
12
|
+
export async function fetchLinkContent(url, options, deps) {
|
|
13
|
+
const timeoutMs = resolveTimeoutMs(options);
|
|
14
|
+
const cacheMode = resolveCacheMode(options);
|
|
15
|
+
const maxCharacters = resolveMaxCharacters(options);
|
|
16
|
+
const youtubeTranscriptMode = options?.youtubeTranscript ?? 'auto';
|
|
17
|
+
const firecrawlMode = resolveFirecrawlMode(options);
|
|
18
|
+
const markdownRequested = (options?.format ?? 'text') === 'markdown';
|
|
19
|
+
const markdownMode = options?.markdownMode ?? 'auto';
|
|
20
|
+
const canUseFirecrawl = firecrawlMode !== 'off' && deps.scrapeWithFirecrawl !== null && !isYouTubeUrl(url);
|
|
21
|
+
const spotifyEpisodeId = extractSpotifyEpisodeId(url);
|
|
22
|
+
if (spotifyEpisodeId) {
|
|
23
|
+
if (!deps.openaiApiKey && !deps.falApiKey) {
|
|
24
|
+
throw new Error('Spotify episode transcription requires OPENAI_API_KEY or FAL_KEY (Whisper); otherwise you may only get a captcha/recaptcha HTML page.');
|
|
25
|
+
}
|
|
26
|
+
const transcriptResolution = await resolveTranscriptForLink(url, null, deps, {
|
|
27
|
+
youtubeTranscriptMode,
|
|
28
|
+
cacheMode,
|
|
29
|
+
});
|
|
30
|
+
if (!transcriptResolution.text) {
|
|
31
|
+
const notes = transcriptResolution.diagnostics?.notes;
|
|
32
|
+
const suffix = notes ? ` (${notes})` : '';
|
|
33
|
+
throw new Error(`Failed to transcribe Spotify episode${suffix}`);
|
|
34
|
+
}
|
|
35
|
+
const transcriptDiagnostics = ensureTranscriptDiagnostics(transcriptResolution, cacheMode ?? 'default');
|
|
36
|
+
transcriptDiagnostics.notes = appendNote(transcriptDiagnostics.notes, 'Spotify episode: skipped HTML fetch to avoid captcha pages');
|
|
37
|
+
return finalizeExtractedLinkContent({
|
|
38
|
+
url,
|
|
39
|
+
baseContent: selectBaseContent('', transcriptResolution.text),
|
|
40
|
+
maxCharacters,
|
|
41
|
+
title: null,
|
|
42
|
+
description: null,
|
|
43
|
+
siteName: 'Spotify',
|
|
44
|
+
transcriptResolution,
|
|
45
|
+
video: null,
|
|
46
|
+
isVideoOnly: false,
|
|
47
|
+
diagnostics: {
|
|
48
|
+
strategy: 'html',
|
|
49
|
+
firecrawl: {
|
|
50
|
+
attempted: false,
|
|
51
|
+
used: false,
|
|
52
|
+
cacheMode,
|
|
53
|
+
cacheStatus: cacheMode === 'bypass' ? 'bypassed' : 'unknown',
|
|
54
|
+
notes: 'Spotify short-circuit skipped HTML/Firecrawl',
|
|
55
|
+
},
|
|
56
|
+
markdown: {
|
|
57
|
+
requested: markdownRequested,
|
|
58
|
+
used: false,
|
|
59
|
+
provider: null,
|
|
60
|
+
notes: 'Spotify short-circuit uses transcript content',
|
|
61
|
+
},
|
|
62
|
+
transcript: transcriptDiagnostics,
|
|
63
|
+
},
|
|
64
|
+
});
|
|
65
|
+
}
|
|
66
|
+
const appleIds = extractApplePodcastIds(url);
|
|
67
|
+
if (appleIds) {
|
|
68
|
+
if (!deps.openaiApiKey && !deps.falApiKey) {
|
|
69
|
+
throw new Error('Apple Podcasts transcription requires OPENAI_API_KEY or FAL_KEY (Whisper); otherwise you may only get a slow/blocked HTML page.');
|
|
70
|
+
}
|
|
71
|
+
const transcriptResolution = await resolveTranscriptForLink(url, null, deps, {
|
|
72
|
+
youtubeTranscriptMode,
|
|
73
|
+
cacheMode,
|
|
74
|
+
});
|
|
75
|
+
if (!transcriptResolution.text) {
|
|
76
|
+
const notes = transcriptResolution.diagnostics?.notes;
|
|
77
|
+
const suffix = notes ? ` (${notes})` : '';
|
|
78
|
+
throw new Error(`Failed to transcribe Apple Podcasts episode${suffix}`);
|
|
79
|
+
}
|
|
80
|
+
const transcriptDiagnostics = ensureTranscriptDiagnostics(transcriptResolution, cacheMode ?? 'default');
|
|
81
|
+
transcriptDiagnostics.notes = appendNote(transcriptDiagnostics.notes, 'Apple Podcasts: skipped HTML fetch (prefer iTunes lookup / enclosures)');
|
|
82
|
+
return finalizeExtractedLinkContent({
|
|
83
|
+
url,
|
|
84
|
+
baseContent: selectBaseContent('', transcriptResolution.text),
|
|
85
|
+
maxCharacters,
|
|
86
|
+
title: null,
|
|
87
|
+
description: null,
|
|
88
|
+
siteName: 'Apple Podcasts',
|
|
89
|
+
transcriptResolution,
|
|
90
|
+
video: null,
|
|
91
|
+
isVideoOnly: false,
|
|
92
|
+
diagnostics: {
|
|
93
|
+
strategy: 'html',
|
|
94
|
+
firecrawl: {
|
|
95
|
+
attempted: false,
|
|
96
|
+
used: false,
|
|
97
|
+
cacheMode,
|
|
98
|
+
cacheStatus: cacheMode === 'bypass' ? 'bypassed' : 'unknown',
|
|
99
|
+
notes: 'Apple Podcasts short-circuit skipped HTML/Firecrawl',
|
|
100
|
+
},
|
|
101
|
+
markdown: {
|
|
102
|
+
requested: markdownRequested,
|
|
103
|
+
used: false,
|
|
104
|
+
provider: null,
|
|
105
|
+
notes: 'Apple Podcasts short-circuit uses transcript content',
|
|
106
|
+
},
|
|
107
|
+
transcript: transcriptDiagnostics,
|
|
108
|
+
},
|
|
109
|
+
});
|
|
110
|
+
}
|
|
111
|
+
let firecrawlAttempted = false;
|
|
112
|
+
let firecrawlPayload = null;
|
|
113
|
+
const firecrawlDiagnostics = {
|
|
114
|
+
attempted: false,
|
|
115
|
+
used: false,
|
|
116
|
+
cacheMode,
|
|
117
|
+
cacheStatus: cacheMode === 'bypass' ? 'bypassed' : 'unknown',
|
|
118
|
+
notes: null,
|
|
119
|
+
};
|
|
120
|
+
const twitterStatus = isTwitterStatusUrl(url);
|
|
121
|
+
const nitterUrls = twitterStatus ? toNitterUrls(url) : [];
|
|
122
|
+
let birdError = null;
|
|
123
|
+
let nitterError = null;
|
|
124
|
+
const attemptFirecrawl = async (reason) => {
|
|
125
|
+
if (!canUseFirecrawl) {
|
|
126
|
+
return null;
|
|
127
|
+
}
|
|
128
|
+
if (!firecrawlAttempted) {
|
|
129
|
+
const attempt = await fetchWithFirecrawl(url, deps.scrapeWithFirecrawl, {
|
|
130
|
+
timeoutMs,
|
|
131
|
+
cacheMode,
|
|
132
|
+
onProgress: deps.onProgress ?? null,
|
|
133
|
+
reason,
|
|
134
|
+
});
|
|
135
|
+
firecrawlAttempted = true;
|
|
136
|
+
firecrawlPayload = attempt.payload;
|
|
137
|
+
firecrawlDiagnostics.attempted = attempt.diagnostics.attempted;
|
|
138
|
+
firecrawlDiagnostics.used = attempt.diagnostics.used;
|
|
139
|
+
firecrawlDiagnostics.cacheMode = attempt.diagnostics.cacheMode;
|
|
140
|
+
firecrawlDiagnostics.cacheStatus = attempt.diagnostics.cacheStatus;
|
|
141
|
+
firecrawlDiagnostics.notes = attempt.diagnostics.notes ?? null;
|
|
142
|
+
}
|
|
143
|
+
firecrawlDiagnostics.notes = appendNote(firecrawlDiagnostics.notes, reason);
|
|
144
|
+
if (!firecrawlPayload) {
|
|
145
|
+
return null;
|
|
146
|
+
}
|
|
147
|
+
const firecrawlResult = await buildResultFromFirecrawl({
|
|
148
|
+
url,
|
|
149
|
+
payload: firecrawlPayload,
|
|
150
|
+
cacheMode,
|
|
151
|
+
maxCharacters,
|
|
152
|
+
youtubeTranscriptMode,
|
|
153
|
+
firecrawlDiagnostics,
|
|
154
|
+
markdownRequested,
|
|
155
|
+
deps,
|
|
156
|
+
});
|
|
157
|
+
if (firecrawlResult) {
|
|
158
|
+
return firecrawlResult;
|
|
159
|
+
}
|
|
160
|
+
firecrawlDiagnostics.notes = appendNote(firecrawlDiagnostics.notes, 'Firecrawl returned empty content');
|
|
161
|
+
return null;
|
|
162
|
+
};
|
|
163
|
+
const attemptBird = async () => {
|
|
164
|
+
if (!deps.readTweetWithBird || !twitterStatus) {
|
|
165
|
+
return null;
|
|
166
|
+
}
|
|
167
|
+
deps.onProgress?.({ kind: 'bird-start', url });
|
|
168
|
+
try {
|
|
169
|
+
const tweet = await deps.readTweetWithBird({ url, timeoutMs });
|
|
170
|
+
const text = tweet?.text?.trim() ?? '';
|
|
171
|
+
if (text.length === 0) {
|
|
172
|
+
deps.onProgress?.({ kind: 'bird-done', url, ok: false, textBytes: null });
|
|
173
|
+
return null;
|
|
174
|
+
}
|
|
175
|
+
const title = tweet?.author?.username ? `@${tweet.author.username}` : null;
|
|
176
|
+
const description = null;
|
|
177
|
+
const siteName = 'X';
|
|
178
|
+
const transcriptResolution = { text: null, source: null };
|
|
179
|
+
const transcriptDiagnostics = ensureTranscriptDiagnostics(transcriptResolution, cacheMode ?? 'default');
|
|
180
|
+
const result = finalizeExtractedLinkContent({
|
|
181
|
+
url,
|
|
182
|
+
baseContent: text,
|
|
183
|
+
maxCharacters,
|
|
184
|
+
title,
|
|
185
|
+
description,
|
|
186
|
+
siteName,
|
|
187
|
+
transcriptResolution,
|
|
188
|
+
video: null,
|
|
189
|
+
isVideoOnly: false,
|
|
190
|
+
diagnostics: {
|
|
191
|
+
strategy: 'bird',
|
|
192
|
+
firecrawl: firecrawlDiagnostics,
|
|
193
|
+
markdown: {
|
|
194
|
+
requested: markdownRequested,
|
|
195
|
+
used: false,
|
|
196
|
+
provider: null,
|
|
197
|
+
notes: 'Bird tweet fetch provides plain text',
|
|
198
|
+
},
|
|
199
|
+
transcript: transcriptDiagnostics,
|
|
200
|
+
},
|
|
201
|
+
});
|
|
202
|
+
deps.onProgress?.({
|
|
203
|
+
kind: 'bird-done',
|
|
204
|
+
url,
|
|
205
|
+
ok: true,
|
|
206
|
+
textBytes: Buffer.byteLength(result.content, 'utf8'),
|
|
207
|
+
});
|
|
208
|
+
return result;
|
|
209
|
+
}
|
|
210
|
+
catch (error) {
|
|
211
|
+
birdError = error;
|
|
212
|
+
deps.onProgress?.({ kind: 'bird-done', url, ok: false, textBytes: null });
|
|
213
|
+
return null;
|
|
214
|
+
}
|
|
215
|
+
};
|
|
216
|
+
const birdResult = await attemptBird();
|
|
217
|
+
if (birdResult) {
|
|
218
|
+
return birdResult;
|
|
219
|
+
}
|
|
220
|
+
const attemptNitter = async () => {
|
|
221
|
+
if (nitterUrls.length === 0) {
|
|
222
|
+
return null;
|
|
223
|
+
}
|
|
224
|
+
for (const nitterUrl of nitterUrls) {
|
|
225
|
+
deps.onProgress?.({ kind: 'nitter-start', url: nitterUrl });
|
|
226
|
+
try {
|
|
227
|
+
const nitterHtml = await fetchHtmlDocument(deps.fetch, nitterUrl, { timeoutMs });
|
|
228
|
+
if (!nitterHtml.trim()) {
|
|
229
|
+
nitterError = new Error(`Nitter returned empty body from ${new URL(nitterUrl).host}`);
|
|
230
|
+
deps.onProgress?.({ kind: 'nitter-done', url: nitterUrl, ok: false, textBytes: null });
|
|
231
|
+
continue;
|
|
232
|
+
}
|
|
233
|
+
if (isAnubisHtml(nitterHtml)) {
|
|
234
|
+
nitterError = new Error(`Nitter returned Anubis challenge from ${new URL(nitterUrl).host}`);
|
|
235
|
+
deps.onProgress?.({ kind: 'nitter-done', url: nitterUrl, ok: false, textBytes: null });
|
|
236
|
+
continue;
|
|
237
|
+
}
|
|
238
|
+
deps.onProgress?.({
|
|
239
|
+
kind: 'nitter-done',
|
|
240
|
+
url: nitterUrl,
|
|
241
|
+
ok: true,
|
|
242
|
+
textBytes: Buffer.byteLength(nitterHtml, 'utf8'),
|
|
243
|
+
});
|
|
244
|
+
return nitterHtml;
|
|
245
|
+
}
|
|
246
|
+
catch (error) {
|
|
247
|
+
nitterError = error;
|
|
248
|
+
deps.onProgress?.({ kind: 'nitter-done', url: nitterUrl, ok: false, textBytes: null });
|
|
249
|
+
}
|
|
250
|
+
}
|
|
251
|
+
return null;
|
|
252
|
+
};
|
|
253
|
+
const nitterHtml = await attemptNitter();
|
|
254
|
+
if (nitterHtml) {
|
|
255
|
+
const nitterResult = await buildResultFromHtmlDocument({
|
|
256
|
+
url,
|
|
257
|
+
html: nitterHtml,
|
|
258
|
+
cacheMode,
|
|
259
|
+
maxCharacters,
|
|
260
|
+
youtubeTranscriptMode,
|
|
261
|
+
firecrawlDiagnostics,
|
|
262
|
+
markdownRequested,
|
|
263
|
+
markdownMode,
|
|
264
|
+
timeoutMs,
|
|
265
|
+
deps,
|
|
266
|
+
readabilityCandidate: null,
|
|
267
|
+
});
|
|
268
|
+
if (!isBlockedTwitterContent(nitterResult.content)) {
|
|
269
|
+
nitterResult.diagnostics.strategy = 'nitter';
|
|
270
|
+
return nitterResult;
|
|
271
|
+
}
|
|
272
|
+
nitterError = new Error('Nitter returned blocked or empty content');
|
|
273
|
+
}
|
|
274
|
+
if (firecrawlMode === 'always') {
|
|
275
|
+
const firecrawlResult = await attemptFirecrawl('Firecrawl forced via options');
|
|
276
|
+
if (firecrawlResult) {
|
|
277
|
+
return firecrawlResult;
|
|
278
|
+
}
|
|
279
|
+
}
|
|
280
|
+
let html = null;
|
|
281
|
+
let htmlError = null;
|
|
282
|
+
try {
|
|
283
|
+
html = await fetchHtmlDocument(deps.fetch, url, {
|
|
284
|
+
timeoutMs,
|
|
285
|
+
onProgress: deps.onProgress ?? null,
|
|
286
|
+
});
|
|
287
|
+
}
|
|
288
|
+
catch (error) {
|
|
289
|
+
htmlError = error;
|
|
290
|
+
}
|
|
291
|
+
if (!html) {
|
|
292
|
+
if (!canUseFirecrawl) {
|
|
293
|
+
throw htmlError instanceof Error ? htmlError : new Error('Failed to fetch HTML document');
|
|
294
|
+
}
|
|
295
|
+
const firecrawlResult = await attemptFirecrawl('HTML fetch failed; falling back to Firecrawl');
|
|
296
|
+
if (firecrawlResult) {
|
|
297
|
+
return firecrawlResult;
|
|
298
|
+
}
|
|
299
|
+
const firecrawlError = firecrawlDiagnostics.notes
|
|
300
|
+
? `; Firecrawl notes: ${firecrawlDiagnostics.notes}`
|
|
301
|
+
: '';
|
|
302
|
+
throw new Error(`Failed to fetch HTML document${firecrawlError}${htmlError instanceof Error ? `; HTML error: ${htmlError.message}` : ''}`);
|
|
303
|
+
}
|
|
304
|
+
let readabilityCandidate = null;
|
|
305
|
+
if (firecrawlMode === 'auto' && shouldFallbackToFirecrawl(html)) {
|
|
306
|
+
readabilityCandidate = await extractReadabilityFromHtml(html, url);
|
|
307
|
+
const readabilityText = readabilityCandidate?.text
|
|
308
|
+
? normalizeForPrompt(readabilityCandidate.text)
|
|
309
|
+
: '';
|
|
310
|
+
if (readabilityText.length < MIN_READABILITY_CONTENT_CHARACTERS) {
|
|
311
|
+
const firecrawlResult = await attemptFirecrawl('HTML content looked blocked/thin; falling back to Firecrawl');
|
|
312
|
+
if (firecrawlResult) {
|
|
313
|
+
return firecrawlResult;
|
|
314
|
+
}
|
|
315
|
+
}
|
|
316
|
+
}
|
|
317
|
+
const htmlResult = await buildResultFromHtmlDocument({
|
|
318
|
+
url,
|
|
319
|
+
html,
|
|
320
|
+
cacheMode,
|
|
321
|
+
maxCharacters,
|
|
322
|
+
youtubeTranscriptMode,
|
|
323
|
+
firecrawlDiagnostics,
|
|
324
|
+
markdownRequested,
|
|
325
|
+
markdownMode,
|
|
326
|
+
timeoutMs,
|
|
327
|
+
deps,
|
|
328
|
+
readabilityCandidate,
|
|
329
|
+
});
|
|
330
|
+
if (twitterStatus && isBlockedTwitterContent(htmlResult.content)) {
|
|
331
|
+
const birdNote = !deps.readTweetWithBird
|
|
332
|
+
? 'Bird not available'
|
|
333
|
+
: birdError
|
|
334
|
+
? `Bird failed: ${birdError instanceof Error ? birdError.message : String(birdError)}`
|
|
335
|
+
: 'Bird returned no text';
|
|
336
|
+
const nitterNote = nitterUrls.length > 0
|
|
337
|
+
? nitterError
|
|
338
|
+
? `Nitter failed: ${nitterError instanceof Error ? nitterError.message : String(nitterError)}`
|
|
339
|
+
: 'Nitter returned no text'
|
|
340
|
+
: 'Nitter not available';
|
|
341
|
+
throw new Error(`Unable to fetch tweet content from X. ${birdNote}. ${nitterNote}.`);
|
|
342
|
+
}
|
|
343
|
+
return htmlResult;
|
|
344
|
+
}
|
|
345
|
+
//# sourceMappingURL=index.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,wBAAwB,EAAE,MAAM,2BAA2B,CAAA;AACpE,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AAGxD,OAAO,EAAE,kBAAkB,EAAE,MAAM,cAAc,CAAA;AACjD,OAAO,EAAE,kCAAkC,EAAE,MAAM,gBAAgB,CAAA;AACnE,OAAO,EAAE,iBAAiB,EAAE,kBAAkB,EAAE,MAAM,cAAc,CAAA;AACpE,OAAO,EAAE,wBAAwB,EAAE,yBAAyB,EAAE,MAAM,gBAAgB,CAAA;AACpF,OAAO,EAAE,2BAA2B,EAAE,MAAM,WAAW,CAAA;AACvD,OAAO,EAAE,sBAAsB,EAAE,uBAAuB,EAAE,MAAM,oBAAoB,CAAA;AACpF,OAAO,EAAE,0BAA0B,EAAE,MAAM,kBAAkB,CAAA;AAC7D,OAAO,EACL,YAAY,EACZ,uBAAuB,EACvB,kBAAkB,EAClB,YAAY,GACb,MAAM,oBAAoB,CAAA;AAE3B,OAAO,EACL,UAAU,EACV,2BAA2B,EAC3B,4BAA4B,EAC5B,gBAAgB,EAChB,oBAAoB,EACpB,oBAAoB,EACpB,gBAAgB,EAChB,iBAAiB,GAClB,MAAM,YAAY,CAAA;AAEnB,MAAM,CAAC,KAAK,UAAU,gBAAgB,CACpC,GAAW,EACX,OAA4C,EAC5C,IAAqB;IAErB,MAAM,SAAS,GAAG,gBAAgB,CAAC,OAAO,CAAC,CAAA;IAC3C,MAAM,SAAS,GAAG,gBAAgB,CAAC,OAAO,CAAC,CAAA;IAC3C,MAAM,aAAa,GAAG,oBAAoB,CAAC,OAAO,CAAC,CAAA;IACnD,MAAM,qBAAqB,GAAG,OAAO,EAAE,iBAAiB,IAAI,MAAM,CAAA;IAClE,MAAM,aAAa,GAAG,oBAAoB,CAAC,OAAO,CAAC,CAAA;IACnD,MAAM,iBAAiB,GAAG,CAAC,OAAO,EAAE,MAAM,IAAI,MAAM,CAAC,KAAK,UAAU,CAAA;IACpE,MAAM,YAAY,GAAiB,OAAO,EAAE,YAAY,IAAI,MAAM,CAAA;IAElE,MAAM,eAAe,GACnB,aAAa,KAAK,KAAK,IAAI,IAAI,CAAC,mBAAmB,KAAK,IAAI,IAAI,CAAC,YAAY,CAAC,GAAG,CAAC,CAAA;IAEpF,MAAM,gBAAgB,GAAG,uBAAuB,CAAC,GAAG,CAAC,CAAA;IACrD,IAAI,gBAAgB,EAAE,CAAC;QACrB,IAAI,CAAC,IAAI,CAAC,YAAY,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,CAAC;YAC1C,MAAM,IAAI,KAAK,CACb,uIAAuI,CACxI,CAAA;QACH,CAAC;QAED,MAAM,oBAAoB,GAAG,MAAM,wBAAwB,CAAC,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE;YAC3E,qBAAqB;YACrB,SAAS;SACV,CAAC,CAAA;QACF,IAAI,CAAC,oBAAoB,CAAC,IAAI,EAAE,CAAC;YAC/B,MAAM,KAAK,GAAG,oBAAoB,CAAC,WAAW,EAAE,KAAK,CAAA;YACrD,MAAM,MAAM,GAAG,KAAK,CAAC,CAAC,CAAC,KAAK,KAAK,GAAG,CAAC,CAAC,CAAC,EAAE,CAAA;YACzC,MAAM,IAAI,KAAK,CAAC,uCAAuC,MAAM,EAAE,CAAC,CAAA;QAClE,CAAC;QAED,MAAM,qBAAqB,GAAG,2BAA2B,CACvD,oBAAoB,EACpB,SAAS,IAAI,SAAS,CACvB,CAAA;QACD,qBAAqB,CAAC,KAAK,GAAG,UAAU,CACtC,qBAAqB,CAAC,KAAK,EAC3B,4DAA4D,CAC7D,CAAA;QAED,OAAO,4BAA4B,CAAC;YAClC,GAAG;YACH,WAAW,EAAE,iBAAiB,CAAC,EAAE,EAAE,oBAAoB,CAAC,IAAI,CAAC;YAC7D,aAAa;YACb,KAAK,EAAE,IAAI;YACX,WAAW,EAAE,IAAI;YACjB,QAAQ,EAAE,SAAS;YACnB,oBAAoB;YACpB,KAAK,EAAE,IAAI;YACX,WAAW,EAAE,KAAK;YAClB,WAAW,EAAE;gBACX,QAAQ,EAAE,MAAM;gBAChB,SAAS,EAAE;oBACT,SAAS,EAAE,KAAK;oBAChB,IAAI,EAAE,KAAK;oBACX,SAAS;oBACT,WAAW,EAAE,SAAS,KAAK,QAAQ,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,SAAS;oBAC5D,KAAK,EAAE,8CAA8C;iBACtD;gBACD,QAAQ,EAAE;oBACR,SAAS,EAAE,iBAAiB;oBAC5B,IAAI,EAAE,KAAK;oBACX,QAAQ,EAAE,IAAI;oBACd,KAAK,EAAE,+CAA+C;iBACvD;gBACD,UAAU,EAAE,qBAAqB;aAClC;SACF,CAAC,CAAA;IACJ,CAAC;IAED,MAAM,QAAQ,GAAG,sBAAsB,CAAC,GAAG,CAAC,CAAA;IAC5C,IAAI,QAAQ,EAAE,CAAC;QACb,IAAI,CAAC,IAAI,CAAC,YAAY,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,CAAC;YAC1C,MAAM,IAAI,KAAK,CACb,iIAAiI,CAClI,CAAA;QACH,CAAC;QAED,MAAM,oBAAoB,GAAG,MAAM,wBAAwB,CAAC,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE;YAC3E,qBAAqB;YACrB,SAAS;SACV,CAAC,CAAA;QACF,IAAI,CAAC,oBAAoB,CAAC,IAAI,EAAE,CAAC;YAC/B,MAAM,KAAK,GAAG,oBAAoB,CAAC,WAAW,EAAE,KAAK,CAAA;YACrD,MAAM,MAAM,GAAG,KAAK,CAAC,CAAC,CAAC,KAAK,KAAK,GAAG,CAAC,CAAC,CAAC,EAAE,CAAA;YACzC,MAAM,IAAI,KAAK,CAAC,8CAA8C,MAAM,EAAE,CAAC,CAAA;QACzE,CAAC;QAED,MAAM,qBAAqB,GAAG,2BAA2B,CACvD,oBAAoB,EACpB,SAAS,IAAI,SAAS,CACvB,CAAA;QACD,qBAAqB,CAAC,KAAK,GAAG,UAAU,CACtC,qBAAqB,CAAC,KAAK,EAC3B,wEAAwE,CACzE,CAAA;QAED,OAAO,4BAA4B,CAAC;YAClC,GAAG;YACH,WAAW,EAAE,iBAAiB,CAAC,EAAE,EAAE,oBAAoB,CAAC,IAAI,CAAC;YAC7D,aAAa;YACb,KAAK,EAAE,IAAI;YACX,WAAW,EAAE,IAAI;YACjB,QAAQ,EAAE,gBAAgB;YAC1B,oBAAoB;YACpB,KAAK,EAAE,IAAI;YACX,WAAW,EAAE,KAAK;YAClB,WAAW,EAAE;gBACX,QAAQ,EAAE,MAAM;gBAChB,SAAS,EAAE;oBACT,SAAS,EAAE,KAAK;oBAChB,IAAI,EAAE,KAAK;oBACX,SAAS;oBACT,WAAW,EAAE,SAAS,KAAK,QAAQ,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,SAAS;oBAC5D,KAAK,EAAE,qDAAqD;iBAC7D;gBACD,QAAQ,EAAE;oBACR,SAAS,EAAE,iBAAiB;oBAC5B,IAAI,EAAE,KAAK;oBACX,QAAQ,EAAE,IAAI;oBACd,KAAK,EAAE,sDAAsD;iBAC9D;gBACD,UAAU,EAAE,qBAAqB;aAClC;SACF,CAAC,CAAA;IACJ,CAAC;IAED,IAAI,kBAAkB,GAAG,KAAK,CAAA;IAC9B,IAAI,gBAAgB,GAAiC,IAAI,CAAA;IACzD,MAAM,oBAAoB,GAAyB;QACjD,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,MAAM,aAAa,GAAG,kBAAkB,CAAC,GAAG,CAAC,CAAA;IAC7C,MAAM,UAAU,GAAG,aAAa,CAAC,CAAC,CAAC,YAAY,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;IACzD,IAAI,SAAS,GAAY,IAAI,CAAA;IAC7B,IAAI,WAAW,GAAY,IAAI,CAAA;IAE/B,MAAM,gBAAgB,GAAG,KAAK,EAAE,MAAc,EAAwC,EAAE;QACtF,IAAI,CAAC,eAAe,EAAE,CAAC;YACrB,OAAO,IAAI,CAAA;QACb,CAAC;QAED,IAAI,CAAC,kBAAkB,EAAE,CAAC;YACxB,MAAM,OAAO,GAAG,MAAM,kBAAkB,CAAC,GAAG,EAAE,IAAI,CAAC,mBAAmB,EAAE;gBACtE,SAAS;gBACT,SAAS;gBACT,UAAU,EAAE,IAAI,CAAC,UAAU,IAAI,IAAI;gBACnC,MAAM;aACP,CAAC,CAAA;YACF,kBAAkB,GAAG,IAAI,CAAA;YACzB,gBAAgB,GAAG,OAAO,CAAC,OAAO,CAAA;YAClC,oBAAoB,CAAC,SAAS,GAAG,OAAO,CAAC,WAAW,CAAC,SAAS,CAAA;YAC9D,oBAAoB,CAAC,IAAI,GAAG,OAAO,CAAC,WAAW,CAAC,IAAI,CAAA;YACpD,oBAAoB,CAAC,SAAS,GAAG,OAAO,CAAC,WAAW,CAAC,SAAS,CAAA;YAC9D,oBAAoB,CAAC,WAAW,GAAG,OAAO,CAAC,WAAW,CAAC,WAAW,CAAA;YAClE,oBAAoB,CAAC,KAAK,GAAG,OAAO,CAAC,WAAW,CAAC,KAAK,IAAI,IAAI,CAAA;QAChE,CAAC;QAED,oBAAoB,CAAC,KAAK,GAAG,UAAU,CAAC,oBAAoB,CAAC,KAAK,EAAE,MAAM,CAAC,CAAA;QAE3E,IAAI,CAAC,gBAAgB,EAAE,CAAC;YACtB,OAAO,IAAI,CAAA;QACb,CAAC;QAED,MAAM,eAAe,GAAG,MAAM,wBAAwB,CAAC;YACrD,GAAG;YACH,OAAO,EAAE,gBAAgB;YACzB,SAAS;YACT,aAAa;YACb,qBAAqB;YACrB,oBAAoB;YACpB,iBAAiB;YACjB,IAAI;SACL,CAAC,CAAA;QACF,IAAI,eAAe,EAAE,CAAC;YACpB,OAAO,eAAe,CAAA;QACxB,CAAC;QAED,oBAAoB,CAAC,KAAK,GAAG,UAAU,CACrC,oBAAoB,CAAC,KAAK,EAC1B,kCAAkC,CACnC,CAAA;QACD,OAAO,IAAI,CAAA;IACb,CAAC,CAAA;IAED,MAAM,WAAW,GAAG,KAAK,IAA0C,EAAE;QACnE,IAAI,CAAC,IAAI,CAAC,iBAAiB,IAAI,CAAC,aAAa,EAAE,CAAC;YAC9C,OAAO,IAAI,CAAA;QACb,CAAC;QAED,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,YAAY,EAAE,GAAG,EAAE,CAAC,CAAA;QAC9C,IAAI,CAAC;YACH,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,iBAAiB,CAAC,EAAE,GAAG,EAAE,SAAS,EAAE,CAAC,CAAA;YAC9D,MAAM,IAAI,GAAG,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,CAAA;YACtC,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBACtB,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,GAAG,EAAE,EAAE,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;gBACzE,OAAO,IAAI,CAAA;YACb,CAAC;YAED,MAAM,KAAK,GAAG,KAAK,EAAE,MAAM,EAAE,QAAQ,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,MAAM,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC,IAAI,CAAA;YAC1E,MAAM,WAAW,GAAG,IAAI,CAAA;YACxB,MAAM,QAAQ,GAAG,GAAG,CAAA;YACpB,MAAM,oBAAoB,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,CAAA;YACzD,MAAM,qBAAqB,GAAG,2BAA2B,CACvD,oBAAoB,EACpB,SAAS,IAAI,SAAS,CACvB,CAAA;YACD,MAAM,MAAM,GAAG,4BAA4B,CAAC;gBAC1C,GAAG;gBACH,WAAW,EAAE,IAAI;gBACjB,aAAa;gBACb,KAAK;gBACL,WAAW;gBACX,QAAQ;gBACR,oBAAoB;gBACpB,KAAK,EAAE,IAAI;gBACX,WAAW,EAAE,KAAK;gBAClB,WAAW,EAAE;oBACX,QAAQ,EAAE,MAAM;oBAChB,SAAS,EAAE,oBAAoB;oBAC/B,QAAQ,EAAE;wBACR,SAAS,EAAE,iBAAiB;wBAC5B,IAAI,EAAE,KAAK;wBACX,QAAQ,EAAE,IAAI;wBACd,KAAK,EAAE,sCAAsC;qBAC9C;oBACD,UAAU,EAAE,qBAAqB;iBAClC;aACF,CAAC,CAAA;YACF,IAAI,CAAC,UAAU,EAAE,CAAC;gBAChB,IAAI,EAAE,WAAW;gBACjB,GAAG;gBACH,EAAE,EAAE,IAAI;gBACR,SAAS,EAAE,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,OAAO,EAAE,MAAM,CAAC;aACrD,CAAC,CAAA;YACF,OAAO,MAAM,CAAA;QACf,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,SAAS,GAAG,KAAK,CAAA;YACjB,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,GAAG,EAAE,EAAE,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;YACzE,OAAO,IAAI,CAAA;QACb,CAAC;IACH,CAAC,CAAA;IAED,MAAM,UAAU,GAAG,MAAM,WAAW,EAAE,CAAA;IACtC,IAAI,UAAU,EAAE,CAAC;QACf,OAAO,UAAU,CAAA;IACnB,CAAC;IAED,MAAM,aAAa,GAAG,KAAK,IAA4B,EAAE;QACvD,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC5B,OAAO,IAAI,CAAA;QACb,CAAC;QACD,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;YACnC,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,cAAc,EAAE,GAAG,EAAE,SAAS,EAAE,CAAC,CAAA;YAC3D,IAAI,CAAC;gBACH,MAAM,UAAU,GAAG,MAAM,iBAAiB,CAAC,IAAI,CAAC,KAAK,EAAE,SAAS,EAAE,EAAE,SAAS,EAAE,CAAC,CAAA;gBAChF,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,EAAE,CAAC;oBACvB,WAAW,GAAG,IAAI,KAAK,CAAC,mCAAmC,IAAI,GAAG,CAAC,SAAS,CAAC,CAAC,IAAI,EAAE,CAAC,CAAA;oBACrF,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,GAAG,EAAE,SAAS,EAAE,EAAE,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;oBACtF,SAAQ;gBACV,CAAC;gBACD,IAAI,YAAY,CAAC,UAAU,CAAC,EAAE,CAAC;oBAC7B,WAAW,GAAG,IAAI,KAAK,CACrB,yCAAyC,IAAI,GAAG,CAAC,SAAS,CAAC,CAAC,IAAI,EAAE,CACnE,CAAA;oBACD,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,GAAG,EAAE,SAAS,EAAE,EAAE,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;oBACtF,SAAQ;gBACV,CAAC;gBACD,IAAI,CAAC,UAAU,EAAE,CAAC;oBAChB,IAAI,EAAE,aAAa;oBACnB,GAAG,EAAE,SAAS;oBACd,EAAE,EAAE,IAAI;oBACR,SAAS,EAAE,MAAM,CAAC,UAAU,CAAC,UAAU,EAAE,MAAM,CAAC;iBACjD,CAAC,CAAA;gBACF,OAAO,UAAU,CAAA;YACnB,CAAC;YAAC,OAAO,KAAK,EAAE,CAAC;gBACf,WAAW,GAAG,KAAK,CAAA;gBACnB,IAAI,CAAC,UAAU,EAAE,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,GAAG,EAAE,SAAS,EAAE,EAAE,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;YACxF,CAAC;QACH,CAAC;QACD,OAAO,IAAI,CAAA;IACb,CAAC,CAAA;IAED,MAAM,UAAU,GAAG,MAAM,aAAa,EAAE,CAAA;IACxC,IAAI,UAAU,EAAE,CAAC;QACf,MAAM,YAAY,GAAG,MAAM,2BAA2B,CAAC;YACrD,GAAG;YACH,IAAI,EAAE,UAAU;YAChB,SAAS;YACT,aAAa;YACb,qBAAqB;YACrB,oBAAoB;YACpB,iBAAiB;YACjB,YAAY;YACZ,SAAS;YACT,IAAI;YACJ,oBAAoB,EAAE,IAAI;SAC3B,CAAC,CAAA;QACF,IAAI,CAAC,uBAAuB,CAAC,YAAY,CAAC,OAAO,CAAC,EAAE,CAAC;YACnD,YAAY,CAAC,WAAW,CAAC,QAAQ,GAAG,QAAQ,CAAA;YAC5C,OAAO,YAAY,CAAA;QACrB,CAAC;QACD,WAAW,GAAG,IAAI,KAAK,CAAC,0CAA0C,CAAC,CAAA;IACrE,CAAC;IAED,IAAI,aAAa,KAAK,QAAQ,EAAE,CAAC;QAC/B,MAAM,eAAe,GAAG,MAAM,gBAAgB,CAAC,8BAA8B,CAAC,CAAA;QAC9E,IAAI,eAAe,EAAE,CAAC;YACpB,OAAO,eAAe,CAAA;QACxB,CAAC;IACH,CAAC;IAED,IAAI,IAAI,GAAkB,IAAI,CAAA;IAC9B,IAAI,SAAS,GAAY,IAAI,CAAA;IAE7B,IAAI,CAAC;QACH,IAAI,GAAG,MAAM,iBAAiB,CAAC,IAAI,CAAC,KAAK,EAAE,GAAG,EAAE;YAC9C,SAAS;YACT,UAAU,EAAE,IAAI,CAAC,UAAU,IAAI,IAAI;SACpC,CAAC,CAAA;IACJ,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,SAAS,GAAG,KAAK,CAAA;IACnB,CAAC;IAED,IAAI,CAAC,IAAI,EAAE,CAAC;QACV,IAAI,CAAC,eAAe,EAAE,CAAC;YACrB,MAAM,SAAS,YAAY,KAAK,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,+BAA+B,CAAC,CAAA;QAC3F,CAAC;QAED,MAAM,eAAe,GAAG,MAAM,gBAAgB,CAAC,8CAA8C,CAAC,CAAA;QAC9F,IAAI,eAAe,EAAE,CAAC;YACpB,OAAO,eAAe,CAAA;QACxB,CAAC;QAED,MAAM,cAAc,GAAG,oBAAoB,CAAC,KAAK;YAC/C,CAAC,CAAC,sBAAsB,oBAAoB,CAAC,KAAK,EAAE;YACpD,CAAC,CAAC,EAAE,CAAA;QACN,MAAM,IAAI,KAAK,CACb,gCAAgC,cAAc,GAC5C,SAAS,YAAY,KAAK,CAAC,CAAC,CAAC,iBAAiB,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC,EACtE,EAAE,CACH,CAAA;IACH,CAAC;IAED,IAAI,oBAAoB,GAAkE,IAAI,CAAA;IAE9F,IAAI,aAAa,KAAK,MAAM,IAAI,yBAAyB,CAAC,IAAI,CAAC,EAAE,CAAC;QAChE,oBAAoB,GAAG,MAAM,0BAA0B,CAAC,IAAI,EAAE,GAAG,CAAC,CAAA;QAClE,MAAM,eAAe,GAAG,oBAAoB,EAAE,IAAI;YAChD,CAAC,CAAC,kBAAkB,CAAC,oBAAoB,CAAC,IAAI,CAAC;YAC/C,CAAC,CAAC,EAAE,CAAA;QACN,IAAI,eAAe,CAAC,MAAM,GAAG,kCAAkC,EAAE,CAAC;YAChE,MAAM,eAAe,GAAG,MAAM,gBAAgB,CAC5C,6DAA6D,CAC9D,CAAA;YACD,IAAI,eAAe,EAAE,CAAC;gBACpB,OAAO,eAAe,CAAA;YACxB,CAAC;QACH,CAAC;IACH,CAAC;IAED,MAAM,UAAU,GAAG,MAAM,2BAA2B,CAAC;QACnD,GAAG;QACH,IAAI;QACJ,SAAS;QACT,aAAa;QACb,qBAAqB;QACrB,oBAAoB;QACpB,iBAAiB;QACjB,YAAY;QACZ,SAAS;QACT,IAAI;QACJ,oBAAoB;KACrB,CAAC,CAAA;IACF,IAAI,aAAa,IAAI,uBAAuB,CAAC,UAAU,CAAC,OAAO,CAAC,EAAE,CAAC;QACjE,MAAM,QAAQ,GAAG,CAAC,IAAI,CAAC,iBAAiB;YACtC,CAAC,CAAC,oBAAoB;YACtB,CAAC,CAAC,SAAS;gBACT,CAAC,CAAC,gBAAgB,SAAS,YAAY,KAAK,CAAC,CAAC,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,EAAE;gBACtF,CAAC,CAAC,uBAAuB,CAAA;QAC7B,MAAM,UAAU,GACd,UAAU,CAAC,MAAM,GAAG,CAAC;YACnB,CAAC,CAAC,WAAW;gBACX,CAAC,CAAC,kBAAkB,WAAW,YAAY,KAAK,CAAC,CAAC,CAAC,WAAW,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,WAAW,CAAC,EAAE;gBAC9F,CAAC,CAAC,yBAAyB;YAC7B,CAAC,CAAC,sBAAsB,CAAA;QAC5B,MAAM,IAAI,KAAK,CAAC,yCAAyC,QAAQ,KAAK,UAAU,GAAG,CAAC,CAAA;IACtF,CAAC;IACD,OAAO,UAAU,CAAA;AACnB,CAAC"}
|
|
@@ -0,0 +1,77 @@
|
|
|
1
|
+
import { load } from 'cheerio';
|
|
2
|
+
import { normalizeCandidate } from './cleaner.js';
|
|
3
|
+
export function extractJsonLdContent(html) {
|
|
4
|
+
try {
|
|
5
|
+
const $ = load(html);
|
|
6
|
+
const scripts = $('script[type="application/ld+json"]').toArray();
|
|
7
|
+
const candidates = [];
|
|
8
|
+
for (const script of scripts) {
|
|
9
|
+
const raw = $(script).text();
|
|
10
|
+
if (!raw)
|
|
11
|
+
continue;
|
|
12
|
+
try {
|
|
13
|
+
const data = JSON.parse(raw);
|
|
14
|
+
collectCandidates(data, candidates);
|
|
15
|
+
}
|
|
16
|
+
catch {
|
|
17
|
+
// ignore malformed jsonld
|
|
18
|
+
}
|
|
19
|
+
}
|
|
20
|
+
if (candidates.length === 0)
|
|
21
|
+
return null;
|
|
22
|
+
const sorted = candidates
|
|
23
|
+
.map((c) => ({
|
|
24
|
+
title: c.title ? normalizeCandidate(c.title) : null,
|
|
25
|
+
description: c.description ? normalizeCandidate(c.description) : null,
|
|
26
|
+
type: c.type ? normalizeCandidate(c.type) : null,
|
|
27
|
+
}))
|
|
28
|
+
.filter((c) => c.title || c.description)
|
|
29
|
+
.sort((a, b) => (b.description?.length ?? 0) - (a.description?.length ?? 0));
|
|
30
|
+
return sorted[0] ?? null;
|
|
31
|
+
}
|
|
32
|
+
catch {
|
|
33
|
+
return null;
|
|
34
|
+
}
|
|
35
|
+
}
|
|
36
|
+
function collectCandidates(input, out) {
|
|
37
|
+
if (!input)
|
|
38
|
+
return;
|
|
39
|
+
if (Array.isArray(input)) {
|
|
40
|
+
for (const item of input)
|
|
41
|
+
collectCandidates(item, out);
|
|
42
|
+
return;
|
|
43
|
+
}
|
|
44
|
+
if (typeof input !== 'object')
|
|
45
|
+
return;
|
|
46
|
+
const record = input;
|
|
47
|
+
if (record['@graph'] && Array.isArray(record['@graph'])) {
|
|
48
|
+
collectCandidates(record['@graph'], out);
|
|
49
|
+
}
|
|
50
|
+
const type = extractType(record);
|
|
51
|
+
if (type) {
|
|
52
|
+
const title = firstString(record, ['name', 'headline', 'title']);
|
|
53
|
+
const description = firstString(record, ['description', 'summary']);
|
|
54
|
+
if (title || description) {
|
|
55
|
+
out.push({ title: title ?? null, description: description ?? null, type });
|
|
56
|
+
}
|
|
57
|
+
}
|
|
58
|
+
}
|
|
59
|
+
function extractType(record) {
|
|
60
|
+
const raw = record['@type'];
|
|
61
|
+
if (typeof raw === 'string')
|
|
62
|
+
return raw.toLowerCase();
|
|
63
|
+
if (Array.isArray(raw)) {
|
|
64
|
+
const found = raw.find((entry) => typeof entry === 'string');
|
|
65
|
+
return typeof found === 'string' ? found.toLowerCase() : null;
|
|
66
|
+
}
|
|
67
|
+
return null;
|
|
68
|
+
}
|
|
69
|
+
function firstString(record, keys) {
|
|
70
|
+
for (const key of keys) {
|
|
71
|
+
const value = record[key];
|
|
72
|
+
if (typeof value === 'string' && value.trim())
|
|
73
|
+
return value.trim();
|
|
74
|
+
}
|
|
75
|
+
return null;
|
|
76
|
+
}
|
|
77
|
+
//# sourceMappingURL=jsonld.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"jsonld.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/jsonld.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,IAAI,EAAE,MAAM,SAAS,CAAA;AAE9B,OAAO,EAAE,kBAAkB,EAAE,MAAM,cAAc,CAAA;AAQjD,MAAM,UAAU,oBAAoB,CAAC,IAAY;IAC/C,IAAI,CAAC;QACH,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAA;QACpB,MAAM,OAAO,GAAG,CAAC,CAAC,oCAAoC,CAAC,CAAC,OAAO,EAAE,CAAA;QACjE,MAAM,UAAU,GAAoB,EAAE,CAAA;QAEtC,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;YAC7B,MAAM,GAAG,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,CAAA;YAC5B,IAAI,CAAC,GAAG;gBAAE,SAAQ;YAClB,IAAI,CAAC;gBACH,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAA;gBAC5B,iBAAiB,CAAC,IAAI,EAAE,UAAU,CAAC,CAAA;YACrC,CAAC;YAAC,MAAM,CAAC;gBACP,0BAA0B;YAC5B,CAAC;QACH,CAAC;QAED,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,IAAI,CAAA;QAExC,MAAM,MAAM,GAAG,UAAU;aACtB,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;YACX,KAAK,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,kBAAkB,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI;YACnD,WAAW,EAAE,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,kBAAkB,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,IAAI;YACrE,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,kBAAkB,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI;SACjD,CAAC,CAAC;aACF,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,WAAW,CAAC;aACvC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,MAAM,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,WAAW,EAAE,MAAM,IAAI,CAAC,CAAC,CAAC,CAAA;QAE9E,OAAO,MAAM,CAAC,CAAC,CAAC,IAAI,IAAI,CAAA;IAC1B,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,IAAI,CAAA;IACb,CAAC;AACH,CAAC;AAED,SAAS,iBAAiB,CAAC,KAAc,EAAE,GAAoB;IAC7D,IAAI,CAAC,KAAK;QAAE,OAAM;IAElB,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QACzB,KAAK,MAAM,IAAI,IAAI,KAAK;YAAE,iBAAiB,CAAC,IAAI,EAAE,GAAG,CAAC,CAAA;QACtD,OAAM;IACR,CAAC;IAED,IAAI,OAAO,KAAK,KAAK,QAAQ;QAAE,OAAM;IAErC,MAAM,MAAM,GAAG,KAAgC,CAAA;IAC/C,IAAI,MAAM,CAAC,QAAQ,CAAC,IAAI,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,CAAC;QACxD,iBAAiB,CAAC,MAAM,CAAC,QAAQ,CAAC,EAAE,GAAG,CAAC,CAAA;IAC1C,CAAC;IAED,MAAM,IAAI,GAAG,WAAW,CAAC,MAAM,CAAC,CAAA;IAChC,IAAI,IAAI,EAAE,CAAC;QACT,MAAM,KAAK,GAAG,WAAW,CAAC,MAAM,EAAE,CAAC,MAAM,EAAE,UAAU,EAAE,OAAO,CAAC,CAAC,CAAA;QAChE,MAAM,WAAW,GAAG,WAAW,CAAC,MAAM,EAAE,CAAC,aAAa,EAAE,SAAS,CAAC,CAAC,CAAA;QACnE,IAAI,KAAK,IAAI,WAAW,EAAE,CAAC;YACzB,GAAG,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,KAAK,IAAI,IAAI,EAAE,WAAW,EAAE,WAAW,IAAI,IAAI,EAAE,IAAI,EAAE,CAAC,CAAA;QAC5E,CAAC;IACH,CAAC;AACH,CAAC;AAED,SAAS,WAAW,CAAC,MAA+B;IAClD,MAAM,GAAG,GAAG,MAAM,CAAC,OAAO,CAAC,CAAA;IAC3B,IAAI,OAAO,GAAG,KAAK,QAAQ;QAAE,OAAO,GAAG,CAAC,WAAW,EAAE,CAAA;IACrD,IAAI,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;QACvB,MAAM,KAAK,GAAG,GAAG,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAA;QAC5D,OAAO,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,IAAI,CAAA;IAC/D,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC;AAED,SAAS,WAAW,CAAC,MAA+B,EAAE,IAAc;IAClE,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,MAAM,KAAK,GAAG,MAAM,CAAC,GAAG,CAAC,CAAA;QACzB,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,IAAI,EAAE;YAAE,OAAO,KAAK,CAAC,IAAI,EAAE,CAAA;IACpE,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC"}
|