@steipete/summarize-core 0.7.1 → 0.8.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/esm/content/cache/types.js +2 -0
- package/dist/esm/content/cache/types.js.map +1 -0
- package/dist/esm/content/index.js +1 -0
- package/dist/esm/content/index.js.map +1 -1
- package/dist/esm/content/link-preview/client.js +3 -0
- package/dist/esm/content/link-preview/client.js.map +1 -1
- package/dist/esm/content/link-preview/content/fetcher.js +1 -1
- package/dist/esm/content/link-preview/content/fetcher.js.map +1 -1
- package/dist/esm/content/link-preview/content/html.js +1 -1
- package/dist/esm/content/link-preview/content/html.js.map +1 -1
- package/dist/esm/content/link-preview/content/index.js +22 -3
- package/dist/esm/content/link-preview/content/index.js.map +1 -1
- package/dist/esm/content/link-preview/deps.js.map +1 -1
- package/dist/esm/content/transcript/index.js +1 -0
- package/dist/esm/content/transcript/index.js.map +1 -1
- package/dist/esm/content/transcript/providers/generic.js +84 -4
- package/dist/esm/content/transcript/providers/generic.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast.js +1 -0
- package/dist/esm/content/transcript/providers/podcast.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube/captions.js +35 -14
- package/dist/esm/content/transcript/providers/youtube/captions.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube/yt-dlp.js +84 -12
- package/dist/esm/content/transcript/providers/youtube/yt-dlp.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube.js +38 -2
- package/dist/esm/content/transcript/providers/youtube.js.map +1 -1
- package/dist/esm/content/transcript/utils.js +25 -69
- package/dist/esm/content/transcript/utils.js.map +1 -1
- package/dist/esm/content/url.js +76 -0
- package/dist/esm/content/url.js.map +1 -0
- package/dist/esm/prompts/cli.js +25 -5
- package/dist/esm/prompts/cli.js.map +1 -1
- package/dist/esm/prompts/file.js +51 -12
- package/dist/esm/prompts/file.js.map +1 -1
- package/dist/esm/prompts/format.js +26 -0
- package/dist/esm/prompts/format.js.map +1 -0
- package/dist/esm/prompts/link-summary.js +51 -22
- package/dist/esm/prompts/link-summary.js.map +1 -1
- package/dist/types/content/cache/types.d.ts +25 -0
- package/dist/types/content/index.d.ts +3 -1
- package/dist/types/content/link-preview/client.d.ts +6 -1
- package/dist/types/content/link-preview/content/types.d.ts +1 -1
- package/dist/types/content/link-preview/deps.d.ts +11 -20
- package/dist/types/content/transcript/cache.d.ts +1 -1
- package/dist/types/content/transcript/providers/generic.d.ts +1 -1
- package/dist/types/content/transcript/providers/youtube/captions.d.ts +3 -1
- package/dist/types/content/transcript/providers/youtube/yt-dlp.d.ts +3 -1
- package/dist/types/content/transcript/types.d.ts +2 -1
- package/dist/types/content/transcript/utils.d.ts +1 -3
- package/dist/types/content/url.d.ts +8 -0
- package/dist/types/prompts/cli.d.ts +4 -1
- package/dist/types/prompts/file.d.ts +9 -2
- package/dist/types/prompts/format.d.ts +14 -0
- package/dist/types/prompts/link-summary.d.ts +4 -1
- package/package.json +10 -10
- package/dist/cjs/content/index.js +0 -14
- package/dist/cjs/content/index.js.map +0 -1
- package/dist/cjs/content/link-preview/client.js +0 -31
- package/dist/cjs/content/link-preview/client.js.map +0 -1
- package/dist/cjs/content/link-preview/content/article.js +0 -164
- package/dist/cjs/content/link-preview/content/article.js.map +0 -1
- package/dist/cjs/content/link-preview/content/cleaner.js +0 -63
- package/dist/cjs/content/link-preview/content/cleaner.js.map +0 -1
- package/dist/cjs/content/link-preview/content/constants.js +0 -10
- package/dist/cjs/content/link-preview/content/constants.js.map +0 -1
- package/dist/cjs/content/link-preview/content/fetcher.js +0 -128
- package/dist/cjs/content/link-preview/content/fetcher.js.map +0 -1
- package/dist/cjs/content/link-preview/content/firecrawl.js +0 -90
- package/dist/cjs/content/link-preview/content/firecrawl.js.map +0 -1
- package/dist/cjs/content/link-preview/content/html.js +0 -165
- package/dist/cjs/content/link-preview/content/html.js.map +0 -1
- package/dist/cjs/content/link-preview/content/index.js +0 -348
- package/dist/cjs/content/link-preview/content/index.js.map +0 -1
- package/dist/cjs/content/link-preview/content/jsonld.js +0 -80
- package/dist/cjs/content/link-preview/content/jsonld.js.map +0 -1
- package/dist/cjs/content/link-preview/content/parsers.js +0 -81
- package/dist/cjs/content/link-preview/content/parsers.js.map +0 -1
- package/dist/cjs/content/link-preview/content/podcast-utils.js +0 -85
- package/dist/cjs/content/link-preview/content/podcast-utils.js.map +0 -1
- package/dist/cjs/content/link-preview/content/readability.js +0 -90
- package/dist/cjs/content/link-preview/content/readability.js.map +0 -1
- package/dist/cjs/content/link-preview/content/twitter-utils.js +0 -74
- package/dist/cjs/content/link-preview/content/twitter-utils.js.map +0 -1
- package/dist/cjs/content/link-preview/content/types.js +0 -7
- package/dist/cjs/content/link-preview/content/types.js.map +0 -1
- package/dist/cjs/content/link-preview/content/utils.js +0 -177
- package/dist/cjs/content/link-preview/content/utils.js.map +0 -1
- package/dist/cjs/content/link-preview/content/video.js +0 -99
- package/dist/cjs/content/link-preview/content/video.js.map +0 -1
- package/dist/cjs/content/link-preview/content/youtube.js +0 -85
- package/dist/cjs/content/link-preview/content/youtube.js.map +0 -1
- package/dist/cjs/content/link-preview/deps.js +0 -23
- package/dist/cjs/content/link-preview/deps.js.map +0 -1
- package/dist/cjs/content/link-preview/fetch-with-timeout.js +0 -38
- package/dist/cjs/content/link-preview/fetch-with-timeout.js.map +0 -1
- package/dist/cjs/content/link-preview/types.js +0 -5
- package/dist/cjs/content/link-preview/types.js.map +0 -1
- package/dist/cjs/content/transcript/cache.js +0 -85
- package/dist/cjs/content/transcript/cache.js.map +0 -1
- package/dist/cjs/content/transcript/index.js +0 -134
- package/dist/cjs/content/transcript/index.js.map +0 -1
- package/dist/cjs/content/transcript/normalize.js +0 -49
- package/dist/cjs/content/transcript/normalize.js.map +0 -1
- package/dist/cjs/content/transcript/providers/generic.js +0 -16
- package/dist/cjs/content/transcript/providers/generic.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/apple-flow.js +0 -226
- package/dist/cjs/content/transcript/providers/podcast/apple-flow.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/apple.js +0 -43
- package/dist/cjs/content/transcript/providers/podcast/apple.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/constants.js +0 -11
- package/dist/cjs/content/transcript/providers/podcast/constants.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/flow-context.js +0 -3
- package/dist/cjs/content/transcript/providers/podcast/flow-context.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/itunes.js +0 -139
- package/dist/cjs/content/transcript/providers/podcast/itunes.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/json.js +0 -43
- package/dist/cjs/content/transcript/providers/podcast/json.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/media.js +0 -355
- package/dist/cjs/content/transcript/providers/podcast/media.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/results.js +0 -32
- package/dist/cjs/content/transcript/providers/podcast/results.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/rss.js +0 -262
- package/dist/cjs/content/transcript/providers/podcast/rss.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/spotify-flow.js +0 -221
- package/dist/cjs/content/transcript/providers/podcast/spotify-flow.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast/spotify.js +0 -119
- package/dist/cjs/content/transcript/providers/podcast/spotify.js.map +0 -1
- package/dist/cjs/content/transcript/providers/podcast.js +0 -260
- package/dist/cjs/content/transcript/providers/podcast.js.map +0 -1
- package/dist/cjs/content/transcript/providers/youtube/api.js +0 -264
- package/dist/cjs/content/transcript/providers/youtube/api.js.map +0 -1
- package/dist/cjs/content/transcript/providers/youtube/apify.js +0 -59
- package/dist/cjs/content/transcript/providers/youtube/apify.js.map +0 -1
- package/dist/cjs/content/transcript/providers/youtube/captions.js +0 -413
- package/dist/cjs/content/transcript/providers/youtube/captions.js.map +0 -1
- package/dist/cjs/content/transcript/providers/youtube/yt-dlp.js +0 -170
- package/dist/cjs/content/transcript/providers/youtube/yt-dlp.js.map +0 -1
- package/dist/cjs/content/transcript/providers/youtube.js +0 -178
- package/dist/cjs/content/transcript/providers/youtube.js.map +0 -1
- package/dist/cjs/content/transcript/types.js +0 -3
- package/dist/cjs/content/transcript/types.js.map +0 -1
- package/dist/cjs/content/transcript/utils.js +0 -303
- package/dist/cjs/content/transcript/utils.js.map +0 -1
- package/dist/cjs/index.js +0 -22
- package/dist/cjs/index.js.map +0 -1
- package/dist/cjs/language.js +0 -132
- package/dist/cjs/language.js.map +0 -1
- package/dist/cjs/package.json +0 -3
- package/dist/cjs/prompts/cli.js +0 -23
- package/dist/cjs/prompts/cli.js.map +0 -1
- package/dist/cjs/prompts/file.js +0 -52
- package/dist/cjs/prompts/file.js.map +0 -1
- package/dist/cjs/prompts/index.js +0 -14
- package/dist/cjs/prompts/index.js.map +0 -1
- package/dist/cjs/prompts/link-summary.js +0 -122
- package/dist/cjs/prompts/link-summary.js.map +0 -1
- package/dist/cjs/shared/contracts.js +0 -5
- package/dist/cjs/shared/contracts.js.map +0 -1
- package/dist/cjs/transcription/whisper/constants.js +0 -11
- package/dist/cjs/transcription/whisper/constants.js.map +0 -1
- package/dist/cjs/transcription/whisper/core.js +0 -307
- package/dist/cjs/transcription/whisper/core.js.map +0 -1
- package/dist/cjs/transcription/whisper/fal.js +0 -44
- package/dist/cjs/transcription/whisper/fal.js.map +0 -1
- package/dist/cjs/transcription/whisper/ffmpeg.js +0 -187
- package/dist/cjs/transcription/whisper/ffmpeg.js.map +0 -1
- package/dist/cjs/transcription/whisper/openai.js +0 -51
- package/dist/cjs/transcription/whisper/openai.js.map +0 -1
- package/dist/cjs/transcription/whisper/types.js +0 -3
- package/dist/cjs/transcription/whisper/types.js.map +0 -1
- package/dist/cjs/transcription/whisper/utils.js +0 -70
- package/dist/cjs/transcription/whisper/utils.js.map +0 -1
- package/dist/cjs/transcription/whisper/whisper-cpp.js +0 -232
- package/dist/cjs/transcription/whisper/whisper-cpp.js.map +0 -1
- package/dist/cjs/transcription/whisper.js +0 -15
- package/dist/cjs/transcription/whisper.js.map +0 -1
|
@@ -1,165 +0,0 @@
|
|
|
1
|
-
"use strict";
|
|
2
|
-
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
-
exports.buildResultFromHtmlDocument = buildResultFromHtmlDocument;
|
|
4
|
-
const index_js_1 = require("../../transcript/index.js");
|
|
5
|
-
const utils_js_1 = require("../../transcript/utils.js");
|
|
6
|
-
const article_js_1 = require("./article.js");
|
|
7
|
-
const cleaner_js_1 = require("./cleaner.js");
|
|
8
|
-
const constants_js_1 = require("./constants.js");
|
|
9
|
-
const jsonld_js_1 = require("./jsonld.js");
|
|
10
|
-
const parsers_js_1 = require("./parsers.js");
|
|
11
|
-
const podcast_utils_js_1 = require("./podcast-utils.js");
|
|
12
|
-
const readability_js_1 = require("./readability.js");
|
|
13
|
-
const utils_js_2 = require("./utils.js");
|
|
14
|
-
const video_js_1 = require("./video.js");
|
|
15
|
-
const youtube_js_1 = require("./youtube.js");
|
|
16
|
-
const LEADING_CONTROL_PATTERN = /^[\s\p{Cc}]+/u;
|
|
17
|
-
function stripLeadingTitle(content, title) {
|
|
18
|
-
if (!(content && title)) {
|
|
19
|
-
return content;
|
|
20
|
-
}
|
|
21
|
-
const normalizedTitle = title.trim();
|
|
22
|
-
if (normalizedTitle.length === 0) {
|
|
23
|
-
return content;
|
|
24
|
-
}
|
|
25
|
-
const trimmedContent = content.trimStart();
|
|
26
|
-
if (!trimmedContent.toLowerCase().startsWith(normalizedTitle.toLowerCase())) {
|
|
27
|
-
return content;
|
|
28
|
-
}
|
|
29
|
-
const remainderOriginal = trimmedContent.slice(normalizedTitle.length);
|
|
30
|
-
const remainder = remainderOriginal.replace(LEADING_CONTROL_PATTERN, '');
|
|
31
|
-
return remainder;
|
|
32
|
-
}
|
|
33
|
-
async function buildResultFromHtmlDocument({ url, html, cacheMode, maxCharacters, youtubeTranscriptMode, firecrawlDiagnostics, markdownRequested, markdownMode, timeoutMs, deps, readabilityCandidate, }) {
|
|
34
|
-
if ((0, utils_js_1.isYouTubeVideoUrl)(url) && !(0, utils_js_1.extractYouTubeVideoId)(url)) {
|
|
35
|
-
throw new Error('Invalid YouTube video id in URL');
|
|
36
|
-
}
|
|
37
|
-
const { title, description, siteName } = (0, parsers_js_1.extractMetadataFromHtml)(html, url);
|
|
38
|
-
const jsonLd = (0, jsonld_js_1.extractJsonLdContent)(html);
|
|
39
|
-
const mergedTitle = (0, utils_js_2.pickFirstText)([jsonLd?.title, title]);
|
|
40
|
-
const mergedDescription = (0, utils_js_2.pickFirstText)([jsonLd?.description, description]);
|
|
41
|
-
const isPodcastJsonLd = (0, podcast_utils_js_1.isPodcastLikeJsonLdType)(jsonLd?.type);
|
|
42
|
-
const readability = readabilityCandidate ?? (await (0, readability_js_1.extractReadabilityFromHtml)(html, url));
|
|
43
|
-
const readabilityText = readability?.text ? (0, cleaner_js_1.normalizeForPrompt)(readability.text) : '';
|
|
44
|
-
const readabilityHtml = (0, readability_js_1.toReadabilityHtml)(readability);
|
|
45
|
-
const normalizedSegmentsFromHtml = (0, cleaner_js_1.normalizeForPrompt)((0, article_js_1.extractArticleContent)(html));
|
|
46
|
-
const normalizedSegmentsFromReadabilityHtml = readabilityHtml
|
|
47
|
-
? (0, cleaner_js_1.normalizeForPrompt)((0, article_js_1.extractArticleContent)(readabilityHtml))
|
|
48
|
-
: '';
|
|
49
|
-
const preferReadabilityHtml = normalizedSegmentsFromReadabilityHtml.length >= constants_js_1.MIN_READABILITY_CONTENT_CHARACTERS &&
|
|
50
|
-
(normalizedSegmentsFromHtml.length < constants_js_1.MIN_HTML_CONTENT_CHARACTERS ||
|
|
51
|
-
normalizedSegmentsFromReadabilityHtml.length >=
|
|
52
|
-
normalizedSegmentsFromHtml.length * constants_js_1.READABILITY_RELATIVE_THRESHOLD);
|
|
53
|
-
const normalizedSegments = preferReadabilityHtml
|
|
54
|
-
? normalizedSegmentsFromReadabilityHtml
|
|
55
|
-
: normalizedSegmentsFromHtml;
|
|
56
|
-
const preferReadabilityText = !preferReadabilityHtml &&
|
|
57
|
-
readabilityText.length >= constants_js_1.MIN_READABILITY_CONTENT_CHARACTERS &&
|
|
58
|
-
(normalizedSegmentsFromHtml.length < constants_js_1.MIN_HTML_CONTENT_CHARACTERS ||
|
|
59
|
-
readabilityText.length >= normalizedSegmentsFromHtml.length * constants_js_1.READABILITY_RELATIVE_THRESHOLD);
|
|
60
|
-
const preferReadability = preferReadabilityHtml || preferReadabilityText;
|
|
61
|
-
const effectiveNormalized = preferReadabilityText ? readabilityText : normalizedSegments;
|
|
62
|
-
const descriptionCandidate = mergedDescription ? (0, cleaner_js_1.normalizeForPrompt)(mergedDescription) : '';
|
|
63
|
-
const preferDescription = descriptionCandidate.length >= constants_js_1.MIN_METADATA_DESCRIPTION_CHARACTERS &&
|
|
64
|
-
(isPodcastJsonLd ||
|
|
65
|
-
(0, podcast_utils_js_1.isPodcastHost)(url) ||
|
|
66
|
-
(!preferReadability &&
|
|
67
|
-
(effectiveNormalized.length < constants_js_1.MIN_HTML_CONTENT_CHARACTERS ||
|
|
68
|
-
descriptionCandidate.length >=
|
|
69
|
-
effectiveNormalized.length * constants_js_1.READABILITY_RELATIVE_THRESHOLD)));
|
|
70
|
-
const effectiveNormalizedWithDescription = preferDescription
|
|
71
|
-
? descriptionCandidate
|
|
72
|
-
: effectiveNormalized;
|
|
73
|
-
const transcriptResolution = await (0, index_js_1.resolveTranscriptForLink)(url, html, deps, {
|
|
74
|
-
youtubeTranscriptMode,
|
|
75
|
-
cacheMode,
|
|
76
|
-
});
|
|
77
|
-
const youtubeDescription = transcriptResolution.text === null ? (0, youtube_js_1.extractYouTubeShortDescription)(html) : null;
|
|
78
|
-
const baseCandidate = youtubeDescription
|
|
79
|
-
? (0, cleaner_js_1.normalizeForPrompt)(youtubeDescription)
|
|
80
|
-
: effectiveNormalizedWithDescription;
|
|
81
|
-
let baseContent = (0, utils_js_2.selectBaseContent)(baseCandidate, transcriptResolution.text);
|
|
82
|
-
if (baseContent === normalizedSegments) {
|
|
83
|
-
baseContent = stripLeadingTitle(baseContent, mergedTitle ?? title);
|
|
84
|
-
}
|
|
85
|
-
const transcriptDiagnostics = (0, utils_js_2.ensureTranscriptDiagnostics)(transcriptResolution, cacheMode ?? 'default');
|
|
86
|
-
const markdownDiagnostics = await (async () => {
|
|
87
|
-
if (!markdownRequested) {
|
|
88
|
-
return { requested: false, used: false, provider: null, notes: null };
|
|
89
|
-
}
|
|
90
|
-
if ((0, utils_js_1.isYouTubeUrl)(url)) {
|
|
91
|
-
return {
|
|
92
|
-
requested: true,
|
|
93
|
-
used: false,
|
|
94
|
-
provider: null,
|
|
95
|
-
notes: 'Skipping Markdown conversion for YouTube URLs',
|
|
96
|
-
};
|
|
97
|
-
}
|
|
98
|
-
if (!deps.convertHtmlToMarkdown) {
|
|
99
|
-
return {
|
|
100
|
-
requested: true,
|
|
101
|
-
used: false,
|
|
102
|
-
provider: null,
|
|
103
|
-
notes: 'No HTML→Markdown converter configured',
|
|
104
|
-
};
|
|
105
|
-
}
|
|
106
|
-
try {
|
|
107
|
-
const htmlForMarkdown = markdownMode === 'readability' && readabilityHtml ? readabilityHtml : html;
|
|
108
|
-
const sanitizedHtml = (0, article_js_1.sanitizeHtmlForMarkdownConversion)(htmlForMarkdown);
|
|
109
|
-
const markdown = await deps.convertHtmlToMarkdown({
|
|
110
|
-
url,
|
|
111
|
-
html: sanitizedHtml,
|
|
112
|
-
title: mergedTitle ?? title,
|
|
113
|
-
siteName,
|
|
114
|
-
timeoutMs,
|
|
115
|
-
});
|
|
116
|
-
const normalizedMarkdown = (0, cleaner_js_1.normalizeForPrompt)(markdown);
|
|
117
|
-
if (normalizedMarkdown.length === 0) {
|
|
118
|
-
return {
|
|
119
|
-
requested: true,
|
|
120
|
-
used: false,
|
|
121
|
-
provider: null,
|
|
122
|
-
notes: 'HTML→Markdown conversion returned empty content',
|
|
123
|
-
};
|
|
124
|
-
}
|
|
125
|
-
baseContent = normalizedMarkdown;
|
|
126
|
-
return {
|
|
127
|
-
requested: true,
|
|
128
|
-
used: true,
|
|
129
|
-
provider: 'llm',
|
|
130
|
-
notes: markdownMode === 'readability' && readabilityHtml
|
|
131
|
-
? 'Readability HTML used for markdown input'
|
|
132
|
-
: null,
|
|
133
|
-
};
|
|
134
|
-
}
|
|
135
|
-
catch (error) {
|
|
136
|
-
const message = error instanceof Error ? error.message : String(error);
|
|
137
|
-
return {
|
|
138
|
-
requested: true,
|
|
139
|
-
used: false,
|
|
140
|
-
provider: null,
|
|
141
|
-
notes: `HTML→Markdown conversion failed: ${message}`,
|
|
142
|
-
};
|
|
143
|
-
}
|
|
144
|
-
})();
|
|
145
|
-
const video = (0, video_js_1.detectPrimaryVideoFromHtml)(html, url);
|
|
146
|
-
const isVideoOnly = !transcriptResolution.text && baseContent.length < constants_js_1.MIN_HTML_CONTENT_CHARACTERS && video !== null;
|
|
147
|
-
return (0, utils_js_2.finalizeExtractedLinkContent)({
|
|
148
|
-
url,
|
|
149
|
-
baseContent,
|
|
150
|
-
maxCharacters,
|
|
151
|
-
title: mergedTitle ?? title,
|
|
152
|
-
description: mergedDescription ?? description,
|
|
153
|
-
siteName,
|
|
154
|
-
transcriptResolution,
|
|
155
|
-
video,
|
|
156
|
-
isVideoOnly,
|
|
157
|
-
diagnostics: {
|
|
158
|
-
strategy: 'html',
|
|
159
|
-
firecrawl: firecrawlDiagnostics,
|
|
160
|
-
markdown: markdownDiagnostics,
|
|
161
|
-
transcript: transcriptDiagnostics,
|
|
162
|
-
},
|
|
163
|
-
});
|
|
164
|
-
}
|
|
165
|
-
//# sourceMappingURL=html.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"html.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/html.ts"],"names":[],"mappings":";;AAgDA,kEAiLC;AAjOD,wDAAoE;AACpE,wDAAkG;AAGlG,6CAAuF;AACvF,6CAAiD;AACjD,iDAKuB;AACvB,2CAAkD;AAClD,6CAAsD;AACtD,yDAA2E;AAC3E,qDAAgF;AAEhF,yCAKmB;AACnB,yCAAuD;AACvD,6CAA6D;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;AAEM,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,IAAA,4BAAiB,EAAC,GAAG,CAAC,IAAI,CAAC,IAAA,gCAAqB,EAAC,GAAG,CAAC,EAAE,CAAC;QAC1D,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAA;IACpD,CAAC;IAED,MAAM,EAAE,KAAK,EAAE,WAAW,EAAE,QAAQ,EAAE,GAAG,IAAA,oCAAuB,EAAC,IAAI,EAAE,GAAG,CAAC,CAAA;IAC3E,MAAM,MAAM,GAAG,IAAA,gCAAoB,EAAC,IAAI,CAAC,CAAA;IACzC,MAAM,WAAW,GAAG,IAAA,wBAAa,EAAC,CAAC,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAA;IACzD,MAAM,iBAAiB,GAAG,IAAA,wBAAa,EAAC,CAAC,MAAM,EAAE,WAAW,EAAE,WAAW,CAAC,CAAC,CAAA;IAC3E,MAAM,eAAe,GAAG,IAAA,0CAAuB,EAAC,MAAM,EAAE,IAAI,CAAC,CAAA;IAC7D,MAAM,WAAW,GAAG,oBAAoB,IAAI,CAAC,MAAM,IAAA,2CAA0B,EAAC,IAAI,EAAE,GAAG,CAAC,CAAC,CAAA;IACzF,MAAM,eAAe,GAAG,WAAW,EAAE,IAAI,CAAC,CAAC,CAAC,IAAA,+BAAkB,EAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;IACrF,MAAM,eAAe,GAAG,IAAA,kCAAiB,EAAC,WAAW,CAAC,CAAA;IAEtD,MAAM,0BAA0B,GAAG,IAAA,+BAAkB,EAAC,IAAA,kCAAqB,EAAC,IAAI,CAAC,CAAC,CAAA;IAClF,MAAM,qCAAqC,GAAG,eAAe;QAC3D,CAAC,CAAC,IAAA,+BAAkB,EAAC,IAAA,kCAAqB,EAAC,eAAe,CAAC,CAAC;QAC5D,CAAC,CAAC,EAAE,CAAA;IACN,MAAM,qBAAqB,GACzB,qCAAqC,CAAC,MAAM,IAAI,iDAAkC;QAClF,CAAC,0BAA0B,CAAC,MAAM,GAAG,0CAA2B;YAC9D,qCAAqC,CAAC,MAAM;gBAC1C,0BAA0B,CAAC,MAAM,GAAG,6CAA8B,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,iDAAkC;QAC5D,CAAC,0BAA0B,CAAC,MAAM,GAAG,0CAA2B;YAC9D,eAAe,CAAC,MAAM,IAAI,0BAA0B,CAAC,MAAM,GAAG,6CAA8B,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,IAAA,+BAAkB,EAAC,iBAAiB,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;IAC3F,MAAM,iBAAiB,GACrB,oBAAoB,CAAC,MAAM,IAAI,kDAAmC;QAClE,CAAC,eAAe;YACd,IAAA,gCAAa,EAAC,GAAG,CAAC;YAClB,CAAC,CAAC,iBAAiB;gBACjB,CAAC,mBAAmB,CAAC,MAAM,GAAG,0CAA2B;oBACvD,oBAAoB,CAAC,MAAM;wBACzB,mBAAmB,CAAC,MAAM,GAAG,6CAA8B,CAAC,CAAC,CAAC,CAAA;IACxE,MAAM,kCAAkC,GAAG,iBAAiB;QAC1D,CAAC,CAAC,oBAAoB;QACtB,CAAC,CAAC,mBAAmB,CAAA;IACvB,MAAM,oBAAoB,GAAG,MAAM,IAAA,mCAAwB,EAAC,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,IAAA,2CAA8B,EAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;IAClF,MAAM,aAAa,GAAG,kBAAkB;QACtC,CAAC,CAAC,IAAA,+BAAkB,EAAC,kBAAkB,CAAC;QACxC,CAAC,CAAC,kCAAkC,CAAA;IAEtC,IAAI,WAAW,GAAG,IAAA,4BAAiB,EAAC,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,IAAA,sCAA2B,EACvD,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,IAAA,uBAAY,EAAC,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,IAAA,8CAAiC,EAAC,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,IAAA,+BAAkB,EAAC,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,IAAA,qCAA0B,EAAC,IAAI,EAAE,GAAG,CAAC,CAAA;IACnD,MAAM,WAAW,GACf,CAAC,oBAAoB,CAAC,IAAI,IAAI,WAAW,CAAC,MAAM,GAAG,0CAA2B,IAAI,KAAK,KAAK,IAAI,CAAA;IAElG,OAAO,IAAA,uCAA4B,EAAC;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"}
|
|
@@ -1,348 +0,0 @@
|
|
|
1
|
-
"use strict";
|
|
2
|
-
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
-
exports.fetchLinkContent = fetchLinkContent;
|
|
4
|
-
const index_js_1 = require("../../transcript/index.js");
|
|
5
|
-
const utils_js_1 = require("../../transcript/utils.js");
|
|
6
|
-
const cleaner_js_1 = require("./cleaner.js");
|
|
7
|
-
const constants_js_1 = require("./constants.js");
|
|
8
|
-
const fetcher_js_1 = require("./fetcher.js");
|
|
9
|
-
const firecrawl_js_1 = require("./firecrawl.js");
|
|
10
|
-
const html_js_1 = require("./html.js");
|
|
11
|
-
const podcast_utils_js_1 = require("./podcast-utils.js");
|
|
12
|
-
const readability_js_1 = require("./readability.js");
|
|
13
|
-
const twitter_utils_js_1 = require("./twitter-utils.js");
|
|
14
|
-
const utils_js_2 = require("./utils.js");
|
|
15
|
-
async function fetchLinkContent(url, options, deps) {
|
|
16
|
-
const timeoutMs = (0, utils_js_2.resolveTimeoutMs)(options);
|
|
17
|
-
const cacheMode = (0, utils_js_2.resolveCacheMode)(options);
|
|
18
|
-
const maxCharacters = (0, utils_js_2.resolveMaxCharacters)(options);
|
|
19
|
-
const youtubeTranscriptMode = options?.youtubeTranscript ?? 'auto';
|
|
20
|
-
const firecrawlMode = (0, utils_js_2.resolveFirecrawlMode)(options);
|
|
21
|
-
const markdownRequested = (options?.format ?? 'text') === 'markdown';
|
|
22
|
-
const markdownMode = options?.markdownMode ?? 'auto';
|
|
23
|
-
const canUseFirecrawl = firecrawlMode !== 'off' && deps.scrapeWithFirecrawl !== null && !(0, utils_js_1.isYouTubeUrl)(url);
|
|
24
|
-
const spotifyEpisodeId = (0, podcast_utils_js_1.extractSpotifyEpisodeId)(url);
|
|
25
|
-
if (spotifyEpisodeId) {
|
|
26
|
-
if (!deps.openaiApiKey && !deps.falApiKey) {
|
|
27
|
-
throw new Error('Spotify episode transcription requires OPENAI_API_KEY or FAL_KEY (Whisper); otherwise you may only get a captcha/recaptcha HTML page.');
|
|
28
|
-
}
|
|
29
|
-
const transcriptResolution = await (0, index_js_1.resolveTranscriptForLink)(url, null, deps, {
|
|
30
|
-
youtubeTranscriptMode,
|
|
31
|
-
cacheMode,
|
|
32
|
-
});
|
|
33
|
-
if (!transcriptResolution.text) {
|
|
34
|
-
const notes = transcriptResolution.diagnostics?.notes;
|
|
35
|
-
const suffix = notes ? ` (${notes})` : '';
|
|
36
|
-
throw new Error(`Failed to transcribe Spotify episode${suffix}`);
|
|
37
|
-
}
|
|
38
|
-
const transcriptDiagnostics = (0, utils_js_2.ensureTranscriptDiagnostics)(transcriptResolution, cacheMode ?? 'default');
|
|
39
|
-
transcriptDiagnostics.notes = (0, utils_js_2.appendNote)(transcriptDiagnostics.notes, 'Spotify episode: skipped HTML fetch to avoid captcha pages');
|
|
40
|
-
return (0, utils_js_2.finalizeExtractedLinkContent)({
|
|
41
|
-
url,
|
|
42
|
-
baseContent: (0, utils_js_2.selectBaseContent)('', transcriptResolution.text),
|
|
43
|
-
maxCharacters,
|
|
44
|
-
title: null,
|
|
45
|
-
description: null,
|
|
46
|
-
siteName: 'Spotify',
|
|
47
|
-
transcriptResolution,
|
|
48
|
-
video: null,
|
|
49
|
-
isVideoOnly: false,
|
|
50
|
-
diagnostics: {
|
|
51
|
-
strategy: 'html',
|
|
52
|
-
firecrawl: {
|
|
53
|
-
attempted: false,
|
|
54
|
-
used: false,
|
|
55
|
-
cacheMode,
|
|
56
|
-
cacheStatus: cacheMode === 'bypass' ? 'bypassed' : 'unknown',
|
|
57
|
-
notes: 'Spotify short-circuit skipped HTML/Firecrawl',
|
|
58
|
-
},
|
|
59
|
-
markdown: {
|
|
60
|
-
requested: markdownRequested,
|
|
61
|
-
used: false,
|
|
62
|
-
provider: null,
|
|
63
|
-
notes: 'Spotify short-circuit uses transcript content',
|
|
64
|
-
},
|
|
65
|
-
transcript: transcriptDiagnostics,
|
|
66
|
-
},
|
|
67
|
-
});
|
|
68
|
-
}
|
|
69
|
-
const appleIds = (0, podcast_utils_js_1.extractApplePodcastIds)(url);
|
|
70
|
-
if (appleIds) {
|
|
71
|
-
if (!deps.openaiApiKey && !deps.falApiKey) {
|
|
72
|
-
throw new Error('Apple Podcasts transcription requires OPENAI_API_KEY or FAL_KEY (Whisper); otherwise you may only get a slow/blocked HTML page.');
|
|
73
|
-
}
|
|
74
|
-
const transcriptResolution = await (0, index_js_1.resolveTranscriptForLink)(url, null, deps, {
|
|
75
|
-
youtubeTranscriptMode,
|
|
76
|
-
cacheMode,
|
|
77
|
-
});
|
|
78
|
-
if (!transcriptResolution.text) {
|
|
79
|
-
const notes = transcriptResolution.diagnostics?.notes;
|
|
80
|
-
const suffix = notes ? ` (${notes})` : '';
|
|
81
|
-
throw new Error(`Failed to transcribe Apple Podcasts episode${suffix}`);
|
|
82
|
-
}
|
|
83
|
-
const transcriptDiagnostics = (0, utils_js_2.ensureTranscriptDiagnostics)(transcriptResolution, cacheMode ?? 'default');
|
|
84
|
-
transcriptDiagnostics.notes = (0, utils_js_2.appendNote)(transcriptDiagnostics.notes, 'Apple Podcasts: skipped HTML fetch (prefer iTunes lookup / enclosures)');
|
|
85
|
-
return (0, utils_js_2.finalizeExtractedLinkContent)({
|
|
86
|
-
url,
|
|
87
|
-
baseContent: (0, utils_js_2.selectBaseContent)('', transcriptResolution.text),
|
|
88
|
-
maxCharacters,
|
|
89
|
-
title: null,
|
|
90
|
-
description: null,
|
|
91
|
-
siteName: 'Apple Podcasts',
|
|
92
|
-
transcriptResolution,
|
|
93
|
-
video: null,
|
|
94
|
-
isVideoOnly: false,
|
|
95
|
-
diagnostics: {
|
|
96
|
-
strategy: 'html',
|
|
97
|
-
firecrawl: {
|
|
98
|
-
attempted: false,
|
|
99
|
-
used: false,
|
|
100
|
-
cacheMode,
|
|
101
|
-
cacheStatus: cacheMode === 'bypass' ? 'bypassed' : 'unknown',
|
|
102
|
-
notes: 'Apple Podcasts short-circuit skipped HTML/Firecrawl',
|
|
103
|
-
},
|
|
104
|
-
markdown: {
|
|
105
|
-
requested: markdownRequested,
|
|
106
|
-
used: false,
|
|
107
|
-
provider: null,
|
|
108
|
-
notes: 'Apple Podcasts short-circuit uses transcript content',
|
|
109
|
-
},
|
|
110
|
-
transcript: transcriptDiagnostics,
|
|
111
|
-
},
|
|
112
|
-
});
|
|
113
|
-
}
|
|
114
|
-
let firecrawlAttempted = false;
|
|
115
|
-
let firecrawlPayload = null;
|
|
116
|
-
const firecrawlDiagnostics = {
|
|
117
|
-
attempted: false,
|
|
118
|
-
used: false,
|
|
119
|
-
cacheMode,
|
|
120
|
-
cacheStatus: cacheMode === 'bypass' ? 'bypassed' : 'unknown',
|
|
121
|
-
notes: null,
|
|
122
|
-
};
|
|
123
|
-
const twitterStatus = (0, twitter_utils_js_1.isTwitterStatusUrl)(url);
|
|
124
|
-
const nitterUrls = twitterStatus ? (0, twitter_utils_js_1.toNitterUrls)(url) : [];
|
|
125
|
-
let birdError = null;
|
|
126
|
-
let nitterError = null;
|
|
127
|
-
const attemptFirecrawl = async (reason) => {
|
|
128
|
-
if (!canUseFirecrawl) {
|
|
129
|
-
return null;
|
|
130
|
-
}
|
|
131
|
-
if (!firecrawlAttempted) {
|
|
132
|
-
const attempt = await (0, fetcher_js_1.fetchWithFirecrawl)(url, deps.scrapeWithFirecrawl, {
|
|
133
|
-
timeoutMs,
|
|
134
|
-
cacheMode,
|
|
135
|
-
onProgress: deps.onProgress ?? null,
|
|
136
|
-
reason,
|
|
137
|
-
});
|
|
138
|
-
firecrawlAttempted = true;
|
|
139
|
-
firecrawlPayload = attempt.payload;
|
|
140
|
-
firecrawlDiagnostics.attempted = attempt.diagnostics.attempted;
|
|
141
|
-
firecrawlDiagnostics.used = attempt.diagnostics.used;
|
|
142
|
-
firecrawlDiagnostics.cacheMode = attempt.diagnostics.cacheMode;
|
|
143
|
-
firecrawlDiagnostics.cacheStatus = attempt.diagnostics.cacheStatus;
|
|
144
|
-
firecrawlDiagnostics.notes = attempt.diagnostics.notes ?? null;
|
|
145
|
-
}
|
|
146
|
-
firecrawlDiagnostics.notes = (0, utils_js_2.appendNote)(firecrawlDiagnostics.notes, reason);
|
|
147
|
-
if (!firecrawlPayload) {
|
|
148
|
-
return null;
|
|
149
|
-
}
|
|
150
|
-
const firecrawlResult = await (0, firecrawl_js_1.buildResultFromFirecrawl)({
|
|
151
|
-
url,
|
|
152
|
-
payload: firecrawlPayload,
|
|
153
|
-
cacheMode,
|
|
154
|
-
maxCharacters,
|
|
155
|
-
youtubeTranscriptMode,
|
|
156
|
-
firecrawlDiagnostics,
|
|
157
|
-
markdownRequested,
|
|
158
|
-
deps,
|
|
159
|
-
});
|
|
160
|
-
if (firecrawlResult) {
|
|
161
|
-
return firecrawlResult;
|
|
162
|
-
}
|
|
163
|
-
firecrawlDiagnostics.notes = (0, utils_js_2.appendNote)(firecrawlDiagnostics.notes, 'Firecrawl returned empty content');
|
|
164
|
-
return null;
|
|
165
|
-
};
|
|
166
|
-
const attemptBird = async () => {
|
|
167
|
-
if (!deps.readTweetWithBird || !twitterStatus) {
|
|
168
|
-
return null;
|
|
169
|
-
}
|
|
170
|
-
deps.onProgress?.({ kind: 'bird-start', url });
|
|
171
|
-
try {
|
|
172
|
-
const tweet = await deps.readTweetWithBird({ url, timeoutMs });
|
|
173
|
-
const text = tweet?.text?.trim() ?? '';
|
|
174
|
-
if (text.length === 0) {
|
|
175
|
-
deps.onProgress?.({ kind: 'bird-done', url, ok: false, textBytes: null });
|
|
176
|
-
return null;
|
|
177
|
-
}
|
|
178
|
-
const title = tweet?.author?.username ? `@${tweet.author.username}` : null;
|
|
179
|
-
const description = null;
|
|
180
|
-
const siteName = 'X';
|
|
181
|
-
const transcriptResolution = { text: null, source: null };
|
|
182
|
-
const transcriptDiagnostics = (0, utils_js_2.ensureTranscriptDiagnostics)(transcriptResolution, cacheMode ?? 'default');
|
|
183
|
-
const result = (0, utils_js_2.finalizeExtractedLinkContent)({
|
|
184
|
-
url,
|
|
185
|
-
baseContent: text,
|
|
186
|
-
maxCharacters,
|
|
187
|
-
title,
|
|
188
|
-
description,
|
|
189
|
-
siteName,
|
|
190
|
-
transcriptResolution,
|
|
191
|
-
video: null,
|
|
192
|
-
isVideoOnly: false,
|
|
193
|
-
diagnostics: {
|
|
194
|
-
strategy: 'bird',
|
|
195
|
-
firecrawl: firecrawlDiagnostics,
|
|
196
|
-
markdown: {
|
|
197
|
-
requested: markdownRequested,
|
|
198
|
-
used: false,
|
|
199
|
-
provider: null,
|
|
200
|
-
notes: 'Bird tweet fetch provides plain text',
|
|
201
|
-
},
|
|
202
|
-
transcript: transcriptDiagnostics,
|
|
203
|
-
},
|
|
204
|
-
});
|
|
205
|
-
deps.onProgress?.({
|
|
206
|
-
kind: 'bird-done',
|
|
207
|
-
url,
|
|
208
|
-
ok: true,
|
|
209
|
-
textBytes: Buffer.byteLength(result.content, 'utf8'),
|
|
210
|
-
});
|
|
211
|
-
return result;
|
|
212
|
-
}
|
|
213
|
-
catch (error) {
|
|
214
|
-
birdError = error;
|
|
215
|
-
deps.onProgress?.({ kind: 'bird-done', url, ok: false, textBytes: null });
|
|
216
|
-
return null;
|
|
217
|
-
}
|
|
218
|
-
};
|
|
219
|
-
const birdResult = await attemptBird();
|
|
220
|
-
if (birdResult) {
|
|
221
|
-
return birdResult;
|
|
222
|
-
}
|
|
223
|
-
const attemptNitter = async () => {
|
|
224
|
-
if (nitterUrls.length === 0) {
|
|
225
|
-
return null;
|
|
226
|
-
}
|
|
227
|
-
for (const nitterUrl of nitterUrls) {
|
|
228
|
-
deps.onProgress?.({ kind: 'nitter-start', url: nitterUrl });
|
|
229
|
-
try {
|
|
230
|
-
const nitterHtml = await (0, fetcher_js_1.fetchHtmlDocument)(deps.fetch, nitterUrl, { timeoutMs });
|
|
231
|
-
if (!nitterHtml.trim()) {
|
|
232
|
-
nitterError = new Error(`Nitter returned empty body from ${new URL(nitterUrl).host}`);
|
|
233
|
-
deps.onProgress?.({ kind: 'nitter-done', url: nitterUrl, ok: false, textBytes: null });
|
|
234
|
-
continue;
|
|
235
|
-
}
|
|
236
|
-
if ((0, twitter_utils_js_1.isAnubisHtml)(nitterHtml)) {
|
|
237
|
-
nitterError = new Error(`Nitter returned Anubis challenge from ${new URL(nitterUrl).host}`);
|
|
238
|
-
deps.onProgress?.({ kind: 'nitter-done', url: nitterUrl, ok: false, textBytes: null });
|
|
239
|
-
continue;
|
|
240
|
-
}
|
|
241
|
-
deps.onProgress?.({
|
|
242
|
-
kind: 'nitter-done',
|
|
243
|
-
url: nitterUrl,
|
|
244
|
-
ok: true,
|
|
245
|
-
textBytes: Buffer.byteLength(nitterHtml, 'utf8'),
|
|
246
|
-
});
|
|
247
|
-
return nitterHtml;
|
|
248
|
-
}
|
|
249
|
-
catch (error) {
|
|
250
|
-
nitterError = error;
|
|
251
|
-
deps.onProgress?.({ kind: 'nitter-done', url: nitterUrl, ok: false, textBytes: null });
|
|
252
|
-
}
|
|
253
|
-
}
|
|
254
|
-
return null;
|
|
255
|
-
};
|
|
256
|
-
const nitterHtml = await attemptNitter();
|
|
257
|
-
if (nitterHtml) {
|
|
258
|
-
const nitterResult = await (0, html_js_1.buildResultFromHtmlDocument)({
|
|
259
|
-
url,
|
|
260
|
-
html: nitterHtml,
|
|
261
|
-
cacheMode,
|
|
262
|
-
maxCharacters,
|
|
263
|
-
youtubeTranscriptMode,
|
|
264
|
-
firecrawlDiagnostics,
|
|
265
|
-
markdownRequested,
|
|
266
|
-
markdownMode,
|
|
267
|
-
timeoutMs,
|
|
268
|
-
deps,
|
|
269
|
-
readabilityCandidate: null,
|
|
270
|
-
});
|
|
271
|
-
if (!(0, twitter_utils_js_1.isBlockedTwitterContent)(nitterResult.content)) {
|
|
272
|
-
nitterResult.diagnostics.strategy = 'nitter';
|
|
273
|
-
return nitterResult;
|
|
274
|
-
}
|
|
275
|
-
nitterError = new Error('Nitter returned blocked or empty content');
|
|
276
|
-
}
|
|
277
|
-
if (firecrawlMode === 'always') {
|
|
278
|
-
const firecrawlResult = await attemptFirecrawl('Firecrawl forced via options');
|
|
279
|
-
if (firecrawlResult) {
|
|
280
|
-
return firecrawlResult;
|
|
281
|
-
}
|
|
282
|
-
}
|
|
283
|
-
let html = null;
|
|
284
|
-
let htmlError = null;
|
|
285
|
-
try {
|
|
286
|
-
html = await (0, fetcher_js_1.fetchHtmlDocument)(deps.fetch, url, {
|
|
287
|
-
timeoutMs,
|
|
288
|
-
onProgress: deps.onProgress ?? null,
|
|
289
|
-
});
|
|
290
|
-
}
|
|
291
|
-
catch (error) {
|
|
292
|
-
htmlError = error;
|
|
293
|
-
}
|
|
294
|
-
if (!html) {
|
|
295
|
-
if (!canUseFirecrawl) {
|
|
296
|
-
throw htmlError instanceof Error ? htmlError : new Error('Failed to fetch HTML document');
|
|
297
|
-
}
|
|
298
|
-
const firecrawlResult = await attemptFirecrawl('HTML fetch failed; falling back to Firecrawl');
|
|
299
|
-
if (firecrawlResult) {
|
|
300
|
-
return firecrawlResult;
|
|
301
|
-
}
|
|
302
|
-
const firecrawlError = firecrawlDiagnostics.notes
|
|
303
|
-
? `; Firecrawl notes: ${firecrawlDiagnostics.notes}`
|
|
304
|
-
: '';
|
|
305
|
-
throw new Error(`Failed to fetch HTML document${firecrawlError}${htmlError instanceof Error ? `; HTML error: ${htmlError.message}` : ''}`);
|
|
306
|
-
}
|
|
307
|
-
let readabilityCandidate = null;
|
|
308
|
-
if (firecrawlMode === 'auto' && (0, firecrawl_js_1.shouldFallbackToFirecrawl)(html)) {
|
|
309
|
-
readabilityCandidate = await (0, readability_js_1.extractReadabilityFromHtml)(html, url);
|
|
310
|
-
const readabilityText = readabilityCandidate?.text
|
|
311
|
-
? (0, cleaner_js_1.normalizeForPrompt)(readabilityCandidate.text)
|
|
312
|
-
: '';
|
|
313
|
-
if (readabilityText.length < constants_js_1.MIN_READABILITY_CONTENT_CHARACTERS) {
|
|
314
|
-
const firecrawlResult = await attemptFirecrawl('HTML content looked blocked/thin; falling back to Firecrawl');
|
|
315
|
-
if (firecrawlResult) {
|
|
316
|
-
return firecrawlResult;
|
|
317
|
-
}
|
|
318
|
-
}
|
|
319
|
-
}
|
|
320
|
-
const htmlResult = await (0, html_js_1.buildResultFromHtmlDocument)({
|
|
321
|
-
url,
|
|
322
|
-
html,
|
|
323
|
-
cacheMode,
|
|
324
|
-
maxCharacters,
|
|
325
|
-
youtubeTranscriptMode,
|
|
326
|
-
firecrawlDiagnostics,
|
|
327
|
-
markdownRequested,
|
|
328
|
-
markdownMode,
|
|
329
|
-
timeoutMs,
|
|
330
|
-
deps,
|
|
331
|
-
readabilityCandidate,
|
|
332
|
-
});
|
|
333
|
-
if (twitterStatus && (0, twitter_utils_js_1.isBlockedTwitterContent)(htmlResult.content)) {
|
|
334
|
-
const birdNote = !deps.readTweetWithBird
|
|
335
|
-
? 'Bird not available'
|
|
336
|
-
: birdError
|
|
337
|
-
? `Bird failed: ${birdError instanceof Error ? birdError.message : String(birdError)}`
|
|
338
|
-
: 'Bird returned no text';
|
|
339
|
-
const nitterNote = nitterUrls.length > 0
|
|
340
|
-
? nitterError
|
|
341
|
-
? `Nitter failed: ${nitterError instanceof Error ? nitterError.message : String(nitterError)}`
|
|
342
|
-
: 'Nitter returned no text'
|
|
343
|
-
: 'Nitter not available';
|
|
344
|
-
throw new Error(`Unable to fetch tweet content from X. ${birdNote}. ${nitterNote}.`);
|
|
345
|
-
}
|
|
346
|
-
return htmlResult;
|
|
347
|
-
}
|
|
348
|
-
//# sourceMappingURL=index.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/index.ts"],"names":[],"mappings":";;AA6BA,4CA6YC;AA1aD,wDAAoE;AACpE,wDAAwD;AAGxD,6CAAiD;AACjD,iDAAmE;AACnE,6CAAoE;AACpE,iDAAoF;AACpF,uCAAuD;AACvD,yDAAoF;AACpF,qDAA6D;AAC7D,yDAK2B;AAE3B,yCASmB;AAEZ,KAAK,UAAU,gBAAgB,CACpC,GAAW,EACX,OAA4C,EAC5C,IAAqB;IAErB,MAAM,SAAS,GAAG,IAAA,2BAAgB,EAAC,OAAO,CAAC,CAAA;IAC3C,MAAM,SAAS,GAAG,IAAA,2BAAgB,EAAC,OAAO,CAAC,CAAA;IAC3C,MAAM,aAAa,GAAG,IAAA,+BAAoB,EAAC,OAAO,CAAC,CAAA;IACnD,MAAM,qBAAqB,GAAG,OAAO,EAAE,iBAAiB,IAAI,MAAM,CAAA;IAClE,MAAM,aAAa,GAAG,IAAA,+BAAoB,EAAC,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,IAAA,uBAAY,EAAC,GAAG,CAAC,CAAA;IAEpF,MAAM,gBAAgB,GAAG,IAAA,0CAAuB,EAAC,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,IAAA,mCAAwB,EAAC,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,IAAA,sCAA2B,EACvD,oBAAoB,EACpB,SAAS,IAAI,SAAS,CACvB,CAAA;QACD,qBAAqB,CAAC,KAAK,GAAG,IAAA,qBAAU,EACtC,qBAAqB,CAAC,KAAK,EAC3B,4DAA4D,CAC7D,CAAA;QAED,OAAO,IAAA,uCAA4B,EAAC;YAClC,GAAG;YACH,WAAW,EAAE,IAAA,4BAAiB,EAAC,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,IAAA,yCAAsB,EAAC,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,IAAA,mCAAwB,EAAC,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,IAAA,sCAA2B,EACvD,oBAAoB,EACpB,SAAS,IAAI,SAAS,CACvB,CAAA;QACD,qBAAqB,CAAC,KAAK,GAAG,IAAA,qBAAU,EACtC,qBAAqB,CAAC,KAAK,EAC3B,wEAAwE,CACzE,CAAA;QAED,OAAO,IAAA,uCAA4B,EAAC;YAClC,GAAG;YACH,WAAW,EAAE,IAAA,4BAAiB,EAAC,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,IAAA,qCAAkB,EAAC,GAAG,CAAC,CAAA;IAC7C,MAAM,UAAU,GAAG,aAAa,CAAC,CAAC,CAAC,IAAA,+BAAY,EAAC,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,IAAA,+BAAkB,EAAC,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,IAAA,qBAAU,EAAC,oBAAoB,CAAC,KAAK,EAAE,MAAM,CAAC,CAAA;QAE3E,IAAI,CAAC,gBAAgB,EAAE,CAAC;YACtB,OAAO,IAAI,CAAA;QACb,CAAC;QAED,MAAM,eAAe,GAAG,MAAM,IAAA,uCAAwB,EAAC;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,IAAA,qBAAU,EACrC,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,IAAA,sCAA2B,EACvD,oBAAoB,EACpB,SAAS,IAAI,SAAS,CACvB,CAAA;YACD,MAAM,MAAM,GAAG,IAAA,uCAA4B,EAAC;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,IAAA,8BAAiB,EAAC,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,IAAA,+BAAY,EAAC,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,IAAA,qCAA2B,EAAC;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,IAAA,0CAAuB,EAAC,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,IAAA,8BAAiB,EAAC,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,IAAA,wCAAyB,EAAC,IAAI,CAAC,EAAE,CAAC;QAChE,oBAAoB,GAAG,MAAM,IAAA,2CAA0B,EAAC,IAAI,EAAE,GAAG,CAAC,CAAA;QAClE,MAAM,eAAe,GAAG,oBAAoB,EAAE,IAAI;YAChD,CAAC,CAAC,IAAA,+BAAkB,EAAC,oBAAoB,CAAC,IAAI,CAAC;YAC/C,CAAC,CAAC,EAAE,CAAA;QACN,IAAI,eAAe,CAAC,MAAM,GAAG,iDAAkC,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,IAAA,qCAA2B,EAAC;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,IAAA,0CAAuB,EAAC,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"}
|
|
@@ -1,80 +0,0 @@
|
|
|
1
|
-
"use strict";
|
|
2
|
-
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
-
exports.extractJsonLdContent = extractJsonLdContent;
|
|
4
|
-
const cheerio_1 = require("cheerio");
|
|
5
|
-
const cleaner_js_1 = require("./cleaner.js");
|
|
6
|
-
function extractJsonLdContent(html) {
|
|
7
|
-
try {
|
|
8
|
-
const $ = (0, cheerio_1.load)(html);
|
|
9
|
-
const scripts = $('script[type="application/ld+json"]').toArray();
|
|
10
|
-
const candidates = [];
|
|
11
|
-
for (const script of scripts) {
|
|
12
|
-
const raw = $(script).text();
|
|
13
|
-
if (!raw)
|
|
14
|
-
continue;
|
|
15
|
-
try {
|
|
16
|
-
const data = JSON.parse(raw);
|
|
17
|
-
collectCandidates(data, candidates);
|
|
18
|
-
}
|
|
19
|
-
catch {
|
|
20
|
-
// ignore malformed jsonld
|
|
21
|
-
}
|
|
22
|
-
}
|
|
23
|
-
if (candidates.length === 0)
|
|
24
|
-
return null;
|
|
25
|
-
const sorted = candidates
|
|
26
|
-
.map((c) => ({
|
|
27
|
-
title: c.title ? (0, cleaner_js_1.normalizeCandidate)(c.title) : null,
|
|
28
|
-
description: c.description ? (0, cleaner_js_1.normalizeCandidate)(c.description) : null,
|
|
29
|
-
type: c.type ? (0, cleaner_js_1.normalizeCandidate)(c.type) : null,
|
|
30
|
-
}))
|
|
31
|
-
.filter((c) => c.title || c.description)
|
|
32
|
-
.sort((a, b) => (b.description?.length ?? 0) - (a.description?.length ?? 0));
|
|
33
|
-
return sorted[0] ?? null;
|
|
34
|
-
}
|
|
35
|
-
catch {
|
|
36
|
-
return null;
|
|
37
|
-
}
|
|
38
|
-
}
|
|
39
|
-
function collectCandidates(input, out) {
|
|
40
|
-
if (!input)
|
|
41
|
-
return;
|
|
42
|
-
if (Array.isArray(input)) {
|
|
43
|
-
for (const item of input)
|
|
44
|
-
collectCandidates(item, out);
|
|
45
|
-
return;
|
|
46
|
-
}
|
|
47
|
-
if (typeof input !== 'object')
|
|
48
|
-
return;
|
|
49
|
-
const record = input;
|
|
50
|
-
if (record['@graph'] && Array.isArray(record['@graph'])) {
|
|
51
|
-
collectCandidates(record['@graph'], out);
|
|
52
|
-
}
|
|
53
|
-
const type = extractType(record);
|
|
54
|
-
if (type) {
|
|
55
|
-
const title = firstString(record, ['name', 'headline', 'title']);
|
|
56
|
-
const description = firstString(record, ['description', 'summary']);
|
|
57
|
-
if (title || description) {
|
|
58
|
-
out.push({ title: title ?? null, description: description ?? null, type });
|
|
59
|
-
}
|
|
60
|
-
}
|
|
61
|
-
}
|
|
62
|
-
function extractType(record) {
|
|
63
|
-
const raw = record['@type'];
|
|
64
|
-
if (typeof raw === 'string')
|
|
65
|
-
return raw.toLowerCase();
|
|
66
|
-
if (Array.isArray(raw)) {
|
|
67
|
-
const found = raw.find((entry) => typeof entry === 'string');
|
|
68
|
-
return typeof found === 'string' ? found.toLowerCase() : null;
|
|
69
|
-
}
|
|
70
|
-
return null;
|
|
71
|
-
}
|
|
72
|
-
function firstString(record, keys) {
|
|
73
|
-
for (const key of keys) {
|
|
74
|
-
const value = record[key];
|
|
75
|
-
if (typeof value === 'string' && value.trim())
|
|
76
|
-
return value.trim();
|
|
77
|
-
}
|
|
78
|
-
return null;
|
|
79
|
-
}
|
|
80
|
-
//# sourceMappingURL=jsonld.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"jsonld.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/jsonld.ts"],"names":[],"mappings":";;AAUA,oDAgCC;AA1CD,qCAA8B;AAE9B,6CAAiD;AAQjD,SAAgB,oBAAoB,CAAC,IAAY;IAC/C,IAAI,CAAC;QACH,MAAM,CAAC,GAAG,IAAA,cAAI,EAAC,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,IAAA,+BAAkB,EAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI;YACnD,WAAW,EAAE,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,IAAA,+BAAkB,EAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,IAAI;YACrE,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAA,+BAAkB,EAAC,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"}
|