@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
|
@@ -1,355 +0,0 @@
|
|
|
1
|
-
"use strict";
|
|
2
|
-
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
-
exports.transcribeMediaUrl = transcribeMediaUrl;
|
|
4
|
-
exports.probeRemoteMedia = probeRemoteMedia;
|
|
5
|
-
exports.downloadCappedBytes = downloadCappedBytes;
|
|
6
|
-
exports.downloadToFile = downloadToFile;
|
|
7
|
-
exports.normalizeHeaderType = normalizeHeaderType;
|
|
8
|
-
exports.parseContentLength = parseContentLength;
|
|
9
|
-
exports.filenameFromUrl = filenameFromUrl;
|
|
10
|
-
exports.formatBytes = formatBytes;
|
|
11
|
-
const node_crypto_1 = require("node:crypto");
|
|
12
|
-
const node_fs_1 = require("node:fs");
|
|
13
|
-
const node_os_1 = require("node:os");
|
|
14
|
-
const node_path_1 = require("node:path");
|
|
15
|
-
const whisper_js_1 = require("../../../../transcription/whisper.js");
|
|
16
|
-
const constants_js_1 = require("./constants.js");
|
|
17
|
-
async function transcribeMediaUrl({ fetchImpl, url, filenameHint, durationSecondsHint, openaiApiKey, falApiKey, notes, progress, }) {
|
|
18
|
-
const canChunk = await (0, whisper_js_1.isFfmpegAvailable)();
|
|
19
|
-
const providerHint = (await (0, whisper_js_1.isWhisperCppReady)())
|
|
20
|
-
? 'cpp'
|
|
21
|
-
: openaiApiKey && falApiKey
|
|
22
|
-
? 'openai->fal'
|
|
23
|
-
: openaiApiKey
|
|
24
|
-
? 'openai'
|
|
25
|
-
: falApiKey
|
|
26
|
-
? 'fal'
|
|
27
|
-
: 'unknown';
|
|
28
|
-
const head = await probeRemoteMedia(fetchImpl, url);
|
|
29
|
-
if (head.contentLength !== null && head.contentLength > constants_js_1.MAX_REMOTE_MEDIA_BYTES) {
|
|
30
|
-
throw new Error(`Remote media too large (${formatBytes(head.contentLength)}). Limit is ${formatBytes(constants_js_1.MAX_REMOTE_MEDIA_BYTES)}.`);
|
|
31
|
-
}
|
|
32
|
-
const mediaType = head.mediaType ?? 'application/octet-stream';
|
|
33
|
-
const filename = head.filename ?? filenameHint;
|
|
34
|
-
const totalBytes = head.contentLength;
|
|
35
|
-
progress?.onProgress?.({
|
|
36
|
-
kind: 'transcript-media-download-start',
|
|
37
|
-
url: progress.url,
|
|
38
|
-
service: progress.service,
|
|
39
|
-
mediaUrl: url,
|
|
40
|
-
totalBytes,
|
|
41
|
-
});
|
|
42
|
-
const modelId = providerHint === 'cpp'
|
|
43
|
-
? ((await (0, whisper_js_1.resolveWhisperCppModelNameForDisplay)()) ?? 'whisper.cpp')
|
|
44
|
-
: openaiApiKey && falApiKey
|
|
45
|
-
? 'whisper-1->fal-ai/wizper'
|
|
46
|
-
: openaiApiKey
|
|
47
|
-
? 'whisper-1'
|
|
48
|
-
: falApiKey
|
|
49
|
-
? 'fal-ai/wizper'
|
|
50
|
-
: null;
|
|
51
|
-
if (!canChunk) {
|
|
52
|
-
const bytes = await downloadCappedBytes(fetchImpl, url, whisper_js_1.MAX_OPENAI_UPLOAD_BYTES, {
|
|
53
|
-
totalBytes,
|
|
54
|
-
onProgress: (downloadedBytes) => progress?.onProgress?.({
|
|
55
|
-
kind: 'transcript-media-download-progress',
|
|
56
|
-
url: progress.url,
|
|
57
|
-
service: progress.service,
|
|
58
|
-
downloadedBytes,
|
|
59
|
-
totalBytes,
|
|
60
|
-
}),
|
|
61
|
-
});
|
|
62
|
-
progress?.onProgress?.({
|
|
63
|
-
kind: 'transcript-media-download-done',
|
|
64
|
-
url: progress.url,
|
|
65
|
-
service: progress.service,
|
|
66
|
-
downloadedBytes: bytes.byteLength,
|
|
67
|
-
totalBytes,
|
|
68
|
-
});
|
|
69
|
-
progress?.onProgress?.({
|
|
70
|
-
kind: 'transcript-whisper-start',
|
|
71
|
-
url: progress.url,
|
|
72
|
-
service: progress.service,
|
|
73
|
-
providerHint,
|
|
74
|
-
modelId,
|
|
75
|
-
totalDurationSeconds: durationSecondsHint,
|
|
76
|
-
parts: null,
|
|
77
|
-
});
|
|
78
|
-
notes.push(`Transcribed first ${formatBytes(bytes.byteLength)} only (ffmpeg not available)`);
|
|
79
|
-
const transcript = await (0, whisper_js_1.transcribeMediaWithWhisper)({
|
|
80
|
-
bytes,
|
|
81
|
-
mediaType,
|
|
82
|
-
filename,
|
|
83
|
-
openaiApiKey,
|
|
84
|
-
falApiKey,
|
|
85
|
-
totalDurationSeconds: durationSecondsHint,
|
|
86
|
-
onProgress: (event) => {
|
|
87
|
-
progress?.onProgress?.({
|
|
88
|
-
kind: 'transcript-whisper-progress',
|
|
89
|
-
url: progress.url,
|
|
90
|
-
service: progress.service,
|
|
91
|
-
processedDurationSeconds: event.processedDurationSeconds,
|
|
92
|
-
totalDurationSeconds: event.totalDurationSeconds,
|
|
93
|
-
partIndex: event.partIndex,
|
|
94
|
-
parts: event.parts,
|
|
95
|
-
});
|
|
96
|
-
},
|
|
97
|
-
});
|
|
98
|
-
if (transcript.notes.length > 0)
|
|
99
|
-
notes.push(...transcript.notes);
|
|
100
|
-
return { text: transcript.text, provider: transcript.provider, error: transcript.error };
|
|
101
|
-
}
|
|
102
|
-
if (head.contentLength !== null && head.contentLength <= whisper_js_1.MAX_OPENAI_UPLOAD_BYTES) {
|
|
103
|
-
const bytes = await downloadCappedBytes(fetchImpl, url, whisper_js_1.MAX_OPENAI_UPLOAD_BYTES, {
|
|
104
|
-
totalBytes,
|
|
105
|
-
onProgress: (downloadedBytes) => progress?.onProgress?.({
|
|
106
|
-
kind: 'transcript-media-download-progress',
|
|
107
|
-
url: progress.url,
|
|
108
|
-
service: progress.service,
|
|
109
|
-
downloadedBytes,
|
|
110
|
-
totalBytes,
|
|
111
|
-
}),
|
|
112
|
-
});
|
|
113
|
-
progress?.onProgress?.({
|
|
114
|
-
kind: 'transcript-media-download-done',
|
|
115
|
-
url: progress.url,
|
|
116
|
-
service: progress.service,
|
|
117
|
-
downloadedBytes: bytes.byteLength,
|
|
118
|
-
totalBytes,
|
|
119
|
-
});
|
|
120
|
-
progress?.onProgress?.({
|
|
121
|
-
kind: 'transcript-whisper-start',
|
|
122
|
-
url: progress.url,
|
|
123
|
-
service: progress.service,
|
|
124
|
-
providerHint,
|
|
125
|
-
modelId,
|
|
126
|
-
totalDurationSeconds: durationSecondsHint,
|
|
127
|
-
parts: null,
|
|
128
|
-
});
|
|
129
|
-
const transcript = await (0, whisper_js_1.transcribeMediaWithWhisper)({
|
|
130
|
-
bytes,
|
|
131
|
-
mediaType,
|
|
132
|
-
filename,
|
|
133
|
-
openaiApiKey,
|
|
134
|
-
falApiKey,
|
|
135
|
-
totalDurationSeconds: durationSecondsHint,
|
|
136
|
-
onProgress: (event) => {
|
|
137
|
-
progress?.onProgress?.({
|
|
138
|
-
kind: 'transcript-whisper-progress',
|
|
139
|
-
url: progress.url,
|
|
140
|
-
service: progress.service,
|
|
141
|
-
processedDurationSeconds: event.processedDurationSeconds,
|
|
142
|
-
totalDurationSeconds: event.totalDurationSeconds,
|
|
143
|
-
partIndex: event.partIndex,
|
|
144
|
-
parts: event.parts,
|
|
145
|
-
});
|
|
146
|
-
},
|
|
147
|
-
});
|
|
148
|
-
if (transcript.notes.length > 0)
|
|
149
|
-
notes.push(...transcript.notes);
|
|
150
|
-
return { text: transcript.text, provider: transcript.provider, error: transcript.error };
|
|
151
|
-
}
|
|
152
|
-
const tmpFile = (0, node_path_1.join)((0, node_os_1.tmpdir)(), `summarize-podcast-${(0, node_crypto_1.randomUUID)()}.bin`);
|
|
153
|
-
try {
|
|
154
|
-
const downloadedBytes = await downloadToFile(fetchImpl, url, tmpFile, {
|
|
155
|
-
totalBytes,
|
|
156
|
-
onProgress: (nextDownloadedBytes) => progress?.onProgress?.({
|
|
157
|
-
kind: 'transcript-media-download-progress',
|
|
158
|
-
url: progress.url,
|
|
159
|
-
service: progress.service,
|
|
160
|
-
downloadedBytes: nextDownloadedBytes,
|
|
161
|
-
totalBytes,
|
|
162
|
-
}),
|
|
163
|
-
});
|
|
164
|
-
progress?.onProgress?.({
|
|
165
|
-
kind: 'transcript-media-download-done',
|
|
166
|
-
url: progress.url,
|
|
167
|
-
service: progress.service,
|
|
168
|
-
downloadedBytes,
|
|
169
|
-
totalBytes,
|
|
170
|
-
});
|
|
171
|
-
const probedDurationSeconds = durationSecondsHint ?? (await (0, whisper_js_1.probeMediaDurationSecondsWithFfprobe)(tmpFile));
|
|
172
|
-
progress?.onProgress?.({
|
|
173
|
-
kind: 'transcript-whisper-start',
|
|
174
|
-
url: progress.url,
|
|
175
|
-
service: progress.service,
|
|
176
|
-
providerHint,
|
|
177
|
-
modelId,
|
|
178
|
-
totalDurationSeconds: probedDurationSeconds,
|
|
179
|
-
parts: null,
|
|
180
|
-
});
|
|
181
|
-
const transcript = await (0, whisper_js_1.transcribeMediaFileWithWhisper)({
|
|
182
|
-
filePath: tmpFile,
|
|
183
|
-
mediaType,
|
|
184
|
-
filename,
|
|
185
|
-
openaiApiKey,
|
|
186
|
-
falApiKey,
|
|
187
|
-
totalDurationSeconds: probedDurationSeconds,
|
|
188
|
-
onProgress: (event) => {
|
|
189
|
-
progress?.onProgress?.({
|
|
190
|
-
kind: 'transcript-whisper-progress',
|
|
191
|
-
url: progress.url,
|
|
192
|
-
service: progress.service,
|
|
193
|
-
processedDurationSeconds: event.processedDurationSeconds,
|
|
194
|
-
totalDurationSeconds: event.totalDurationSeconds,
|
|
195
|
-
partIndex: event.partIndex,
|
|
196
|
-
parts: event.parts,
|
|
197
|
-
});
|
|
198
|
-
},
|
|
199
|
-
});
|
|
200
|
-
if (transcript.notes.length > 0)
|
|
201
|
-
notes.push(...transcript.notes);
|
|
202
|
-
return { text: transcript.text, provider: transcript.provider, error: transcript.error };
|
|
203
|
-
}
|
|
204
|
-
finally {
|
|
205
|
-
await node_fs_1.promises.unlink(tmpFile).catch(() => { });
|
|
206
|
-
}
|
|
207
|
-
}
|
|
208
|
-
async function probeRemoteMedia(fetchImpl, url) {
|
|
209
|
-
try {
|
|
210
|
-
const res = await fetchImpl(url, {
|
|
211
|
-
method: 'HEAD',
|
|
212
|
-
redirect: 'follow',
|
|
213
|
-
signal: AbortSignal.timeout(constants_js_1.TRANSCRIPTION_TIMEOUT_MS),
|
|
214
|
-
});
|
|
215
|
-
if (!res.ok)
|
|
216
|
-
throw new Error('head failed');
|
|
217
|
-
const contentLength = parseContentLength(res.headers.get('content-length'));
|
|
218
|
-
const mediaType = normalizeHeaderType(res.headers.get('content-type'));
|
|
219
|
-
const filename = filenameFromUrl(url);
|
|
220
|
-
return { contentLength, mediaType, filename };
|
|
221
|
-
}
|
|
222
|
-
catch {
|
|
223
|
-
return { contentLength: null, mediaType: null, filename: filenameFromUrl(url) };
|
|
224
|
-
}
|
|
225
|
-
}
|
|
226
|
-
async function downloadCappedBytes(fetchImpl, url, maxBytes, options) {
|
|
227
|
-
const res = await fetchImpl(url, {
|
|
228
|
-
redirect: 'follow',
|
|
229
|
-
headers: { Range: `bytes=0-${maxBytes - 1}` },
|
|
230
|
-
signal: AbortSignal.timeout(constants_js_1.TRANSCRIPTION_TIMEOUT_MS),
|
|
231
|
-
});
|
|
232
|
-
if (!res.ok) {
|
|
233
|
-
throw new Error(`Download failed (${res.status})`);
|
|
234
|
-
}
|
|
235
|
-
const body = res.body;
|
|
236
|
-
if (!body) {
|
|
237
|
-
const arrayBuffer = await res.arrayBuffer();
|
|
238
|
-
return new Uint8Array(arrayBuffer.slice(0, maxBytes));
|
|
239
|
-
}
|
|
240
|
-
const reader = body.getReader();
|
|
241
|
-
const chunks = [];
|
|
242
|
-
let total = 0;
|
|
243
|
-
let lastReported = 0;
|
|
244
|
-
try {
|
|
245
|
-
while (total < maxBytes) {
|
|
246
|
-
const { value, done } = await reader.read();
|
|
247
|
-
if (done)
|
|
248
|
-
break;
|
|
249
|
-
if (!value)
|
|
250
|
-
continue;
|
|
251
|
-
const remaining = maxBytes - total;
|
|
252
|
-
const next = value.byteLength > remaining ? value.slice(0, remaining) : value;
|
|
253
|
-
chunks.push(next);
|
|
254
|
-
total += next.byteLength;
|
|
255
|
-
if (total - lastReported >= 64 * 1024) {
|
|
256
|
-
lastReported = total;
|
|
257
|
-
options?.onProgress?.(total);
|
|
258
|
-
}
|
|
259
|
-
if (total >= maxBytes)
|
|
260
|
-
break;
|
|
261
|
-
}
|
|
262
|
-
}
|
|
263
|
-
finally {
|
|
264
|
-
await reader.cancel().catch(() => { });
|
|
265
|
-
}
|
|
266
|
-
options?.onProgress?.(total);
|
|
267
|
-
const out = new Uint8Array(total);
|
|
268
|
-
let offset = 0;
|
|
269
|
-
for (const chunk of chunks) {
|
|
270
|
-
out.set(chunk, offset);
|
|
271
|
-
offset += chunk.byteLength;
|
|
272
|
-
}
|
|
273
|
-
return out;
|
|
274
|
-
}
|
|
275
|
-
async function downloadToFile(fetchImpl, url, filePath, options) {
|
|
276
|
-
const res = await fetchImpl(url, {
|
|
277
|
-
redirect: 'follow',
|
|
278
|
-
signal: AbortSignal.timeout(constants_js_1.TRANSCRIPTION_TIMEOUT_MS),
|
|
279
|
-
});
|
|
280
|
-
if (!res.ok) {
|
|
281
|
-
throw new Error(`Download failed (${res.status})`);
|
|
282
|
-
}
|
|
283
|
-
const body = res.body;
|
|
284
|
-
if (!body) {
|
|
285
|
-
const bytes = new Uint8Array(await res.arrayBuffer());
|
|
286
|
-
await node_fs_1.promises.writeFile(filePath, bytes);
|
|
287
|
-
options?.onProgress?.(bytes.byteLength);
|
|
288
|
-
return bytes.byteLength;
|
|
289
|
-
}
|
|
290
|
-
const handle = await node_fs_1.promises.open(filePath, 'w');
|
|
291
|
-
let downloadedBytes = 0;
|
|
292
|
-
let lastReported = 0;
|
|
293
|
-
try {
|
|
294
|
-
const reader = body.getReader();
|
|
295
|
-
try {
|
|
296
|
-
while (true) {
|
|
297
|
-
const { value, done } = await reader.read();
|
|
298
|
-
if (done)
|
|
299
|
-
break;
|
|
300
|
-
if (!value)
|
|
301
|
-
continue;
|
|
302
|
-
await handle.write(value);
|
|
303
|
-
downloadedBytes += value.byteLength;
|
|
304
|
-
if (downloadedBytes - lastReported >= 128 * 1024) {
|
|
305
|
-
lastReported = downloadedBytes;
|
|
306
|
-
options?.onProgress?.(downloadedBytes);
|
|
307
|
-
}
|
|
308
|
-
}
|
|
309
|
-
options?.onProgress?.(downloadedBytes);
|
|
310
|
-
}
|
|
311
|
-
finally {
|
|
312
|
-
await reader.cancel().catch(() => { });
|
|
313
|
-
}
|
|
314
|
-
}
|
|
315
|
-
finally {
|
|
316
|
-
await handle.close().catch(() => { });
|
|
317
|
-
}
|
|
318
|
-
return downloadedBytes;
|
|
319
|
-
}
|
|
320
|
-
function normalizeHeaderType(value) {
|
|
321
|
-
if (!value)
|
|
322
|
-
return null;
|
|
323
|
-
const trimmed = value.trim();
|
|
324
|
-
if (!trimmed)
|
|
325
|
-
return null;
|
|
326
|
-
return trimmed.split(';')[0]?.trim().toLowerCase() ?? null;
|
|
327
|
-
}
|
|
328
|
-
function parseContentLength(value) {
|
|
329
|
-
if (!value)
|
|
330
|
-
return null;
|
|
331
|
-
const parsed = Number(value);
|
|
332
|
-
return Number.isFinite(parsed) && parsed > 0 ? Math.floor(parsed) : null;
|
|
333
|
-
}
|
|
334
|
-
function filenameFromUrl(url) {
|
|
335
|
-
try {
|
|
336
|
-
const parsed = new URL(url);
|
|
337
|
-
const base = parsed.pathname.split('/').pop() ?? '';
|
|
338
|
-
return base.trim().length > 0 ? base : null;
|
|
339
|
-
}
|
|
340
|
-
catch {
|
|
341
|
-
return null;
|
|
342
|
-
}
|
|
343
|
-
}
|
|
344
|
-
function formatBytes(bytes) {
|
|
345
|
-
const units = ['B', 'KB', 'MB', 'GB'];
|
|
346
|
-
let value = bytes;
|
|
347
|
-
let idx = 0;
|
|
348
|
-
while (value >= 1024 && idx < units.length - 1) {
|
|
349
|
-
value /= 1024;
|
|
350
|
-
idx += 1;
|
|
351
|
-
}
|
|
352
|
-
const decimals = value >= 10 || idx === 0 ? 0 : 1;
|
|
353
|
-
return `${value.toFixed(decimals)}${units[idx]}`;
|
|
354
|
-
}
|
|
355
|
-
//# sourceMappingURL=media.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"media.js","sourceRoot":"","sources":["../../../../../../src/content/transcript/providers/podcast/media.ts"],"names":[],"mappings":";;AA4BA,gDA+NC;AAED,4CAkBC;AAED,kDAmDC;AAED,wCA8CC;AAED,kDAKC;AAED,gDAIC;AAED,0CAQC;AAED,kCAUC;AAvZD,6CAAwC;AACxC,qCAAwC;AACxC,qCAAgC;AAChC,yCAAgC;AAChC,qEAQ6C;AAE7C,iDAAiF;AAc1E,KAAK,UAAU,kBAAkB,CAAC,EACvC,SAAS,EACT,GAAG,EACH,YAAY,EACZ,mBAAmB,EACnB,YAAY,EACZ,SAAS,EACT,KAAK,EACL,QAAQ,GAcT;IACC,MAAM,QAAQ,GAAG,MAAM,IAAA,8BAAiB,GAAE,CAAA;IAC1C,MAAM,YAAY,GAChB,CAAC,MAAM,IAAA,8BAAiB,GAAE,CAAC;QACzB,CAAC,CAAC,KAAK;QACP,CAAC,CAAC,YAAY,IAAI,SAAS;YACzB,CAAC,CAAC,aAAa;YACf,CAAC,CAAC,YAAY;gBACZ,CAAC,CAAC,QAAQ;gBACV,CAAC,CAAC,SAAS;oBACT,CAAC,CAAC,KAAK;oBACP,CAAC,CAAC,SAAS,CAAA;IAErB,MAAM,IAAI,GAAG,MAAM,gBAAgB,CAAC,SAAS,EAAE,GAAG,CAAC,CAAA;IACnD,IAAI,IAAI,CAAC,aAAa,KAAK,IAAI,IAAI,IAAI,CAAC,aAAa,GAAG,qCAAsB,EAAE,CAAC;QAC/E,MAAM,IAAI,KAAK,CACb,2BAA2B,WAAW,CAAC,IAAI,CAAC,aAAa,CAAC,eAAe,WAAW,CAAC,qCAAsB,CAAC,GAAG,CAChH,CAAA;IACH,CAAC;IAED,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,IAAI,0BAA0B,CAAA;IAC9D,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,YAAY,CAAA;IAC9C,MAAM,UAAU,GAAG,IAAI,CAAC,aAAa,CAAA;IAErC,QAAQ,EAAE,UAAU,EAAE,CAAC;QACrB,IAAI,EAAE,iCAAiC;QACvC,GAAG,EAAE,QAAQ,CAAC,GAAG;QACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;QACzB,QAAQ,EAAE,GAAG;QACb,UAAU;KACX,CAAC,CAAA;IAEF,MAAM,OAAO,GACX,YAAY,KAAK,KAAK;QACpB,CAAC,CAAC,CAAC,CAAC,MAAM,IAAA,iDAAoC,GAAE,CAAC,IAAI,aAAa,CAAC;QACnE,CAAC,CAAC,YAAY,IAAI,SAAS;YACzB,CAAC,CAAC,0BAA0B;YAC5B,CAAC,CAAC,YAAY;gBACZ,CAAC,CAAC,WAAW;gBACb,CAAC,CAAC,SAAS;oBACT,CAAC,CAAC,eAAe;oBACjB,CAAC,CAAC,IAAI,CAAA;IAChB,IAAI,CAAC,QAAQ,EAAE,CAAC;QACd,MAAM,KAAK,GAAG,MAAM,mBAAmB,CAAC,SAAS,EAAE,GAAG,EAAE,oCAAuB,EAAE;YAC/E,UAAU;YACV,UAAU,EAAE,CAAC,eAAe,EAAE,EAAE,CAC9B,QAAQ,EAAE,UAAU,EAAE,CAAC;gBACrB,IAAI,EAAE,oCAAoC;gBAC1C,GAAG,EAAE,QAAQ,CAAC,GAAG;gBACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;gBACzB,eAAe;gBACf,UAAU;aACX,CAAC;SACL,CAAC,CAAA;QACF,QAAQ,EAAE,UAAU,EAAE,CAAC;YACrB,IAAI,EAAE,gCAAgC;YACtC,GAAG,EAAE,QAAQ,CAAC,GAAG;YACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;YACzB,eAAe,EAAE,KAAK,CAAC,UAAU;YACjC,UAAU;SACX,CAAC,CAAA;QACF,QAAQ,EAAE,UAAU,EAAE,CAAC;YACrB,IAAI,EAAE,0BAA0B;YAChC,GAAG,EAAE,QAAQ,CAAC,GAAG;YACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;YACzB,YAAY;YACZ,OAAO;YACP,oBAAoB,EAAE,mBAAmB;YACzC,KAAK,EAAE,IAAI;SACZ,CAAC,CAAA;QACF,KAAK,CAAC,IAAI,CAAC,qBAAqB,WAAW,CAAC,KAAK,CAAC,UAAU,CAAC,8BAA8B,CAAC,CAAA;QAC5F,MAAM,UAAU,GAAG,MAAM,IAAA,uCAA0B,EAAC;YAClD,KAAK;YACL,SAAS;YACT,QAAQ;YACR,YAAY;YACZ,SAAS;YACT,oBAAoB,EAAE,mBAAmB;YACzC,UAAU,EAAE,CAAC,KAAK,EAAE,EAAE;gBACpB,QAAQ,EAAE,UAAU,EAAE,CAAC;oBACrB,IAAI,EAAE,6BAA6B;oBACnC,GAAG,EAAE,QAAQ,CAAC,GAAG;oBACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;oBACzB,wBAAwB,EAAE,KAAK,CAAC,wBAAwB;oBACxD,oBAAoB,EAAE,KAAK,CAAC,oBAAoB;oBAChD,SAAS,EAAE,KAAK,CAAC,SAAS;oBAC1B,KAAK,EAAE,KAAK,CAAC,KAAK;iBACnB,CAAC,CAAA;YACJ,CAAC;SACF,CAAC,CAAA;QACF,IAAI,UAAU,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC;YAAE,KAAK,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,KAAK,CAAC,CAAA;QAChE,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,IAAI,EAAE,QAAQ,EAAE,UAAU,CAAC,QAAQ,EAAE,KAAK,EAAE,UAAU,CAAC,KAAK,EAAE,CAAA;IAC1F,CAAC;IAED,IAAI,IAAI,CAAC,aAAa,KAAK,IAAI,IAAI,IAAI,CAAC,aAAa,IAAI,oCAAuB,EAAE,CAAC;QACjF,MAAM,KAAK,GAAG,MAAM,mBAAmB,CAAC,SAAS,EAAE,GAAG,EAAE,oCAAuB,EAAE;YAC/E,UAAU;YACV,UAAU,EAAE,CAAC,eAAe,EAAE,EAAE,CAC9B,QAAQ,EAAE,UAAU,EAAE,CAAC;gBACrB,IAAI,EAAE,oCAAoC;gBAC1C,GAAG,EAAE,QAAQ,CAAC,GAAG;gBACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;gBACzB,eAAe;gBACf,UAAU;aACX,CAAC;SACL,CAAC,CAAA;QACF,QAAQ,EAAE,UAAU,EAAE,CAAC;YACrB,IAAI,EAAE,gCAAgC;YACtC,GAAG,EAAE,QAAQ,CAAC,GAAG;YACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;YACzB,eAAe,EAAE,KAAK,CAAC,UAAU;YACjC,UAAU;SACX,CAAC,CAAA;QACF,QAAQ,EAAE,UAAU,EAAE,CAAC;YACrB,IAAI,EAAE,0BAA0B;YAChC,GAAG,EAAE,QAAQ,CAAC,GAAG;YACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;YACzB,YAAY;YACZ,OAAO;YACP,oBAAoB,EAAE,mBAAmB;YACzC,KAAK,EAAE,IAAI;SACZ,CAAC,CAAA;QACF,MAAM,UAAU,GAAG,MAAM,IAAA,uCAA0B,EAAC;YAClD,KAAK;YACL,SAAS;YACT,QAAQ;YACR,YAAY;YACZ,SAAS;YACT,oBAAoB,EAAE,mBAAmB;YACzC,UAAU,EAAE,CAAC,KAAK,EAAE,EAAE;gBACpB,QAAQ,EAAE,UAAU,EAAE,CAAC;oBACrB,IAAI,EAAE,6BAA6B;oBACnC,GAAG,EAAE,QAAQ,CAAC,GAAG;oBACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;oBACzB,wBAAwB,EAAE,KAAK,CAAC,wBAAwB;oBACxD,oBAAoB,EAAE,KAAK,CAAC,oBAAoB;oBAChD,SAAS,EAAE,KAAK,CAAC,SAAS;oBAC1B,KAAK,EAAE,KAAK,CAAC,KAAK;iBACnB,CAAC,CAAA;YACJ,CAAC;SACF,CAAC,CAAA;QACF,IAAI,UAAU,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC;YAAE,KAAK,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,KAAK,CAAC,CAAA;QAChE,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,IAAI,EAAE,QAAQ,EAAE,UAAU,CAAC,QAAQ,EAAE,KAAK,EAAE,UAAU,CAAC,KAAK,EAAE,CAAA;IAC1F,CAAC;IAED,MAAM,OAAO,GAAG,IAAA,gBAAI,EAAC,IAAA,gBAAM,GAAE,EAAE,qBAAqB,IAAA,wBAAU,GAAE,MAAM,CAAC,CAAA;IACvE,IAAI,CAAC;QACH,MAAM,eAAe,GAAG,MAAM,cAAc,CAAC,SAAS,EAAE,GAAG,EAAE,OAAO,EAAE;YACpE,UAAU;YACV,UAAU,EAAE,CAAC,mBAAmB,EAAE,EAAE,CAClC,QAAQ,EAAE,UAAU,EAAE,CAAC;gBACrB,IAAI,EAAE,oCAAoC;gBAC1C,GAAG,EAAE,QAAQ,CAAC,GAAG;gBACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;gBACzB,eAAe,EAAE,mBAAmB;gBACpC,UAAU;aACX,CAAC;SACL,CAAC,CAAA;QACF,QAAQ,EAAE,UAAU,EAAE,CAAC;YACrB,IAAI,EAAE,gCAAgC;YACtC,GAAG,EAAE,QAAQ,CAAC,GAAG;YACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;YACzB,eAAe;YACf,UAAU;SACX,CAAC,CAAA;QAEF,MAAM,qBAAqB,GACzB,mBAAmB,IAAI,CAAC,MAAM,IAAA,iDAAoC,EAAC,OAAO,CAAC,CAAC,CAAA;QAC9E,QAAQ,EAAE,UAAU,EAAE,CAAC;YACrB,IAAI,EAAE,0BAA0B;YAChC,GAAG,EAAE,QAAQ,CAAC,GAAG;YACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;YACzB,YAAY;YACZ,OAAO;YACP,oBAAoB,EAAE,qBAAqB;YAC3C,KAAK,EAAE,IAAI;SACZ,CAAC,CAAA;QACF,MAAM,UAAU,GAAG,MAAM,IAAA,2CAA8B,EAAC;YACtD,QAAQ,EAAE,OAAO;YACjB,SAAS;YACT,QAAQ;YACR,YAAY;YACZ,SAAS;YACT,oBAAoB,EAAE,qBAAqB;YAC3C,UAAU,EAAE,CAAC,KAAK,EAAE,EAAE;gBACpB,QAAQ,EAAE,UAAU,EAAE,CAAC;oBACrB,IAAI,EAAE,6BAA6B;oBACnC,GAAG,EAAE,QAAQ,CAAC,GAAG;oBACjB,OAAO,EAAE,QAAQ,CAAC,OAAO;oBACzB,wBAAwB,EAAE,KAAK,CAAC,wBAAwB;oBACxD,oBAAoB,EAAE,KAAK,CAAC,oBAAoB;oBAChD,SAAS,EAAE,KAAK,CAAC,SAAS;oBAC1B,KAAK,EAAE,KAAK,CAAC,KAAK;iBACnB,CAAC,CAAA;YACJ,CAAC;SACF,CAAC,CAAA;QACF,IAAI,UAAU,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC;YAAE,KAAK,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,KAAK,CAAC,CAAA;QAChE,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,IAAI,EAAE,QAAQ,EAAE,UAAU,CAAC,QAAQ,EAAE,KAAK,EAAE,UAAU,CAAC,KAAK,EAAE,CAAA;IAC1F,CAAC;YAAS,CAAC;QACT,MAAM,kBAAE,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAA;IAC1C,CAAC;AACH,CAAC;AAEM,KAAK,UAAU,gBAAgB,CACpC,SAAuB,EACvB,GAAW;IAEX,IAAI,CAAC;QACH,MAAM,GAAG,GAAG,MAAM,SAAS,CAAC,GAAG,EAAE;YAC/B,MAAM,EAAE,MAAM;YACd,QAAQ,EAAE,QAAQ;YAClB,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,uCAAwB,CAAC;SACtD,CAAC,CAAA;QACF,IAAI,CAAC,GAAG,CAAC,EAAE;YAAE,MAAM,IAAI,KAAK,CAAC,aAAa,CAAC,CAAA;QAC3C,MAAM,aAAa,GAAG,kBAAkB,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,gBAAgB,CAAC,CAAC,CAAA;QAC3E,MAAM,SAAS,GAAG,mBAAmB,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,cAAc,CAAC,CAAC,CAAA;QACtE,MAAM,QAAQ,GAAG,eAAe,CAAC,GAAG,CAAC,CAAA;QACrC,OAAO,EAAE,aAAa,EAAE,SAAS,EAAE,QAAQ,EAAE,CAAA;IAC/C,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,EAAE,aAAa,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,QAAQ,EAAE,eAAe,CAAC,GAAG,CAAC,EAAE,CAAA;IACjF,CAAC;AACH,CAAC;AAEM,KAAK,UAAU,mBAAmB,CACvC,SAAuB,EACvB,GAAW,EACX,QAAgB,EAChB,OAAgG;IAEhG,MAAM,GAAG,GAAG,MAAM,SAAS,CAAC,GAAG,EAAE;QAC/B,QAAQ,EAAE,QAAQ;QAClB,OAAO,EAAE,EAAE,KAAK,EAAE,WAAW,QAAQ,GAAG,CAAC,EAAE,EAAE;QAC7C,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,uCAAwB,CAAC;KACtD,CAAC,CAAA;IACF,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC;QACZ,MAAM,IAAI,KAAK,CAAC,oBAAoB,GAAG,CAAC,MAAM,GAAG,CAAC,CAAA;IACpD,CAAC;IACD,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAA;IACrB,IAAI,CAAC,IAAI,EAAE,CAAC;QACV,MAAM,WAAW,GAAG,MAAM,GAAG,CAAC,WAAW,EAAE,CAAA;QAC3C,OAAO,IAAI,UAAU,CAAC,WAAW,CAAC,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC,CAAA;IACvD,CAAC;IAED,MAAM,MAAM,GAAG,IAAI,CAAC,SAAS,EAAE,CAAA;IAC/B,MAAM,MAAM,GAAiB,EAAE,CAAA;IAC/B,IAAI,KAAK,GAAG,CAAC,CAAA;IACb,IAAI,YAAY,GAAG,CAAC,CAAA;IACpB,IAAI,CAAC;QACH,OAAO,KAAK,GAAG,QAAQ,EAAE,CAAC;YACxB,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAA;YAC3C,IAAI,IAAI;gBAAE,MAAK;YACf,IAAI,CAAC,KAAK;gBAAE,SAAQ;YACpB,MAAM,SAAS,GAAG,QAAQ,GAAG,KAAK,CAAA;YAClC,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,GAAG,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,CAAC,CAAC,KAAK,CAAA;YAC7E,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YACjB,KAAK,IAAI,IAAI,CAAC,UAAU,CAAA;YACxB,IAAI,KAAK,GAAG,YAAY,IAAI,EAAE,GAAG,IAAI,EAAE,CAAC;gBACtC,YAAY,GAAG,KAAK,CAAA;gBACpB,OAAO,EAAE,UAAU,EAAE,CAAC,KAAK,CAAC,CAAA;YAC9B,CAAC;YACD,IAAI,KAAK,IAAI,QAAQ;gBAAE,MAAK;QAC9B,CAAC;IACH,CAAC;YAAS,CAAC;QACT,MAAM,MAAM,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAA;IACvC,CAAC;IACD,OAAO,EAAE,UAAU,EAAE,CAAC,KAAK,CAAC,CAAA;IAE5B,MAAM,GAAG,GAAG,IAAI,UAAU,CAAC,KAAK,CAAC,CAAA;IACjC,IAAI,MAAM,GAAG,CAAC,CAAA;IACd,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,GAAG,CAAC,GAAG,CAAC,KAAK,EAAE,MAAM,CAAC,CAAA;QACtB,MAAM,IAAI,KAAK,CAAC,UAAU,CAAA;IAC5B,CAAC;IACD,OAAO,GAAG,CAAA;AACZ,CAAC;AAEM,KAAK,UAAU,cAAc,CAClC,SAAuB,EACvB,GAAW,EACX,QAAgB,EAChB,OAAgG;IAEhG,MAAM,GAAG,GAAG,MAAM,SAAS,CAAC,GAAG,EAAE;QAC/B,QAAQ,EAAE,QAAQ;QAClB,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,uCAAwB,CAAC;KACtD,CAAC,CAAA;IACF,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC;QACZ,MAAM,IAAI,KAAK,CAAC,oBAAoB,GAAG,CAAC,MAAM,GAAG,CAAC,CAAA;IACpD,CAAC;IACD,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAA;IACrB,IAAI,CAAC,IAAI,EAAE,CAAC;QACV,MAAM,KAAK,GAAG,IAAI,UAAU,CAAC,MAAM,GAAG,CAAC,WAAW,EAAE,CAAC,CAAA;QACrD,MAAM,kBAAE,CAAC,SAAS,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAA;QACnC,OAAO,EAAE,UAAU,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,CAAA;QACvC,OAAO,KAAK,CAAC,UAAU,CAAA;IACzB,CAAC;IAED,MAAM,MAAM,GAAG,MAAM,kBAAE,CAAC,IAAI,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAA;IAC3C,IAAI,eAAe,GAAG,CAAC,CAAA;IACvB,IAAI,YAAY,GAAG,CAAC,CAAA;IACpB,IAAI,CAAC;QACH,MAAM,MAAM,GAAG,IAAI,CAAC,SAAS,EAAE,CAAA;QAC/B,IAAI,CAAC;YACH,OAAO,IAAI,EAAE,CAAC;gBACZ,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAA;gBAC3C,IAAI,IAAI;oBAAE,MAAK;gBACf,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBACpB,MAAM,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,CAAA;gBACzB,eAAe,IAAI,KAAK,CAAC,UAAU,CAAA;gBACnC,IAAI,eAAe,GAAG,YAAY,IAAI,GAAG,GAAG,IAAI,EAAE,CAAC;oBACjD,YAAY,GAAG,eAAe,CAAA;oBAC9B,OAAO,EAAE,UAAU,EAAE,CAAC,eAAe,CAAC,CAAA;gBACxC,CAAC;YACH,CAAC;YACD,OAAO,EAAE,UAAU,EAAE,CAAC,eAAe,CAAC,CAAA;QACxC,CAAC;gBAAS,CAAC;YACT,MAAM,MAAM,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAA;QACvC,CAAC;IACH,CAAC;YAAS,CAAC;QACT,MAAM,MAAM,CAAC,KAAK,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAA;IACtC,CAAC;IACD,OAAO,eAAe,CAAA;AACxB,CAAC;AAED,SAAgB,mBAAmB,CAAC,KAAoB;IACtD,IAAI,CAAC,KAAK;QAAE,OAAO,IAAI,CAAA;IACvB,MAAM,OAAO,GAAG,KAAK,CAAC,IAAI,EAAE,CAAA;IAC5B,IAAI,CAAC,OAAO;QAAE,OAAO,IAAI,CAAA;IACzB,OAAO,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,WAAW,EAAE,IAAI,IAAI,CAAA;AAC5D,CAAC;AAED,SAAgB,kBAAkB,CAAC,KAAoB;IACrD,IAAI,CAAC,KAAK;QAAE,OAAO,IAAI,CAAA;IACvB,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,CAAA;IAC5B,OAAO,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;AAC1E,CAAC;AAED,SAAgB,eAAe,CAAC,GAAW;IACzC,IAAI,CAAC;QACH,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAA;QAC3B,MAAM,IAAI,GAAG,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,IAAI,EAAE,CAAA;QACnD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAA;IAC7C,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,IAAI,CAAA;IACb,CAAC;AACH,CAAC;AAED,SAAgB,WAAW,CAAC,KAAa;IACvC,MAAM,KAAK,GAAG,CAAC,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAA;IACrC,IAAI,KAAK,GAAG,KAAK,CAAA;IACjB,IAAI,GAAG,GAAG,CAAC,CAAA;IACX,OAAO,KAAK,IAAI,IAAI,IAAI,GAAG,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC/C,KAAK,IAAI,IAAI,CAAA;QACb,GAAG,IAAI,CAAC,CAAA;IACV,CAAC;IACD,MAAM,QAAQ,GAAG,KAAK,IAAI,EAAE,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;IACjD,OAAO,GAAG,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,GAAG,KAAK,CAAC,GAAG,CAAC,EAAE,CAAA;AAClD,CAAC"}
|
|
@@ -1,32 +0,0 @@
|
|
|
1
|
-
"use strict";
|
|
2
|
-
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
-
exports.joinNotes = joinNotes;
|
|
4
|
-
exports.buildWhisperResult = buildWhisperResult;
|
|
5
|
-
function joinNotes(notes) {
|
|
6
|
-
return notes.length > 0 ? notes.join('; ') : null;
|
|
7
|
-
}
|
|
8
|
-
function buildWhisperResult({ attemptedProviders, notes, outcome, metadata, includeProviderOnFailure = false, }) {
|
|
9
|
-
if (outcome.text) {
|
|
10
|
-
return {
|
|
11
|
-
text: outcome.text,
|
|
12
|
-
source: 'whisper',
|
|
13
|
-
attemptedProviders,
|
|
14
|
-
notes: joinNotes(notes),
|
|
15
|
-
metadata: {
|
|
16
|
-
...metadata,
|
|
17
|
-
transcriptionProvider: outcome.provider,
|
|
18
|
-
},
|
|
19
|
-
};
|
|
20
|
-
}
|
|
21
|
-
const failureMetadata = includeProviderOnFailure && outcome.provider
|
|
22
|
-
? { ...metadata, transcriptionProvider: outcome.provider }
|
|
23
|
-
: metadata;
|
|
24
|
-
return {
|
|
25
|
-
text: null,
|
|
26
|
-
source: null,
|
|
27
|
-
attemptedProviders,
|
|
28
|
-
notes: outcome.error?.message ?? null,
|
|
29
|
-
metadata: failureMetadata,
|
|
30
|
-
};
|
|
31
|
-
}
|
|
32
|
-
//# sourceMappingURL=results.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"results.js","sourceRoot":"","sources":["../../../../../../src/content/transcript/providers/podcast/results.ts"],"names":[],"mappings":";;AAGA,8BAEC;AAED,gDAsCC;AA1CD,SAAgB,SAAS,CAAC,KAAe;IACvC,OAAO,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;AACnD,CAAC;AAED,SAAgB,kBAAkB,CAAC,EACjC,kBAAkB,EAClB,KAAK,EACL,OAAO,EACP,QAAQ,EACR,wBAAwB,GAAG,KAAK,GAOjC;IACC,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC;QACjB,OAAO;YACL,IAAI,EAAE,OAAO,CAAC,IAAI;YAClB,MAAM,EAAE,SAAS;YACjB,kBAAkB;YAClB,KAAK,EAAE,SAAS,CAAC,KAAK,CAAC;YACvB,QAAQ,EAAE;gBACR,GAAG,QAAQ;gBACX,qBAAqB,EAAE,OAAO,CAAC,QAAQ;aACxC;SACF,CAAA;IACH,CAAC;IAED,MAAM,eAAe,GACnB,wBAAwB,IAAI,OAAO,CAAC,QAAQ;QAC1C,CAAC,CAAC,EAAE,GAAG,QAAQ,EAAE,qBAAqB,EAAE,OAAO,CAAC,QAAQ,EAAE;QAC1D,CAAC,CAAC,QAAQ,CAAA;IAEd,OAAO;QACL,IAAI,EAAE,IAAI;QACV,MAAM,EAAE,IAAI;QACZ,kBAAkB;QAClB,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,OAAO,IAAI,IAAI;QACrC,QAAQ,EAAE,eAAe;KAC1B,CAAA;AACH,CAAC"}
|
|
@@ -1,262 +0,0 @@
|
|
|
1
|
-
"use strict";
|
|
2
|
-
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
-
exports.looksLikeRssOrAtomFeed = looksLikeRssOrAtomFeed;
|
|
4
|
-
exports.extractEnclosureFromFeed = extractEnclosureFromFeed;
|
|
5
|
-
exports.extractEnclosureForEpisode = extractEnclosureForEpisode;
|
|
6
|
-
exports.extractItemDurationSeconds = extractItemDurationSeconds;
|
|
7
|
-
exports.decodeXmlEntities = decodeXmlEntities;
|
|
8
|
-
exports.normalizeLooseTitle = normalizeLooseTitle;
|
|
9
|
-
exports.tryFetchTranscriptFromFeedXml = tryFetchTranscriptFromFeedXml;
|
|
10
|
-
const constants_js_1 = require("./constants.js");
|
|
11
|
-
function looksLikeRssOrAtomFeed(xml) {
|
|
12
|
-
const head = xml.slice(0, 4096).trimStart().toLowerCase();
|
|
13
|
-
if (head.startsWith('<rss') || head.includes('<rss'))
|
|
14
|
-
return true;
|
|
15
|
-
if (head.startsWith('<?xml') && (head.includes('<rss') || head.includes('<feed')))
|
|
16
|
-
return true;
|
|
17
|
-
if (head.startsWith('<feed') || head.includes('<feed'))
|
|
18
|
-
return true;
|
|
19
|
-
return false;
|
|
20
|
-
}
|
|
21
|
-
function extractEnclosureFromFeed(xml) {
|
|
22
|
-
const items = xml.match(/<item\b[\s\S]*?<\/item>/gi) ?? [];
|
|
23
|
-
for (const item of items) {
|
|
24
|
-
const enclosureUrl = extractEnclosureUrlFromItem(item);
|
|
25
|
-
if (!enclosureUrl)
|
|
26
|
-
continue;
|
|
27
|
-
return { enclosureUrl, durationSeconds: extractItemDurationSeconds(item) };
|
|
28
|
-
}
|
|
29
|
-
const enclosureMatch = xml.match(/<enclosure\b[^>]*\burl\s*=\s*(['"])([^'"]+)\1/i);
|
|
30
|
-
if (enclosureMatch?.[2]) {
|
|
31
|
-
return { enclosureUrl: enclosureMatch[2], durationSeconds: extractItemDurationSeconds(xml) };
|
|
32
|
-
}
|
|
33
|
-
const atomMatch = xml.match(/<link\b[^>]*\brel\s*=\s*(['"])enclosure\1[^>]*\bhref\s*=\s*(['"])([^'"]+)\2/i);
|
|
34
|
-
if (atomMatch?.[3]) {
|
|
35
|
-
return { enclosureUrl: atomMatch[3], durationSeconds: extractItemDurationSeconds(xml) };
|
|
36
|
-
}
|
|
37
|
-
return null;
|
|
38
|
-
}
|
|
39
|
-
function extractEnclosureForEpisode(feedXml, episodeTitle) {
|
|
40
|
-
const normalizedTarget = normalizeLooseTitle(episodeTitle);
|
|
41
|
-
const items = feedXml.match(/<item\b[\s\S]*?<\/item>/gi) ?? [];
|
|
42
|
-
for (const item of items) {
|
|
43
|
-
const title = extractItemTitle(item);
|
|
44
|
-
if (!title)
|
|
45
|
-
continue;
|
|
46
|
-
if (normalizeLooseTitle(title) !== normalizedTarget)
|
|
47
|
-
continue;
|
|
48
|
-
const enclosureUrl = extractEnclosureUrlFromItem(item);
|
|
49
|
-
if (!enclosureUrl)
|
|
50
|
-
continue;
|
|
51
|
-
return { enclosureUrl, durationSeconds: extractItemDurationSeconds(item) };
|
|
52
|
-
}
|
|
53
|
-
return null;
|
|
54
|
-
}
|
|
55
|
-
function extractItemDurationSeconds(itemXml) {
|
|
56
|
-
const match = itemXml.match(/<itunes:duration>([\s\S]*?)<\/itunes:duration>/i);
|
|
57
|
-
if (!match?.[1])
|
|
58
|
-
return null;
|
|
59
|
-
const raw = match[1]
|
|
60
|
-
.replaceAll(/<!\[CDATA\[/gi, '')
|
|
61
|
-
.replaceAll(/\]\]>/g, '')
|
|
62
|
-
.trim();
|
|
63
|
-
if (!raw)
|
|
64
|
-
return null;
|
|
65
|
-
// common forms: "HH:MM:SS", "MM:SS", "SS"
|
|
66
|
-
if (/^\d+$/.test(raw)) {
|
|
67
|
-
const seconds = Number(raw);
|
|
68
|
-
return Number.isFinite(seconds) && seconds > 0 ? seconds : null;
|
|
69
|
-
}
|
|
70
|
-
const parts = raw
|
|
71
|
-
.split(':')
|
|
72
|
-
.map((value) => value.trim())
|
|
73
|
-
.filter(Boolean);
|
|
74
|
-
if (parts.length < 2 || parts.length > 3)
|
|
75
|
-
return null;
|
|
76
|
-
const nums = parts.map((value) => Number(value));
|
|
77
|
-
if (nums.some((n) => !Number.isFinite(n) || n < 0))
|
|
78
|
-
return null;
|
|
79
|
-
const seconds = (() => {
|
|
80
|
-
if (nums.length === 3) {
|
|
81
|
-
const [hours, minutes, secondsRaw] = nums;
|
|
82
|
-
if (hours === undefined || minutes === undefined || secondsRaw === undefined)
|
|
83
|
-
return null;
|
|
84
|
-
return Math.round(hours * 3600 + minutes * 60 + secondsRaw);
|
|
85
|
-
}
|
|
86
|
-
const [minutes, secondsRaw] = nums;
|
|
87
|
-
if (minutes === undefined || secondsRaw === undefined)
|
|
88
|
-
return null;
|
|
89
|
-
return Math.round(minutes * 60 + secondsRaw);
|
|
90
|
-
})();
|
|
91
|
-
if (seconds === null)
|
|
92
|
-
return null;
|
|
93
|
-
return seconds > 0 ? seconds : null;
|
|
94
|
-
}
|
|
95
|
-
function decodeXmlEntities(value) {
|
|
96
|
-
return value
|
|
97
|
-
.replaceAll(/&/gi, '&')
|
|
98
|
-
.replaceAll(/&/g, '&')
|
|
99
|
-
.replaceAll(/</gi, '<')
|
|
100
|
-
.replaceAll(/>/gi, '>')
|
|
101
|
-
.replaceAll(/"/gi, '"')
|
|
102
|
-
.replaceAll(/'/gi, "'");
|
|
103
|
-
}
|
|
104
|
-
function normalizeLooseTitle(value) {
|
|
105
|
-
return value
|
|
106
|
-
.toLowerCase()
|
|
107
|
-
.normalize('NFKD')
|
|
108
|
-
.replaceAll(/\p{Diacritic}+/gu, '')
|
|
109
|
-
.replaceAll(/[^a-z0-9]+/g, ' ')
|
|
110
|
-
.trim();
|
|
111
|
-
}
|
|
112
|
-
async function tryFetchTranscriptFromFeedXml({ fetchImpl, feedXml, episodeTitle, notes, }) {
|
|
113
|
-
const items = feedXml.match(/<item\b[\s\S]*?<\/item>/gi) ?? [];
|
|
114
|
-
const normalizedTarget = episodeTitle ? normalizeLooseTitle(episodeTitle) : null;
|
|
115
|
-
for (const item of items) {
|
|
116
|
-
if (normalizedTarget) {
|
|
117
|
-
const title = extractItemTitle(item);
|
|
118
|
-
if (!title || normalizeLooseTitle(title) !== normalizedTarget)
|
|
119
|
-
continue;
|
|
120
|
-
}
|
|
121
|
-
const candidates = extractPodcastTranscriptCandidatesFromItem(item);
|
|
122
|
-
const preferred = selectPreferredTranscriptCandidate(candidates);
|
|
123
|
-
if (!preferred) {
|
|
124
|
-
if (normalizedTarget)
|
|
125
|
-
return null;
|
|
126
|
-
continue;
|
|
127
|
-
}
|
|
128
|
-
const transcriptUrl = decodeXmlEntities(preferred.url);
|
|
129
|
-
try {
|
|
130
|
-
const res = await fetchImpl(transcriptUrl, {
|
|
131
|
-
redirect: 'follow',
|
|
132
|
-
signal: AbortSignal.timeout(constants_js_1.TRANSCRIPTION_TIMEOUT_MS),
|
|
133
|
-
headers: { accept: 'text/vtt,text/plain,application/json;q=0.9,*/*;q=0.8' },
|
|
134
|
-
});
|
|
135
|
-
if (!res.ok)
|
|
136
|
-
throw new Error(`transcript fetch failed (${res.status})`);
|
|
137
|
-
const contentType = res.headers.get('content-type')?.toLowerCase().split(';')[0]?.trim() ?? null;
|
|
138
|
-
const hintedType = preferred.type?.toLowerCase().split(';')[0]?.trim() ?? null;
|
|
139
|
-
const effectiveType = hintedType ?? contentType;
|
|
140
|
-
const body = await res.text();
|
|
141
|
-
const text = (() => {
|
|
142
|
-
if (effectiveType === 'application/json' || transcriptUrl.toLowerCase().endsWith('.json')) {
|
|
143
|
-
try {
|
|
144
|
-
return jsonTranscriptToPlainText(JSON.parse(body));
|
|
145
|
-
}
|
|
146
|
-
catch {
|
|
147
|
-
return null;
|
|
148
|
-
}
|
|
149
|
-
}
|
|
150
|
-
if (effectiveType === 'text/vtt' || transcriptUrl.toLowerCase().endsWith('.vtt')) {
|
|
151
|
-
const plain = vttToPlainText(body);
|
|
152
|
-
return plain.length > 0 ? plain : null;
|
|
153
|
-
}
|
|
154
|
-
const plain = body.trim();
|
|
155
|
-
return plain.length > 0 ? plain : null;
|
|
156
|
-
})();
|
|
157
|
-
if (!text) {
|
|
158
|
-
if (normalizedTarget)
|
|
159
|
-
return null;
|
|
160
|
-
continue;
|
|
161
|
-
}
|
|
162
|
-
notes.push('Used RSS <podcast:transcript> (skipped Whisper)');
|
|
163
|
-
return { text, transcriptUrl, transcriptType: effectiveType };
|
|
164
|
-
}
|
|
165
|
-
catch (error) {
|
|
166
|
-
if (normalizedTarget) {
|
|
167
|
-
notes.push(`RSS <podcast:transcript> fetch failed: ${error instanceof Error ? error.message : String(error)}`);
|
|
168
|
-
return null;
|
|
169
|
-
}
|
|
170
|
-
}
|
|
171
|
-
}
|
|
172
|
-
return null;
|
|
173
|
-
}
|
|
174
|
-
function extractEnclosureUrlFromItem(xml) {
|
|
175
|
-
const enclosureMatch = xml.match(/<enclosure\b[^>]*\burl\s*=\s*(['"])([^'"]+)\1/i);
|
|
176
|
-
if (enclosureMatch?.[2])
|
|
177
|
-
return enclosureMatch[2];
|
|
178
|
-
const atomMatch = xml.match(/<link\b[^>]*\brel\s*=\s*(['"])enclosure\1[^>]*\bhref\s*=\s*(['"])([^'"]+)\2/i);
|
|
179
|
-
if (atomMatch?.[3])
|
|
180
|
-
return atomMatch[3];
|
|
181
|
-
return null;
|
|
182
|
-
}
|
|
183
|
-
function extractItemTitle(itemXml) {
|
|
184
|
-
const match = itemXml.match(/<title>([\s\S]*?)<\/title>/i);
|
|
185
|
-
if (!match?.[1])
|
|
186
|
-
return null;
|
|
187
|
-
const raw = match[1]
|
|
188
|
-
.replaceAll(/<!\[CDATA\[/gi, '')
|
|
189
|
-
.replaceAll(/\]\]>/g, '')
|
|
190
|
-
.trim();
|
|
191
|
-
return raw.length > 0 ? raw : null;
|
|
192
|
-
}
|
|
193
|
-
function extractPodcastTranscriptCandidatesFromItem(itemXml) {
|
|
194
|
-
const matches = itemXml.matchAll(/<podcast:transcript\b[^>]*\burl\s*=\s*(['"])([^'"]+)\1[^>]*>/gi);
|
|
195
|
-
const results = [];
|
|
196
|
-
for (const match of matches) {
|
|
197
|
-
const tag = match[0];
|
|
198
|
-
const url = match[2]?.trim();
|
|
199
|
-
if (!url)
|
|
200
|
-
continue;
|
|
201
|
-
const type = tag.match(/\btype\s*=\s*(['"])([^'"]+)\1/i)?.[2]?.trim() ?? null;
|
|
202
|
-
results.push({ url, type });
|
|
203
|
-
}
|
|
204
|
-
return results;
|
|
205
|
-
}
|
|
206
|
-
function selectPreferredTranscriptCandidate(candidates) {
|
|
207
|
-
if (candidates.length === 0)
|
|
208
|
-
return null;
|
|
209
|
-
const normalized = candidates.map((c) => ({
|
|
210
|
-
...c,
|
|
211
|
-
type: c.type?.toLowerCase().split(';')[0]?.trim() ?? null,
|
|
212
|
-
}));
|
|
213
|
-
const json = normalized.find((c) => c.type === 'application/json' || c.url.toLowerCase().endsWith('.json'));
|
|
214
|
-
if (json)
|
|
215
|
-
return json;
|
|
216
|
-
const vtt = normalized.find((c) => c.type === 'text/vtt' || c.url.toLowerCase().endsWith('.vtt'));
|
|
217
|
-
if (vtt)
|
|
218
|
-
return vtt;
|
|
219
|
-
return normalized[0] ?? null;
|
|
220
|
-
}
|
|
221
|
-
function vttToPlainText(raw) {
|
|
222
|
-
const lines = raw
|
|
223
|
-
.replace(/\r\n/g, '\n')
|
|
224
|
-
.split('\n')
|
|
225
|
-
.map((line) => line.trim())
|
|
226
|
-
.filter((line) => line.length > 0)
|
|
227
|
-
.filter((line) => line.toUpperCase() !== 'WEBVTT')
|
|
228
|
-
.filter((line) => !/^\d{2}:\d{2}:\d{2}\.\d{3}\s+-->\s+\d{2}:\d{2}:\d{2}\.\d{3}/.test(line))
|
|
229
|
-
.filter((line) => !/^\d+$/.test(line))
|
|
230
|
-
.filter((line) => !/^(NOTE|STYLE|REGION)\b/i.test(line));
|
|
231
|
-
return lines.join('\n').trim();
|
|
232
|
-
}
|
|
233
|
-
function jsonTranscriptToPlainText(payload) {
|
|
234
|
-
if (Array.isArray(payload)) {
|
|
235
|
-
const parts = payload
|
|
236
|
-
.map((row) => (row && typeof row === 'object' ? row.text : null))
|
|
237
|
-
.filter((t) => typeof t === 'string')
|
|
238
|
-
.map((t) => t.trim())
|
|
239
|
-
.filter(Boolean);
|
|
240
|
-
const text = parts.join('\n').trim();
|
|
241
|
-
return text.length > 0 ? text : null;
|
|
242
|
-
}
|
|
243
|
-
if (payload && typeof payload === 'object') {
|
|
244
|
-
const record = payload;
|
|
245
|
-
if (typeof record.transcript === 'string' && record.transcript.trim())
|
|
246
|
-
return record.transcript.trim();
|
|
247
|
-
if (typeof record.text === 'string' && record.text.trim())
|
|
248
|
-
return record.text.trim();
|
|
249
|
-
const segments = record.segments;
|
|
250
|
-
if (Array.isArray(segments)) {
|
|
251
|
-
const parts = segments
|
|
252
|
-
.map((row) => row && typeof row === 'object' ? row.text : null)
|
|
253
|
-
.filter((t) => typeof t === 'string')
|
|
254
|
-
.map((t) => t.trim())
|
|
255
|
-
.filter(Boolean);
|
|
256
|
-
const text = parts.join('\n').trim();
|
|
257
|
-
return text.length > 0 ? text : null;
|
|
258
|
-
}
|
|
259
|
-
}
|
|
260
|
-
return null;
|
|
261
|
-
}
|
|
262
|
-
//# sourceMappingURL=rss.js.map
|