Allow returning page if transcript extraction failed
| @@ -22,6 +22,72 @@ const visitHeaders = { | ||
| 22 | 22 | 'Sec-Fetch-User': '?1', |
| 23 | 23 | }; |
| 24 | 24 | |
| 25 | +/** | |
| 26 | + * Extract the transcript of a YouTube video | |
| 27 | + * @param {string} videoPageBody HTML of the video page | |
| 28 | + * @param {string} lang Language code | |
| 29 | + * @returns {Promise<string>} Transcript text | |
| 30 | + */ | |
| 31 | +async function extractTranscript(videoPageBody, lang) { | |
| 32 | + const he = require('he'); | |
| 33 | + const RE_XML_TRANSCRIPT = /<text start="([^"]*)" dur="([^"]*)">([^<]*)<\/text>/g; | |
| 34 | + const splittedHTML = videoPageBody.split('"captions":'); | |
| 35 | + | |
| 36 | + if (splittedHTML.length <= 1) { | |
| 37 | + if (videoPageBody.includes('class="g-recaptcha"')) { | |
| 38 | + throw new Error('Too many requests'); | |
| 39 | + } | |
| 40 | + if (!videoPageBody.includes('"playabilityStatus":')) { | |
| 41 | + throw new Error('Video is not available'); | |
| 42 | + } | |
| 43 | + throw new Error('Transcript not available'); | |
| 44 | + } | |
| 45 | + | |
| 46 | + const captions = (() => { | |
| 47 | + try { | |
| 48 | + return JSON.parse(splittedHTML[1].split(',"videoDetails')[0].replace('\n', '')); | |
| 49 | + } catch (e) { | |
| 50 | + return undefined; | |
| 51 | + } | |
| 52 | + })()?.['playerCaptionsTracklistRenderer']; | |
| 53 | + | |
| 54 | + if (!captions) { | |
| 55 | + throw new Error('Transcript disabled'); | |
| 56 | + } | |
| 57 | + | |
| 58 | + if (!('captionTracks' in captions)) { | |
| 59 | + throw new Error('Transcript not available'); | |
| 60 | + } | |
| 61 | + | |
| 62 | + if (lang && !captions.captionTracks.some(track => track.languageCode === lang)) { | |
| 63 | + throw new Error('Transcript not available in this language'); | |
| 64 | + } | |
| 65 | + | |
| 66 | + const transcriptURL = (lang ? captions.captionTracks.find(track => track.languageCode === lang) : captions.captionTracks[0]).baseUrl; | |
| 67 | + const transcriptResponse = await fetch(transcriptURL, { | |
| 68 | + headers: { | |
| 69 | + ...(lang && { 'Accept-Language': lang }), | |
| 70 | + 'User-Agent': visitHeaders['User-Agent'], | |
| 71 | + }, | |
| 72 | + }); | |
| 73 | + | |
| 74 | + if (!transcriptResponse.ok) { | |
| 75 | + throw new Error('Transcript request failed'); | |
| 76 | + } | |
| 77 | + | |
| 78 | + const transcriptBody = await transcriptResponse.text(); | |
| 79 | + const results = [...transcriptBody.matchAll(RE_XML_TRANSCRIPT)]; | |
| 80 | + const transcript = results.map((result) => ({ | |
| 81 | + text: result[3], | |
| 82 | + duration: parseFloat(result[2]), | |
| 83 | + offset: parseFloat(result[1]), | |
| 84 | + lang: lang ?? captions.captionTracks[0].languageCode, | |
| 85 | + })); | |
| 86 | + // The text is double-encoded | |
| 87 | + const transcriptText = transcript.map((line) => he.decode(he.decode(line.text))).join(' '); | |
| 88 | + return transcriptText; | |
| 89 | +} | |
| 90 | + | |
| 25 | 91 | router.post('/serpapi', jsonParser, async (request, response) => { |
| 26 | 92 | try { |
| 27 | 93 | const key = readSecret(request.user.directories, SECRET_KEYS.SERPAPI); |
| @@ -56,8 +122,6 @@ router.post('/serpapi', jsonParser, async (request, response) => { | ||
| 56 | 122 | */ |
| 57 | 123 | router.post('/transcript', jsonParser, async (request, response) => { |
| 58 | 124 | try { |
| 59 | - const he = require('he'); | |
| 60 | - const RE_XML_TRANSCRIPT = /<text start="([^"]*)" dur="([^"]*)">([^<]*)<\/text>/g; | |
| 61 | 125 | const id = request.body.id; |
| 62 | 126 | const lang = request.body.lang; |
| 63 | 127 | const json = request.body.json; |
| @@ -75,64 +139,18 @@ router.post('/transcript', jsonParser, async (request, response) => { | ||
| 75 | 139 | }); |
| 76 | 140 | |
| 77 | 141 | const videoPageBody = await videoPageResponse.text(); |
| 78 | - const splittedHTML = videoPageBody.split('"captions":'); | |
| 79 | 142 | |
| 80 | - if (splittedHTML.length <= 1) { | |
| 143 | + try { | |
| 81 | - if (videoPageBody.includes('class="g-recaptcha"')) { | |
| 144 | + const transcriptText = await extractTranscript(videoPageBody, lang); | |
| 82 | - throw new Error('Too many requests'); | |
| 145 | + return json | |
| 83 | - } | |
| 146 | + ? response.json({ transcript: transcriptText, html: videoPageBody }) | |
| 84 | - if (!videoPageBody.includes('"playabilityStatus":')) { | |
| 147 | + : response.send(transcriptText); | |
| 85 | - throw new Error('Video is not available'); | |
| 148 | + } catch (error) { | |
| 86 | - } | |
| 149 | + if (json) { | |
| 87 | - throw new Error('Transcript not available'); | |
| 150 | + return response.json({ html: videoPageBody, transcript: '' }); | |
| 88 | - } | |
| 89 | - | |
| 90 | - const captions = (() => { | |
| 91 | - try { | |
| 92 | - return JSON.parse(splittedHTML[1].split(',"videoDetails')[0].replace('\n', '')); | |
| 93 | - } catch (e) { | |
| 94 | - return undefined; | |
| 95 | 151 | } |
| 96 | - })()?.['playerCaptionsTracklistRenderer']; | |
| 152 | + throw error; | |
| 97 | - | |
| 98 | - if (!captions) { | |
| 99 | - throw new Error('Transcript disabled'); | |
| 100 | - } | |
| 101 | - | |
| 102 | - if (!('captionTracks' in captions)) { | |
| 103 | - throw new Error('Transcript not available'); | |
| 104 | - } | |
| 105 | - | |
| 106 | - if (lang && !captions.captionTracks.some(track => track.languageCode === lang)) { | |
| 107 | - throw new Error('Transcript not available in this language'); | |
| 108 | 153 | } |
| 109 | - | |
| 110 | - const transcriptURL = (lang ? captions.captionTracks.find(track => track.languageCode === lang) : captions.captionTracks[0]).baseUrl; | |
| 111 | - const transcriptResponse = await fetch(transcriptURL, { | |
| 112 | - headers: { | |
| 113 | - ...(lang && { 'Accept-Language': lang }), | |
| 114 | - 'User-Agent': visitHeaders['User-Agent'], | |
| 115 | - }, | |
| 116 | - }); | |
| 117 | - | |
| 118 | - if (!transcriptResponse.ok) { | |
| 119 | - throw new Error('Transcript request failed'); | |
| 120 | - } | |
| 121 | - | |
| 122 | - const transcriptBody = await transcriptResponse.text(); | |
| 123 | - const results = [...transcriptBody.matchAll(RE_XML_TRANSCRIPT)]; | |
| 124 | - const transcript = results.map((result) => ({ | |
| 125 | - text: result[3], | |
| 126 | - duration: parseFloat(result[2]), | |
| 127 | - offset: parseFloat(result[1]), | |
| 128 | - lang: lang ?? captions.captionTracks[0].languageCode, | |
| 129 | - })); | |
| 130 | - // The text is double-encoded | |
| 131 | - const transcriptText = transcript.map((line) => he.decode(he.decode(line.text))).join(' '); | |
| 132 | - | |
| 133 | - return json | |
| 134 | - ? response.json({ transcript: transcriptText, html: videoPageBody }) | |
| 135 | - : response.send(transcriptText); | |
| 136 | 154 | } catch (error) { |
| 137 | 155 | console.log(error); |
| 138 | 156 | return response.sendStatus(500); |