identification mostly working

index.js

@@ -1,66 +1,80 @@
-const spawn = require('child_process').spawn
+const { spawn, exec } = require('child_process')
 const fs = require('fs')
 const youtubedl = require('youtube-dl-exec')
 const tmp = require('tmp')
+const playlist = require('./playlist-search')
 
 const STREAM_URL = "https://www.youtube.com/henrikomagnifico/live"
+const DURATION_REGEX = /(\d{1,2}:\d{2})\/(\d{1,2}:\d{2})/
+const EXPIRATION_REGEX = /.+\/expire\/([0-9]{10})\/.+/
+
+let currentTrack = {}
+let nextTrackTimestamp = 0
+
+let resolvedUrl = {
+	url: '',
+	expires: 0
+}
+
 
 function readText(tmpfile, charWhitelist) {
 	return new Promise((resolve, reject) => {
-		const tesseract = charWhitelist ? spawn('tesseract', [tmpfile.name, '-', '-c', `tessedit_char_whitelist="${charWhitelist}`]) : spawn('tesseract', [tmpfile.name, '-'])
-		let recognizedString = ''
-
-		tesseract.stdout.on('data', (data) => {
-			recognizedString += data;
-		})
-
-		tesseract.on('exit', (code) => {
-			tmpfile.removeCallback()
-			if (code !== 0) {
-				reject()
-			} else {
-				resolve(recognizedString)
+	    let command = `tesseract ${tmpfile.name} -`
+		if (charWhitelist) {
+			command += ` -c tessedit_char_whitelist="${charWhitelist}"`
 		}
-		})
+		exec(command, ((error, stdout, stderr) => {
+			tmpfile.removeCallback()
+			if (error) {
+				console.error(stderr)
+				reject(error)
+			} else {
+				resolve(stdout)
+			}
+		}))
 	})
 }
 
 function thresholdImage(image, threshold) {
+    return new Promise(((resolve, reject) => {
+    	const tmpFile = tmp.fileSync()
+        exec(`convert png:${image.name} -white-threshold ${threshold}% -colorspace HSB -channel B -separate ${tmpFile.name}`, ((error, stdout, stderr) => {
+        	if (error) {
+        		console.error(stderr)
+				tmpFile.removeCallback()
+				reject(error)
+			} else {
+        		resolve(tmpFile)
+			}
+		}))
+	}))
 
 }
 
 function getRegion(frame, x, y, width, height) { // Returns Promise<Buffer>
-    console.log("retrieving region")
-	const threshold = "15%"
 	return new Promise(function (resolve, reject) {
 		const tmpFile = tmp.fileSync()
-		const convert = spawn('convert', [`png:${frame.name}`, '-negate', '-crop', `${width}x${height}+${x}+${y}`, '-white-threshold', threshold, `png:${tmpFile.name}`])
-
-		convert.stderr.on('data', data => console.error(data.toString()))
-
-		convert.on('exit', (code) => {
-			if (code !== 0) {
-				reject()
+		exec(`convert png:${frame.name} -negate -crop ${width}x${height}+${x}+${y} png:${tmpFile.name}`, ((error, stdout, stderr) => {
+			if (error) {
+				console.error(stderr)
+				tmpFile.removeCallback()
+				reject(error)
 			} else {
-				console.log("got region", tmpFile.name)
-				resolve(tmpFile);
+				resolve(tmpFile)
 			}
-		});
+		}))
 	});
 }
 
 function getFrame(url) {
-	return new Promise((resolve, reject) => { // Returns Promise<Buffer>
+	return new Promise((resolve, reject) => {
 	    const frameFile = tmp.fileSync()
-		const ffmpeg = spawn('ffmpeg', ['-i', url, '-y', '-f', 'image2', '-c:v', 'png', '-frames:v', '1', frameFile.name])
-
-		ffmpeg.stderr.on('data', data => console.error(data.toString()))
-
-		ffmpeg.on('exit', (code) => {
-			if (code !== 0) {
-				reject()
+		exec(`ffmpeg -i ${url} -y -f image2 -c:v png -frames:v 1 ${frameFile.name}`, (err, stdout, stderr) => {
+			if (err) {
+				console.log(stderr)
+				frameFile.removeCallback()
+				reject(err)
 			} else {
-				console.log("got frame", frameFile.name)
 				resolve(frameFile)
 			}
 		})
@@ -77,24 +91,172 @@ function getYoutubeStream() {
 	})
 }
 
-getYoutubeStream()
-	.then(url => getFrame(url))
-	.then(frame => {
-		let title = '',
-			album = '',
-			duration = ''
+function getYoutubeAudioUrl() {
 
-		const titlePromise = getRegion(frame, 432, 906, 1487, 54).then(titleRegion => readText(titleRegion))
-		const albumPromise = getRegion(frame, 440, 957, 1487, 32).then(albumRegion => readText(albumRegion))
-		const durationPromise = getRegion(frame, 0, 1028, 235, 34).then(durationRegion => readText(durationRegion, "1234567890:/"))
+}
 
-		console.log("waiting for processors")
-		Promise.all([titlePromise, albumPromise, durationPromise]).then(values => {
-			frame.removeCallback()
-			title = values[0].trim()
-			album = values[1].trim()
-			duration = values[2].trim()
+function getVotes(ocrResults) {
+	return ocrResults.map((text, index, arr) => { // Actually perform the search. Map returns array of search results, result.refIndex is of interest
+		const title = text[0].trim()
+		const album = text[1].trim()
+		const position_duration = text[2].trim()
+		let capturedDuration = DURATION_REGEX.test(position_duration) ? position_duration.match(DURATION_REGEX) : ['', '', '']
+		const retval = {
+			result: playlist.search(album, title, capturedDuration[2]),
+			position: capturedDuration[1]
+		}
 
-			console.log(`${album}: ${title} (${duration})`)
-		}).catch(err => console.error(err))
-	}).catch(err => console.error(err))
+		return retval
+	})
+}
+
+function tallyVotes(votes) {
+	const trackResults = {}
+	const positionResults = {}
+	votes.forEach(vote => {
+		if (vote.result != null) {
+			if (trackResults.hasOwnProperty(vote.result.refIndex)) {
+				trackResults[`${vote.result.refIndex}`] = trackResults[`${vote.result.refIndex}`] + 1
+			} else {
+				trackResults[`${vote.result.refIndex}`] = 1
+			}
+		}
+		if (vote.position) {
+			if (positionResults.hasOwnProperty(vote.position)) {
+				positionResults[`${vote.position}`] = positionResults[`${vote.position}`] + 1
+			} else {
+				positionResults[`${vote.position}`] = 1
+			}
+		}
+	})
+	if (trackResults.length === 0) {
+		console.log("votes resulted in no results!")
+		return {
+			position: null,
+			positionConfidence: 0,
+			result: null,
+			resultConfidence: 0
+		}
+	}
+
+	// Retrieve votes
+	let trackIndex = Object.keys(trackResults)[0]
+	let trackMaxVotes = trackResults[trackIndex]
+	let position = Object.keys(positionResults)[0]
+	let positionMaxVotes = positionResults[position]
+	for (const index in trackResults) {
+		if (trackResults[index] > trackMaxVotes) {
+			trackIndex = index
+			trackMaxVotes = trackResults[index]
+		}
+	}
+	for (const positionKey in positionResults) {
+		if (positionResults[positionKey] > positionMaxVotes) {
+			position = positionKey
+			positionMaxVotes = positionResults[positionKey]
+		}
+	}
+
+	// console.log(JSON.stringify(votes))
+	console.debug(`Vote was index ${trackIndex}`)
+	const actualResult = votes.find(v => v.result != null && `${v.result.refIndex}` === trackIndex).result.item
+	return {
+		position: position,
+		positionConfidence: positionMaxVotes / votes.length,
+		result: actualResult,
+		resultConfidence: trackMaxVotes / votes.length
+	}
+}
+
+function timeUntilNextTrack(position, duration) {
+	const timeRegex = /(\d{1,2}):(\d{2})/
+	const durationValues = duration.match(timeRegex)
+	const positionValues = position.match(timeRegex)
+	const durationInSeconds = (parseInt(durationValues[1]) * 60) + parseInt(durationValues[2])
+	const positionInSeconds = (parseInt(positionValues[1]) * 60) + parseInt(positionValues[2])
+	return durationInSeconds - positionInSeconds
+}
+
+function performQuorumProcessing(titleCroppedPromise, albumCroppedPromise, durationCroppedPromise) {
+	return new Promise(((resolve, reject) => {
+		Promise.all([titleCroppedPromise, albumCroppedPromise, durationCroppedPromise]) // Wait for all promises to be resolved, returns paths to cropped segments of frame
+			.then(images => {
+				const thresholds = [15, 12.5, 10, 7.5, 5]
+				Promise.all(thresholds.map(t => { // Threshold all segments at determined levels, returns text determined by each threshold level as [title, album, position/duration]
+					return Promise.all([
+						thresholdImage(images[0], t).then(file => readText(file)),
+						thresholdImage(images[1], t).then(file => readText(file)),
+						thresholdImage(images[2], t).then(file => readText(file, "1234567890:/"))
+					])
+				})).then(allOcrResults => { // Perform search based on resolved text. Perform voting here between thresholds
+				    // console.debug(allOcrResults)
+					// Cast votes
+					const votes = getVotes(allOcrResults) // Actually perform the search. Map returns array of search results, result.refIndex is of interest
+					// console.debug(JSON.stringify(votes))
+
+					// Tally votes
+					const winner = tallyVotes(votes)
+
+					console.debug(`Position confidence: ${winner.positionConfidence}`)
+					console.debug(`Result confidence: ${winner.resultConfidence}`)
+
+					resolve(winner)
+				})
+			}).catch(error => reject(error))
+	}))
+}
+
+async function updateStreamUrl() {
+	const ytStreamUrl = await getYoutubeStream()
+	resolvedUrl.url = ytStreamUrl
+	resolvedUrl.expires = parseInt(ytStreamUrl.match(EXPIRATION_REGEX)[1]) * 1000
+}
+
+async function updateTrackInfo() {
+	const urlExpiresSoon = resolvedUrl.expires < (Date.now() + (30 * 60 * 1000))
+	if (resolvedUrl.expires < Date.now()) {
+		await updateStreamUrl()
+        console.log("Got stream info")
+	} else if (urlExpiresSoon) {
+		updateStreamUrl().then(() => console.log("Updated stream info in the background"))
+	}
+	console.log(`Now: ${Date.now()}\tExpiration: ${resolvedUrl.expires}`)
+	const frame = await getFrame(resolvedUrl.url)
+    const frameTime = Date.now()
+
+	const titlePromise = getRegion(frame, 432, 906, 1487, 54)
+	const albumPromise = getRegion(frame, 440, 957, 1487, 32)
+	const durationPromise = getRegion(frame, 0, 1028, 235, 34)
+
+	const result = await performQuorumProcessing(titlePromise, albumPromise, durationPromise)
+	if (result.resultConfidence < 0.5) {
+		setTimeout(() => updateTrackInfo(), 250)
+		return;
+	}
+	currentTrack = result.result
+	console.log(`${currentTrack.album}: ${currentTrack.track} (${result.position}/${currentTrack.duration})`)
+	const secondsUntilNext = timeUntilNextTrack(result.position, currentTrack.duration)
+	const processingTime = Date.now() - frameTime
+	nextTrackTimestamp = frameTime + (secondsUntilNext * 1000)
+	setTimeout(() => updateTrackInfo(), (secondsUntilNext * 1000) - processingTime + 500)
+	console.log(`Frame processing took ${processingTime}ms`)
+	console.log(`Next track should be at: ${new Date(nextTrackTimestamp).toLocaleString()}`)
+}
+
+// getYoutubeStream()
+// 	.then(url => getFrame(url))
+// 	.then(frame => {
+// 		const titlePromise = getRegion(frame, 432, 906, 1487, 54)
+// 		const albumPromise = getRegion(frame, 440, 957, 1487, 32)
+// 		const durationPromise = getRegion(frame, 0, 1028, 235, 34)
+//
+// 		return performQuorumProcessing(titlePromise, albumPromise, durationPromise)
+// 	})
+// 	.then(value => {
+// 		console.log('Processing complete, voted result was:')
+// 		console.log(`${value.result.album}: ${value.result.track} (${value.position}/${value.result.duration})`)
+// 		console.log(`Time until next track: ${timeUntilNextTrack(value.position, value.result.duration)}`)
+// 	})
+// 	.catch(err => console.error(err))
+
+updateTrackInfo()

playlist-scrape.js

@@ -1,18 +1,29 @@
 const got = require('got')
 const cheerio = require('cheerio')
 const fs = require('fs')
+const PLAYLIST_URL = 'https://nintendoradioplaylist2.000webhostapp.com/Livestream%20Playlist.html'
 
-got('https://nintendoradioplaylist2.000webhostapp.com/Livestream%20Playlist.html').then(res => {
+function scrapePlaylist(url) {
+	got(url).then(res => {
 		const $ = cheerio.load(res.body)
 		const playlist = []
 		$('#playlistTableBody tr').each((i, row) => {
-		cells = $(row).find('td')
-		album = $(cells[1]).text()
-		track = $(cells[2]).text()
-		duration = $(cells[3]).text()
+			const cells = $(row).find('td')
+			const album = $(cells[1]).text()
+			const track = $(cells[2]).text()
+			const duration = $(cells[3]).text()
 			playlist.push({album, track, duration})
 		})
 		fs.writeFileSync('playlist.json', JSON.stringify(playlist))
-}).catch(err => {
+	}).catch(err => {
 		console.log(err)
-})
+	});
+}
+
+module.exports = {
+	scrapePlaylist
+}
+
+if (require.main === module) {
+	scrapePlaylist(PLAYLIST_URL)
+}

playlist-search.js

@@ -1,5 +1,38 @@
 const Fuse = require('fuse.js')
-const playlist = require('../radioScraper/playlist.json')
+const fs = require('fs')
+const scraper = require('./playlist-scrape')
+const PLAYLIST_URL = 'https://nintendoradioplaylist2.000webhostapp.com/Livestream%20Playlist.html'
+const playlist = require('./playlist.json')
+
+const albumValidCharacters = new Set()
+const trackValidCharacters = new Set()
+
+function buildAlbumValidCharacters() {
+	albumValidCharacters.clear()
+	new Set(playlist.map(entry => entry.album)) // Get set of albums
+		.forEach(album => album.split('').forEach(letter => albumValidCharacters.add(letter)))
+}
+
+function buildTrackValidCharacters() {
+	trackValidCharacters.clear()
+	new Set(playlist.map(entry => entry.track))
+		.forEach(track => track.split('').forEach(letter => trackValidCharacters.add(letter)))
+}
+
+function getValidAlbumCharacters() {
+	if (albumValidCharacters.size === 0) {
+		buildAlbumValidCharacters()
+	}
+	return albumValidCharacters
+}
+
+function getValidTrackCharacters() {
+	if (trackValidCharacters.size === 0) {
+		console.log("Building index of valid track characters, this may take some time...")
+		buildTrackValidCharacters()
+	}
+	return trackValidCharacters
+}
 
 const fuseByAlbum = new Fuse(playlist, {
 	keys: ['album']
@@ -9,33 +42,70 @@ const fuseByTrack = new Fuse(playlist, {
 })
 
 function searchAlbums(query) {
-	return fuseByAlbum.search(query).map(result => result.item)
+	return fuseByAlbum.search(query)
 }
 
 function searchTracks(query) {
-	return fuseByTrack.search(query).map(result => result.item)
+	return fuseByTrack.search(query)
+}
+
+function fullSearch(album, track, duration) {
+	const inAlbumSearch = new Fuse(searchAlbums(album), {keys: ['item.track']})
+	const trackResults = inAlbumSearch.search(track).map(result => result.item)
+	if (duration) {
+	    return filterResultsByDuration(trackResults, duration)
+	} else {
+		// No duration present, treat as album+track lookup
+		return trackResults[0]
+	}
+}
+
+function albumDurationSearch(album, duration) {
+	const albumResults = searchAlbums(album)
+	return filterResultsByDuration(albumResults, duration, true)
+}
+
+function trackDurationSearch(track, duration) {
+	const trackResults = searchTracks(track)
+	return filterResultsByDuration(trackResults, duration)
+}
+
+function filterResultsByDuration(results, duration, failSearch = false) {
+	const durationResults = results.filter(result => result.item.duration === duration)
+	if (durationResults.length > 0) {
+		return durationResults[0]
+	} else {
+		return failSearch ? null : results[0]
+	}
 }
 
 function search(albumQuery, trackQuery, duration) {
-	if (albumQuery) {
-		const inAlbumSearch = new Fuse(searchAlbums(albumQuery), {keys: ['track']})
-		const trackResults = inAlbumSearch.search(trackQuery)
-		console.log(`initial results: ${trackResults.length}`)
-		trackResults.filter(result => result.item.duration === duration)
-		if (trackResults.length > 0) {
-			return trackResults[0].item
-		}
+	if (albumQuery && trackQuery && duration) {
+		return fullSearch(albumQuery, trackQuery, duration)
+	} else if (albumQuery && duration) {
+		return albumDurationSearch(albumQuery, duration)
+	} else if (trackQuery && duration) {
+		return trackDurationSearch(trackQuery, duration)
 	} else {
-		const trackResults = searchTracks(trackQuery)
-		console.log(`initial results: ${trackResults.length}`)
-		const durationResults = trackResults.filter(result => result.duration === duration)
-		console.log(`secondary results: ${durationResults.length}`)
-		if (durationResults.length > 0) {
-			return durationResults[0]
-		}
+	    // Not enough info to perform a decent search
+		return null
 	}
 }
 
 function trySearch() {
-	console.log(JSON.stringify(search("The Lede", "Lanayru Mining Facility", "3:10")))
+	console.log(JSON.stringify(search("Final Fantasy VI", "Dancing Mad", "\f")))
+}
+
+// These shouldn't be called asynchronously, manually call it ourselves
+buildTrackValidCharacters()
+buildAlbumValidCharacters()
+
+module.exports = {
+	search,
+	getValidAlbumCharacters,
+	getValidTrackCharacters
+}
+
+if (require.main === module) {
+	trySearch()
 }