minstrel/internal/release/parse.go

// Package release reads music release names and judges them against a
// wanted album: what a name says it is (format, bitrate, bit
// depth, source, year, edition), whether it is the album wanted, and which
// quality tier it sits in. Nothing here touches the network or the database,
// so every rule is testable on a string.
package release

import (
	"regexp"
	"strconv"
	"strings"
)

// Formats a name can state.
const (
	FormatFLAC   = "FLAC"
	FormatALAC   = "ALAC"
	FormatAPE    = "APE"
	FormatWV     = "WavPack"
	FormatWAV    = "WAV"
	FormatAIFF   = "AIFF"
	FormatDSD    = "DSD"
	FormatMP3    = "MP3"
	FormatAAC    = "AAC"
	FormatVorbis = "Vorbis"
	FormatOpus   = "Opus"
)

// Sources a name can state.
const (
	SourceWEB      = "WEB"
	SourceCD       = "CD"
	SourceVinyl    = "Vinyl"
	SourceSACD     = "SACD"
	SourceCassette = "Cassette"
)

// Release is what a release name says about itself. Zero values mean the
// name does not say.
type Release struct {
	Name   string
	Year   int
	Format string
	// Bitrate is a constant bitrate in kbps; VBR is a LAME preset (V0, V1, V2).
	Bitrate int
	VBR     string
	// BitDepth (16, 24) and SampleRate (kHz) for lossless.
	BitDepth   int
	SampleRate float64
	Source     string
	// Video is the marker that says this is a video ("x264", "1080p"), or "".
	Video string
	// Discography: a year range or a word naming many albums.
	Discography string
	// Tokens are the name's words, normalised for matching.
	Tokens []string
}

var (
	formatRe = regexp.MustCompile(`\b(flac|alac|ape|wavpack|wv|wav|aiff|dsd|dsf|dff|mp3|aac|m4a|ogg|vorbis|opus)\b`)
	// "24-96", "24/192", "16-44 1" (44.1 after dots become spaces).
	depthRateRe = regexp.MustCompile(`\b(16|24|32) ?[-/ ] ?(44 1|44|48|88 2|88|96|176 4|176|192|352 8|384)\b`)
	depthRe     = regexp.MustCompile(`\b(16|24|32) ?-?bits?\b`)
	rateRe      = regexp.MustCompile(`\b(44 1|48|88 2|96|176 4|192) ?khz\b`)
	hiResRe     = regexp.MustCompile(`\bhi ?-?res\b`)
	codecRateRe = regexp.MustCompile(`\b(?:mp3|aac|m4a|cbr|ogg|opus)[ \-@]*(320|256|224|192|160|128|112|96|64)\b`)
	kbpsRe      = regexp.MustCompile(`\b(320|256|224|192|160|128|112|96|64) ?(?:kbps|kbit|kb/s|k)\b`)
	vbrRe       = regexp.MustCompile(`\bv([0-2])\b`)
	webRe       = regexp.MustCompile(`\bweb(?: ?-?dl)?\b`)
	cdRe        = regexp.MustCompile(`\bcd ?(?:rip)?\b`)
	vinylRe     = regexp.MustCompile(`\b(?:vinyl|vinylrip|lp ?rip|needle ?drop)\b`)
	sacdRe      = regexp.MustCompile(`\bsacd\b`)
	cassetteRe  = regexp.MustCompile(`\b(?:cassette|tape ?rip)\b`)
	yearRe      = regexp.MustCompile(`\b(19[3-9]\d|20[0-4]\d)\b`)
	bracketYear = regexp.MustCompile(`[(\[{] ?(19[3-9]\d|20[0-4]\d) ?[)\]}]`)
	yearRangeRe = regexp.MustCompile(`\b(19[3-9]\d|20[0-4]\d) ?(?:-|–|to) ?(19[3-9]\d|20[0-4]\d)\b`)
	videoRe     = regexp.MustCompile(`\b(x264|x265|h ?264|h ?265|hevc|xvid|divx|1080p|1080i|720p|2160p|bluray|blu ?ray|bdrip|brrip|bdremux|dvdrip|dvd9|dvd5|webrip|hdtv|mkv|avi|m2ts)\b`)
	discogRe    = regexp.MustCompile(`\b(discography|discografia|discographie|diskografie|discografie)\b`)
)

// Parse reads what a release name says.
func Parse(name string) Release {
	r := Release{Name: name}
	s := spaced(name)

	r.Format = bestFormat(formatRe.FindAllStringSubmatch(s, -1))

	if m := depthRateRe.FindStringSubmatch(s); m != nil {
		r.BitDepth, _ = strconv.Atoi(m[1])
		r.SampleRate = rate(m[2])
	}
	if r.BitDepth == 0 {
		if m := depthRe.FindStringSubmatch(s); m != nil {
			r.BitDepth, _ = strconv.Atoi(m[1])
		}
	}
	if r.SampleRate == 0 {
		if m := rateRe.FindStringSubmatch(s); m != nil {
			r.SampleRate = rate(m[1])
		}
	}
	if r.BitDepth == 0 && hiResRe.MatchString(s) {
		r.BitDepth = 24
	}

	if m := codecRateRe.FindStringSubmatch(s); m != nil {
		r.Bitrate, _ = strconv.Atoi(m[1])
	} else if m := kbpsRe.FindStringSubmatch(s); m != nil {
		r.Bitrate, _ = strconv.Atoi(m[1])
	}
	// A bare "V0" is a LAME preset only beside MP3 or VBR; elsewhere it can
	// be part of a title.
	if m := vbrRe.FindStringSubmatch(s); m != nil && (r.Format == FormatMP3 || strings.Contains(s, "vbr")) {
		r.VBR = "V" + m[1]
	}
	if r.Format == "" && (r.Bitrate > 0 || r.VBR != "") {
		r.Format = FormatMP3
	}

	switch {
	case sacdRe.MatchString(s):
		r.Source = SourceSACD
	case vinylRe.MatchString(s):
		r.Source = SourceVinyl
	case webRe.MatchString(s):
		r.Source = SourceWEB
	case cassetteRe.MatchString(s):
		r.Source = SourceCassette
	case cdRe.MatchString(s):
		r.Source = SourceCD
	}

	if m := videoRe.FindStringSubmatch(s); m != nil {
		r.Video = m[1]
	}
	if m := discogRe.FindStringSubmatch(s); m != nil {
		r.Discography = m[1]
	} else if m := yearRangeRe.FindStringSubmatch(s); m != nil && m[2] > m[1] {
		r.Discography = m[1] + "–" + m[2]
	}
	r.Year = year(s)
	r.Tokens = Tokens(name)
	return r
}

// spaced lowercases a name and turns its separators into spaces, keeping
// the slash in "24/96" and "kb/s".
func spaced(name string) string {
	var b strings.Builder
	for _, c := range strings.ToLower(name) {
		switch c {
		case '_', '.', ',':
			b.WriteByte(' ')
		default:
			b.WriteRune(c)
		}
	}
	return b.String()
}

// rate reads a sample rate in kHz; "44", "88" and "176" are the short
// spellings of 44.1, 88.2 and 176.4.
func rate(s string) float64 {
	switch s {
	case "44":
		return 44.1
	case "88":
		return 88.2
	case "176":
		return 176.4
	}
	f, _ := strconv.ParseFloat(strings.ReplaceAll(s, " ", "."), 64)
	return f
}

// formatRank orders formats a name might state together ("FLAC + MP3"):
// the best is the one it is described by.
var formatRank = map[string]int{
	FormatDSD: 11, FormatFLAC: 10, FormatALAC: 9, FormatWV: 8, FormatAPE: 7, FormatWAV: 6, FormatAIFF: 5,
	FormatOpus: 4, FormatAAC: 3, FormatVorbis: 2, FormatMP3: 1,
}

func bestFormat(found [][]string) string {
	best := ""
	for _, m := range found {
		f := formatOf(m[1])
		if formatRank[f] > formatRank[best] {
			best = f
		}
	}
	return best
}

// formatOf names the format a word or file extension stands for.
func formatOf(word string) string {
	switch word {
	case "flac":
		return FormatFLAC
	case "alac":
		return FormatALAC
	case "ape":
		return FormatAPE
	case "wavpack", "wv":
		return FormatWV
	case "wav":
		return FormatWAV
	case "aiff", "aif":
		return FormatAIFF
	case "dsd", "dsf", "dff":
		return FormatDSD
	case "mp3":
		return FormatMP3
	case "aac", "m4a":
		return FormatAAC
	case "ogg", "vorbis":
		return FormatVorbis
	case "opus":
		return FormatOpus
	}
	return ""
}

// year is the release year a name states: one in brackets when there is
// one (so the title "1989" is not read as its year), else the last.
func year(s string) int {
	if m := bracketYear.FindStringSubmatch(s); m != nil {
		y, _ := strconv.Atoi(m[1])
		return y
	}
	all := yearRe.FindAllString(s, -1)
	if len(all) == 0 {
		return 0
	}
	y, _ := strconv.Atoi(all[len(all)-1])
	return y
}

// fold maps the accented letters music names carry to plain ones.
var fold = map[rune]string{
	'à': "a", 'á': "a", 'â': "a", 'ã': "a", 'ä': "a", 'å': "a", 'ā': "a", 'ą': "a",
	'æ': "ae", 'ç': "c", 'ć': "c", 'č': "c",
	'è': "e", 'é': "e", 'ê': "e", 'ë': "e", 'ē': "e", 'ė': "e", 'ę': "e", 'ě': "e",
	'ì': "i", 'í': "i", 'î': "i", 'ï': "i", 'ī': "i", 'ı': "i",
	'ñ': "n", 'ń': "n", 'ň': "n",
	'ò': "o", 'ó': "o", 'ô': "o", 'õ': "o", 'ö': "o", 'ø': "o", 'ō': "o", 'ő': "o",
	'œ': "oe", 'ù': "u", 'ú': "u", 'û': "u", 'ü': "u", 'ū': "u", 'ů': "u", 'ű': "u",
	'ý': "y", 'ÿ': "y", 'ß': "ss", 'š': "s", 'ś': "s", 'ş': "s",
	'ž': "z", 'ź': "z", 'ż': "z", 'ł': "l", 'đ': "d", 'ð': "d", 'þ': "th", 'ř': "r",
}

// abbreviations are spelled out on both sides before comparing.
var abbreviations = map[string]string{"vol": "volume", "pt": "part", "n": "and"}

// Tokens are a name's words for matching: lowercase, accents folded, "&"
// and "+" read as "and", apostrophes dropped (so "Don't" is "dont"), every
// other mark a word break.
func Tokens(s string) []string {
	var b strings.Builder
	for _, c := range strings.ToLower(s) {
		switch {
		case c >= 'a' && c <= 'z', c >= '0' && c <= '9':
			b.WriteRune(c)
		case c == '\'' || c == '’' || c == '`' || c == '´':
		case c == '&' || c == '+':
			b.WriteString(" and ")
		default:
			if f, ok := fold[c]; ok {
				b.WriteString(f)
			} else if c > 127 && isLetter(c) {
				b.WriteRune(c)
			} else {
				b.WriteByte(' ')
			}
		}
	}
	words := strings.Fields(b.String())
	for i, w := range words {
		if full, ok := abbreviations[w]; ok {
			words[i] = full
		}
	}
	return words
}

// isLetter keeps letters of scripts the fold table does not cover (Cyrillic,
// CJK) as letters rather than breaks.
func isLetter(c rune) bool {
	return (c >= 0x0400 && c <= 0x04FF) || (c >= 0x0370 && c <= 0x03FF) || c >= 0x3040
}