minstrel/internal/release/match.go

package release

import "strings"

// Want is the album a search is for, as the catalogue describes it.
type Want struct {
	Artist         string
	Title          string
	Year           int
	PrimaryType    string // Album, EP, Single, Broadcast, Other
	SecondaryTypes []string
	// MinTracks and MaxTracks span the group's editions' track counts; zero
	// when the catalogue has not read them.
	MinTracks int
	MaxTracks int
}

// Veto codes: why a name is not the album. Kept on a candidate as its
// reason's code.
const (
	VetoArtist      = "wrong_artist"
	VetoTitle       = "wrong_title"
	VetoVideo       = "video"
	VetoDiscography = "discography"
	VetoVersion     = "other_version"
	VetoType        = "wrong_type"
	VetoNoAudio     = "no_audio"
	VetoIncomplete  = "incomplete"
)

// Match is what a name says about whether it is the album wanted.
type Match struct {
	// Veto is a code from above when the name is not the album, with Reason
	// saying why in a sentence.
	Veto   string
	Reason string
	// Doubt is set when the name may be the album and cannot say for sure: a
	// person decides. It is a sentence.
	Doubt string
	// YearAgrees: the name states the album's year, give or take one.
	YearAgrees bool
}

// versions are words that make a release a different recording of the
// album, with what to call it. A word the wanted title or artist holds does
// not count ("Live Through This", "Instrumental Core").
var versions = []struct {
	words []string
	what  string
	// typ is the MusicBrainz secondary type that wants this version.
	typ string
}{
	{[]string{"instrumental", "instrumentals"}, "an instrumental version", ""},
	{[]string{"drumless"}, "a drumless edition", ""},
	{[]string{"karaoke"}, "a karaoke version", ""},
	{[]string{"acapella", "acappella", "accapella", "acapellas"}, "an a cappella version", ""},
	{[]string{"remix", "remixes", "remixed"}, "a remix release", "Remix"},
	{[]string{"live", "bootleg"}, "a live recording", "Live"},
	{[]string{"demo", "demos"}, "demo recordings", "Demo"},
	// Not "covers": in a release name that is usually the artwork scans.
	{[]string{"tribute", "lullaby", "lullabies"}, "a tribute release", ""},
	{[]string{"nightcore", "slowed", "sped"}, "a slowed or sped-up edit", ""},
	{[]string{"commentary", "interview", "interviews"}, "commentary or interviews", "Interview"},
}

// collections are words naming more than one album.
var collections = []string{"collection", "anthology", "albums", "boxset", "discography"}

// Judge says whether a release name is the album wanted. It looks only at
// the name; JudgeFiles looks at what a torrent holds.
func Judge(w Want, r Release) Match {
	var m Match
	m.YearAgrees = r.Year != 0 && w.Year != 0 && abs(r.Year-w.Year) <= 1
	if r.Video != "" {
		m.Veto, m.Reason = VetoVideo, "A video, not audio (“"+r.Video+"”)."
		return m
	}
	artist := Tokens(w.Artist)
	title := Tokens(w.Title)
	if len(artist) > 0 && !contains(r.Tokens, artist) && !contains(r.Tokens, withoutThe(artist)) {
		m.Veto, m.Reason = VetoArtist, "The name doesn’t mention "+w.Artist+"."
		return m
	}
	if len(title) > 0 && !contains(r.Tokens, title) && !contains(r.Tokens, bare(w.Title)) {
		m.Veto, m.Reason = VetoTitle, "The name doesn’t contain the title “"+w.Title+"”."
		return m
	}
	own := map[string]bool{}
	for _, t := range append(append([]string{}, title...), artist...) {
		own[t] = true
	}
	if r.Discography != "" && !own[r.Discography] {
		m.Veto, m.Reason = VetoDiscography, "A discography or collection ("+r.Discography+"), and one album is wanted."
		return m
	}
	for _, c := range collections {
		if !own[c] && has(r.Tokens, c) {
			m.Veto, m.Reason = VetoDiscography, "A collection of albums (“"+c+"”), and one album is wanted."
			return m
		}
	}
	for _, v := range versions {
		if v.typ != "" && hasType(w.SecondaryTypes, v.typ) {
			continue
		}
		for _, word := range v.words {
			if !own[word] && has(r.Tokens, word) {
				m.Veto, m.Reason = VetoVersion, "This is "+v.what+" (“"+word+"”)."
				return m
			}
		}
	}
	if w.PrimaryType == "Album" || w.PrimaryType == "EP" {
		if !own["single"] && has(r.Tokens, "single") {
			m.Veto, m.Reason = VetoType, "A single, and the "+strings.ToLower(w.PrimaryType)+" is wanted."
			return m
		}
	}
	if w.PrimaryType == "Album" && !own["ep"] && has(r.Tokens, "ep") {
		m.Veto, m.Reason = VetoType, "An EP, and the album is wanted."
		return m
	}
	// A self-titled album: every release of the artist names it once. Two
	// mentions, or its year, say it is this one.
	if len(title) > 0 && equal(title, artist) && count(r.Tokens, title) < 2 && !m.YearAgrees {
		m.Doubt = "The album shares the artist’s name, and nothing else in the name says it is this album."
	}
	return m
}

func abs(n int) int {
	if n < 0 {
		return -n
	}
	return n
}

func withoutThe(t []string) []string {
	if len(t) > 1 && t[0] == "the" {
		return t[1:]
	}
	return t
}

// bare is a title without a trailing bracketed part ("Album (Remastered)"),
// for names that leave it off.
func bare(title string) []string {
	if i := strings.IndexAny(title, "(["); i > 0 {
		return Tokens(title[:i])
	}
	return nil
}

// contains says whether want's words appear in name in order and together,
// or, for two or more words, run together ("ACDC" for AC/DC). A single word
// is never looked for run together: "muse" is inside "museum".
func contains(name, want []string) bool {
	if len(want) == 0 {
		return false
	}
	if count(name, want) > 0 {
		return true
	}
	joined := strings.Join(want, "")
	if len(want) < 2 || len(joined) < 4 {
		return false
	}
	return strings.Contains(strings.Join(name, ""), joined)
}

// count is how often want's words appear in name in order and together.
func count(name, want []string) int {
	n := 0
	for i := 0; i+len(want) <= len(name); i++ {
		if equal(name[i:i+len(want)], want) {
			n++
		}
	}
	return n
}

func equal(a, b []string) bool {
	if len(a) != len(b) {
		return false
	}
	for i := range a {
		if a[i] != b[i] {
			return false
		}
	}
	return true
}

func has(tokens []string, word string) bool {
	for _, t := range tokens {
		if t == word {
			return true
		}
	}
	return false
}

func hasType(types []string, t string) bool {
	for _, x := range types {
		if strings.EqualFold(x, t) {
			return true
		}
	}
	return false
}