playlink/internal/engine/fixer_steps.go

package engine

import (
	"context"
	"fmt"
	"path/filepath"
	"regexp"
	"sort"
	"strconv"
	"strings"

	"github.com/cdnutter/playlink/internal/fsops"
	"github.com/cdnutter/playlink/internal/tags"
)

// discNum recognizes disc folders. "Vol. 2" is left out on purpose: volumes
// are usually albums of their own.
var discNum = regexp.MustCompile(`(?i)^(?:disc|disk|cd|dvd|part)[\s._-]*(\d{1,2})(?:\b|$)`)

// junkFolder matches extra folders inside an album that only describe the
// files ("FLAC", "MP3 320", "24bit"), whose songs belong in the album folder.
var junkFolder = regexp.MustCompile(`(?i)^[\[(]?\s*(?:flac|mp3|aac|m4a|alac|ogg|opus|wav|lossless|lossy|hi-?res|web|` +
	`(?:mp3\s*)?(?:320|256|192|v0|v2)(?:\s*kbps)?|\d{2}\s*-?\s*bit.*|audio|tracks?|files?|songs?)\s*[\])]?$`)

// folderNoise is release information in folder names: "Nightmare (2010)",
// "Nightmare [FLAC]", "2010 - Nightmare".
var (
	folderNoise = regexp.MustCompile(`(?i)\s*[\[({]\s*(?:\d{4}(?:[-.]\d{2}){0,2}|flac|mp3|aac|alac|ogg|opus|web|cd|vinyl|lossless|` +
		`deluxe|remaster(?:ed)?|\d{2}\s*-?\s*bit[^\])}]*|\d+(?:\.\d+)?\s*k(?:hz|bps)[^\])}]*|v0|v2|320|\d{2}-\d{2,3})\s*[\])}]\s*$`)
	yearPrefix = regexp.MustCompile(`^\s*[\[(]?\d{4}[\])]?\s*[-–.]\s+`)
)

// cleanFolderName turns a folder name into an album or artist name.
func cleanFolderName(s string) string {
	for {
		t := folderNoise.ReplaceAllString(s, "")
		t = yearPrefix.ReplaceAllString(t, "")
		t = strings.TrimSpace(t)
		if t == s || t == "" {
			return s
		}
		s = t
	}
}

// layoutFor returns the formats that apply at rel and the folder they are
// relative to ("" for the library, or a folder with its own naming).
func (c NamingConfig) layoutFor(rel string) (string, formats) {
	if rule, ok := c.folderOf(rel); ok {
		return rel[:len(rule.Folder)], rule.formats()
	}
	return "", c.formats()
}

// albumDepth is how many folders deep albums sit: Artist/Album or Album.
func (f formats) albumDepth() int {
	if f.artist == "" {
		return 1
	}
	return 2
}

// renderComponent renders a one-folder format from a song's tags.
func renderComponent(c NamingConfig, format string, f map[string]string) string {
	v := valuesOf(f)
	return finishComponent(tokenPat.ReplaceAllStringFunc(format, func(tok string) string { return renderToken(tok[1:len(tok)-1], v, c) }))
}

// renderAlbumDir is the album folder a song at rel belongs in, from its tags.
func renderAlbumDir(c NamingConfig, rel string, f map[string]string) (string, string) {
	flat := c
	flat.TrackFormat, flat.MultiDiscFormat = "{Track Title}", "{Track Title}"
	flat.Folders = make([]FolderNaming, len(c.Folders))
	for i, r := range c.Folders {
		r.TrackFormat, r.MultiDiscFormat = "{Track Title}", "{Track Title}"
		flat.Folders[i] = r
	}
	g := map[string]string{}
	for k, v := range f {
		g[k] = v
	}
	if strings.TrimSpace(g[tags.Title]) == "" {
		g[tags.Title] = "x" // the title doesn't name folders
	}
	p, why := renderPath(flat, rel, g)
	if why != "" {
		return "", why
	}
	return filepath.Dir(p), ""
}

func depthOf(rel string) int {
	if rel == "" || rel == "." {
		return 0
	}
	return strings.Count(rel, "/") + 1
}

// majority is the most common non-empty value of a field.
func majority(fields map[string]map[string]string, songs []string, keys ...string) string {
	for _, key := range keys {
		count := map[string]int{}
		best, bestN := "", 0
		for _, s := range songs {
			v := strings.TrimSpace(fields[s][key])
			if v == "" {
				continue
			}
			count[v]++
			if count[v] > bestN || (count[v] == bestN && v < best) {
				best, bestN = v, count[v]
			}
		}
		if best != "" {
			return best
		}
	}
	return ""
}

func similar(a, b string) bool {
	ka, kb := matchKey(a), matchKey(b)
	return ka != "" && kb != "" && (ka == kb || levenshteinRatio(ka, kb) >= 0.8)
}

// readLibrary lists a library's songs and reads their tags.
func (e *Engine) readLibrary(ctx context.Context, j *Job, l *libs, side string) (fsops.Root, map[string]fsops.Info, map[string]map[string]string, error) {
	root, err := sideRoot(l, side)
	if err != nil {
		return root, nil, nil, err
	}
	j.phase("Listing "+side+" files", 0)
	files := map[string]fsops.Info{}
	if _, err := walk(ctx, l, root, files); err != nil {
		return root, nil, nil, err
	}
	j.phase("Reading "+side+" tags", int64(len(files)))
	fields := make(map[string]map[string]string, len(files))
	for _, rel := range sortedKeys(files) {
		if ctx.Err() != nil {
			return root, nil, nil, ctx.Err()
		}
		f := map[string]string{}
		if _, ok := tags.FormatOf(rel); ok {
			if got, err := e.tagCache.get(ctx, e.tagTools, filepath.Join(root.Path, rel), files[rel]); err == nil && got != nil {
				f = got
			}
		}
		fields[rel] = f
		j.tick()
	}
	return root, files, fields, nil
}

func (e *Engine) moveItem(root fsops.Root, side, from, to string, in fsops.Info) *organizeItem {
	return &organizeItem{Side: side, From: from, To: to, Move: true, ino: in.Ino, size: in.Size, mtime: in.MtimeNs,
		Sidecars: sidecarsOf(root, from)}
}

// ---- step 1: structure ----

func (e *Engine) checkStructure(ctx context.Context, j *Job, l *libs, plan *FixPlan) error {
	root, files, fields, err := e.readLibrary(ctx, j, l, plan.Side)
	if err != nil {
		return err
	}
	naming := NamingOf(l.pair)
	albums := albumsOf(files)
	plan.Songs, plan.Albums = len(files), len(albums)
	claimed := map[string]string{} // lower(new path) -> old path
	claim := func(from, to string) string {
		k := strings.ToLower(to)
		if c := claimed[k]; c != "" && c != from {
			return "another item gets the same name (" + c + ")"
		}
		claimed[k] = from
		return ""
	}
	// Songs that change tags here, so a move of the same song can re-read it.
	type deep struct {
		item  int
		extra string
	}
	var deeps []deep
	for _, a := range albums {
		if ctx.Err() != nil {
			return ctx.Err()
		}
		prefix, fm := naming.layoutFor(filepath.Join(a.Dir, "x"))
		rest := strings.TrimPrefix(strings.TrimPrefix(a.Dir, prefix), "/")
		if rest == "." {
			rest = ""
		}
		want, depth := fm.albumDepth(), depthOf(rest)
		target := a.Dir // where the album will be, for filling tags from names

		// looseInto moves songs into album folders named from their tags,
		// under base (or where the naming puts them when base is "").
		looseInto := func(songs []string, base, why string) {
			for _, s := range songs {
				var dir, blocked string
				if base == "" {
					dir, blocked = renderAlbumDir(naming, s, fields[s])
				} else if strings.TrimSpace(fields[s][tags.Album]) == "" {
					blocked = "missing album tag"
				} else {
					dir = filepath.Join(base, renderComponent(naming, fm.album, fields[s]))
				}
				it := FixItem{Kind: FixMoveFile, From: s, Why: why, Group: dir}
				if blocked != "" {
					it.Blocked = blocked + ": set it in the Tags tab, then check again"
					it.Group = filepath.Dir(s)
				} else {
					it.To = filepath.Join(dir, filepath.Base(s))
					if targetTaken(root, s, it.To) {
						it.Blocked = "a different file already has that name"
					} else if c := claim(s, it.To); c != "" {
						it.Blocked = c
					} else {
						it.org = e.moveItem(root, plan.Side, s, it.To, files[s])
					}
				}
				plan.add(it)
			}
		}

		switch {
		case depth == 0:
			where := "the top of the library"
			if prefix != "" {
				where = "the top of " + prefix
			}
			looseInto(a.Songs, "", "Not in an album folder (at "+where+")")
			continue
		case depth < want:
			// Music with one folder: an album without an artist folder, or an
			// artist folder with songs directly inside it.
			album := majority(fields, a.Songs, tags.Album)
			artist := majority(fields, a.Songs, tags.AlbumArtist, tags.Artist)
			name := filepath.Base(a.Dir)
			isArtist := a.Nested || (artist != "" && similar(artist, cleanFolderName(name)) && !similar(album, cleanFolderName(name)))
			if isArtist {
				looseInto(a.Songs, a.Dir, "Songs directly in the artist folder, not in an album folder")
				continue
			}
			it := FixItem{Kind: FixMoveFolder, From: a.Dir, Why: "Album folder is not inside an artist folder"}
			if artist == "" {
				it.Blocked = "its songs have no artist tag: set the album artist (Edit album in the Tags tab), then check again"
				it.Group = a.Dir
			} else {
				f := map[string]string{tags.AlbumArtist: artist}
				it.To = filepath.Join(prefix, renderComponent(naming, fm.artist, f), name)
				it.Group = it.To
				if why := checkDirMove(l, root, it.From, it.To); why != "" {
					it.Blocked = why
				} else if c := claim(it.From, it.To); c != "" {
					it.Blocked = c
				}
				target = it.To
			}
			plan.add(it)
		case depth > want:
			comps := strings.Split(rest, "/")
			last := comps[len(comps)-1]
			if junkFolder.MatchString(last) && !a.Nested {
				parent := filepath.Dir(a.Dir)
				why := fmt.Sprintf("Extra folder %q inside the album", last)
				for _, s := range a.Songs {
					to := filepath.Join(parent, strings.TrimPrefix(s, a.Dir+"/"))
					it := FixItem{Kind: FixMoveFile, From: s, To: to, Why: why, Group: parent}
					if targetTaken(root, s, to) {
						it.Blocked = "a different file already has that name"
					} else if c := claim(s, to); c != "" {
						it.Blocked = c
					} else {
						it.org = e.moveItem(root, plan.Side, s, to, files[s])
					}
					plan.add(it)
				}
				target = parent
				break
			}
			extra := comps[:len(comps)-want]
			to := filepath.Join(prefix, filepath.Join(comps[len(comps)-want:]...))
			kind := "Album"
			if want == 1 {
				kind = "Soundtrack"
			}
			it := FixItem{Kind: FixMoveFolder, From: a.Dir, To: to, Group: to,
				Why: fmt.Sprintf("%s is inside extra folders (%s)", kind, strings.Join(extra, "/"))}
			switch {
			case a.Nested:
				it.Blocked = "other album folders are inside it"
			default:
				if why := checkDirMove(l, root, it.From, it.To); why != "" {
					it.Blocked = why
				} else if c := claim(it.From, it.To); c != "" {
					it.Blocked = c
				}
			}
			plan.add(it)
			deeps = append(deeps, deep{item: len(plan.Items) - 1, extra: filepath.Join(prefix, extra[0])})
			target = to
		}

		// Disc folders and tags filled from folder and file names.
		if len(a.Discs) > 0 && a.Loose {
			plan.add(FixItem{Kind: FixProblem, Group: a.Dir, From: a.Dir, Why: "Songs are both in the album folder and in disc folders",
				Blocked: "move them into the right disc folder by hand"})
		}
		discTotal := 0
		for _, n := range a.Discs {
			discTotal = max(discTotal, n)
		}
		albumName := cleanFolderName(filepath.Base(target))
		artistName := ""
		if want == 2 && depthOf(strings.TrimPrefix(strings.TrimPrefix(target, prefix), "/")) == 2 {
			artistName = cleanFolderName(filepath.Base(filepath.Dir(target)))
		}
		for _, s := range a.Songs {
			f := fields[s]
			if _, ok := tags.FormatOf(s); !ok {
				continue
			}
			ch := map[string][2]string{}
			var why []string
			set := func(k, v string) {
				if v != "" && strings.TrimSpace(f[k]) != v {
					ch[k] = [2]string{f[k], v}
				}
			}
			if n := a.Discs[filepath.Dir(s)]; n > 0 {
				if leadingNumber(f[tags.Disc]) != n {
					set(tags.Disc, strconv.Itoa(n))
				}
				if leadingNumber(f[tags.DiscTotal]) != discTotal {
					set(tags.DiscTotal, strconv.Itoa(discTotal))
				}
				if len(ch) > 0 {
					why = append(why, "disc number from its disc folder")
				}
			}
			filled := len(ch)
			if strings.TrimSpace(f[tags.Album]) == "" {
				set(tags.Album, albumName)
			}
			if artistName != "" && strings.TrimSpace(f[tags.Artist]) == "" && strings.TrimSpace(f[tags.AlbumArtist]) == "" {
				set(tags.Artist, artistName)
			}
			stem := strings.TrimSuffix(filepath.Base(s), filepath.Ext(s))
			if strings.TrimSpace(f[tags.Title]) == "" {
				if t := strings.TrimSpace(trackPrefix.ReplaceAllString(stem, "")); t != "" {
					set(tags.Title, t)
				} else {
					set(tags.Title, stem)
				}
			}
			if strings.TrimSpace(f[tags.Track]) == "" {
				if m := trackNum.FindStringSubmatch(stem); m != nil {
					set(tags.Track, strconv.Itoa(leadingNumber(m[1])))
				}
			}
			if len(ch) > filled {
				why = append(why, "missing tags filled from the folder and file names")
			}
			if len(ch) > 0 {
				w := strings.Join(why, "; ")
				plan.add(FixItem{Kind: FixTags, From: s, Group: target, Changes: ch, Why: strings.ToUpper(w[:1]) + w[1:]})
			}
		}
	}
	// Several albums inside the same extra folder look like a category
	// ("Video Games/…", "Rock/…") rather than a mistake: don't select those.
	byExtra := map[string]int{}
	for _, d := range deeps {
		byExtra[d.extra]++
	}
	var cats []string
	for _, d := range deeps {
		if byExtra[d.extra] >= 3 {
			plan.Items[d.item].Unsure = true
		}
	}
	for x, n := range byExtra {
		if n >= 3 {
			cats = append(cats, fmt.Sprintf("%s (%d albums)", x, n))
		}
	}
	if len(cats) > 0 {
		sort.Strings(cats)
		plan.Notices = append(plan.Notices, "Some folders hold several albums one level deeper than expected: "+strings.Join(cats, ", ")+
			". If they are categories you want to keep (like Video Games or Movies), add them under Organize → Folders with their own naming "+
			"instead of moving the albums out. Those moves are not selected.")
	}
	return nil
}

// ---- step 3: folder names ----

func (e *Engine) checkFolders(ctx context.Context, j *Job, l *libs, plan *FixPlan) error {
	root, files, fields, err := e.readLibrary(ctx, j, l, plan.Side)
	if err != nil {
		return err
	}
	naming := NamingOf(l.pair)
	albums := albumsOf(files)
	plan.Songs, plan.Albums = len(files), len(albums)
	claimed := map[string]string{}
	for _, a := range albums {
		prefix, _ := naming.layoutFor(filepath.Join(a.Dir, "x"))
		if a.Dir == "." || a.Dir == prefix {
			continue // loose songs: the file step puts them in album folders
		}
		targets := map[string]string{}
		var missing string
		untagged := 0
		for _, s := range a.Songs {
			t, why := renderAlbumDir(naming, s, fields[s])
			if why != "" {
				missing = why
				untagged++
				continue
			}
			targets[strings.ToLower(t)] = t
		}
		switch {
		case len(targets) == 0:
			plan.add(FixItem{Kind: FixProblem, From: a.Dir, Group: a.Dir, Why: "Can't name this album folder",
				Blocked: missing + " (fix it with Edit album in the Tags tab, then check again)"})
			continue
		case len(targets) > 1:
			var names []string
			for _, t := range targets {
				names = append(names, t)
			}
			sort.Strings(names)
			plan.add(FixItem{Kind: FixProblem, From: a.Dir, Group: a.Dir,
				Why:     fmt.Sprintf("Songs belong to %d different album folders (%s)", len(names), strings.Join(names, ", ")),
				Blocked: "the file step sorts them into their own folders; or make the album tags agree with Edit album"})
			continue
		}
		var to string
		for _, t := range targets {
			to = t
		}
		if to == a.Dir {
			continue
		}
		why := "Renamed from its tags"
		if filepath.Dir(to) != filepath.Dir(a.Dir) {
			why = "Moved to the folder its tags name"
		}
		if untagged > 0 {
			why += fmt.Sprintf(" (%d songs without tags move along)", untagged)
		}
		it := FixItem{Kind: FixMoveFolder, From: a.Dir, To: to, Group: to, Why: why}
		switch {
		case a.Nested:
			it.Blocked = "other album folders are inside it"
		default:
			if b := checkDirMove(l, root, a.Dir, to); b != "" {
				it.Blocked = b
				if strings.Contains(b, "already exists") {
					it.Blocked = "that folder already exists; its songs are moved into it in the file step"
				}
			} else if c := claimed[strings.ToLower(to)]; c != "" {
				it.Blocked = "another folder gets the same name (" + c + ")"
			} else {
				claimed[strings.ToLower(to)] = a.Dir
			}
		}
		plan.add(it)
	}
	return nil
}

// ---- step 4: file names ----

func (e *Engine) checkFiles(ctx context.Context, j *Job, l *libs, plan *FixPlan) error {
	op, err := e.planOrganize(ctx, j, l, []string{plan.Side})
	if err != nil {
		return err
	}
	plan.Songs = op.Scanned
	for _, it := range op.Items {
		fi := FixItem{Kind: FixMoveFile, From: it.From, To: it.To, Why: "Renamed from its tags"}
		if it.Move {
			org := it
			fi.org = &org
			fi.Group = filepath.Dir(it.To)
			if filepath.Dir(it.To) != filepath.Dir(it.From) {
				fi.Why = "Moved and renamed from its tags"
			}
		} else {
			fi.Kind, fi.Blocked, fi.Why, fi.Group = FixProblem, it.Reason, "Can't be renamed", filepath.Dir(it.From)
		}
		plan.add(fi)
	}
	return nil
}