playlink/internal/engine/bulk.go

package engine

import (
	"context"
	"encoding/json"
	"errors"
	"fmt"
	"path/filepath"
	"sort"
	"strings"

	"github.com/cdnutter/playlink/internal/probe"
	"github.com/cdnutter/playlink/internal/store"
)

// Bulk actions apply one decision to many matches, typically whole albums.
// Every match still goes through the same single-item code path (and its
// preflight checks); bulk only adds planning, a preview and a job.

const (
	BulkLink     = "link"
	BulkPromote  = "promote"
	BulkIgnore   = "ignore"
	BulkUnignore = "unignore"
)

// ErrInvalidBulk marks a malformed bulk request (reported as a 400).
var ErrInvalidBulk = errors.New("invalid bulk request")

// MaxBulk caps one request so a mistaken "select all" stays reviewable.
const MaxBulk = 20000

type BulkOptions struct {
	Action string `json:"action"`
	// AllowDowngrade includes matches where the kept copy is lower quality:
	// for link, the master is worse; for promote, the slave is worse.
	AllowDowngrade bool `json:"allow_downgrade"`
	KeepFolder     bool `json:"keep_folder"`
}

// AlbumFolder is the folder a slave file is grouped under: its directory
// with disc folders ("Disc 1", "CD2") folded into the album.
func AlbumFolder(rel string) string {
	dir := filepath.Dir(rel)
	for dir != "." && discFolder.MatchString(filepath.Base(dir)) {
		dir = filepath.Dir(dir)
	}
	if dir == "." {
		return ""
	}
	return dir
}

func verdict(m store.Match) probe.Verdict {
	var v probe.Verdict
	_ = json.Unmarshal(m.Verdict, &v)
	return v
}

// planBulk decides whether a match takes part in a bulk action. It returns
// the reason when it does not.
func planBulk(pairID int64, m store.Match, opt BulkOptions) (bool, string) {
	if m.PairID != pairID {
		return false, "belongs to another library pair"
	}
	v := verdict(m)
	switch opt.Action {
	case BulkIgnore:
		if m.Ignored {
			return false, "already ignored"
		}
		return true, ""
	case BulkUnignore:
		if !m.Ignored {
			return false, "not ignored"
		}
		return true, ""
	}
	if m.Ignored {
		return false, "ignored"
	}
	switch m.Status {
	case store.StatusLinked:
		return false, "already linked"
	case store.StatusUnmatched:
		return false, "no master copy"
	case store.StatusError:
		return false, "has an error"
	}
	switch opt.Action {
	case BulkLink:
		if v.Better == "slave" && !opt.AllowDowngrade {
			return false, "master copy is lower quality"
		}
		return true, ""
	case BulkPromote:
		if m.Status == store.StatusIdentical {
			return false, "identical, so link instead"
		}
		if v.Better == "master" && !opt.AllowDowngrade {
			return false, "master copy is better"
		}
		return true, ""
	}
	return false, "unknown action"
}

// BulkPreview summarises a bulk action before it runs.
type BulkPreview struct {
	Action   string         `json:"action"`
	Selected int            `json:"selected"`
	Albums   int            `json:"albums"`
	Process  int            `json:"process"`
	Skip     map[string]int `json:"skip"` // reason -> count

	// Among the matches that will be processed:
	Identical        int   `json:"identical"`         // linked without quarantine
	Quarantined      int   `json:"quarantined"`       // replaced file is kept in quarantine
	Upgrades         int   `json:"upgrades"`          // the copy kept is better
	Downgrades       int   `json:"downgrades"`        // the copy kept is worse (only with AllowDowngrade)
	Approximate      int   `json:"approximate"`       // approximate matches
	MovesFolder      int   `json:"moves_folder"`      // link moves the slave file into the master's folders
	Renames          int   `json:"renames"`           // file name changes
	DurationMismatch int   `json:"duration_mismatch"` // durations differ by more than 2 seconds
	BytesFreed       int64 `json:"bytes_freed"`       // identical slave files replaced by links

	// A few example paths per warning, so the user can spot-check.
	Examples map[string][]string `json:"examples"`
}

func (p *BulkPreview) example(kind, path string) {
	if len(p.Examples[kind]) < 3 {
		p.Examples[kind] = append(p.Examples[kind], path)
	}
}

func (e *Engine) PreviewBulk(ctx context.Context, pairID int64, ids []int64, opt BulkOptions) (BulkPreview, error) {
	ids = uniqueIDs(ids)
	p := BulkPreview{Action: opt.Action, Selected: len(ids), Skip: map[string]int{}, Examples: map[string][]string{}}
	if err := validBulk(ids, opt); err != nil {
		return p, err
	}
	ms, err := e.st.MatchesByIDs(ctx, ids)
	if err != nil {
		return p, err
	}
	albums := map[string]bool{}
	for _, m := range ms {
		albums[AlbumFolder(m.RelSlave)] = true
		ok, why := planBulk(pairID, m, opt)
		if !ok {
			p.Skip[why]++
			continue
		}
		p.Process++
		if opt.Action != BulkLink && opt.Action != BulkPromote {
			continue
		}
		v := verdict(m)
		kept := "master"
		if opt.Action == BulkPromote {
			kept = "slave"
		}
		switch {
		case v.Better == kept:
			p.Upgrades++
		case v.Better != "" && v.Better != "equal" && v.Better != "unknown":
			p.Downgrades++
			p.example("downgrades", m.RelSlave)
		}
		if m.Kind == store.KindApprox {
			p.Approximate++
			p.example("approximate", m.RelSlave)
		}
		if v.DurationDelta > 2 {
			p.DurationMismatch++
			p.example("duration_mismatch", m.RelSlave)
		}
		if filepath.Base(m.RelSlave) != filepath.Base(m.RelMaster) {
			p.Renames++
		}
		if opt.Action == BulkLink {
			if m.Status == store.StatusIdentical {
				p.Identical++
				p.BytesFreed += m.SlaveSize
			} else {
				p.Quarantined++
			}
			if target := LinkTarget(m.RelSlave, m.RelMaster, opt.KeepFolder); filepath.Dir(target) != filepath.Dir(m.RelSlave) {
				p.MovesFolder++
				p.example("moves_folder", m.RelSlave+" → "+target)
			}
		} else {
			p.Quarantined++ // every promoted master file is quarantined
			p.example("master_replaced", m.RelMaster)
		}
	}
	if missing := len(ids) - len(ms); missing > 0 {
		p.Skip["no longer exists (rescan)"] += missing
	}
	p.Albums = len(albums)
	return p, nil
}

// uniqueIDs drops duplicates (a song selected on its own and via its album).
func uniqueIDs(ids []int64) []int64 {
	seen := make(map[int64]bool, len(ids))
	out := make([]int64, 0, len(ids))
	for _, id := range ids {
		if !seen[id] {
			seen[id] = true
			out = append(out, id)
		}
	}
	return out
}

func validBulk(ids []int64, opt BulkOptions) error {
	switch opt.Action {
	case BulkLink, BulkPromote, BulkIgnore, BulkUnignore:
	default:
		return fmt.Errorf("%w: unknown action %q", ErrInvalidBulk, opt.Action)
	}
	if len(ids) == 0 {
		return fmt.Errorf("%w: nothing selected", ErrInvalidBulk)
	}
	if len(ids) > MaxBulk {
		return fmt.Errorf("%w: select at most %d songs at a time", ErrInvalidBulk, MaxBulk)
	}
	return nil
}

// ApplyIgnore ignores or un-ignores matches immediately (no files change).
func (e *Engine) ApplyIgnore(ctx context.Context, pairID int64, ids []int64, opt BulkOptions) (int, error) {
	ids = uniqueIDs(ids)
	if err := validBulk(ids, opt); err != nil {
		return 0, err
	}
	if opt.Action != BulkIgnore && opt.Action != BulkUnignore {
		return 0, fmt.Errorf("%w: not an ignore action", ErrInvalidBulk)
	}
	ms, err := e.st.MatchesByIDs(ctx, ids)
	if err != nil {
		return 0, err
	}
	n := 0
	for _, m := range ms {
		if ok, _ := planBulk(pairID, m, opt); ok {
			if err := e.st.SetIgnored(ctx, m.ID, opt.Action == BulkIgnore); err != nil {
				return n, err
			}
			n++
		}
	}
	return n, nil
}

// StartBulk links or promotes the selected matches in a background job. Each
// match is re-planned when its turn comes (the scan may have changed it) and
// then goes through the normal single-item operation.
func (e *Engine) StartBulk(pairID int64, ids []int64, opt BulkOptions) (*Job, error) {
	if err := validBulk(ids, opt); err != nil {
		return nil, err
	}
	if opt.Action != BulkLink && opt.Action != BulkPromote {
		return nil, fmt.Errorf("%w: only link and promote run as a job", ErrInvalidBulk)
	}
	sorted := uniqueIDs(ids)
	sort.Slice(sorted, func(i, j int) bool { return sorted[i] < sorted[j] })
	return e.startJob(pairID, "bulk-"+opt.Action, func(ctx context.Context, j *Job) error {
		verb := "Linking"
		if opt.Action == BulkPromote {
			verb = "Promoting slave copies"
		}
		j.phase(verb, int64(len(sorted)))
		for _, id := range sorted {
			if ctx.Err() != nil {
				return ctx.Err()
			}
			e.bulkOne(ctx, j, pairID, id, opt)
			j.tick()
		}
		return nil
	})
}

func (e *Engine) bulkOne(ctx context.Context, j *Job, pairID, id int64, opt BulkOptions) {
	m, err := e.st.Match(ctx, id)
	if err != nil {
		j.skip(fmt.Sprintf("match %d: no longer exists (rescan)", id))
		return
	}
	if ok, why := planBulk(pairID, m, opt); !ok {
		j.skip(m.RelSlave + ": " + why)
		return
	}
	var op store.Operation
	if opt.Action == BulkLink {
		op, err = e.Link(ctx, id, LinkOptions{AllowDowngrade: opt.AllowDowngrade, KeepFolder: opt.KeepFolder, Trigger: "bulk"})
	} else {
		op, err = e.promote(ctx, id, "bulk")
	}
	switch {
	case err != nil:
		j.skip(m.RelSlave + ": " + strings.TrimSpace(err.Error()))
	case op.DryRun:
		j.count("planned")
	case opt.Action == BulkLink:
		j.count("linked")
	default:
		j.count("promoted")
	}
}

// AlbumGroup is one album folder of the slave library in the Review list.
type AlbumGroup struct {
	Folder      string  `json:"folder"`
	IDs         []int64 `json:"ids"`
	Songs       int     `json:"songs"`
	Identical   int     `json:"identical"`
	Review      int     `json:"review"` // different + similar
	Upgrades    int     `json:"upgrades"`
	Downgrades  int     `json:"downgrades"`
	Approximate int     `json:"approximate"`
	Other       int     `json:"other"` // linked, unmatched, error, ignored
	Bytes       int64   `json:"bytes"`
	// MasterFolders lists where the matched master files live (usually one).
	MasterFolders []string `json:"master_folders"`
}

// Groups returns the matches of a filter grouped by album folder.
func (e *Engine) Groups(ctx context.Context, pairID int64, f store.MatchFilter) ([]AlbumGroup, error) {
	byFolder := map[string]*AlbumGroup{}
	masters := map[string]map[string]bool{}
	err := e.st.ExportMatches(ctx, pairID, f, func(m store.Match) error {
		folder := AlbumFolder(m.RelSlave)
		g := byFolder[folder]
		if g == nil {
			g = &AlbumGroup{Folder: folder, IDs: []int64{}}
			byFolder[folder] = g
			masters[folder] = map[string]bool{}
		}
		g.IDs = append(g.IDs, m.ID)
		g.Songs++
		g.Bytes += m.SlaveSize
		switch {
		case m.Ignored || m.Status == store.StatusLinked || m.Status == store.StatusUnmatched || m.Status == store.StatusError:
			g.Other++
		case m.Status == store.StatusIdentical:
			g.Identical++
		default:
			g.Review++
			switch verdict(m).Better {
			case "master":
				g.Upgrades++
			case "slave":
				g.Downgrades++
			}
		}
		if m.Kind == store.KindApprox {
			g.Approximate++
		}
		if m.RelMaster != "" {
			masters[folder][AlbumFolder(m.RelMaster)] = true
		}
		return nil
	})
	if err != nil {
		return nil, err
	}
	out := make([]AlbumGroup, 0, len(byFolder))
	for folder, g := range byFolder {
		for mf := range masters[folder] {
			g.MasterFolders = append(g.MasterFolders, mf)
		}
		sort.Strings(g.MasterFolders)
		if g.MasterFolders == nil {
			g.MasterFolders = []string{}
		}
		out = append(out, *g)
	}
	sort.Slice(out, func(i, j int) bool { return strings.ToLower(out[i].Folder) < strings.ToLower(out[j].Folder) })
	return out, nil
}