minstrel/internal/acquire/search.go

package acquire

import (
	"context"
	"errors"
	"sort"
	"strings"

	"github.com/jackc/pgx/v5/pgtype"

	"git.fabledsword.com/bvandeusen/minstrel/internal/indexer"
	"git.fabledsword.com/bvandeusen/minstrel/internal/release"
)

// One search for one want, and what could be done about each result (M502
// #5521), ported from Fabled Librarian's internal/acquire. internal/release
// says whether a name is the album and what quality it is; this package
// asks in the right order and says why in words a person can act on.

// Verdict is what can be done about one result.
type Verdict string

const (
	// Grab: the album, in a quality the profile takes.
	Grab Verdict = "grab"
	// Review: maybe the album; a person decides.
	Review Verdict = "review"
	// Refused: the album, in a quality the profile does not take. The only
	// verdict that names a setting.
	Refused Verdict = "refused"
	// Rejected: not the album.
	Rejected Verdict = "rejected"
	// Blocked: a release set aside, by a person or by its files.
	Blocked Verdict = "blocked"
)

// order sorts verdicts: what can be taken first, what was given up on last.
var order = map[Verdict]int{Grab: 0, Review: 1, Refused: 2, Rejected: 3, Blocked: 4}

// Reason kinds: a hard reason is a fact about the bytes, a soft one about the
// moment or a setting.
const (
	Hard = "hard"
	Soft = "soft"
)

// Reason codes beyond internal/release's vetoes.
const (
	CodeBlocked   = "blocked"
	CodeProfile   = "profile"
	CodeNoSeeders = "no_seeders"
	CodeDoubt     = "doubt"
)

// Reason is one thing a verdict rests on.
type Reason struct {
	Kind string `json:"kind"`
	Code string `json:"code"`
	Text string `json:"text"`
}

// Want is the album searched for: its wants row and how the catalogue
// describes it.
type Want struct {
	ID pgtype.UUID
	release.Want
}

// Candidate is one result with everything decided about it.
type Candidate struct {
	release.Release
	// Source is the indexer's own result. It never leaves the server: its
	// DownloadURL can carry the indexer's key.
	Source     indexer.Result
	InfoHash   string
	Indexer    string
	SizeBytes  int64
	Seeders    int
	SwarmKnown bool
	Category   int
	FilesCount int
	// Files is the torrent's file list once its .torrent has been read.
	Files      []release.File
	FoldedFrom []string

	Tier       release.Tier
	YearAgrees bool
	Verdict    Verdict
	Reason     string
	Reasons    []Reason
}

// Key is how a candidate is named across a search and a grab: its infohash,
// or its name when the source gave none.
func Key(c Candidate) string {
	if c.InfoHash != "" {
		return c.InfoHash
	}
	return c.Name
}

// Result is one search.
type Result struct {
	Query string
	// Unconfigured: no search source is set up.
	Unconfigured bool
	// Error is a search that failed outright.
	Error string
	// Narrowed names sources that did not answer on a search that otherwise
	// did.
	Narrowed   []Narrowing
	Considered int
	Folded     int
	// Looked is how many .torrent files were read for their file lists.
	Looked     int
	Candidates []Candidate
	// KeepError: the candidates could not be stored.
	KeepError string
}

// Grabbable is what may be taken, best first.
func (r Result) Grabbable() []Candidate {
	var out []Candidate
	for _, c := range r.Candidates {
		if c.Verdict == Grab {
			out = append(out, c)
		}
	}
	return out
}

// ForAPerson says whether anything found is worth a person's look: a
// release that may be the album, whatever its quality.
func (r Result) ForAPerson() bool {
	for _, c := range r.Candidates {
		if c.Verdict == Grab || c.Verdict == Review || c.Verdict == Refused {
			return true
		}
	}
	return false
}

// Service searches and grabs.
type Service struct {
	// Search is nil when no source is configured.
	Search Searcher
	// Grabber is nil when no download client is configured.
	Grabber *Downloader
	Profile release.Profile
	// Memory stores candidates, grabs and blocks. Nil keeps nothing.
	Memory *Memory
}

// Query is the text sent to the indexers: the artist and the title, without
// a trailing bracketed part the release names usually leave off.
func Query(w release.Want) string {
	title := w.Title
	if i := strings.IndexAny(title, "(["); i > 0 {
		title = title[:i]
	}
	return strings.Join(strings.Fields(strings.TrimSpace(w.Artist)+" "+strings.TrimSpace(title)), " ")
}

// For searches for one want and judges everything that comes back. It never
// returns an error: a search that could not be made is a Result saying so.
func (s *Service) For(ctx context.Context, want Want) Result {
	r := Result{Query: Query(want.Want)}
	if s == nil || s.Search == nil {
		r.Unconfigured = true
		return r
	}
	found, err := s.Search.Search(ctx, indexer.Query{Text: r.Query, Categories: indexer.MusicCategories})
	// A partial search arrives as results and an error together.
	if narrowed, answered := narrowings(err); narrowed != nil && answered {
		r.Narrowed, err = narrowed, nil
	}
	switch {
	case errors.Is(err, indexer.ErrUnconfigured):
		r.Unconfigured = true
		return r
	case err != nil:
		r.Error = indexer.FirstLine(err.Error())
		return r
	}
	r.Considered = len(found)
	cands, folded := fold(found)
	r.Folded = folded

	blocked := s.Memory.blocked(ctx)
	for i := range cands {
		decide(&cands[i], want.Want, s.Profile, blocked)
	}
	sortCandidates(cands)
	for _, i := range s.peek(ctx, cands) {
		decide(&cands[i], want.Want, s.Profile, blocked)
		r.Looked++
	}
	sortCandidates(cands)
	r.Candidates = cands
	if s.Memory != nil && want.ID.Valid {
		if err := s.Memory.Keep(ctx, want.ID, cands); err != nil {
			r.KeepError = indexer.FirstLine(err.Error())
		}
	}
	return r
}

// decide works out one candidate's verdict, in order: given up on, then
// whether it is the album, then whether the profile takes its quality, then
// whether anything leaves it to a person. A quality refusal said about the
// wrong album would be noise, so identity comes first.
func decide(c *Candidate, w release.Want, profile release.Profile, blocked map[string]string) {
	c.Verdict, c.Reason, c.Reasons = "", "", nil
	if why, ok := blocked[c.InfoHash]; ok && c.InfoHash != "" {
		set(c, Blocked, Hard, CodeBlocked, why)
		return
	}
	m := release.Judge(w, c.Release)
	c.YearAgrees = m.YearAgrees
	if m.Veto != "" {
		set(c, Rejected, Hard, m.Veto, m.Reason)
		return
	}
	doubt := m.Doubt
	if len(c.Files) > 0 {
		if in := release.Inspect(c.Files); in.Format != "" {
			c.Format = in.Format
		}
		veto, reason, fileDoubt := release.JudgeFiles(w, c.Files)
		if veto != "" {
			set(c, Rejected, Hard, veto, reason)
			return
		}
		if doubt == "" {
			doubt = fileDoubt
		}
	}
	c.Tier = release.TierOf(c.Release, c.Category)
	if !profile.Allows(c.Tier) {
		set(c, Refused, Soft, CodeProfile, "Your quality settings don’t take "+strings.ToLower(release.Label(c.Tier))+".")
		return
	}
	if doubt != "" {
		set(c, Review, Soft, CodeDoubt, doubt)
		return
	}
	if c.Seeders == 0 {
		set(c, Review, Soft, CodeNoSeeders, "Nobody is sharing it right now.")
		return
	}
	c.Verdict = Grab
	c.Reason = "The album, " + strings.ToLower(release.Label(c.Tier)) + "."
	if d := release.Describe(c.Release); d != "" {
		c.Reason = "The album, " + d + "."
	}
}

func set(c *Candidate, v Verdict, kind, code, text string) {
	c.Verdict, c.Reason = v, text
	c.Reasons = append(c.Reasons, Reason{Kind: kind, Code: code, Text: text})
}

// sortCandidates puts the best first, stably so the same search proposes the
// same grab twice.
func sortCandidates(cands []Candidate) {
	sort.SliceStable(cands, func(a, b int) bool { return better(cands[a], cands[b]) })
}

// better says whether x ranks above y: verdict, then quality, then how many
// share it, then whether its year agrees.
func better(x, y Candidate) bool {
	if order[x.Verdict] != order[y.Verdict] {
		return order[x.Verdict] < order[y.Verdict]
	}
	if rx, ry := release.Rank(x.Tier), release.Rank(y.Tier); rx != ry {
		return rx > ry
	}
	if x.Seeders != y.Seeders {
		return x.Seeders > y.Seeders
	}
	return x.YearAgrees && !y.YearAgrees
}

// fold merges results that are the same torrent, by infohash. Prowlarr
// aggregates, so one release arrives once per tracker that lists it. A
// result without a hash is never folded: an absent hash is not a shared one.
func fold(found []indexer.Result) ([]Candidate, int) {
	out := make([]Candidate, 0, len(found))
	at := make(map[string]int, len(found))
	folded := 0
	for _, f := range found {
		if strings.TrimSpace(f.Title) == "" {
			continue
		}
		hash := indexer.NormalizeHash(f.InfoHash)
		if hash == "" && f.Magnet != "" {
			hash = indexer.MagnetHash(f.Magnet)
		}
		if hash != "" {
			if i, seen := at[hash]; seen {
				folded++
				if f.Indexer != "" {
					out[i].FoldedFrom = append(out[i].FoldedFrom, f.Indexer)
				}
				// A measured swarm beats a reported one; among like, the
				// highest count.
				switch {
				case f.SwarmKnown && !out[i].SwarmKnown:
					out[i].Seeders, out[i].SwarmKnown = f.Seeders, true
				case f.SwarmKnown == out[i].SwarmKnown && f.Seeders > out[i].Seeders:
					out[i].Seeders = f.Seeders
				}
				if f.FilesCount > 0 && out[i].FilesCount == 0 {
					out[i].FilesCount = f.FilesCount
				}
				continue
			}
			at[hash] = len(out)
		}
		out = append(out, Candidate{
			Release: release.Parse(f.Title), Source: f, InfoHash: hash, Indexer: f.Indexer,
			SizeBytes: f.SizeBytes, Seeders: f.Seeders, SwarmKnown: f.SwarmKnown,
			Category: f.Category, FilesCount: f.FilesCount,
		})
	}
	return out, folded
}

// maxPeeks is how many .torrent files one search reads. Few: each goes
// through Prowlarr to a tracker, and Prowlarr answers 429 to a burst.
const maxPeeks = 3

// peek reads the .torrent of candidates that are the album but whose name
// states no format, so their files can say what they are; it returns which
// now carry a file list. A failed fetch leaves the candidate as it was.
func (s *Service) peek(ctx context.Context, cands []Candidate) []int {
	f, ok := s.Search.(Fetcher)
	if !ok {
		return nil
	}
	var looked []int
	for i := range cands {
		if len(looked) >= maxPeeks || ctx.Err() != nil {
			break
		}
		c := &cands[i]
		if c.Tier != release.TierUnknown || (c.Verdict != Grab && c.Verdict != Refused) ||
			c.Files != nil || c.Source.DownloadURL == "" || c.Source.Magnet != "" {
			continue
		}
		body, err := f.Download(ctx, c.Source)
		if err != nil {
			continue
		}
		meta, err := release.ParseTorrent(body)
		if err != nil {
			continue
		}
		c.Files = meta.Files
		if c.InfoHash == "" {
			c.InfoHash = meta.InfoHash
		}
		if c.FilesCount == 0 {
			c.FilesCount = len(meta.Files)
		}
		looked = append(looked, i)
	}
	return looked
}

// pick finds a candidate by key: the infohash first, the exact name for a
// result that has none.
func pick(cands []Candidate, key string) *Candidate {
	key = strings.TrimSpace(key)
	if key == "" {
		return nil
	}
	hash := indexer.NormalizeHash(key)
	for i := range cands {
		if cands[i].InfoHash != "" && cands[i].InfoHash == hash {
			return &cands[i]
		}
	}
	for i := range cands {
		if cands[i].InfoHash == "" && cands[i].Name == key {
			return &cands[i]
		}
	}
	return nil
}