cdnutter/playlink
1
0 Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Add opt-in approximate matching and upgrade recommendations

Browse Source 
A per-pair "Approximate matching" setting adds a second pass for slave
files the strict rules could not place. It searches the same album with
disc folders (Disc 02, CD2, ...) ignored, then the top-level folder, and
scores titles by edit distance and word containment against a configurable
threshold (default 85%), with a small track-number bonus.

Guards against false matches: version words (live, remix, reprise,
instrumental, ...) must agree, near-tied candidates are left unmatched,
and approximate matches always go to manual review, never auto-link.
Each match records a human-readable reason.

Review now flags matches where the master copy is better as upgrades,
with an Upgrades filter and count and an "Upgrade to master copy" action.
The redundant "master is lossless" warning note is dropped.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
cdnutterandClaude Opus 5.5 committed 2026-09-22 16:58:53 -07:00
1 parent 5b088bb4a6
commit 381b804742
12 files changed
+552 -51
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

internal/store/models.go
+38 -20
View File
@@ -104,6 +104,8 @@ type Pair struct {
ScanIntervalMinutes int `json:"scan_interval_minutes"`
RetentionDays int `json:"retention_days"`
SettleMinutes int `json:"settle_minutes"`
ApproxMatching bool `json:"approx_matching"`
ApproxThreshold int `json:"approx_threshold"` // percent
LastScanAt *int64 `json:"last_scan_at"`
LastScanError string `json:"last_scan_error"`
LastScanWarnings []string `json:"last_scan_warnings"`
@@ -111,7 +113,8 @@ type Pair struct {
}
const pairCols = `id, name, master_root, slave_root, quarantine_dir, extensions, auto_link_identical,
scan_interval_minutes, retention_days, settle_minutes, last_scan_at, last_scan_error, last_scan_warnings, created_at`
scan_interval_minutes, retention_days, settle_minutes, approx_matching, approx_threshold, last_scan_at, last_scan_error,
last_scan_warnings, created_at`
type scanner interface{ Scan(...any) error }
@@ -120,7 +123,8 @@ func scanPair(r scanner) (Pair, error) {
var exts, warns string
var last sql.NullInt64
err := r.Scan(&p.ID, &p.Name, &p.MasterRoot, &p.SlaveRoot, &p.QuarantineDir, &exts, &p.AutoLinkIdentical,
&p.ScanIntervalMinutes, &p.RetentionDays, &p.SettleMinutes, &last, &p.LastScanError, &warns, &p.CreatedAt)
&p.ScanIntervalMinutes, &p.RetentionDays, &p.SettleMinutes, &p.ApproxMatching, &p.ApproxThreshold, &last, &p.LastScanError,
&warns, &p.CreatedAt)
if err != nil {
return p, err
}
@@ -164,10 +168,10 @@ func (s *Store) SavePair(ctx context.Context, p *Pair) error {
exts := strings.Join(p.Extensions, ",")
if p.ID == 0 {
res, err := s.DB.ExecContext(ctx, `INSERT INTO pairs (name, master_root, slave_root, quarantine_dir, extensions,
auto_link_identical, scan_interval_minutes, retention_days, settle_minutes, created_at)
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
auto_link_identical, scan_interval_minutes, retention_days, settle_minutes, approx_matching, approx_threshold, created_at)
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
p.Name, p.MasterRoot, p.SlaveRoot, p.QuarantineDir, exts, p.AutoLinkIdentical,
p.ScanIntervalMinutes, p.RetentionDays, p.SettleMinutes, now())
p.ScanIntervalMinutes, p.RetentionDays, p.SettleMinutes, p.ApproxMatching, p.ApproxThreshold, now())
if err != nil {
return err
}
@@ -175,10 +179,10 @@ func (s *Store) SavePair(ctx context.Context, p *Pair) error {
return nil
}
_, err := s.DB.ExecContext(ctx, `UPDATE pairs SET name = ?, master_root = ?, slave_root = ?, quarantine_dir = ?,
extensions = ?, auto_link_identical = ?, scan_interval_minutes = ?, retention_days = ?, settle_minutes = ?
WHERE id = ?`,
extensions = ?, auto_link_identical = ?, scan_interval_minutes = ?, retention_days = ?, settle_minutes = ?,
approx_matching = ?, approx_threshold = ? WHERE id = ?`,
p.Name, p.MasterRoot, p.SlaveRoot, p.QuarantineDir, exts, p.AutoLinkIdentical,
p.ScanIntervalMinutes, p.RetentionDays, p.SettleMinutes, p.ID)
p.ScanIntervalMinutes, p.RetentionDays, p.SettleMinutes, p.ApproxMatching, p.ApproxThreshold, p.ID)
return err
}
@@ -216,11 +220,12 @@ const (
// Match kinds (how the master candidate was found).
const (
KindPath = "path" // identical relative path
KindPathCase = "path_case" // relative path differs only by case
KindExtension = "extension" // same directory and title, different extension
KindFuzzy = "fuzzy" // same directory, normalized title matches
KindElsewhere = "elsewhere" // same top-level folder and title, different subfolder
KindPath = "path" // identical relative path
KindPathCase = "path_case" // relative path differs only by case
KindExtension = "extension" // same directory and title, different extension
KindFuzzy = "fuzzy" // same directory, normalized title matches
KindElsewhere = "elsewhere" // same top-level folder and title, different subfolder
KindApprox = "approximate" // opt-in: similar title, disc folders ignored
KindNone = "none"
)
@@ -241,18 +246,19 @@ type Match struct {
SlaveProbe json.RawMessage `json:"slave_probe,omitempty"`
Verdict json.RawMessage `json:"verdict,omitempty"`
Note string `json:"note"`
Reason string `json:"reason"` // why an approximate match was chosen
Ignored bool `json:"ignored"`
UpdatedAt int64 `json:"updated_at"`
}
const matchCols = `id, pair_id, rel_slave, rel_master, kind, status, master_ino, master_size, master_mtime,
slave_ino, slave_size, slave_mtime, master_probe, slave_probe, verdict, note, ignored, updated_at`
slave_ino, slave_size, slave_mtime, master_probe, slave_probe, verdict, note, reason, ignored, updated_at`
func scanMatch(r scanner) (Match, error) {
var m Match
var mp, sp, v string
err := r.Scan(&m.ID, &m.PairID, &m.RelSlave, &m.RelMaster, &m.Kind, &m.Status, &m.MasterIno, &m.MasterSize, &m.MasterMtime,
&m.SlaveIno, &m.SlaveSize, &m.SlaveMtime, &mp, &sp, &v, &m.Note, &m.Ignored, &m.UpdatedAt)
&m.SlaveIno, &m.SlaveSize, &m.SlaveMtime, &mp, &sp, &v, &m.Note, &m.Reason, &m.Ignored, &m.UpdatedAt)
if mp != "" {
m.MasterProbe = json.RawMessage(mp)
}
@@ -316,6 +322,8 @@ func (s *Store) Matches(ctx context.Context, pairID int64, f MatchFilter) ([]Mat
where = append(where, "status IN ('different','similar') AND ignored = 0")
case "ignored":
where = append(where, "ignored = 1")
case "upgrade":
where = append(where, "status IN ('different','similar') AND ignored = 0 AND "+upgradeCond)
default:
where = append(where, "status = ? AND ignored = 0")
args = append(args, f.Status)
@@ -371,15 +379,15 @@ func (s *Store) IdenticalIDs(ctx context.Context, pairID int64) ([]int64, error)
func (s *Store) UpsertMatch(ctx context.Context, m *Match) error {
nullable := func(r json.RawMessage) string { return string(r) }
_, err := s.DB.ExecContext(ctx, `INSERT INTO matches (pair_id, rel_slave, rel_master, kind, status, master_ino, master_size,
master_mtime, slave_ino, slave_size, slave_mtime, master_probe, slave_probe, verdict, note, ignored, updated_at)
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
master_mtime, slave_ino, slave_size, slave_mtime, master_probe, slave_probe, verdict, note, reason, ignored, updated_at)
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
ON CONFLICT(pair_id, rel_slave) DO UPDATE SET rel_master = excluded.rel_master, kind = excluded.kind,
status = excluded.status, master_ino = excluded.master_ino, master_size = excluded.master_size,
master_mtime = excluded.master_mtime, slave_ino = excluded.slave_ino, slave_size = excluded.slave_size,
slave_mtime = excluded.slave_mtime, master_probe = excluded.master_probe, slave_probe = excluded.slave_probe,
verdict = excluded.verdict, note = excluded.note, ignored = excluded.ignored, updated_at = excluded.updated_at`,
verdict = excluded.verdict, note = excluded.note, reason = excluded.reason, ignored = excluded.ignored, updated_at = excluded.updated_at`,
m.PairID, m.RelSlave, m.RelMaster, m.Kind, m.Status, m.MasterIno, m.MasterSize, m.MasterMtime, m.SlaveIno, m.SlaveSize,
m.SlaveMtime, nullable(m.MasterProbe), nullable(m.SlaveProbe), nullable(m.Verdict), m.Note, m.Ignored, now())
m.SlaveMtime, nullable(m.MasterProbe), nullable(m.SlaveProbe), nullable(m.Verdict), m.Note, m.Reason, m.Ignored, now())
return err
}
@@ -433,9 +441,14 @@ func (s *Store) LinkedMasterPaths(ctx context.Context, pairID int64, prefix stri
return out, rows.Err()
}
// upgradeCond selects review matches where the master copy scored better, so
// linking upgrades the slave.
const upgradeCond = `verdict LIKE '%"better":"master"%'`
type Summary struct {
Counts map[string]int `json:"counts"`
Review int `json:"review"`
Upgrades int `json:"upgrades"`
Ignored int `json:"ignored"`
ReclaimableSize int64 `json:"reclaimable_bytes"`
LinkedSize int64 `json:"linked_bytes"`
@@ -470,7 +483,12 @@ func (s *Store) Summary(ctx context.Context, pairID int64) (Summary, error) {
sum.Review += n
}
}
return sum, rows.Err()
if err := rows.Err(); err != nil {
return sum, err
}
err = s.DB.QueryRowContext(ctx, `SELECT COUNT(*) FROM matches WHERE pair_id = ? AND ignored = 0
AND status IN ('different','similar') AND `+upgradeCond, pairID).Scan(&sum.Upgrades)
return sum, err
}
// ---- operations ----