Music duplicate finder

music-dup-finder.py

#!/usr/bin/python
# -*- coding: utf-8 -*-

import os, sys

import acoustid
import xdg.BaseDirectory
import apsw

def setwal(db):
    db.cursor().execute("pragma journal_mode=wal")
    db.wal_autocheckpoint(10)

apsw.connection_hooks.append(setwal)


cache = xdg.BaseDirectory.save_cache_path('music-dup-finder') + "/cache.sqlite"
db = apsw.Connection(cache)

cursor = db.cursor()
row = cursor.execute("SELECT count(*) FROM sqlite_master WHERE type='table' AND name='files'").next()
if row[0] == 0:
    cursor.execute("CREATE TABLE files (id INTEGER PRIMARY KEY, name TEXT NOT NULL, fingerprint TEXT NOT NULL)")
    cursor.execute("CREATE UNIQUE INDEX files_name ON files (name)")
    cursor.execute("CREATE INDEX files_fingerprint ON files (fingerprint)")

# New Fingerprints
for (path, _, files) in os.walk(unicode(sys.argv[1])):
    p = os.path.abspath(path)
    for f in files:
        name = os.path.join(p, f)
        if os.path.splitext(f)[1].lower() not in (u'.mp3', u'.ogg', u'.wav', u'.wma', u'.flac', u'.m4a', u'.mp4', u'.mp2', u'.mpg'):
            continue
        exist = cursor.execute("SELECT count(*) FROM files WHERE name = ?", (name,)).next()
        if exist[0] == 1:
            continue
        try:
            sys.stdout.write(u"Fingerprinting »%s«... " % (f,))
            sys.stdout.flush()
            _, fingerprint = acoustid.fingerprint_file(name.encode('utf-8'))
            cursor.execute("INSERT INTO files (name, fingerprint) VALUES (?,?)", (name, fingerprint))
        except acoustid.FingerprintGenerationError as e:
            print u"ERR: %s" % (e,)
        else:
            print u"OK"

# Remove old fingerprints
cursor2 = db.cursor()
for f in cursor.execute("SELECT name, id FROM files"):
    if not os.access(f[0], os.F_OK):
        cursor2.execute("DELETE FROM files WHERE id = ?", (f[1],))

# Find duplicates
for dup in cursor.execute("SELECT count(*) AS no, fingerprint FROM files GROUP BY fingerprint HAVING no > 1"):
    print "Possible duplicate:"
    for f in cursor2.execute("SELECT name FROM files WHERE fingerprint = ?", (dup[1],)):
        print u"\t%s" % f

db.close()