def fingerprint_file(self, file_path: str, song_name: str = None) -> None:
"""
Given a path to a file the method generates hashes for it and stores them in the database
for later be queried.
:param file_path: path to the file.
:param song_name: song name associated to the audio file.
"""
song_name_from_path = decoder.get_audio_name_from_path(file_path)
song_hash = decoder.unique_hash(file_path)
song_name = song_name or song_name_from_path
# don't refingerprint already fingerprinted files
if song_hash in self.songhashes_set:
print(f"{song_name} already fingerprinted, continuing...")
else:
song_name, hashes, file_hash = Dejavu._fingerprint_worker(
file_path, self.limit, song_name=song_name)
sid = self.db.insert_song(song_name, file_hash)
self.db.insert_hashes(sid, hashes)
self.db.set_song_fingerprinted(sid)
self.__load_fingerprinted_audio_hashes()
def begin(self):
for f in self.test_files:
log_msg('--------------------------------------------------')
log_msg(f'file: {f}')
# get column
col = self.get_column_id([
x for x in re.findall("[0-9]sec", f) if x in self.test_seconds
][0])
# format: XXXX_offset_length.mp3, we also take into account underscores within XXXX
splits = get_audio_name_from_path(f).split("_")
song = "_".join(
splits[0:len(get_audio_name_from_path(f).split("_")) - 2])
line = self.get_line_id(song)
result = subprocess.check_output([
"python", "dejavu.py", '-r', 'file',
join(self.test_folder, f)
])
if result.strip() == "None":
log_msg('No match')
self.result_match[line][col] = 'no'
self.result_matching_times[line][col] = 0
self.result_query_duration[line][col] = 0
self.result_match_confidence[line][col] = 0
else:
result = result.strip()
# we parse the output song back to a json
result = json.loads(
result.decode('utf-8').replace("'",
'"').replace(': b"', ':"'))
# which song did we predict? We consider only the first match.
match = result[RESULTS][0]
song_result = match[SONG_NAME]
log_msg(f'song: {song}')
log_msg(f'song_result: {song_result}')
if song_result != song:
log_msg('invalid match')
self.result_match[line][col] = 'invalid'
self.result_matching_times[line][col] = 0
self.result_query_duration[line][col] = 0
self.result_match_confidence[line][col] = 0
else:
log_msg('correct match')
print(self.result_match)
self.result_match[line][col] = 'yes'
self.result_query_duration[line][col] = round(
result[TOTAL_TIME], 3)
self.result_match_confidence[line][col] = match[
HASHES_MATCHED]
# using replace in f for getting rid of underscores in name
song_start_time = re.findall("_[^_]+", f.replace(song, ""))
song_start_time = song_start_time[0].lstrip("_ ")
result_start_time = round(
(match[OFFSET] * DEFAULT_WINDOW_SIZE *
DEFAULT_OVERLAP_RATIO) / DEFAULT_FS, 0)
self.result_matching_times[line][col] = int(
result_start_time) - int(song_start_time)
if abs(self.result_matching_times[line][col]) == 1:
self.result_matching_times[line][col] = 0
log_msg(f'query duration: {round(result[TOTAL_TIME], 3)}')
log_msg(f'confidence: {match[HASHES_MATCHED]}')
log_msg(f'song start_time: {song_start_time}')
log_msg(f'result start time: {result_start_time}')
if self.result_matching_times[line][col] == 0:
log_msg('accurate match')
else:
log_msg('inaccurate match')
log_msg('--------------------------------------------------\n')