def findAllSeqs()

in data_preparation/make_vad_inputs.py [0:0]

• 37 lines of code
• 13 McCabe index (conditional complexity)

def findAllSeqs(dirName,
                extension='.flac',
                loadCache=False):
    r"""
    Lists all the sequences with the given extension in the dirName directory.
    Output:
        outSequences, speakers

        outSequence
        A list of tuples seq_path, speaker where:
            - seq_path is the relative path of each sequence relative to the
            parent directory
            - speaker is the corresponding speaker index

        outSpeakers
        The speaker labels (in order)

    The speaker labels are organized the following way
    \dirName
        \speaker_label
            \..
                ...
                seqName.extension
    """
    cache_path = os.path.join(dirName, '_seqs_cache.txt')
    if loadCache:
        try:
            outSequences, speakers = torch.load(cache_path)
            print(f'Loaded from cache {cache_path} successfully')
            return outSequences, speakers
        except OSError as err:
            print(f'Ran in an error while loading {cache_path}: {err}')
        print('Could not load cache, rebuilding')

    if dirName[-1] != os.sep:
        dirName += os.sep
    prefixSize = len(dirName)
    speakersTarget = {}
    outSequences = []
    for root, dirs, filenames in tqdm.tqdm(os.walk(dirName)):
        filtered_files = [f for f in filenames if f.endswith(extension)]

        if len(filtered_files) > 0:
            speakerStr = root[prefixSize:].split(os.sep)[0]
            if speakerStr not in speakersTarget:
                speakersTarget[speakerStr] = len(speakersTarget)
            speaker = speakersTarget[speakerStr]
            for filename in filtered_files:
                full_path = os.path.join(root[prefixSize:], filename)
                outSequences.append((speaker, full_path))
    outSpeakers = [None for x in speakersTarget]
    for key, index in speakersTarget.items():
        outSpeakers[index] = key
    try:
        torch.save((outSequences, outSpeakers), cache_path)
        print(f'Saved cache file at {cache_path}')
    except OSError as err:
        print(f'Ran in an error while saving {cache_path}: {err}')
    return outSequences, outSpeakers