def read_chunk(file: BinaryIO) -> 'ExplicitVocab':
length, ngram_length, min_n, max_n = _read_required_binary(
file, "<QQII")
words = _read_items(file, length)
ngram_list, ngram_index = _read_items_with_indices(file, ngram_length)
indexer = ExplicitIndexer(ngram_list, min_n, max_n, ngram_index)
return ExplicitVocab(words, indexer)
def read_chunk(file: BinaryIO) -> 'FastTextVocab':
length, min_n, max_n, buckets = _read_required_binary(file, "<QIII")
words = _read_items(file, length)
indexer = FastTextIndexer(buckets, min_n, max_n)
return FastTextVocab(words, indexer)
def read_chunk(file: BinaryIO) -> 'SimpleVocab':
length = _read_required_binary(file, "<Q")[0]
words = _read_items(file, length)
return SimpleVocab(words)