test_tokenization_xglm.py

# coding=utf-8
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import pickle
import shutil
import tempfile
import unittest

from transformers import SPIECE_UNDERLINE, XGLMTokenizer, XGLMTokenizerFast
from transformers.testing_utils import get_tests_dir, require_sentencepiece, require_tokenizers, slow
from transformers.utils import cached_property

from ...test_tokenization_common import TokenizerTesterMixin


SAMPLE_VOCAB = get_tests_dir("fixtures/test_sentencepiece.model")


@require_sentencepiece
@require_tokenizers
class XGLMTokenizationTest(TokenizerTesterMixin, unittest.TestCase):

    tokenizer_class = XGLMTokenizer
    rust_tokenizer_class = XGLMTokenizerFast
    test_rust_tokenizer = True
    test_sentencepiece = True

    def setUp(self):
        super().setUp()

        # We have a SentencePiece fixture for testing
        tokenizer = XGLMTokenizer(SAMPLE_VOCAB, keep_accents=True)
        tokenizer.save_pretrained(self.tmpdirname)

    def test_convert_token_and_id(self):
        """Test ``_convert_token_to_id`` and ``_convert_id_to_token``."""
        token = "<pad>"
        token_id = 1

        self.assertEqual(self.get_tokenizer()._convert_token_to_id(token), token_id)
        self.assertEqual(self.get_tokenizer()._convert_id_to_token(token_id), token)

    def test_get_vocab(self):
        vocab_keys = list(self.get_tokenizer().get_vocab().keys())

        self.assertEqual(vocab_keys[0], "<s>")
        self.assertEqual(vocab_keys[1], "<pad>")
        self.assertEqual(len(vocab_keys), 1_008)

    def test_vocab_size(self):
        self.assertEqual(self.get_tokenizer().vocab_size, 1_008)

    def test_full_tokenizer(self):
        tokenizer = XGLMTokenizer(SAMPLE_VOCAB, keep_accents=True)

        tokens = tokenizer.tokenize("This is a test")
        self.assertListEqual(tokens, ["▁This", "▁is", "▁a", "▁t", "est"])

        self.assertListEqual(
            tokenizer.convert_tokens_to_ids(tokens),
            [value + tokenizer.fairseq_offset for value in [285, 46, 10, 170, 382]],
        )

        tokens = tokenizer.tokenize("I was born in 92000, and this is falsé.")
        self.assertListEqual(
            tokens,
            [
                SPIECE_UNDERLINE + "I",
                SPIECE_UNDERLINE + "was",
                SPIECE_UNDERLINE + "b",
                "or",
                "n",
                SPIECE_UNDERLINE + "in",
                SPIECE_UNDERLINE + "",
                "9",
                "2",
                "0",
                "0",
                "0",
                ",",
                SPIECE_UNDERLINE + "and",
                SPIECE_UNDERLINE + "this",
                SPIECE_UNDERLINE + "is",
                SPIECE_UNDERLINE + "f",
                "al",
                "s",
                "é",
                ".",
            ],
        )
        ids = tokenizer.convert_tokens_to_ids(tokens)
        self.assertListEqual(
            ids,
            [
                value + tokenizer.fairseq_offset
                for value in [8, 21, 84, 55, 24, 19, 7, 2, 602, 347, 347, 347, 3, 12, 66, 46, 72, 80, 6, 2, 4]
            ],
        )

        back_tokens = tokenizer.convert_ids_to_tokens(ids)
        self.assertListEqual(
            back_tokens,
            [
                SPIECE_UNDERLINE + "I",
                SPIECE_UNDERLINE + "was",
                SPIECE_UNDERLINE + "b",
                "or",
                "n",
                SPIECE_UNDERLINE + "in",
                SPIECE_UNDERLINE + "",
                "<unk>",
                "2",
                "0",
                "0",
                "0",
                ",",
                SPIECE_UNDERLINE + "and",
                SPIECE_UNDERLINE + "this",
                SPIECE_UNDERLINE + "is",
                SPIECE_UNDERLINE + "f",
                "al",
                "s",
                "<unk>",
                ".",
            ],
        )

    @cached_property
    def big_tokenizer(self):
        return XGLMTokenizer.from_pretrained("facebook/xglm-564M")

    def test_picklable_without_disk(self):
        with tempfile.NamedTemporaryFile() as f:
            shutil.copyfile(SAMPLE_VOCAB, f.name)
            tokenizer = XGLMTokenizer(f.name, keep_accents=True)
            pickled_tokenizer = pickle.dumps(tokenizer)
        pickle.loads(pickled_tokenizer)

    def test_rust_and_python_full_tokenizers(self):
        if not self.test_rust_tokenizer:
            return

        tokenizer = self.get_tokenizer()
        rust_tokenizer = self.get_rust_tokenizer()

        sequence = "I was born in 92000, and this is falsé."

        tokens = tokenizer.tokenize(sequence)
        rust_tokens = rust_tokenizer.tokenize(sequence)
        self.assertListEqual(tokens, rust_tokens)

        ids = tokenizer.encode(sequence, add_special_tokens=False)
        rust_ids = rust_tokenizer.encode(sequence, add_special_tokens=False)
        self.assertListEqual(ids, rust_ids)

        rust_tokenizer = self.get_rust_tokenizer()
        ids = tokenizer.encode(sequence)
        rust_ids = rust_tokenizer.encode(sequence)
        self.assertListEqual(ids, rust_ids)

    @slow
    def test_tokenization_base_easy_symbols(self):
        symbols = "Hello World!"
        original_tokenizer_encodings = [2, 31227, 4447, 35]

        self.assertListEqual(original_tokenizer_encodings, self.big_tokenizer.encode(symbols))

    @slow
    def test_tokenization_base_hard_symbols(self):
        symbols = 'This is a very long text with a lot of weird characters, such as: . , ~ ? ( ) " [ ] ! : - . Also we will add words that should not exsist and be tokenized to unk, such as saoneuhaoesuth'
        # fmt: off
        original_tokenizer_encodings = [2, 1018, 67, 11, 1988, 2617, 5631, 278, 11, 3407, 48, 71630, 28085, 4, 3234, 157, 13, 6, 5, 6, 4, 3526, 768, 15, 659, 57, 298, 3983, 864, 129, 21, 6, 5, 13675, 377, 652, 7580, 10341, 155, 2817, 422, 1666, 7, 1674, 53, 113, 202277, 17892, 33, 60, 87, 4, 3234, 157, 61, 2667, 52376, 19, 88, 23, 735]
        # fmt: on

        self.assertListEqual(original_tokenizer_encodings, self.big_tokenizer.encode(symbols))

    @slow
    def test_tokenizer_integration(self):
        # fmt: off
        expected_encoding = {
            'input_ids': [[2, 108825, 1163, 15, 88010, 473, 15898, 157, 13672, 1857, 312, 8, 238021, 1163, 53, 13672, 1857, 312, 8, 53283, 182396, 8, 18566, 16, 36733, 4101, 8, 230, 244017, 122553, 7, 15, 132597, 4, 293, 12511, 7610, 4, 3414, 132597, 9, 4, 32361, 362, 4, 734, 28512, 32569, 18, 4, 32361, 26096, 14982, 73, 18715, 21433, 235261, 15, 492, 12427, 16, 53, 18715, 21433, 65454, 15, 23659, 563, 16, 278, 597, 2843, 595, 7931, 182396, 64186, 22, 886, 595, 132981, 53, 25540, 3449, 43982, 39901, 5951, 878, 330, 4, 27694, 80269, 312, 53, 6517, 11780, 611, 20408, 5], [2, 6, 132597, 67, 42897, 33, 592, 8, 163729, 25540, 361, 136997, 109514, 173230, 7, 501, 60, 102913, 196, 5631, 235, 63243, 473, 6, 231757, 74, 5277, 7905, 53, 3095, 37317, 22, 454, 183874, 5], [2, 268, 31298, 46530, 6, 132935, 43831, 7, 597, 32, 24, 3688, 9865, 5]],
            'attention_mask': [[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]]
        }  # noqa: E501
        # fmt: on

        self.tokenizer_integration_test_util(
            expected_encoding=expected_encoding,
            model_name="facebook/xglm-564M",
            padding=False,
        )