Separate tests which require external resource

[mthrok]

Tests that require dataset (either via network access or cache) as following are moved to test/test_build.py.

test_vocab_vectors_custom_cache (test.test_vocab.TestVocab)
test_vocab_extend (test.test_vocab.TestVocab)
test_vocab_download_glove_vectors (test.test_vocab.TestVocab)
test_vocab_download_fasttext_vectors (test.test_vocab.TestVocab)
test_vocab_download_custom_vectors (test.test_vocab.TestVocab)
test_vocab_download_charngram_vectors (test.test_vocab.TestVocab)
test_vectors_get_vecs (test.test_vocab.TestVocab)
test_get_tokenizer_spacy (test.data.test_utils.TestUtils)
test_json_dataset_one_key_multiple_fields (test.data.test_dataset.TestDataset)
test_csv_file_no_header_one_col_multiple_fields (test.data.test_dataset.TestDataset)
test_build_vocab (test.data.test_field.TestNestedField)

In addition to that, the following test modules contain only such tests. so they can simply exclude in fbcode.

• test/test_utils.py (unchanged)
• test/data/test_subword.py (unchanged)

[zhangguanheng66]

I assume this PR consolidates the tests that require external resource into test_build.py. However, why there are some tests missing? For example, test_vocab_download_fasttext_vectors, test_vocab_extend etc?

[mthrok]

They are not missing. I omitted _vocab_ from the original test names because it was redundant. Search download_fasttext_vectors or test_extend

[zhangguanheng66]

Look good to me!

[mthrok]

Thanks!