| 1 | # SPDX-License-Identifier: Apache-2.0 |
| 2 | # SPDX-FileCopyrightText: Copyright contributors to the vLLM project |
| 3 | |
| 4 | |
| 5 | import tempfile |
| 6 | from pathlib import Path |
| 7 | from unittest.mock import MagicMock, call, patch |
| 8 | |
| 9 | import pytest |
| 10 | from huggingface_hub import _CACHED_NO_EXIST |
| 11 | |
| 12 | from vllm.transformers_utils.repo_utils import ( |
| 13 | any_pattern_in_repo_files, |
| 14 | get_hf_file_to_dict, |
| 15 | is_mistral_model_repo, |
| 16 | list_filtered_repo_files, |
| 17 | with_retry, |
| 18 | ) |
| 19 | |
| 20 | |
| 21 | @pytest.mark.parametrize( |
| 22 | "allow_patterns,expected_relative_files", |
| 23 | [ |
| 24 | ( |
| 25 | ["*.json", "correct*.txt"], |
| 26 | ["json_file.json", "subfolder/correct.txt", "correct_2.txt"], |
| 27 | ), |
| 28 | ], |
| 29 | ) |
| 30 | def test_list_filtered_repo_files( |
| 31 | allow_patterns: list[str], expected_relative_files: list[str] |
| 32 | ): |
| 33 | with tempfile.TemporaryDirectory() as tmp_dir: |
| 34 | # Prep folder and files |
| 35 | path_tmp_dir = Path(tmp_dir) |
| 36 | subfolder = path_tmp_dir / "subfolder" |
| 37 | subfolder.mkdir() |
| 38 | (path_tmp_dir / "json_file.json").touch() |
| 39 | (path_tmp_dir / "correct_2.txt").touch() |
| 40 | (path_tmp_dir / "incorrect.txt").touch() |
| 41 | (path_tmp_dir / "incorrect.jpeg").touch() |
| 42 | (subfolder / "correct.txt").touch() |
| 43 | (subfolder / "incorrect_sub.txt").touch() |
| 44 | |
| 45 | def _glob_path() -> list[str]: |
| 46 | return [ |
| 47 | str(file.relative_to(path_tmp_dir)) |
| 48 | for file in path_tmp_dir.glob("**/*") |
| 49 | if file.is_file() |
| 50 | ] |
| 51 | |
| 52 | # Patch list_repo_files called by fn |
| 53 | with patch( |
| 54 | "vllm.transformers_utils.repo_utils.list_repo_files", |
| 55 | MagicMock(return_value=_glob_path()), |
| 56 | ) as mock_list_repo_files: |
| 57 | out_files = sorted( |
| 58 | list_filtered_repo_files( |
| 59 | tmp_dir, allow_patterns, "revision", "model", "token" |
| 60 | ) |
| 61 | ) |
| 62 | assert out_files == sorted(expected_relative_files) |
| 63 | assert mock_list_repo_files.call_count == 1 |
| 64 | assert mock_list_repo_files.call_args_list[0] == call( |
| 65 | repo_id=tmp_dir, |
| 66 | revision="revision", |
| 67 | repo_type="model", |
| 68 | token="token", |
| 69 | ) |
| 70 | |
| 71 | |
| 72 | @pytest.mark.parametrize( |
| 73 | ("allow_patterns", "expected_bool"), |
| 74 | [ |
| 75 | (["*.json", "correct*.txt"], True), |
| 76 | ( |
| 77 | ["*.jpeg"], |
| 78 | True, |
| 79 | ), |
| 80 | ( |
| 81 | ["not_found.jpeg"], |
| 82 | False, |
| 83 | ), |
| 84 | ], |
| 85 | ) |
| 86 | def test_one_filtered_repo_files(allow_patterns: list[str], expected_bool: bool): |
| 87 | with tempfile.TemporaryDirectory() as tmp_dir: |
| 88 | # Prep folder and files |
| 89 | path_tmp_dir = Path(tmp_dir) |
| 90 | subfolder = path_tmp_dir / "subfolder" |
| 91 | subfolder.mkdir() |
| 92 | (path_tmp_dir / "incorrect.jpeg").touch() |
| 93 | (subfolder / "correct.txt").touch() |
| 94 | |
| 95 | def _glob_path() -> list[str]: |
| 96 | return [ |
| 97 | str(file.relative_to(path_tmp_dir)) |
| 98 | for file in path_tmp_dir.glob("**/*") |
| 99 | if file.is_file() |
| 100 | ] |
| 101 | |
| 102 | # Patch list_repo_files called by fn |
| 103 | with patch( |
| 104 | "vllm.transformers_utils.repo_utils.list_repo_files", |
| 105 | MagicMock(return_value=_glob_path()), |
| 106 | ) as mock_list_repo_files: |
| 107 | assert ( |
| 108 | any_pattern_in_repo_files( |
| 109 | tmp_dir, allow_patterns, "revision", "model", "token" |
| 110 | ) |
| 111 | ) is expected_bool |
| 112 | assert mock_list_repo_files.call_count == 1 |
| 113 | assert mock_list_repo_files.call_args_list[0] == call( |
| 114 | repo_id=tmp_dir, |
| 115 | revision="revision", |
| 116 | repo_type="model", |
| 117 | token="token", |
| 118 | ) |
| 119 | |
| 120 | |
| 121 | @pytest.mark.parametrize( |
| 122 | ("cache_result", "should_download"), |
| 123 | [ |
| 124 | # HF Hub recorded a prior 404: don't re-probe the Hub. |
| 125 | (_CACHED_NO_EXIST, False), |
| 126 | # File not in cache and existence unknown: preserve download behavior. |
| 127 | (None, True), |
| 128 | ], |
| 129 | ) |
| 130 | def test_get_hf_file_to_dict_honors_no_exist_marker( |
| 131 | cache_result: object, should_download: bool |
| 132 | ): |
| 133 | with ( |
| 134 | patch( |
| 135 | "vllm.transformers_utils.repo_utils.try_to_load_from_cache", |
| 136 | MagicMock(return_value=cache_result), |
| 137 | ), |
| 138 | patch( |
| 139 | "vllm.transformers_utils.repo_utils._try_download_from_hf_hub", |
| 140 | MagicMock(return_value=None), |
| 141 | ) as mock_download, |
| 142 | ): |
| 143 | result = get_hf_file_to_dict("processor_config.json", "some/repo") |
| 144 | assert result is None |
| 145 | assert mock_download.call_count == int(should_download) |
| 146 | |
| 147 | |
| 148 | @pytest.mark.parametrize( |
| 149 | ("files", "expected_bool"), |
| 150 | [ |
| 151 | (["consolidated.safetensors", "incorrect.txt"], True), |
| 152 | (["consolidated-1.safetensors", "incorrect.txt"], True), |
| 153 | ( |
| 154 | ["consolidated-1.json"], |
| 155 | False, |
| 156 | ), |
| 157 | ], |
| 158 | ) |
| 159 | def test_is_mistral_model_repo(files: list[str], expected_bool: bool): |
| 160 | with tempfile.TemporaryDirectory() as tmp_dir: |
| 161 | # Prep folder and files |
| 162 | path_tmp_dir = Path(tmp_dir) |
| 163 | for file in files: |
| 164 | (path_tmp_dir / file).touch() |
| 165 | |
| 166 | def _glob_path() -> list[str]: |
| 167 | return [ |
| 168 | str(file.relative_to(path_tmp_dir)) |
| 169 | for file in path_tmp_dir.glob("**/*") |
| 170 | if file.is_file() |
| 171 | ] |
| 172 | |
| 173 | # Patch list_repo_files called by fn |
| 174 | with patch( |
| 175 | "vllm.transformers_utils.repo_utils.list_repo_files", |
| 176 | MagicMock(return_value=_glob_path()), |
| 177 | ) as mock_list_repo_files: |
| 178 | assert ( |
| 179 | is_mistral_model_repo(tmp_dir, "revision", "model", "token") |
| 180 | is expected_bool |
| 181 | ) |
| 182 | assert mock_list_repo_files.call_count == 1 |
| 183 | assert mock_list_repo_files.call_args_list[0] == call( |
| 184 | repo_id=tmp_dir, |
| 185 | revision="revision", |
| 186 | repo_type="model", |
| 187 | token="token", |
| 188 | ) |
| 189 | |
| 190 | |
| 191 | def test_with_retry_does_not_retry_fatal_errors(): |
| 192 | """A definitive answer must not be retried, which costs a delay per attempt.""" |
| 193 | func = MagicMock(side_effect=FileNotFoundError("no such file")) |
| 194 | |
| 195 | with pytest.raises(FileNotFoundError): |
| 196 | with_retry(func, "Error", fatal_errors=(FileNotFoundError,)) |
| 197 | |
| 198 | func.assert_called_once() |