|
1 | 1 | from __future__ import annotations |
2 | 2 |
|
| 3 | +import io |
3 | 4 | from types import SimpleNamespace |
4 | | -from unittest.mock import mock_open, patch |
| 5 | +from unittest.mock import MagicMock, mock_open, patch |
5 | 6 |
|
6 | 7 | import pytest |
7 | 8 | from hotdata.exceptions import ApiException |
@@ -194,16 +195,105 @@ def test_upload_parquet_rejects_non_parquet(): |
194 | 195 | client.upload_parquet("/tmp/data.csv") |
195 | 196 |
|
196 | 197 |
|
197 | | -def test_upload_parquet_returns_upload_id(): |
| 198 | +def _mock_open_bytes(data: bytes) -> MagicMock: |
| 199 | + """Return an open() mock whose file handle supports read(n) via BytesIO.""" |
| 200 | + bio = io.BytesIO(data) |
| 201 | + m = MagicMock() |
| 202 | + m.__enter__ = lambda s: bio |
| 203 | + m.__exit__ = MagicMock(return_value=False) |
| 204 | + return MagicMock(return_value=m) |
| 205 | + |
| 206 | + |
| 207 | +def _session(mode: str, **kw) -> SimpleNamespace: |
| 208 | + defaults = dict( |
| 209 | + upload_id="upl_sess", |
| 210 | + finalize_token="tok", |
| 211 | + headers={}, |
| 212 | + part_size=None, |
| 213 | + part_urls=None, |
| 214 | + url=None, |
| 215 | + ) |
| 216 | + return SimpleNamespace(mode=mode, **{**defaults, **kw}) |
| 217 | + |
| 218 | + |
| 219 | +def _http_resp(status: int = 200, etag: str = '"abc"') -> SimpleNamespace: |
| 220 | + return SimpleNamespace(status=status, headers={"ETag": etag}) |
| 221 | + |
| 222 | + |
| 223 | +def test_upload_parquet_multipart(): |
| 224 | + client = _client() |
| 225 | + data = b"PAR1" + b"\x00" * 6 # 10 bytes -> 2 parts of 5 |
| 226 | + session = _session("multipart", part_size=5, part_urls=["https://s/1", "https://s/2"]) |
| 227 | + finalized = SimpleNamespace(upload_id="upl_final") |
| 228 | + |
| 229 | + with ( |
| 230 | + patch("builtins.open", _mock_open_bytes(data)), |
| 231 | + patch("os.path.getsize", return_value=len(data)), |
| 232 | + patch.object(client, "uploads") as uploads, |
| 233 | + patch("hotdata_framework.client.urllib3.PoolManager") as MockPool, |
| 234 | + ): |
| 235 | + pool = MockPool.return_value |
| 236 | + pool.request.return_value = _http_resp() |
| 237 | + pool.clear.return_value = None |
| 238 | + uploads.return_value.create_upload_session_handler.return_value = session |
| 239 | + uploads.return_value.finalize_upload_handler.return_value = finalized |
| 240 | + |
| 241 | + upload_id = client.upload_parquet("/tmp/data.parquet") |
| 242 | + |
| 243 | + assert upload_id == "upl_final" |
| 244 | + assert pool.request.call_count == 2 |
| 245 | + finalize_call = uploads.return_value.finalize_upload_handler.call_args |
| 246 | + assert finalize_call.kwargs["upload_id"] == "upl_sess" |
| 247 | + assert finalize_call.kwargs["x_upload_finalize_token"] == "tok" |
| 248 | + parts = finalize_call.kwargs["finalize_upload_request"].parts |
| 249 | + assert len(parts) == 2 |
| 250 | + assert parts[0].part_number == 1 |
| 251 | + assert parts[1].part_number == 2 |
| 252 | + |
| 253 | + |
| 254 | +def test_upload_parquet_single_put(): |
198 | 255 | client = _client() |
199 | | - uploaded = SimpleNamespace(id="upl_123") |
| 256 | + data = b"PAR1tiny" |
| 257 | + session = _session("single", url="https://s/put") |
| 258 | + finalized = SimpleNamespace(upload_id="upl_single") |
| 259 | + |
| 260 | + with ( |
| 261 | + patch("builtins.open", mock_open(read_data=data)), |
| 262 | + patch("os.path.getsize", return_value=len(data)), |
| 263 | + patch.object(client, "uploads") as uploads, |
| 264 | + patch("hotdata_framework.client.urllib3.PoolManager") as MockPool, |
| 265 | + ): |
| 266 | + pool = MockPool.return_value |
| 267 | + pool.request.return_value = _http_resp() |
| 268 | + pool.clear.return_value = None |
| 269 | + uploads.return_value.create_upload_session_handler.return_value = session |
| 270 | + uploads.return_value.finalize_upload_handler.return_value = finalized |
| 271 | + |
| 272 | + upload_id = client.upload_parquet("/tmp/data.parquet") |
| 273 | + |
| 274 | + assert upload_id == "upl_single" |
| 275 | + pool.request.assert_called_once() |
| 276 | + call_args = pool.request.call_args |
| 277 | + assert call_args.args[0] == "PUT" |
| 278 | + assert call_args.args[1] == "https://s/put" |
| 279 | + |
| 280 | + |
| 281 | +def test_upload_parquet_fallback_on_501(): |
| 282 | + client = _client() |
| 283 | + uploaded = SimpleNamespace(id="upl_post") |
| 284 | + |
200 | 285 | with ( |
201 | 286 | patch("builtins.open", mock_open(read_data=b"PAR1")), |
| 287 | + patch("os.path.getsize", return_value=4), |
202 | 288 | patch.object(client, "uploads") as uploads, |
203 | 289 | ): |
| 290 | + err = ApiException(status=501) |
| 291 | + uploads.return_value.create_upload_session_handler.side_effect = err |
204 | 292 | uploads.return_value.upload_file.return_value = uploaded |
| 293 | + |
205 | 294 | upload_id = client.upload_parquet("/tmp/data.parquet") |
206 | | - assert upload_id == "upl_123" |
| 295 | + |
| 296 | + assert upload_id == "upl_post" |
207 | 297 |
|
208 | 298 |
|
209 | 299 | def test_load_managed_table_with_upload_id(): |
|
0 commit comments