arkindex-base-worker 0.4.0b1__py3-none-any.whl → 0.4.0b3__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {arkindex_base_worker-0.4.0b1.dist-info → arkindex_base_worker-0.4.0b3.dist-info}/METADATA +1 -1
- {arkindex_base_worker-0.4.0b1.dist-info → arkindex_base_worker-0.4.0b3.dist-info}/RECORD +19 -19
- arkindex_worker/image.py +2 -1
- arkindex_worker/utils.py +81 -0
- arkindex_worker/worker/__init__.py +3 -2
- arkindex_worker/worker/classification.py +31 -15
- arkindex_worker/worker/element.py +71 -10
- arkindex_worker/worker/entity.py +25 -11
- arkindex_worker/worker/metadata.py +18 -8
- arkindex_worker/worker/transcription.py +38 -17
- tests/test_elements_worker/test_classifications.py +107 -60
- tests/test_elements_worker/test_elements.py +318 -49
- tests/test_elements_worker/test_entities.py +102 -33
- tests/test_elements_worker/test_metadata.py +223 -98
- tests/test_elements_worker/test_transcriptions.py +293 -143
- tests/test_utils.py +28 -0
- {arkindex_base_worker-0.4.0b1.dist-info → arkindex_base_worker-0.4.0b3.dist-info}/LICENSE +0 -0
- {arkindex_base_worker-0.4.0b1.dist-info → arkindex_base_worker-0.4.0b3.dist-info}/WHEEL +0 -0
- {arkindex_base_worker-0.4.0b1.dist-info → arkindex_base_worker-0.4.0b3.dist-info}/top_level.txt +0 -0
|
@@ -7,6 +7,7 @@ from apistar.exceptions import ErrorResponse
|
|
|
7
7
|
from arkindex.mock import MockApiClient
|
|
8
8
|
from arkindex_worker.cache import CachedElement
|
|
9
9
|
from arkindex_worker.models import Element
|
|
10
|
+
from arkindex_worker.utils import DEFAULT_BATCH_SIZE
|
|
10
11
|
from arkindex_worker.worker import MetaType
|
|
11
12
|
|
|
12
13
|
from . import BASE_API_CALLS
|
|
@@ -228,143 +229,267 @@ def test_create_metadata_cached_element(responses, mock_elements_worker_with_cac
|
|
|
228
229
|
assert metadata_id == "12345678-1234-1234-1234-123456789123"
|
|
229
230
|
|
|
230
231
|
|
|
231
|
-
@pytest.mark.parametrize(
|
|
232
|
-
|
|
233
|
-
[
|
|
234
|
-
[{"type": MetaType.Text, "name": "fake_name", "value": "fake_value"}],
|
|
235
|
-
[
|
|
236
|
-
{
|
|
237
|
-
"type": MetaType.Text,
|
|
238
|
-
"name": "fake_name",
|
|
239
|
-
"value": "fake_value",
|
|
240
|
-
"entity_id": "fake_entity_id",
|
|
241
|
-
}
|
|
242
|
-
],
|
|
243
|
-
],
|
|
244
|
-
)
|
|
245
|
-
def test_create_metadata_bulk(responses, mock_elements_worker, metadata_list):
|
|
232
|
+
@pytest.mark.parametrize("batch_size", [DEFAULT_BATCH_SIZE, 1])
|
|
233
|
+
def test_create_metadata_bulk(batch_size, responses, mock_elements_worker):
|
|
246
234
|
element = Element({"id": "12341234-1234-1234-1234-123412341234"})
|
|
247
|
-
|
|
248
|
-
|
|
249
|
-
"
|
|
250
|
-
|
|
251
|
-
|
|
252
|
-
"
|
|
253
|
-
"
|
|
254
|
-
|
|
255
|
-
"id": "fake_metadata_id",
|
|
256
|
-
"type": metadata_list[0]["type"].value,
|
|
257
|
-
"name": metadata_list[0]["name"],
|
|
258
|
-
"value": metadata_list[0]["value"],
|
|
259
|
-
"dates": [],
|
|
260
|
-
"entity_id": metadata_list[0].get("entity_id"),
|
|
261
|
-
}
|
|
262
|
-
],
|
|
235
|
+
|
|
236
|
+
metadata_list = [
|
|
237
|
+
{"type": MetaType.Text, "name": "fake_name", "value": "fake_value"},
|
|
238
|
+
{
|
|
239
|
+
"type": MetaType.Text,
|
|
240
|
+
"name": "Year",
|
|
241
|
+
"value": "2024",
|
|
242
|
+
"entity_id": "entity_id",
|
|
263
243
|
},
|
|
264
|
-
|
|
244
|
+
]
|
|
245
|
+
if batch_size > 1:
|
|
246
|
+
responses.add(
|
|
247
|
+
responses.POST,
|
|
248
|
+
"http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/metadata/bulk/",
|
|
249
|
+
status=201,
|
|
250
|
+
json={
|
|
251
|
+
"worker_run_id": "56785678-5678-5678-5678-567856785678",
|
|
252
|
+
"metadata_list": [
|
|
253
|
+
{
|
|
254
|
+
"id": "fake_metadata_id1",
|
|
255
|
+
"type": metadata_list[0]["type"].value,
|
|
256
|
+
"name": metadata_list[0]["name"],
|
|
257
|
+
"value": metadata_list[0]["value"],
|
|
258
|
+
"dates": [],
|
|
259
|
+
"entity_id": None,
|
|
260
|
+
},
|
|
261
|
+
{
|
|
262
|
+
"id": "fake_metadata_id2",
|
|
263
|
+
"type": metadata_list[1]["type"].value,
|
|
264
|
+
"name": metadata_list[1]["name"],
|
|
265
|
+
"value": metadata_list[1]["value"],
|
|
266
|
+
"dates": [],
|
|
267
|
+
"entity_id": metadata_list[1]["entity_id"],
|
|
268
|
+
},
|
|
269
|
+
],
|
|
270
|
+
},
|
|
271
|
+
)
|
|
272
|
+
else:
|
|
273
|
+
for idx, meta in enumerate(metadata_list, start=1):
|
|
274
|
+
responses.add(
|
|
275
|
+
responses.POST,
|
|
276
|
+
"http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/metadata/bulk/",
|
|
277
|
+
status=201,
|
|
278
|
+
json={
|
|
279
|
+
"worker_run_id": "56785678-5678-5678-5678-567856785678",
|
|
280
|
+
"metadata_list": [
|
|
281
|
+
{
|
|
282
|
+
"id": f"fake_metadata_id{idx}",
|
|
283
|
+
"type": meta["type"].value,
|
|
284
|
+
"name": meta["name"],
|
|
285
|
+
"value": meta["value"],
|
|
286
|
+
"dates": [],
|
|
287
|
+
"entity_id": meta.get("entity_id"),
|
|
288
|
+
}
|
|
289
|
+
],
|
|
290
|
+
},
|
|
291
|
+
)
|
|
265
292
|
|
|
266
293
|
created_metadata_list = mock_elements_worker.create_metadata_bulk(
|
|
267
|
-
element, metadata_list
|
|
294
|
+
element, metadata_list, batch_size=batch_size
|
|
268
295
|
)
|
|
269
296
|
|
|
270
|
-
|
|
271
|
-
assert [
|
|
272
|
-
(call.request.method, call.request.url) for call in responses.calls
|
|
273
|
-
] == BASE_API_CALLS + [
|
|
297
|
+
bulk_api_calls = [
|
|
274
298
|
(
|
|
275
299
|
"POST",
|
|
276
300
|
"http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/metadata/bulk/",
|
|
277
|
-
)
|
|
278
|
-
]
|
|
279
|
-
assert json.loads(responses.calls[-1].request.body)["metadata_list"] == [
|
|
280
|
-
{
|
|
281
|
-
"type": metadata_list[0]["type"].value,
|
|
282
|
-
"name": metadata_list[0]["name"],
|
|
283
|
-
"value": metadata_list[0]["value"],
|
|
284
|
-
"entity_id": metadata_list[0].get("entity_id"),
|
|
285
|
-
}
|
|
301
|
+
)
|
|
286
302
|
]
|
|
303
|
+
if batch_size != DEFAULT_BATCH_SIZE:
|
|
304
|
+
bulk_api_calls.append(
|
|
305
|
+
(
|
|
306
|
+
"POST",
|
|
307
|
+
"http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/metadata/bulk/",
|
|
308
|
+
)
|
|
309
|
+
)
|
|
310
|
+
|
|
311
|
+
assert len(responses.calls) == len(BASE_API_CALLS) + len(bulk_api_calls)
|
|
312
|
+
assert [
|
|
313
|
+
(call.request.method, call.request.url) for call in responses.calls
|
|
314
|
+
] == BASE_API_CALLS + bulk_api_calls
|
|
315
|
+
|
|
316
|
+
first_meta = {
|
|
317
|
+
**metadata_list[0],
|
|
318
|
+
"type": metadata_list[0]["type"].value,
|
|
319
|
+
"entity_id": None,
|
|
320
|
+
}
|
|
321
|
+
second_meta = {**metadata_list[1], "type": metadata_list[1]["type"].value}
|
|
322
|
+
empty_payload = {
|
|
323
|
+
"worker_run_id": "56785678-5678-5678-5678-567856785678",
|
|
324
|
+
"metadata_list": [],
|
|
325
|
+
}
|
|
326
|
+
|
|
327
|
+
bodies = []
|
|
328
|
+
first_call_idx = None
|
|
329
|
+
if batch_size > 1:
|
|
330
|
+
first_call_idx = -1
|
|
331
|
+
bodies.append({**empty_payload, "metadata_list": [first_meta, second_meta]})
|
|
332
|
+
else:
|
|
333
|
+
first_call_idx = -2
|
|
334
|
+
bodies.append({**empty_payload, "metadata_list": [first_meta]})
|
|
335
|
+
bodies.append({**empty_payload, "metadata_list": [second_meta]})
|
|
336
|
+
|
|
337
|
+
assert [
|
|
338
|
+
json.loads(bulk_call.request.body)
|
|
339
|
+
for bulk_call in responses.calls[first_call_idx:]
|
|
340
|
+
] == bodies
|
|
341
|
+
|
|
287
342
|
assert created_metadata_list == [
|
|
288
343
|
{
|
|
289
|
-
"id": "
|
|
344
|
+
"id": "fake_metadata_id1",
|
|
290
345
|
"type": metadata_list[0]["type"].value,
|
|
291
346
|
"name": metadata_list[0]["name"],
|
|
292
347
|
"value": metadata_list[0]["value"],
|
|
293
348
|
"dates": [],
|
|
294
|
-
"entity_id":
|
|
295
|
-
}
|
|
349
|
+
"entity_id": None,
|
|
350
|
+
},
|
|
351
|
+
{
|
|
352
|
+
"id": "fake_metadata_id2",
|
|
353
|
+
"type": metadata_list[1]["type"].value,
|
|
354
|
+
"name": metadata_list[1]["name"],
|
|
355
|
+
"value": metadata_list[1]["value"],
|
|
356
|
+
"dates": [],
|
|
357
|
+
"entity_id": metadata_list[1]["entity_id"],
|
|
358
|
+
},
|
|
296
359
|
]
|
|
297
360
|
|
|
298
361
|
|
|
299
|
-
@pytest.mark.parametrize(
|
|
300
|
-
"metadata_list",
|
|
301
|
-
[
|
|
302
|
-
[{"type": MetaType.Text, "name": "fake_name", "value": "fake_value"}],
|
|
303
|
-
[
|
|
304
|
-
{
|
|
305
|
-
"type": MetaType.Text,
|
|
306
|
-
"name": "fake_name",
|
|
307
|
-
"value": "fake_value",
|
|
308
|
-
"entity_id": "fake_entity_id",
|
|
309
|
-
}
|
|
310
|
-
],
|
|
311
|
-
],
|
|
312
|
-
)
|
|
362
|
+
@pytest.mark.parametrize("batch_size", [DEFAULT_BATCH_SIZE, 1])
|
|
313
363
|
def test_create_metadata_bulk_cached_element(
|
|
314
|
-
responses, mock_elements_worker_with_cache
|
|
364
|
+
batch_size, responses, mock_elements_worker_with_cache
|
|
315
365
|
):
|
|
316
366
|
element = CachedElement.create(
|
|
317
367
|
id="12341234-1234-1234-1234-123412341234", type="thing"
|
|
318
368
|
)
|
|
319
|
-
|
|
320
|
-
|
|
321
|
-
"
|
|
322
|
-
|
|
323
|
-
|
|
324
|
-
"
|
|
325
|
-
"
|
|
326
|
-
|
|
327
|
-
"id": "fake_metadata_id",
|
|
328
|
-
"type": metadata_list[0]["type"].value,
|
|
329
|
-
"name": metadata_list[0]["name"],
|
|
330
|
-
"value": metadata_list[0]["value"],
|
|
331
|
-
"dates": [],
|
|
332
|
-
"entity_id": metadata_list[0].get("entity_id"),
|
|
333
|
-
}
|
|
334
|
-
],
|
|
369
|
+
|
|
370
|
+
metadata_list = [
|
|
371
|
+
{"type": MetaType.Text, "name": "fake_name", "value": "fake_value"},
|
|
372
|
+
{
|
|
373
|
+
"type": MetaType.Text,
|
|
374
|
+
"name": "Year",
|
|
375
|
+
"value": "2024",
|
|
376
|
+
"entity_id": "entity_id",
|
|
335
377
|
},
|
|
336
|
-
|
|
378
|
+
]
|
|
379
|
+
if batch_size > 1:
|
|
380
|
+
responses.add(
|
|
381
|
+
responses.POST,
|
|
382
|
+
"http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/metadata/bulk/",
|
|
383
|
+
status=201,
|
|
384
|
+
json={
|
|
385
|
+
"worker_run_id": "56785678-5678-5678-5678-567856785678",
|
|
386
|
+
"metadata_list": [
|
|
387
|
+
{
|
|
388
|
+
"id": "fake_metadata_id1",
|
|
389
|
+
"type": metadata_list[0]["type"].value,
|
|
390
|
+
"name": metadata_list[0]["name"],
|
|
391
|
+
"value": metadata_list[0]["value"],
|
|
392
|
+
"dates": [],
|
|
393
|
+
"entity_id": None,
|
|
394
|
+
},
|
|
395
|
+
{
|
|
396
|
+
"id": "fake_metadata_id2",
|
|
397
|
+
"type": metadata_list[1]["type"].value,
|
|
398
|
+
"name": metadata_list[1]["name"],
|
|
399
|
+
"value": metadata_list[1]["value"],
|
|
400
|
+
"dates": [],
|
|
401
|
+
"entity_id": metadata_list[1]["entity_id"],
|
|
402
|
+
},
|
|
403
|
+
],
|
|
404
|
+
},
|
|
405
|
+
)
|
|
406
|
+
else:
|
|
407
|
+
for idx, meta in enumerate(metadata_list, start=1):
|
|
408
|
+
responses.add(
|
|
409
|
+
responses.POST,
|
|
410
|
+
"http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/metadata/bulk/",
|
|
411
|
+
status=201,
|
|
412
|
+
json={
|
|
413
|
+
"worker_run_id": "56785678-5678-5678-5678-567856785678",
|
|
414
|
+
"metadata_list": [
|
|
415
|
+
{
|
|
416
|
+
"id": f"fake_metadata_id{idx}",
|
|
417
|
+
"type": meta["type"].value,
|
|
418
|
+
"name": meta["name"],
|
|
419
|
+
"value": meta["value"],
|
|
420
|
+
"dates": [],
|
|
421
|
+
"entity_id": meta.get("entity_id"),
|
|
422
|
+
}
|
|
423
|
+
],
|
|
424
|
+
},
|
|
425
|
+
)
|
|
337
426
|
|
|
338
427
|
created_metadata_list = mock_elements_worker_with_cache.create_metadata_bulk(
|
|
339
|
-
element, metadata_list
|
|
428
|
+
element, metadata_list, batch_size=batch_size
|
|
340
429
|
)
|
|
341
430
|
|
|
342
|
-
|
|
343
|
-
assert [
|
|
344
|
-
(call.request.method, call.request.url) for call in responses.calls
|
|
345
|
-
] == BASE_API_CALLS + [
|
|
431
|
+
bulk_api_calls = [
|
|
346
432
|
(
|
|
347
433
|
"POST",
|
|
348
434
|
"http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/metadata/bulk/",
|
|
349
|
-
)
|
|
350
|
-
]
|
|
351
|
-
assert json.loads(responses.calls[-1].request.body)["metadata_list"] == [
|
|
352
|
-
{
|
|
353
|
-
"type": metadata_list[0]["type"].value,
|
|
354
|
-
"name": metadata_list[0]["name"],
|
|
355
|
-
"value": metadata_list[0]["value"],
|
|
356
|
-
"entity_id": metadata_list[0].get("entity_id"),
|
|
357
|
-
}
|
|
435
|
+
)
|
|
358
436
|
]
|
|
437
|
+
if batch_size != DEFAULT_BATCH_SIZE:
|
|
438
|
+
bulk_api_calls.append(
|
|
439
|
+
(
|
|
440
|
+
"POST",
|
|
441
|
+
"http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/metadata/bulk/",
|
|
442
|
+
)
|
|
443
|
+
)
|
|
444
|
+
|
|
445
|
+
assert len(responses.calls) == len(BASE_API_CALLS) + len(bulk_api_calls)
|
|
446
|
+
assert [
|
|
447
|
+
(call.request.method, call.request.url) for call in responses.calls
|
|
448
|
+
] == BASE_API_CALLS + bulk_api_calls
|
|
449
|
+
|
|
450
|
+
first_meta = {
|
|
451
|
+
**metadata_list[0],
|
|
452
|
+
"type": metadata_list[0]["type"].value,
|
|
453
|
+
"entity_id": None,
|
|
454
|
+
}
|
|
455
|
+
second_meta = {**metadata_list[1], "type": metadata_list[1]["type"].value}
|
|
456
|
+
empty_payload = {
|
|
457
|
+
"worker_run_id": "56785678-5678-5678-5678-567856785678",
|
|
458
|
+
"metadata_list": [],
|
|
459
|
+
}
|
|
460
|
+
|
|
461
|
+
bodies = []
|
|
462
|
+
first_call_idx = None
|
|
463
|
+
if batch_size > 1:
|
|
464
|
+
first_call_idx = -1
|
|
465
|
+
bodies.append({**empty_payload, "metadata_list": [first_meta, second_meta]})
|
|
466
|
+
else:
|
|
467
|
+
first_call_idx = -2
|
|
468
|
+
bodies.append({**empty_payload, "metadata_list": [first_meta]})
|
|
469
|
+
bodies.append({**empty_payload, "metadata_list": [second_meta]})
|
|
470
|
+
|
|
471
|
+
assert [
|
|
472
|
+
json.loads(bulk_call.request.body)
|
|
473
|
+
for bulk_call in responses.calls[first_call_idx:]
|
|
474
|
+
] == bodies
|
|
475
|
+
|
|
359
476
|
assert created_metadata_list == [
|
|
360
477
|
{
|
|
361
|
-
"id": "
|
|
478
|
+
"id": "fake_metadata_id1",
|
|
362
479
|
"type": metadata_list[0]["type"].value,
|
|
363
480
|
"name": metadata_list[0]["name"],
|
|
364
481
|
"value": metadata_list[0]["value"],
|
|
365
482
|
"dates": [],
|
|
366
|
-
"entity_id":
|
|
367
|
-
}
|
|
483
|
+
"entity_id": None,
|
|
484
|
+
},
|
|
485
|
+
{
|
|
486
|
+
"id": "fake_metadata_id2",
|
|
487
|
+
"type": metadata_list[1]["type"].value,
|
|
488
|
+
"name": metadata_list[1]["name"],
|
|
489
|
+
"value": metadata_list[1]["value"],
|
|
490
|
+
"dates": [],
|
|
491
|
+
"entity_id": metadata_list[1]["entity_id"],
|
|
492
|
+
},
|
|
368
493
|
]
|
|
369
494
|
|
|
370
495
|
|