deltacat 1.1.17__py3-none-any.whl → 1.1.19__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.
Files changed (61) hide show
  1. deltacat/__init__.py +1 -1
  2. deltacat/aws/constants.py +0 -1
  3. deltacat/compute/compactor/model/compact_partition_params.py +76 -0
  4. deltacat/compute/compactor/model/compaction_session_audit_info.py +26 -0
  5. deltacat/compute/compactor/model/delta_annotated.py +16 -9
  6. deltacat/compute/compactor_v2/constants.py +3 -0
  7. deltacat/compute/compactor_v2/private/compaction_utils.py +9 -5
  8. deltacat/compute/compactor_v2/utils/content_type_params.py +185 -34
  9. deltacat/compute/compactor_v2/utils/io.py +28 -14
  10. deltacat/compute/compactor_v2/utils/primary_key_index.py +9 -4
  11. deltacat/compute/compactor_v2/utils/task_options.py +128 -183
  12. deltacat/compute/resource_estimation/__init__.py +27 -0
  13. deltacat/compute/resource_estimation/delta.py +271 -0
  14. deltacat/compute/resource_estimation/manifest.py +394 -0
  15. deltacat/compute/resource_estimation/model.py +165 -0
  16. deltacat/compute/resource_estimation/parquet.py +108 -0
  17. deltacat/constants.py +5 -0
  18. deltacat/exceptions.py +2 -4
  19. deltacat/logs.py +8 -0
  20. deltacat/tests/compute/compact_partition_multiple_rounds_test_cases.py +77 -0
  21. deltacat/tests/compute/compact_partition_rebase_test_cases.py +308 -0
  22. deltacat/tests/compute/compact_partition_rebase_then_incremental_test_cases.py +159 -0
  23. deltacat/tests/compute/compactor_v2/test_compaction_session.py +157 -0
  24. deltacat/tests/compute/compactor_v2/utils/test_task_options.py +3 -3
  25. deltacat/tests/compute/resource_estimation/test_delta.py +605 -0
  26. deltacat/tests/compute/resource_estimation/test_manifest.py +921 -0
  27. deltacat/tests/compute/test_compact_partition_rebase.py +13 -4
  28. deltacat/tests/compute/test_util_common.py +2 -0
  29. deltacat/tests/compute/test_util_create_table_deltas_repo.py +13 -5
  30. deltacat/tests/test_logs.py +34 -0
  31. deltacat/tests/test_utils/pyarrow.py +15 -5
  32. {deltacat-1.1.17.dist-info → deltacat-1.1.19.dist-info}/METADATA +2 -2
  33. {deltacat-1.1.17.dist-info → deltacat-1.1.19.dist-info}/RECORD +38 -54
  34. deltacat/compute/metastats/meta_stats.py +0 -479
  35. deltacat/compute/metastats/model/__init__.py +0 -0
  36. deltacat/compute/metastats/model/partition_stats_dict.py +0 -34
  37. deltacat/compute/metastats/model/stats_cluster_size_estimator.py +0 -68
  38. deltacat/compute/metastats/stats.py +0 -182
  39. deltacat/compute/metastats/utils/__init__.py +0 -0
  40. deltacat/compute/metastats/utils/constants.py +0 -16
  41. deltacat/compute/metastats/utils/io.py +0 -223
  42. deltacat/compute/metastats/utils/pyarrow_memory_estimation_function.py +0 -18
  43. deltacat/compute/metastats/utils/ray_utils.py +0 -129
  44. deltacat/compute/stats/basic.py +0 -226
  45. deltacat/compute/stats/models/__init__.py +0 -0
  46. deltacat/compute/stats/models/delta_column_stats.py +0 -98
  47. deltacat/compute/stats/models/delta_stats.py +0 -233
  48. deltacat/compute/stats/models/delta_stats_cache_result.py +0 -49
  49. deltacat/compute/stats/models/manifest_entry_stats.py +0 -72
  50. deltacat/compute/stats/models/stats_result.py +0 -104
  51. deltacat/compute/stats/utils/__init__.py +0 -0
  52. deltacat/compute/stats/utils/intervals.py +0 -94
  53. deltacat/compute/stats/utils/io.py +0 -230
  54. deltacat/compute/stats/utils/manifest_stats_file.py +0 -100
  55. deltacat/tests/stats/__init__.py +0 -0
  56. deltacat/tests/stats/test_intervals.py +0 -49
  57. /deltacat/{compute/metastats → tests/compute/resource_estimation}/__init__.py +0 -0
  58. /deltacat/{compute/metastats/config → tests/compute/resource_estimation/data}/__init__.py +0 -0
  59. {deltacat-1.1.17.dist-info → deltacat-1.1.19.dist-info}/LICENSE +0 -0
  60. {deltacat-1.1.17.dist-info → deltacat-1.1.19.dist-info}/WHEEL +0 -0
  61. {deltacat-1.1.17.dist-info → deltacat-1.1.19.dist-info}/top_level.txt +0 -0
@@ -249,6 +249,7 @@ def test_compact_partition_rebase_same_source_and_destination(
249
249
  pgm = PlacementGroupManager(
250
250
  1, total_cpus, worker_instance_cpu, memory_per_bundle=4000000
251
251
  ).pgs[0]
252
+ last_stream_position_to_compact = source_partition.stream_position
252
253
  compact_partition_params = CompactPartitionParams.of(
253
254
  {
254
255
  "compaction_artifact_s3_bucket": TEST_S3_RCF_BUCKET_NAME,
@@ -258,17 +259,19 @@ def test_compact_partition_rebase_same_source_and_destination(
258
259
  "deltacat_storage_kwargs": ds_mock_kwargs,
259
260
  "destination_partition_locator": rebased_partition.locator,
260
261
  "hash_bucket_count": hash_bucket_count_param,
261
- "last_stream_position_to_compact": source_partition.stream_position,
262
+ "last_stream_position_to_compact": last_stream_position_to_compact,
262
263
  "list_deltas_kwargs": {**ds_mock_kwargs, **{"equivalent_table_types": []}},
263
264
  "object_store": RayPlasmaObjectStore(),
264
265
  "pg_config": pgm,
265
266
  "primary_keys": primary_keys,
266
267
  "read_kwargs_provider": read_kwargs_provider_param,
267
268
  "rebase_source_partition_locator": source_partition.locator,
269
+ "rebase_source_partition_high_watermark": rebased_partition.stream_position,
268
270
  "records_per_compacted_file": records_per_compacted_file_param,
269
271
  "s3_client_kwargs": {},
270
272
  "source_partition_locator": rebased_partition.locator,
271
273
  "sort_keys": sort_keys if sort_keys else None,
274
+ "drop_duplicates": drop_duplicates_param,
272
275
  }
273
276
  )
274
277
 
@@ -300,14 +303,20 @@ def test_compact_partition_rebase_same_source_and_destination(
300
303
  compacted_delta_locator: DeltaLocator = get_compacted_delta_locator_from_rcf(
301
304
  s3_resource, rcf_file_s3_uri
302
305
  )
306
+ assert (
307
+ compacted_delta_locator.stream_position == last_stream_position_to_compact
308
+ ), "Compacted delta locator must be equal to last stream position"
303
309
  tables = ds.download_delta(
304
310
  compacted_delta_locator, storage_type=StorageType.LOCAL, **ds_mock_kwargs
305
311
  )
306
312
  actual_rebase_compacted_table = pa.concat_tables(tables)
307
313
  # if no primary key is specified then sort by sort_key for consistent assertion
308
- sorting_cols: List[Any] = (
309
- [(val, "ascending") for val in primary_keys] if primary_keys else sort_keys
310
- )
314
+ sorting_cols: List[Any] = []
315
+ if primary_keys:
316
+ sorting_cols.extend([(val, "ascending") for val in primary_keys])
317
+ if sort_keys:
318
+ sorting_cols.extend(sort_keys)
319
+
311
320
  rebase_expected_compact_partition_result = (
312
321
  rebase_expected_compact_partition_result.combine_chunks().sort_by(sorting_cols)
313
322
  )
@@ -267,6 +267,8 @@ def assert_compaction_audit(
267
267
  compaction_audit.peak_memory_used_bytes_per_task,
268
268
  compaction_audit.pyarrow_version,
269
269
  compaction_audit.telemetry_time_in_seconds,
270
+ compaction_audit.observed_input_inflation,
271
+ compaction_audit.observed_input_average_record_size_bytes,
270
272
  ]
271
273
  for entry in audit_entries:
272
274
  assert entry is not None
@@ -193,8 +193,10 @@ def create_src_w_deltas_destination_rebase_w_deltas_strategy(
193
193
  ds_mock_kwargs: Optional[Dict[str, Any]],
194
194
  ) -> Tuple[Stream, Stream, Optional[Stream]]:
195
195
  import deltacat.tests.local_deltacat_storage as ds
196
- from deltacat.storage import Partition, Stream
196
+ from deltacat.storage import Delta
197
+ from deltacat.utils.common import current_time_ms
197
198
 
199
+ last_stream_position = current_time_ms()
198
200
  source_namespace, source_table_name, source_table_version = create_src_table(
199
201
  primary_keys, sort_keys, partition_keys, ds_mock_kwargs
200
202
  )
@@ -208,10 +210,12 @@ def create_src_w_deltas_destination_rebase_w_deltas_strategy(
208
210
  staged_partition: Partition = ds.stage_partition(
209
211
  source_table_stream, partition_values, **ds_mock_kwargs
210
212
  )
213
+ staged_delta: Delta = ds.stage_delta(
214
+ input_deltas, staged_partition, input_delta_type, **ds_mock_kwargs
215
+ )
216
+ staged_delta.locator.stream_position = last_stream_position
211
217
  ds.commit_delta(
212
- ds.stage_delta(
213
- input_deltas, staged_partition, input_delta_type, **ds_mock_kwargs
214
- ),
218
+ staged_delta,
215
219
  **ds_mock_kwargs,
216
220
  )
217
221
  ds.commit_partition(staged_partition, **ds_mock_kwargs)
@@ -244,8 +248,12 @@ def create_src_w_deltas_destination_rebase_w_deltas_strategy(
244
248
  staged_partition: Partition = ds.stage_partition(
245
249
  rebasing_table_stream, partition_values, **ds_mock_kwargs
246
250
  )
251
+ staged_delta: Delta = ds.stage_delta(
252
+ input_deltas, staged_partition, **ds_mock_kwargs
253
+ )
254
+ staged_delta.locator.stream_position = last_stream_position
247
255
  ds.commit_delta(
248
- ds.stage_delta(input_deltas, staged_partition, **ds_mock_kwargs),
256
+ staged_delta,
249
257
  **ds_mock_kwargs,
250
258
  )
251
259
  ds.commit_partition(staged_partition, **ds_mock_kwargs)
@@ -1,6 +1,7 @@
1
1
  import unittest
2
2
  import json
3
3
  import ray
4
+ from unittest import mock
4
5
  from logging import LogRecord
5
6
  from deltacat.logs import JsonFormatter
6
7
 
@@ -190,3 +191,36 @@ class TestJsonFormatter(unittest.TestCase):
190
191
  )
191
192
  self.assertFalse(ray.is_initialized())
192
193
  self.assertNotIn("ray_runtime_context", json.loads(result))
194
+
195
+ @mock.patch("deltacat.logs.DELTACAT_LOGGER_CONTEXT", '{"DATABASE_URL": "mytemp"}')
196
+ def test_format_with_env_context_kwargs(self):
197
+ ray.shutdown()
198
+ formatter = JsonFormatter(
199
+ {"message": "msg"}, context_kwargs={"custom_key": "custom_val"}
200
+ )
201
+
202
+ record = LogRecord(
203
+ level="INFO",
204
+ name="test",
205
+ pathname="test",
206
+ lineno=0,
207
+ message="test_message",
208
+ msg="test_message",
209
+ args=None,
210
+ exc_info=None,
211
+ )
212
+
213
+ result = formatter.format(record)
214
+
215
+ self.assertEqual(
216
+ {
217
+ "message": "test_message",
218
+ "additional_context": {
219
+ "custom_key": "custom_val",
220
+ "DATABASE_URL": "mytemp",
221
+ },
222
+ },
223
+ json.loads(result),
224
+ )
225
+ self.assertFalse(ray.is_initialized())
226
+ self.assertNotIn("ray_runtime_context", json.loads(result))
@@ -2,7 +2,7 @@ from typing import List, Optional, Union
2
2
  import pyarrow as pa
3
3
  from deltacat.storage import Delta, Partition, PartitionLocator, DeltaLocator
4
4
  import deltacat.tests.local_deltacat_storage as ds
5
- from deltacat.types.media import StorageType
5
+ from deltacat.types.media import StorageType, ContentType
6
6
 
7
7
 
8
8
  def create_delta_from_csv_file(
@@ -10,6 +10,7 @@ def create_delta_from_csv_file(
10
10
  file_paths: List[str],
11
11
  table_name: Optional[str] = None,
12
12
  table_version: int = 1,
13
+ content_type: ContentType = ContentType.PARQUET,
13
14
  *args,
14
15
  **kwargs,
15
16
  ) -> Delta:
@@ -22,7 +23,7 @@ def create_delta_from_csv_file(
22
23
  **kwargs,
23
24
  )
24
25
  committed_delta = commit_delta_to_staged_partition(
25
- staged_partition, file_paths, *args, **kwargs
26
+ staged_partition, file_paths, content_type=content_type, *args, **kwargs
26
27
  )
27
28
  return committed_delta
28
29
 
@@ -45,10 +46,18 @@ def stage_partition_from_file_paths(
45
46
 
46
47
 
47
48
  def commit_delta_to_staged_partition(
48
- staged_partition, file_paths: List[str], *args, **kwargs
49
+ staged_partition,
50
+ file_paths: List[str],
51
+ content_type: ContentType = ContentType.PARQUET,
52
+ *args,
53
+ **kwargs,
49
54
  ) -> Delta:
50
55
  committed_delta = commit_delta_to_partition(
51
- staged_partition, *args, file_paths=file_paths, **kwargs
56
+ staged_partition,
57
+ *args,
58
+ file_paths=file_paths,
59
+ content_type=content_type,
60
+ **kwargs,
52
61
  )
53
62
  ds.commit_partition(staged_partition, **kwargs)
54
63
  return committed_delta
@@ -68,6 +77,7 @@ def download_delta(delta_like: Union[Delta, DeltaLocator], *args, **kwargs) -> D
68
77
  def commit_delta_to_partition(
69
78
  partition: Union[Partition, PartitionLocator],
70
79
  file_paths: List[str],
80
+ content_type: ContentType = ContentType.PARQUET,
71
81
  *args,
72
82
  **kwargs,
73
83
  ) -> Delta:
@@ -83,6 +93,6 @@ def commit_delta_to_partition(
83
93
  tables.append(table)
84
94
 
85
95
  table = pa.concat_tables(tables)
86
- staged_delta = ds.stage_delta(table, partition, **kwargs)
96
+ staged_delta = ds.stage_delta(table, partition, content_type=content_type, **kwargs)
87
97
 
88
98
  return ds.commit_delta(staged_delta, **kwargs)
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: deltacat
3
- Version: 1.1.17
3
+ Version: 1.1.19
4
4
  Summary: A scalable, fast, ACID-compliant Data Catalog powered by Ray.
5
5
  Home-page: https://github.com/ray-project/deltacat
6
6
  Author: Ray Team
@@ -27,7 +27,7 @@ Requires-Dist: tenacity==8.1.0
27
27
  Requires-Dist: typing-extensions==4.4.0
28
28
  Requires-Dist: pymemcache==4.0.0
29
29
  Requires-Dist: redis==4.6.0
30
- Requires-Dist: getdaft==0.2.31
30
+ Requires-Dist: getdaft==0.3.4
31
31
  Requires-Dist: schedule==1.2.0
32
32
 
33
33
  # DeltaCAT
@@ -1,10 +1,10 @@
1
- deltacat/__init__.py,sha256=x9MIXX9uM_Gf7hvnQuf0dQwXgyXztDLmGSbM1F27e10,1778
2
- deltacat/constants.py,sha256=_6oRI-3yp5c8J1qKGQZrt89I9-ttT_gSSvVsJ0h8Duc,1939
3
- deltacat/exceptions.py,sha256=yWM4RXK7uRrQc1VgJv6Lv2UiNZWAx2wolLq7cBwjlkg,12770
4
- deltacat/logs.py,sha256=_UAc_6GiQR3mxccys32Cp2CZOKOVZ9L-AkNUAlzepns,9091
1
+ deltacat/__init__.py,sha256=DbcEZN-_SJRPLtnDLy3HszN0IaqTarQRzgZmMDcmqi4,1778
2
+ deltacat/constants.py,sha256=TUJLXUJ9xq1Ryil72yLkKR8EDH_Irp5wUg56QstbRNE,2181
3
+ deltacat/exceptions.py,sha256=7sjk3BuMY5Oo-6OvAfHncZx_OcvtEL47BblWr2F7waE,12740
4
+ deltacat/logs.py,sha256=EQSDin1deehzz5xlLV1_TrFJrO_IBZ9Ahp7MdL-4cK8,9363
5
5
  deltacat/aws/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
6
6
  deltacat/aws/clients.py,sha256=4eQvpkV1PzFfxog7EriuglOGGwNFHR5hbGYpjsNNPxk,6949
7
- deltacat/aws/constants.py,sha256=1HnDXrSokW-G3YA3qKEiv7fZVntDs1uSk6a7On-VG5k,1223
7
+ deltacat/aws/constants.py,sha256=hcYAUot4ahq9GXCMClQiuYCtiDs5XaOebdUoKg4V84k,1222
8
8
  deltacat/aws/s3u.py,sha256=IdT0XqDXVOkPdo5Em5u3qAkV1UXFpXaE1rTkUDKv4f4,28578
9
9
  deltacat/aws/redshift/__init__.py,sha256=7SvjG-dqox8zZUhFicTsUvpG5vXYDl_QQ3ohlHOgTKc,342
10
10
  deltacat/aws/redshift/model/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -24,11 +24,11 @@ deltacat/compute/compactor/__init__.py,sha256=ivpOPve1yKi3Vz3tVgp-eeFMNEeUSf-dlR
24
24
  deltacat/compute/compactor/compaction_session.py,sha256=YthBYNpj6qvr6SqfVfXTy5ylKFOo8zUKI3bn4tHt0e8,27766
25
25
  deltacat/compute/compactor/repartition_session.py,sha256=AAPwNZtPpC_Mtoja855_alBdXDA6efp7zcvkE-MANaQ,7254
26
26
  deltacat/compute/compactor/model/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
27
- deltacat/compute/compactor/model/compact_partition_params.py,sha256=MOgSmprPw6AN4u0CIDrFTzKwuJEE3QgTkdGcvJwdKVA,16076
28
- deltacat/compute/compactor/model/compaction_session_audit_info.py,sha256=3BipvaREfk4l0DjdkH1ryy1E46NQdyYi33DPv3O5i7U,30462
27
+ deltacat/compute/compactor/model/compact_partition_params.py,sha256=jjvpUiHfGAw-Dy7s4wyTINtruf8Nk4EPMma7Y4KMF2U,19067
28
+ deltacat/compute/compactor/model/compaction_session_audit_info.py,sha256=Jjt4YOEO8lc-kiV4fB7rOD_Xd17_BS6pRDzqbtZp0GI,31350
29
29
  deltacat/compute/compactor/model/compactor_version.py,sha256=RwRvManiCxZmzjAWzm1OPDxjB1BEHu1d0fBJyGhXKxA,87
30
30
  deltacat/compute/compactor/model/dedupe_result.py,sha256=1OCV944qJdLQ_-8scisVKl45ej1eRv9OV539QYZtQ-U,292
31
- deltacat/compute/compactor/model/delta_annotated.py,sha256=NmL1Dgso1OLenduGSmIrFf_IE5n4sbAolrch8ry5ABc,12230
31
+ deltacat/compute/compactor/model/delta_annotated.py,sha256=bCE9H5mrBoHfd1lbL6tYWC4_dbAgucAlFLjOtyPLW14,12515
32
32
  deltacat/compute/compactor/model/delta_file_envelope.py,sha256=6P-3qM4HE1dIGqbKmiyk8cyJAJD1WbwnN22_ppQocHc,3676
33
33
  deltacat/compute/compactor/model/delta_file_locator.py,sha256=AmhPGPDsmahVhp91rohJMx4ByumcIY5feqRLZTrNu4s,1905
34
34
  deltacat/compute/compactor/model/hash_bucket_result.py,sha256=71qGmaT1Mks-r3-aatjNbn2x3yWIgT8RmV0bRWe6pdA,275
@@ -51,7 +51,7 @@ deltacat/compute/compactor/utils/sort_key.py,sha256=oK6otg-CSsma6zlGPaKg-KNEvcZR
51
51
  deltacat/compute/compactor/utils/system_columns.py,sha256=CNIgAGos0xAGEpdaQIH7KfbSRrGZgjRbItXMararqXQ,9399
52
52
  deltacat/compute/compactor_v2/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
53
53
  deltacat/compute/compactor_v2/compaction_session.py,sha256=COtol2s63DRPbd-AN9KCiWr4exLX8x5Tvxea_7cOGEQ,8078
54
- deltacat/compute/compactor_v2/constants.py,sha256=4HkSebuRWlAzOnZ-_nYmMsf6d3koTwfrlBx9KxuoGe4,2417
54
+ deltacat/compute/compactor_v2/constants.py,sha256=AOvnIxQfKOnLubrUsg4g8OPLgqvOT46LE_da9_Dm2KY,2507
55
55
  deltacat/compute/compactor_v2/deletes/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
56
56
  deltacat/compute/compactor_v2/deletes/delete_file_envelope.py,sha256=AeuH9JRMwp6mvQf6P2cqL92hUEtResQq6qUTS0kIKac,3111
57
57
  deltacat/compute/compactor_v2/deletes/delete_strategy.py,sha256=SMEJOxR-5r92kvKNqtu2w6HmwtmhljcZX1wcNEuS-4w,2833
@@ -66,49 +66,31 @@ deltacat/compute/compactor_v2/model/merge_file_group.py,sha256=1o86t9lc3K6ZvtViV
66
66
  deltacat/compute/compactor_v2/model/merge_input.py,sha256=-SxTE0e67z2V7MiMEVz5aMu4E0k8h3-vqohvUUOC0do,5659
67
67
  deltacat/compute/compactor_v2/model/merge_result.py,sha256=_IZTCStpb4UKiRCJYA3g6EhAqjrw0t9vmoDAN8kIK-Y,436
68
68
  deltacat/compute/compactor_v2/private/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
69
- deltacat/compute/compactor_v2/private/compaction_utils.py,sha256=HIr3ikF4iu_ztdy3FbtOw8vUKjc_RBP93ogH8EzMV64,30294
69
+ deltacat/compute/compactor_v2/private/compaction_utils.py,sha256=jkCCK1qbA9Ia_Be92xGjO9jLhX-pfXblVwZMS5BkEHU,30415
70
70
  deltacat/compute/compactor_v2/steps/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
71
71
  deltacat/compute/compactor_v2/steps/hash_bucket.py,sha256=1R5xLUkl7GqL1nY-apAgY1czKDEHjIVYSRi9qLOMass,6726
72
72
  deltacat/compute/compactor_v2/steps/merge.py,sha256=LpktsDPfj7Of6RgUw9w1f3Y3OBkPDjvtyXjzFaIDoSo,21771
73
73
  deltacat/compute/compactor_v2/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
74
- deltacat/compute/compactor_v2/utils/content_type_params.py,sha256=rNKZisxGrLQOkwX8eHUQiFoTR1V-E66pMqWigtrs618,2156
74
+ deltacat/compute/compactor_v2/utils/content_type_params.py,sha256=Ftl8ZEroiKGmPkFY9bv0prpfrDtD-VK7vpQJbP1br70,7395
75
75
  deltacat/compute/compactor_v2/utils/dedupe.py,sha256=62tFCY2iRP7I3-45GCIYs6_SJsQl8C5lBEr8gbNfbsw,1932
76
76
  deltacat/compute/compactor_v2/utils/delta.py,sha256=I7Yvda8NVbpKXG3nM2Ku1utvR2r2OpHvUMqUL2ja3aw,3626
77
- deltacat/compute/compactor_v2/utils/io.py,sha256=5zwJEW_UHv9ttQ2exJ23ZnExwBQXn1KgN7FDx1MGYv0,5262
77
+ deltacat/compute/compactor_v2/utils/io.py,sha256=3m4dorxj-WD6Yu9_3gRE6gz3C-eNJA7nn02sHKwo-J8,6018
78
78
  deltacat/compute/compactor_v2/utils/merge.py,sha256=EV_iKhNc3WflgfLW1Q46dXUvyClx8VebWHGtninEfsI,5311
79
- deltacat/compute/compactor_v2/utils/primary_key_index.py,sha256=ghyIifjXtqXgi8lN3lfnVQ2vi8uk_ny0FE7hsQlLjRQ,11538
80
- deltacat/compute/compactor_v2/utils/task_options.py,sha256=VXvoVVUq5re8NiOoyrfz34qSRiOTB0IkxHJlMqKsBmk,14066
79
+ deltacat/compute/compactor_v2/utils/primary_key_index.py,sha256=QOMwWxGhZ7VWa3oE6InM4thR5pbjmT7ttNXvx_IiKjo,11676
80
+ deltacat/compute/compactor_v2/utils/task_options.py,sha256=W0jyWIIZ0tcSAGp8mhpnu1G8p3rmX4d3juCPpAJxnDM,12649
81
81
  deltacat/compute/merge_on_read/__init__.py,sha256=ckbgngmqPjYBYz_NySsR1vNTOb_hNpeL1sYkZKvBI9M,214
82
82
  deltacat/compute/merge_on_read/daft.py,sha256=1oC38u5ig_aTrq7EzyWBo8Ui54rb6yERYMk-vEFbpxM,1400
83
83
  deltacat/compute/merge_on_read/model/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
84
84
  deltacat/compute/merge_on_read/model/merge_on_read_params.py,sha256=Q51znagh8PtLnsY987Ulx9n20oAydfPq3Zd3Y9ocbTI,2035
85
85
  deltacat/compute/merge_on_read/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
86
86
  deltacat/compute/merge_on_read/utils/delta.py,sha256=e4BtOHa5XPpUnR4r0HqBKjXckBsTI8qBwdUWwpJfkWQ,1367
87
- deltacat/compute/metastats/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
88
- deltacat/compute/metastats/meta_stats.py,sha256=78hN3aN5wLHUFJsZXuv2JLeqA35HZ8mLUWJDMslMj5Q,18731
89
- deltacat/compute/metastats/stats.py,sha256=8iUiSXOAjqiEeNP5RIb5gvhykBgpNHD5IKkB8zsPR0E,7363
90
- deltacat/compute/metastats/config/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
91
- deltacat/compute/metastats/model/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
92
- deltacat/compute/metastats/model/partition_stats_dict.py,sha256=FbfoOxmTZfjRT7iHwc_96gHmB_r6iUvVM9BoTldD5mY,1123
93
- deltacat/compute/metastats/model/stats_cluster_size_estimator.py,sha256=AfH2rsC1DdJ2R_CwOPgjGJ04h-yWROsMfTw83GdpGXM,2849
94
- deltacat/compute/metastats/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
95
- deltacat/compute/metastats/utils/constants.py,sha256=bFUPKmR3FkjEnwpHuToQYZ9QcHqYpd4OMMSwVwnJcaA,869
96
- deltacat/compute/metastats/utils/io.py,sha256=CpTNH3BLsRaQEGTFEAmB_SjLCPH0zrlidhWCDs_wHtQ,9007
97
- deltacat/compute/metastats/utils/pyarrow_memory_estimation_function.py,sha256=-3utoiC9fP2UFiJ-u7KbESNiHCRVzh5NGtSld0xRXX0,1143
98
- deltacat/compute/metastats/utils/ray_utils.py,sha256=sEDzcA0K8DMbQ_i8axBCQiPRrySPM14piaTqzKqhkss,4516
87
+ deltacat/compute/resource_estimation/__init__.py,sha256=4bfBXcq-VAt9JCmjvj3yAmn0lEHVGdGsUCCoMGxjEqA,799
88
+ deltacat/compute/resource_estimation/delta.py,sha256=mzq_0YUGwLMYcTsYhx1QFRyZljC-JYNLjD6WCqUVuXI,9045
89
+ deltacat/compute/resource_estimation/manifest.py,sha256=gSqOyIda-pYq3vRsKFq3IiZvwhV3mMqrWPtsmUH9dD8,13035
90
+ deltacat/compute/resource_estimation/model.py,sha256=psyagFXdpLGt8DfDqy7c8DWiuXCacr0Swe5f0M7DdO4,5465
91
+ deltacat/compute/resource_estimation/parquet.py,sha256=5_apma4EKbKcm-nfV73-qN2nfnCeyhFW23ZHX3jz0Kw,3158
99
92
  deltacat/compute/stats/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
100
- deltacat/compute/stats/basic.py,sha256=m_tDdtLbsyyky-UJ0UULBZDoAAjYr02O0sSvFCKyHGk,8837
101
93
  deltacat/compute/stats/types.py,sha256=cp0lT8nITTKbnkc03OysRjXfcfXzQml9a4wqCnR6kqs,215
102
- deltacat/compute/stats/models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
103
- deltacat/compute/stats/models/delta_column_stats.py,sha256=-wXjB2c0BC1RDheumjL_j5-DfRNql4WsK9GpMFQI1cg,3300
104
- deltacat/compute/stats/models/delta_stats.py,sha256=hBith8_hbF9TVr6HocLAt6RJ_kZZKO4zrGP8VOP05vA,8556
105
- deltacat/compute/stats/models/delta_stats_cache_result.py,sha256=mbJYxpZd5jaER_BWrCD2hROFy3p1nNdBrj66nUpc6io,1624
106
- deltacat/compute/stats/models/manifest_entry_stats.py,sha256=NCDAe2nPDEI4kOkuwNkRFgGPS-rqQaQqLuaLoKk20KQ,2419
107
- deltacat/compute/stats/models/stats_result.py,sha256=XQAlmzhUqRmg4jzEMUAOqcYn1HUOBTMryBH1CCVlet8,3820
108
- deltacat/compute/stats/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
109
- deltacat/compute/stats/utils/intervals.py,sha256=9ezOzIrBGU1fWBuAn1CorJ3uX5COU7vxrfA7kI1cB7I,3094
110
- deltacat/compute/stats/utils/io.py,sha256=vCvtSu8z3rS6taXTLXQpqh_M63BFtZTSmi_859gsqQ4,9059
111
- deltacat/compute/stats/utils/manifest_stats_file.py,sha256=PtqW5Zc5e09HcfiAgvoZHVMJ2gamGdwmynMXOJNJUaY,3693
112
94
  deltacat/io/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
113
95
  deltacat/io/dataset.py,sha256=pFU5UfK-fD9C4fIeffJtrA6yVQSgAx2UPbxzQ4GMFL8,3203
114
96
  deltacat/io/file_object_store.py,sha256=HCFeXu9cWXPXVk54MHel_nw3-wIuzhMt2RI6jKzjRYM,1346
@@ -138,25 +120,25 @@ deltacat/storage/model/transform.py,sha256=t4hg1dKua8VPeMFgyllkWdzq-L5M_DRG0HD9s
138
120
  deltacat/storage/model/types.py,sha256=hj7MmjjVmKT-R9sMUulOWG-FByGZKKaYXNnOWW32mP0,1608
139
121
  deltacat/tests/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
140
122
  deltacat/tests/test_exceptions.py,sha256=V3jUQClHLD24tS18tnGvNIt0psn2WFT3Nf_CIvSqL08,3140
141
- deltacat/tests/test_logs.py,sha256=QmnArioPo0dbOUXjaG-b7HEg1GaYI8NDhinPDmzwjfQ,5889
123
+ deltacat/tests/test_logs.py,sha256=ULmb3OJ8GGEpq_LFgcil-CPjZQpO9341Ws12svoct0s,6909
142
124
  deltacat/tests/aws/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
143
125
  deltacat/tests/aws/test_clients.py,sha256=23GMWfz27WWBDXSqphG9mfputsyS7j3I5P_HRk4YoKE,3790
144
126
  deltacat/tests/aws/test_s3u.py,sha256=FsYCH8K8DsDRPOtTp-w1Nu3ATqt4p1mqDo6aVJV-SbU,7918
145
127
  deltacat/tests/catalog/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
146
128
  deltacat/tests/catalog/test_default_catalog_impl.py,sha256=2l5uwmtLlUJ9yH1LDggtj81fa-pHqbE0-VBt6G4Hyc0,3180
147
129
  deltacat/tests/compute/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
148
- deltacat/tests/compute/compact_partition_multiple_rounds_test_cases.py,sha256=vyqwuKYOb4FtRwC5r1SJf7kcZNYXoiGb-BUrBgr5_Xw,34852
149
- deltacat/tests/compute/compact_partition_rebase_test_cases.py,sha256=NfcB1aTq6HeYMFrtooIIoifzLp5U0xFTN6F7Lpk8cYQ,3143
150
- deltacat/tests/compute/compact_partition_rebase_then_incremental_test_cases.py,sha256=Kl5A7hoJ0pVOE-rZna_4XcuIjhuzQbJudvnfNYKHgGo,75436
130
+ deltacat/tests/compute/compact_partition_multiple_rounds_test_cases.py,sha256=kWyIJQMyF9oBemvgOp3ngGhMpH9zjkznV-67ewELgHQ,37719
131
+ deltacat/tests/compute/compact_partition_rebase_test_cases.py,sha256=8HVr3EIFYFqNaJoqeCuj9xIBjM4Ch2bx-mJcO4BRrLo,16839
132
+ deltacat/tests/compute/compact_partition_rebase_then_incremental_test_cases.py,sha256=l_6-pAKOsRY3NbtfHsYmEaJEkq6IJueYuLsjyJxNgz4,81564
151
133
  deltacat/tests/compute/compact_partition_test_cases.py,sha256=R9eiKvxCLqcoHjAx3iOogdnXZEO9TvLbRf0wA7bcJN4,26170
152
134
  deltacat/tests/compute/test_compact_partition_incremental.py,sha256=Z0hyQGhMZjCaOn1Vk4qUbgDiS7HDhtdNeFQyG1PJhqA,14559
153
135
  deltacat/tests/compute/test_compact_partition_multiple_rounds.py,sha256=xhKCurTA29Y78_1eksUVJ0W35zNNZYm40rMpMM9ynvM,11853
154
136
  deltacat/tests/compute/test_compact_partition_params.py,sha256=Dm5eLyHo8oGMeO3XBbpj1rZqHtPZ1hAB7z2qvzc4Lxk,8497
155
- deltacat/tests/compute/test_compact_partition_rebase.py,sha256=p97zJmEoC2t6R12luSkCKjjBl50l4UGzh-IHdiQdpCs,11445
137
+ deltacat/tests/compute/test_compact_partition_rebase.py,sha256=O_IwZ1Xeaff98V1XYOyVD8PoS_EpVXSQcHWz4In8bK4,11889
156
138
  deltacat/tests/compute/test_compact_partition_rebase_then_incremental.py,sha256=CHHfNFEJW8S1We7NE1Gg6EaoKEWnaOMRxWrLyirrahc,14643
157
- deltacat/tests/compute/test_util_common.py,sha256=oTkTuo6wscVN8hmoQASIKP_DJN-M0um_ySCOcXv9AJA,11699
139
+ deltacat/tests/compute/test_util_common.py,sha256=0mEHo38bgH64y0XZ_zgUL_aZgQMgJOSTlOYvIJxG_MM,11825
158
140
  deltacat/tests/compute/test_util_constant.py,sha256=4o-W3E7r7jhFl1A3OFLLrdKnwcF46zx4lEIDY8ONJ3c,929
159
- deltacat/tests/compute/test_util_create_table_deltas_repo.py,sha256=k9lq_3r_kNMzruTSn4JE7yjdBBUT3Lh-l8khSYdYpYs,12945
141
+ deltacat/tests/compute/test_util_create_table_deltas_repo.py,sha256=Q3HJj1fjoe2JwRUOW8KEjbTqPIIoP2o_T3ZGH6SJnCM,13244
160
142
  deltacat/tests/compute/compactor/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
161
143
  deltacat/tests/compute/compactor/steps/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
162
144
  deltacat/tests/compute/compactor/steps/test_repartition.py,sha256=0uRguPEKeLSYs746Jv8io-HZMWdyXNcOMBu8GO2mA0M,9305
@@ -164,10 +146,14 @@ deltacat/tests/compute/compactor/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JC
164
146
  deltacat/tests/compute/compactor/utils/test_io.py,sha256=st5mlU4cVU-eQl7B4mvPgNA3izuNwbVawYOp-NcoyrI,4326
165
147
  deltacat/tests/compute/compactor/utils/test_round_completion_file.py,sha256=LAQ4usiRF4oTx4cA85L0eOcBa_Z-febc-CuzUijSGrI,7439
166
148
  deltacat/tests/compute/compactor_v2/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
167
- deltacat/tests/compute/compactor_v2/test_compaction_session.py,sha256=Ln3NxUy_2oC8cTfFSNy28lIRK8iNEabtxaqWzIqzyEY,16260
149
+ deltacat/tests/compute/compactor_v2/test_compaction_session.py,sha256=0U8Hmu-qLvqXqLPBPS6qENc1ErolWAaAoUlwms2xLe8,23124
168
150
  deltacat/tests/compute/compactor_v2/test_hashlib.py,sha256=8csF2hFWtBvY2MbX3-6iphCsVXxRp0zP1NTnKhfdmkg,328
169
151
  deltacat/tests/compute/compactor_v2/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
170
- deltacat/tests/compute/compactor_v2/utils/test_task_options.py,sha256=4fc5MJTLm3hFlFHK_-5MfyfzeZtOo8D2kBqDE2b8lh4,862
152
+ deltacat/tests/compute/compactor_v2/utils/test_task_options.py,sha256=37DkR1u_XwhedV9cGed6FFuJTC0XmuiowHJIa_Op6uA,865
153
+ deltacat/tests/compute/resource_estimation/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
154
+ deltacat/tests/compute/resource_estimation/test_delta.py,sha256=fE3UlfF7Oi07SxX7zFkeUFcgzUTWp3yUyaNe2QAKIFw,22520
155
+ deltacat/tests/compute/resource_estimation/test_manifest.py,sha256=yrMvqDjolExdRf6Vtg5XaKDuaKz9ok15PCZ7_aJOYrI,32893
156
+ deltacat/tests/compute/resource_estimation/data/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
171
157
  deltacat/tests/io/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
172
158
  deltacat/tests/io/test_cloudpickle_bug_fix.py,sha256=qnYJg_S-nsLai77a4_I3Qs2Jtr_KWQJOxyl96f9PgHA,1376
173
159
  deltacat/tests/io/test_file_object_store.py,sha256=bHEJRleVHwvk-bbvAlNOFnOA_tbR8i0SxtsllMTb8w0,2559
@@ -177,11 +163,9 @@ deltacat/tests/io/test_redis_object_store.py,sha256=sZrXrYjkw8u_XrvFilhBbLc8PPnZ
177
163
  deltacat/tests/io/test_s3_object_store.py,sha256=4b7PYEfQJnYGUz6fcLFWVVyRHTlH_yd8CIaCv9l33Gg,1900
178
164
  deltacat/tests/local_deltacat_storage/__init__.py,sha256=5T9ubNIS42-BotEH0yrUiWEU92feW7lkoSA1-wMeAnQ,40104
179
165
  deltacat/tests/local_deltacat_storage/exceptions.py,sha256=oxZ0psmrEO0M6P2r8gHQ2E8E-Y8UBfUCBUIwfuHcx38,251
180
- deltacat/tests/stats/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
181
- deltacat/tests/stats/test_intervals.py,sha256=S92DgkALQ1WmbLWcxtvS7RlVGvL-XoPJKUUbkdn9_CQ,1955
182
166
  deltacat/tests/test_utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
183
167
  deltacat/tests/test_utils/constants.py,sha256=UYe--9T_clYjiOpv0M7TtAMGdpje_SMZ-w8n0IeCAjc,214
184
- deltacat/tests/test_utils/pyarrow.py,sha256=N8tjBaCFcOnhiu__oP2QaoJTjPK_WAOzmr2UHhcl1zo,2529
168
+ deltacat/tests/test_utils/pyarrow.py,sha256=pzTBk07xMaAfykXo3GNGwTqaQxrKnSbr-WO3HBszikI,2828
185
169
  deltacat/tests/test_utils/storage.py,sha256=93GEn4A5WbMHWk0Ec4Bd7RxeHoSEnBfSarfWhKOSNtM,972
186
170
  deltacat/tests/test_utils/utils.py,sha256=a32qEwcSSd1lvRi0aJJ4ZLnc1ZyXmoQF_K95zaQRk2M,455
187
171
  deltacat/tests/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -220,8 +204,8 @@ deltacat/utils/ray_utils/concurrency.py,sha256=JDVwMiQWrmuSlyCWAoiq9ctoJ0XADEfDD
220
204
  deltacat/utils/ray_utils/dataset.py,sha256=waHdtH0c835a-2t51HYRHnulfC0_zBxx8mFSAPvPSPM,3274
221
205
  deltacat/utils/ray_utils/performance.py,sha256=d7JFM7vTXHzkGx9qNQcZzUWajnqINvYRwaM088_FpsE,464
222
206
  deltacat/utils/ray_utils/runtime.py,sha256=rB0A-tU9WZHz0J11LzJdANYtL397YyuemcA1l-K9dAw,5029
223
- deltacat-1.1.17.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
224
- deltacat-1.1.17.dist-info/METADATA,sha256=SxYHCGGCCaUP0Yej0eT0BMV3K8lQdOmodQj6VSnALcU,1734
225
- deltacat-1.1.17.dist-info/WHEEL,sha256=eOLhNAGa2EW3wWl_TU484h7q1UNgy0JXjjoqKoxAAQc,92
226
- deltacat-1.1.17.dist-info/top_level.txt,sha256=RWdIcid4Bv2i2ozLVh-70kJpyB61xEKXod9XXGpiono,9
227
- deltacat-1.1.17.dist-info/RECORD,,
207
+ deltacat-1.1.19.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
208
+ deltacat-1.1.19.dist-info/METADATA,sha256=hCnsFTHooSkUetzi8KtsguCKURAvrxW57FUW9xOw-cg,1733
209
+ deltacat-1.1.19.dist-info/WHEEL,sha256=eOLhNAGa2EW3wWl_TU484h7q1UNgy0JXjjoqKoxAAQc,92
210
+ deltacat-1.1.19.dist-info/top_level.txt,sha256=RWdIcid4Bv2i2ozLVh-70kJpyB61xEKXod9XXGpiono,9
211
+ deltacat-1.1.19.dist-info/RECORD,,