acryl-datahub-cloud 0.3.7.9rc1__py3-none-any.whl → 0.3.8__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.
Potentially problematic release.
This version of acryl-datahub-cloud might be problematic. Click here for more details.
- acryl_datahub_cloud/_codegen_config.json +1 -1
- acryl_datahub_cloud/acryl_cs_issues/source.py +0 -1
- acryl_datahub_cloud/api/__init__.py +1 -0
- acryl_datahub_cloud/api/client.py +6 -0
- acryl_datahub_cloud/api/entity_versioning.py +167 -0
- acryl_datahub_cloud/datahub_metadata_sharing/__init__.py +0 -0
- acryl_datahub_cloud/datahub_metadata_sharing/metadata_sharing_source.py +267 -0
- acryl_datahub_cloud/datahub_metadata_sharing/query.py +7 -0
- acryl_datahub_cloud/datahub_metadata_sharing/scroll_shared_entities.gql +204 -0
- acryl_datahub_cloud/datahub_metadata_sharing/share_entity.gql +9 -0
- acryl_datahub_cloud/datahub_reporting/datahub_dataset.py +0 -2
- acryl_datahub_cloud/datahub_reporting/datahub_form_reporting.py +0 -1
- acryl_datahub_cloud/datahub_reporting/extract_graph.py +0 -1
- acryl_datahub_cloud/datahub_reporting/extract_sql.py +0 -1
- acryl_datahub_cloud/datahub_usage_reporting/query_builder.py +79 -57
- acryl_datahub_cloud/datahub_usage_reporting/usage_feature_reporter.py +284 -258
- acryl_datahub_cloud/lineage_features/source.py +22 -5
- acryl_datahub_cloud/metadata/_urns/urn_defs.py +1564 -1465
- acryl_datahub_cloud/metadata/com/linkedin/pegasus2avro/common/__init__.py +6 -0
- acryl_datahub_cloud/metadata/com/linkedin/pegasus2avro/executor/__init__.py +15 -0
- acryl_datahub_cloud/metadata/com/linkedin/pegasus2avro/metadata/key/__init__.py +4 -0
- acryl_datahub_cloud/metadata/com/linkedin/pegasus2avro/ml/metadata/__init__.py +2 -0
- acryl_datahub_cloud/metadata/com/linkedin/pegasus2avro/versionset/__init__.py +17 -0
- acryl_datahub_cloud/metadata/schema.avsc +23777 -22729
- acryl_datahub_cloud/metadata/schema_classes.py +1322 -519
- acryl_datahub_cloud/metadata/schemas/AssertionAnalyticsRunEvent.avsc +1 -1
- acryl_datahub_cloud/metadata/schemas/AssertionInferenceDetails.avsc +1 -1
- acryl_datahub_cloud/metadata/schemas/AssertionInfo.avsc +1 -1
- acryl_datahub_cloud/metadata/schemas/AssertionRunEvent.avsc +1 -1
- acryl_datahub_cloud/metadata/schemas/ContainerKey.avsc +1 -0
- acryl_datahub_cloud/metadata/schemas/DataFlowKey.avsc +1 -0
- acryl_datahub_cloud/metadata/schemas/DataHubIngestionSourceInfo.avsc +6 -0
- acryl_datahub_cloud/metadata/schemas/DataHubViewInfo.avsc +2 -0
- acryl_datahub_cloud/metadata/schemas/DataJobKey.avsc +3 -1
- acryl_datahub_cloud/metadata/schemas/DataProcessInstanceKey.avsc +4 -0
- acryl_datahub_cloud/metadata/schemas/DataProcessInstanceOutput.avsc +2 -1
- acryl_datahub_cloud/metadata/schemas/DataTransformLogic.avsc +63 -0
- acryl_datahub_cloud/metadata/schemas/DatasetKey.avsc +2 -1
- acryl_datahub_cloud/metadata/schemas/Deprecation.avsc +12 -0
- acryl_datahub_cloud/metadata/schemas/DynamicFormAssignment.avsc +2 -0
- acryl_datahub_cloud/metadata/schemas/EntityTypeKey.avsc +1 -0
- acryl_datahub_cloud/metadata/schemas/ExecutionRequestInput.avsc +9 -0
- acryl_datahub_cloud/metadata/schemas/ExecutionRequestResult.avsc +14 -0
- acryl_datahub_cloud/metadata/schemas/Filter.avsc +2 -0
- acryl_datahub_cloud/metadata/schemas/MLFeatureProperties.avsc +51 -0
- acryl_datahub_cloud/metadata/schemas/MLModelDeploymentProperties.avsc +51 -0
- acryl_datahub_cloud/metadata/schemas/MLModelGroupProperties.avsc +155 -0
- acryl_datahub_cloud/metadata/schemas/MLModelKey.avsc +2 -1
- acryl_datahub_cloud/metadata/schemas/MLModelProperties.avsc +155 -47
- acryl_datahub_cloud/metadata/schemas/MLPrimaryKeyProperties.avsc +51 -0
- acryl_datahub_cloud/metadata/schemas/MLTrainingRunProperties.avsc +171 -0
- acryl_datahub_cloud/metadata/schemas/MetadataChangeEvent.avsc +178 -47
- acryl_datahub_cloud/metadata/schemas/MonitorInfo.avsc +10 -1
- acryl_datahub_cloud/metadata/schemas/PostInfo.avsc +23 -0
- acryl_datahub_cloud/metadata/schemas/RecommendationModule.avsc +2 -0
- acryl_datahub_cloud/metadata/schemas/RemoteExecutorKey.avsc +21 -0
- acryl_datahub_cloud/metadata/schemas/RemoteExecutorStatus.avsc +80 -0
- acryl_datahub_cloud/metadata/schemas/SchemaFieldKey.avsc +2 -1
- acryl_datahub_cloud/metadata/schemas/VersionProperties.avsc +216 -0
- acryl_datahub_cloud/metadata/schemas/VersionSetKey.avsc +26 -0
- acryl_datahub_cloud/metadata/schemas/VersionSetProperties.avsc +49 -0
- {acryl_datahub_cloud-0.3.7.9rc1.dist-info → acryl_datahub_cloud-0.3.8.dist-info}/METADATA +57 -47
- {acryl_datahub_cloud-0.3.7.9rc1.dist-info → acryl_datahub_cloud-0.3.8.dist-info}/RECORD +66 -49
- {acryl_datahub_cloud-0.3.7.9rc1.dist-info → acryl_datahub_cloud-0.3.8.dist-info}/WHEEL +1 -1
- {acryl_datahub_cloud-0.3.7.9rc1.dist-info → acryl_datahub_cloud-0.3.8.dist-info}/entry_points.txt +1 -0
- {acryl_datahub_cloud-0.3.7.9rc1.dist-info → acryl_datahub_cloud-0.3.8.dist-info}/top_level.txt +0 -0
|
@@ -180,7 +180,6 @@ class DataHubFormReportingData(FormData):
|
|
|
180
180
|
def form_assigned_date(
|
|
181
181
|
self, search_row: DataHubDatasetSearchRow
|
|
182
182
|
) -> Dict[str, date]:
|
|
183
|
-
|
|
184
183
|
form_assigned_dates: Dict[str, date] = {}
|
|
185
184
|
forms = self.graph.get_aspect(search_row.urn, FormsClass)
|
|
186
185
|
if not forms:
|
|
@@ -3,24 +3,31 @@ from typing import Dict
|
|
|
3
3
|
|
|
4
4
|
class QueryBuilder:
|
|
5
5
|
@staticmethod
|
|
6
|
-
def
|
|
6
|
+
def get_dataset_entities_query() -> Dict:
|
|
7
7
|
return {
|
|
8
8
|
"sort": [{"urn": {"order": "asc"}}],
|
|
9
|
+
"_source": {
|
|
10
|
+
"includes": [
|
|
11
|
+
"urn",
|
|
12
|
+
"lastModifiedAt",
|
|
13
|
+
"removed",
|
|
14
|
+
"siblings",
|
|
15
|
+
"typeNames",
|
|
16
|
+
]
|
|
17
|
+
},
|
|
9
18
|
}
|
|
10
19
|
|
|
11
20
|
@staticmethod
|
|
12
21
|
def get_query_entities_query() -> Dict:
|
|
13
22
|
return {
|
|
14
23
|
"sort": [{"urn": {"order": "asc"}}],
|
|
24
|
+
"_source": {"includes": ["urn", "lastModifiedAt", "platform", "removed"]},
|
|
15
25
|
"query": {
|
|
16
26
|
"bool": {
|
|
17
|
-
"filter":
|
|
18
|
-
"bool": {
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
]
|
|
22
|
-
}
|
|
23
|
-
}
|
|
27
|
+
"filter": [
|
|
28
|
+
{"bool": {"must_not": [{"term": {"source": "MANUAL"}}]}},
|
|
29
|
+
{"exists": {"field": "platform"}},
|
|
30
|
+
]
|
|
24
31
|
}
|
|
25
32
|
},
|
|
26
33
|
}
|
|
@@ -29,6 +36,7 @@ class QueryBuilder:
|
|
|
29
36
|
def get_upstreams_query() -> Dict:
|
|
30
37
|
return {
|
|
31
38
|
"sort": [{"destination.urn": {"order": "asc"}}],
|
|
39
|
+
"_source": {"includes": ["source.urn", "destination.urn"]},
|
|
32
40
|
"query": {
|
|
33
41
|
"bool": {
|
|
34
42
|
"must": [
|
|
@@ -43,23 +51,27 @@ class QueryBuilder:
|
|
|
43
51
|
def get_dashboard_usage_query(days: int) -> Dict:
|
|
44
52
|
return {
|
|
45
53
|
"sort": [{"urn": {"order": "asc"}}],
|
|
54
|
+
"_source": {
|
|
55
|
+
"includes": [
|
|
56
|
+
"timestampMillis",
|
|
57
|
+
"systemMetadata.lastObserved",
|
|
58
|
+
"urn",
|
|
59
|
+
"eventGranularity",
|
|
60
|
+
"viewsCount",
|
|
61
|
+
"uniqueUserCount",
|
|
62
|
+
"event.userCounts",
|
|
63
|
+
]
|
|
64
|
+
},
|
|
46
65
|
"query": {
|
|
47
66
|
"bool": {
|
|
48
|
-
"filter":
|
|
49
|
-
|
|
50
|
-
"
|
|
51
|
-
{
|
|
52
|
-
|
|
53
|
-
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
}
|
|
57
|
-
}
|
|
58
|
-
},
|
|
59
|
-
{"term": {"isExploded": False}},
|
|
60
|
-
]
|
|
61
|
-
}
|
|
62
|
-
}
|
|
67
|
+
"filter": [
|
|
68
|
+
{
|
|
69
|
+
"range": {
|
|
70
|
+
"@timestamp": {"gte": f"now-{days}d", "lt": "now/d"}
|
|
71
|
+
}
|
|
72
|
+
},
|
|
73
|
+
{"term": {"isExploded": False}},
|
|
74
|
+
]
|
|
63
75
|
}
|
|
64
76
|
},
|
|
65
77
|
}
|
|
@@ -68,24 +80,28 @@ class QueryBuilder:
|
|
|
68
80
|
def get_dataset_usage_query(days: int) -> Dict:
|
|
69
81
|
return {
|
|
70
82
|
"sort": [{"urn": {"order": "asc"}}],
|
|
83
|
+
"_source": {
|
|
84
|
+
"includes": [
|
|
85
|
+
"timestampMillis",
|
|
86
|
+
"urn",
|
|
87
|
+
"eventGranularity",
|
|
88
|
+
"totalSqlQueries",
|
|
89
|
+
"uniqueUserCount",
|
|
90
|
+
"event.userCounts",
|
|
91
|
+
"platform",
|
|
92
|
+
]
|
|
93
|
+
},
|
|
71
94
|
"query": {
|
|
72
95
|
"bool": {
|
|
73
|
-
"filter":
|
|
74
|
-
|
|
75
|
-
"
|
|
76
|
-
{
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
|
|
82
|
-
}
|
|
83
|
-
},
|
|
84
|
-
{"term": {"isExploded": False}},
|
|
85
|
-
{"range": {"totalSqlQueries": {"gt": 0}}},
|
|
86
|
-
]
|
|
87
|
-
}
|
|
88
|
-
}
|
|
96
|
+
"filter": [
|
|
97
|
+
{
|
|
98
|
+
"range": {
|
|
99
|
+
"@timestamp": {"gte": f"now-{days}d/d", "lt": "now/d"}
|
|
100
|
+
}
|
|
101
|
+
},
|
|
102
|
+
{"term": {"isExploded": False}},
|
|
103
|
+
{"range": {"totalSqlQueries": {"gt": 0}}},
|
|
104
|
+
]
|
|
89
105
|
}
|
|
90
106
|
},
|
|
91
107
|
}
|
|
@@ -94,6 +110,11 @@ class QueryBuilder:
|
|
|
94
110
|
def get_dataset_write_usage_raw_query(days: int) -> Dict:
|
|
95
111
|
return {
|
|
96
112
|
"sort": [{"urn": {"order": "asc"}}, {"@timestamp": {"order": "asc"}}],
|
|
113
|
+
"_source": {
|
|
114
|
+
"includes": [
|
|
115
|
+
"urn" # Only field needed for platform extraction via regex
|
|
116
|
+
]
|
|
117
|
+
},
|
|
97
118
|
"query": {
|
|
98
119
|
"bool": {
|
|
99
120
|
"must": [
|
|
@@ -106,9 +127,6 @@ class QueryBuilder:
|
|
|
106
127
|
]
|
|
107
128
|
}
|
|
108
129
|
},
|
|
109
|
-
"_source": {
|
|
110
|
-
"includes": ["urn", "@timestamp"],
|
|
111
|
-
},
|
|
112
130
|
}
|
|
113
131
|
|
|
114
132
|
@staticmethod
|
|
@@ -141,23 +159,27 @@ class QueryBuilder:
|
|
|
141
159
|
def get_query_usage_query(days: int) -> Dict:
|
|
142
160
|
return {
|
|
143
161
|
"sort": [{"urn": {"order": "asc"}}],
|
|
162
|
+
"_source": {
|
|
163
|
+
"includes": [
|
|
164
|
+
"timestampMillis",
|
|
165
|
+
"systemMetadata.lastObserved",
|
|
166
|
+
"urn",
|
|
167
|
+
"eventGranularity",
|
|
168
|
+
"queryCount",
|
|
169
|
+
"uniqueUserCount",
|
|
170
|
+
"event.userCounts",
|
|
171
|
+
]
|
|
172
|
+
},
|
|
144
173
|
"query": {
|
|
145
174
|
"bool": {
|
|
146
|
-
"filter":
|
|
147
|
-
|
|
148
|
-
"
|
|
149
|
-
{
|
|
150
|
-
|
|
151
|
-
|
|
152
|
-
|
|
153
|
-
|
|
154
|
-
}
|
|
155
|
-
}
|
|
156
|
-
},
|
|
157
|
-
{"term": {"isExploded": False}},
|
|
158
|
-
]
|
|
159
|
-
}
|
|
160
|
-
}
|
|
175
|
+
"filter": [
|
|
176
|
+
{
|
|
177
|
+
"range": {
|
|
178
|
+
"@timestamp": {"gte": f"now-{days}d/d", "lt": "now/d"}
|
|
179
|
+
}
|
|
180
|
+
},
|
|
181
|
+
{"term": {"isExploded": False}},
|
|
182
|
+
]
|
|
161
183
|
}
|
|
162
184
|
},
|
|
163
185
|
}
|