acryl-datahub-cloud 0.3.8rc8__py3-none-any.whl → 0.3.8rc9__py3-none-any.whl

acryl_datahub_cloud/metadata/schemas/SchemaFieldsInferredMetadata.avsc

@@ -0,0 +1,222 @@
+{
+  "type": "record",
+  "Aspect": {
+    "name": "schemaFieldsInferredMetadata"
+  },
+  "name": "SchemaFieldsInferredMetadata",
+  "namespace": "com.linkedin.pegasus2avro.inferred",
+  "fields": [
+    {
+      "type": {
+        "type": "array",
+        "items": {
+          "type": "record",
+          "name": "SchemaFieldInferredMetadata",
+          "namespace": "com.linkedin.pegasus2avro.inferred",
+          "fields": [
+            {
+              "java": {
+                "class": "com.linkedin.pegasus2avro.common.urn.Urn"
+              },
+              "type": "string",
+              "name": "fieldUrn",
+              "doc": "SchemaField Urn this metadata is associated with",
+              "Urn": "Urn"
+            },
+            {
+              "Searchable": {
+                "/*/description": {
+                  "fieldName": "inferredFieldDescriptions",
+                  "fieldType": "TEXT"
+                }
+              },
+              "type": [
+                "null",
+                {
+                  "type": "array",
+                  "items": {
+                    "type": "record",
+                    "name": "InferredDescription",
+                    "namespace": "com.linkedin.pegasus2avro.inferred",
+                    "fields": [
+                      {
+                        "type": "string",
+                        "name": "description",
+                        "doc": "Inferred description text"
+                      },
+                      {
+                        "type": {
+                          "type": "record",
+                          "name": "InferredMetadataSource",
+                          "namespace": "com.linkedin.pegasus2avro.inferred",
+                          "fields": [
+                            {
+                              "type": "string",
+                              "name": "algorithm",
+                              "doc": "Algorithm used for inference"
+                            },
+                            {
+                              "type": [
+                                "null",
+                                "float"
+                              ],
+                              "name": "score",
+                              "default": null,
+                              "doc": "Inference score - can be used to decide rank of inference\nFor same algorithm, higher score represents higher confidence in inference.\nThis score is not intended to be compared across multiple algorithms."
+                            },
+                            {
+                              "type": [
+                                "null",
+                                {
+                                  "type": "map",
+                                  "values": "string"
+                                }
+                              ],
+                              "name": "context",
+                              "default": null,
+                              "doc": "Additional details about this inference"
+                            },
+                            {
+                              "type": [
+                                "null",
+                                {
+                                  "type": "array",
+                                  "items": {
+                                    "type": "enum",
+                                    "symbolDocs": {
+                                      "DATASET_PLATFORM": "Platform of dataset",
+                                      "DATASET_SCHEMA": "Schema of dataset",
+                                      "ENTITY_DESCRIPTION": "Description of entity",
+                                      "ENTITY_NAME": "Name of entity",
+                                      "FIELD_DATA_TYPE": "Datatype of dataset field",
+                                      "FIELD_DESCRIPTION": "Description of dataset field",
+                                      "FIELD_NAME": "Name of dataset field",
+                                      "FIELD_PARENT_DATASET": "Dataset containing the field",
+                                      "LINEAGE": "Lineage of entity"
+                                    },
+                                    "name": "SimilarityFactor",
+                                    "namespace": "com.linkedin.pegasus2avro.inferred",
+                                    "symbols": [
+                                      "ENTITY_NAME",
+                                      "ENTITY_DESCRIPTION",
+                                      "LINEAGE",
+                                      "DATASET_SCHEMA",
+                                      "DATASET_PLATFORM",
+                                      "FIELD_NAME",
+                                      "FIELD_DESCRIPTION",
+                                      "FIELD_DATA_TYPE",
+                                      "FIELD_PARENT_DATASET"
+                                    ],
+                                    "doc": "Factors related to an entity"
+                                  }
+                                }
+                              ],
+                              "name": "similarityFactors",
+                              "default": null,
+                              "doc": "Primary factors contributing to this inference\nOrdering of the array indicates rank."
+                            }
+                          ],
+                          "doc": "Provenace details corresponding to an inference"
+                        },
+                        "name": "source",
+                        "doc": "Provenance details about this inference"
+                      }
+                    ],
+                    "doc": "Records inferred description and its provenance"
+                  }
+                }
+              ],
+              "name": "inferredDescriptions",
+              "default": null,
+              "doc": "Inferred descriptions for schema field"
+            },
+            {
+              "Searchable": {
+                "/*/tags/*": {
+                  "fieldName": "inferredFieldTags",
+                  "fieldType": "URN"
+                }
+              },
+              "type": [
+                "null",
+                {
+                  "type": "array",
+                  "items": {
+                    "type": "record",
+                    "name": "InferredTags",
+                    "namespace": "com.linkedin.pegasus2avro.inferred",
+                    "fields": [
+                      {
+                        "type": {
+                          "type": "array",
+                          "items": "string"
+                        },
+                        "name": "tags",
+                        "doc": "Urns of inferred Tag",
+                        "Urn": "Urn",
+                        "urn_is_array": true
+                      },
+                      {
+                        "type": "com.linkedin.pegasus2avro.inferred.InferredMetadataSource",
+                        "name": "source",
+                        "doc": "Provenance details about this inference"
+                      }
+                    ],
+                    "doc": "Records inferred tags along with their provenance\nAll tags in this record share same provenance"
+                  }
+                }
+              ],
+              "name": "inferredTags",
+              "default": null,
+              "doc": "Inferred tags for schema field"
+            },
+            {
+              "Searchable": {
+                "/*/glossaryTerms/*": {
+                  "fieldName": "inferredFieldGlossaryTerms",
+                  "fieldType": "URN"
+                }
+              },
+              "type": [
+                "null",
+                {
+                  "type": "array",
+                  "items": {
+                    "type": "record",
+                    "name": "InferredGlossaryTerms",
+                    "namespace": "com.linkedin.pegasus2avro.inferred",
+                    "fields": [
+                      {
+                        "type": {
+                          "type": "array",
+                          "items": "string"
+                        },
+                        "name": "glossaryTerms",
+                        "doc": "Urn of inferred glossary term",
+                        "Urn": "Urn",
+                        "urn_is_array": true
+                      },
+                      {
+                        "type": "com.linkedin.pegasus2avro.inferred.InferredMetadataSource",
+                        "name": "source",
+                        "doc": "Provenance details about this inference"
+                      }
+                    ],
+                    "doc": "Records inferred glossary terms along with their provenance\nAll glossary terms from this record share same provenance"
+                  }
+                }
+              ],
+              "name": "inferredGlossaryTerms",
+              "default": null,
+              "doc": "Inferred glossary terms for schema field"
+            }
+          ],
+          "doc": "Inferred Metadata for dataset schema fields"
+        }
+      },
+      "name": "schemaFieldsInferredMetadata",
+      "doc": "Inferred Metadata for dataset's schema fields"
+    }
+  ],
+  "doc": "Aspect for representing Inferred Metadata for dataset's schema fields"
+}

acryl_datahub_cloud/metadata/schemas/SchemaFieldsInferredNeighbors.avsc

@@ -0,0 +1,136 @@
+{
+  "type": "record",
+  "Aspect": {
+    "name": "schemaFieldsInferredNeighbors"
+  },
+  "name": "SchemaFieldsInferredNeighbors",
+  "namespace": "com.linkedin.pegasus2avro.inferred",
+  "fields": [
+    {
+      "type": {
+        "type": "array",
+        "items": {
+          "type": "record",
+          "name": "SchemaFieldInferredNeighbors",
+          "namespace": "com.linkedin.pegasus2avro.inferred",
+          "fields": [
+            {
+              "java": {
+                "class": "com.linkedin.pegasus2avro.common.urn.Urn"
+              },
+              "type": "string",
+              "name": "fieldUrn",
+              "doc": "SchemaField Urn",
+              "Urn": "Urn"
+            },
+            {
+              "type": {
+                "type": "array",
+                "items": {
+                  "type": "record",
+                  "name": "InferredNeighbor",
+                  "namespace": "com.linkedin.pegasus2avro.inferred",
+                  "fields": [
+                    {
+                      "java": {
+                        "class": "com.linkedin.pegasus2avro.common.urn.Urn"
+                      },
+                      "type": "string",
+                      "name": "urn",
+                      "doc": "Urn of inferred neighbor entity",
+                      "Urn": "Urn"
+                    },
+                    {
+                      "type": {
+                        "type": "record",
+                        "name": "InferredMetadataSource",
+                        "namespace": "com.linkedin.pegasus2avro.inferred",
+                        "fields": [
+                          {
+                            "type": "string",
+                            "name": "algorithm",
+                            "doc": "Algorithm used for inference"
+                          },
+                          {
+                            "type": [
+                              "null",
+                              "float"
+                            ],
+                            "name": "score",
+                            "default": null,
+                            "doc": "Inference score - can be used to decide rank of inference\nFor same algorithm, higher score represents higher confidence in inference.\nThis score is not intended to be compared across multiple algorithms."
+                          },
+                          {
+                            "type": [
+                              "null",
+                              {
+                                "type": "map",
+                                "values": "string"
+                              }
+                            ],
+                            "name": "context",
+                            "default": null,
+                            "doc": "Additional details about this inference"
+                          },
+                          {
+                            "type": [
+                              "null",
+                              {
+                                "type": "array",
+                                "items": {
+                                  "type": "enum",
+                                  "symbolDocs": {
+                                    "DATASET_PLATFORM": "Platform of dataset",
+                                    "DATASET_SCHEMA": "Schema of dataset",
+                                    "ENTITY_DESCRIPTION": "Description of entity",
+                                    "ENTITY_NAME": "Name of entity",
+                                    "FIELD_DATA_TYPE": "Datatype of dataset field",
+                                    "FIELD_DESCRIPTION": "Description of dataset field",
+                                    "FIELD_NAME": "Name of dataset field",
+                                    "FIELD_PARENT_DATASET": "Dataset containing the field",
+                                    "LINEAGE": "Lineage of entity"
+                                  },
+                                  "name": "SimilarityFactor",
+                                  "namespace": "com.linkedin.pegasus2avro.inferred",
+                                  "symbols": [
+                                    "ENTITY_NAME",
+                                    "ENTITY_DESCRIPTION",
+                                    "LINEAGE",
+                                    "DATASET_SCHEMA",
+                                    "DATASET_PLATFORM",
+                                    "FIELD_NAME",
+                                    "FIELD_DESCRIPTION",
+                                    "FIELD_DATA_TYPE",
+                                    "FIELD_PARENT_DATASET"
+                                  ],
+                                  "doc": "Factors related to an entity"
+                                }
+                              }
+                            ],
+                            "name": "similarityFactors",
+                            "default": null,
+                            "doc": "Primary factors contributing to this inference\nOrdering of the array indicates rank."
+                          }
+                        ],
+                        "doc": "Provenace details corresponding to an inference"
+                      },
+                      "name": "source",
+                      "doc": "Provenance details about this inference"
+                    }
+                  ],
+                  "doc": "Records Urn of inferred neighbor entity and its provenance"
+                }
+              },
+              "name": "inferredNeighbors",
+              "doc": "Neighbors of a schema field as determined by algorithm"
+            }
+          ],
+          "doc": "Neighbors for a particular schema field as inferred by various algorithms"
+        }
+      },
+      "name": "schemaFieldsInferredNeighbors",
+      "doc": "Neighbors of dataset's schema fields as determined by algorithm"
+    }
+  ],
+  "doc": "Aspect for representing Neighbors of dataset's schema fields as determined by algorithm"
+}