acryl-datahub-cloud 0.3.8rc7__py3-none-any.whl → 0.3.8rc9__py3-none-any.whl

acryl_datahub_cloud/metadata/schemas/DashboardUsageStatistics.avsc

@@ -0,0 +1,255 @@
+{
+  "type": "record",
+  "Aspect": {
+    "name": "dashboardUsageStatistics",
+    "type": "timeseries"
+  },
+  "name": "DashboardUsageStatistics",
+  "namespace": "com.linkedin.pegasus2avro.dashboard",
+  "fields": [
+    {
+      "type": "long",
+      "name": "timestampMillis",
+      "doc": "The event timestamp field as epoch at UTC in milli seconds."
+    },
+    {
+      "type": [
+        "null",
+        {
+          "type": "record",
+          "name": "TimeWindowSize",
+          "namespace": "com.linkedin.pegasus2avro.timeseries",
+          "fields": [
+            {
+              "type": {
+                "type": "enum",
+                "name": "CalendarInterval",
+                "namespace": "com.linkedin.pegasus2avro.timeseries",
+                "symbols": [
+                  "SECOND",
+                  "MINUTE",
+                  "HOUR",
+                  "DAY",
+                  "WEEK",
+                  "MONTH",
+                  "QUARTER",
+                  "YEAR"
+                ]
+              },
+              "name": "unit",
+              "doc": "Interval unit such as minute/hour/day etc."
+            },
+            {
+              "type": "int",
+              "name": "multiple",
+              "default": 1,
+              "doc": "How many units. Defaults to 1."
+            }
+          ],
+          "doc": "Defines the size of a time window."
+        }
+      ],
+      "name": "eventGranularity",
+      "default": null,
+      "doc": "Granularity of the event if applicable"
+    },
+    {
+      "type": [
+        {
+          "type": "record",
+          "name": "PartitionSpec",
+          "namespace": "com.linkedin.pegasus2avro.timeseries",
+          "fields": [
+            {
+              "TimeseriesField": {},
+              "type": "string",
+              "name": "partition",
+              "doc": "A unique id / value for the partition for which statistics were collected,\ngenerated by applying the key definition to a given row."
+            },
+            {
+              "type": [
+                "null",
+                {
+                  "type": "record",
+                  "name": "TimeWindow",
+                  "namespace": "com.linkedin.pegasus2avro.timeseries",
+                  "fields": [
+                    {
+                      "type": "long",
+                      "name": "startTimeMillis",
+                      "doc": "Start time as epoch at UTC."
+                    },
+                    {
+                      "type": "com.linkedin.pegasus2avro.timeseries.TimeWindowSize",
+                      "name": "length",
+                      "doc": "The length of the window."
+                    }
+                  ]
+                }
+              ],
+              "name": "timePartition",
+              "default": null,
+              "doc": "Time window of the partition, if we are able to extract it from the partition key."
+            },
+            {
+              "deprecated": true,
+              "type": {
+                "type": "enum",
+                "name": "PartitionType",
+                "namespace": "com.linkedin.pegasus2avro.timeseries",
+                "symbols": [
+                  "FULL_TABLE",
+                  "QUERY",
+                  "PARTITION"
+                ]
+              },
+              "name": "type",
+              "default": "PARTITION",
+              "doc": "Unused!"
+            }
+          ],
+          "doc": "A reference to a specific partition in a dataset."
+        },
+        "null"
+      ],
+      "name": "partitionSpec",
+      "default": {
+        "partition": "FULL_TABLE_SNAPSHOT",
+        "type": "FULL_TABLE",
+        "timePartition": null
+      },
+      "doc": "The optional partition specification."
+    },
+    {
+      "type": [
+        "null",
+        "string"
+      ],
+      "name": "messageId",
+      "default": null,
+      "doc": "The optional messageId, if provided serves as a custom user-defined unique identifier for an aspect value."
+    },
+    {
+      "TimeseriesField": {},
+      "type": [
+        "null",
+        "int"
+      ],
+      "name": "viewsCount",
+      "default": null,
+      "doc": "The total number of times dashboard has been viewed"
+    },
+    {
+      "TimeseriesField": {},
+      "type": [
+        "null",
+        "int"
+      ],
+      "name": "executionsCount",
+      "default": null,
+      "doc": "The total number of dashboard executions (refreshes / syncs) "
+    },
+    {
+      "TimeseriesField": {},
+      "type": [
+        "null",
+        "int"
+      ],
+      "name": "uniqueUserCount",
+      "default": null,
+      "doc": "Unique user count"
+    },
+    {
+      "TimeseriesFieldCollection": {
+        "key": "user"
+      },
+      "type": [
+        "null",
+        {
+          "type": "array",
+          "items": {
+            "type": "record",
+            "name": "DashboardUserUsageCounts",
+            "namespace": "com.linkedin.pegasus2avro.dashboard",
+            "fields": [
+              {
+                "java": {
+                  "class": "com.linkedin.pegasus2avro.common.urn.Urn"
+                },
+                "type": "string",
+                "name": "user",
+                "doc": "The unique id of the user.",
+                "Urn": "Urn"
+              },
+              {
+                "TimeseriesField": {},
+                "type": [
+                  "null",
+                  "int"
+                ],
+                "name": "viewsCount",
+                "default": null,
+                "doc": "The number of times the user has viewed the dashboard"
+              },
+              {
+                "TimeseriesField": {},
+                "type": [
+                  "null",
+                  "int"
+                ],
+                "name": "executionsCount",
+                "default": null,
+                "doc": "The number of times the user has executed (refreshed) the dashboard"
+              },
+              {
+                "TimeseriesField": {},
+                "type": [
+                  "null",
+                  "int"
+                ],
+                "name": "usageCount",
+                "default": null,
+                "doc": "Normalized numeric metric representing user's dashboard usage -- the number of times the user executed or viewed the dashboard. "
+              },
+              {
+                "TimeseriesField": {},
+                "type": [
+                  "null",
+                  "string"
+                ],
+                "name": "userEmail",
+                "default": null,
+                "doc": "If user_email is set, we attempt to resolve the user's urn upon ingest"
+              }
+            ],
+            "doc": "Records a single user's usage counts for a given resource"
+          }
+        }
+      ],
+      "name": "userCounts",
+      "default": null,
+      "doc": "Users within this bucket, with frequency counts"
+    },
+    {
+      "TimeseriesField": {},
+      "type": [
+        "null",
+        "int"
+      ],
+      "name": "favoritesCount",
+      "default": null,
+      "doc": "The total number of times that the dashboard has been favorited "
+    },
+    {
+      "TimeseriesField": {},
+      "type": [
+        "null",
+        "long"
+      ],
+      "name": "lastViewedAt",
+      "default": null,
+      "doc": "Last viewed at\n\nThis should not be set in cases where statistics are windowed. "
+    }
+  ],
+  "doc": "Experimental (Subject to breaking change) -- Stats corresponding to dashboard's usage.\n\nIf this aspect represents the latest snapshot of the statistics about a Dashboard, the eventGranularity field should be null. \nIf this aspect represents a bucketed window of usage statistics (e.g. over a day), then the eventGranularity field should be set accordingly. "
+}

acryl_datahub_cloud/metadata/schemas/DataContractKey.avsc

@@ -0,0 +1,23 @@
+{
+  "type": "record",
+  "Aspect": {
+    "name": "dataContractKey",
+    "keyForEntity": "dataContract",
+    "entityCategory": "core",
+    "entityAspects": [
+      "dataContractProperties",
+      "dataContractStatus",
+      "status"
+    ]
+  },
+  "name": "DataContractKey",
+  "namespace": "com.linkedin.pegasus2avro.metadata.key",
+  "fields": [
+    {
+      "type": "string",
+      "name": "id",
+      "doc": "Unique id for the contract"
+    }
+  ],
+  "doc": "Key for a Data Contract"
+}

acryl_datahub_cloud/metadata/schemas/DataContractProperties.avsc

@@ -0,0 +1,201 @@
+{
+  "type": "record",
+  "Aspect": {
+    "name": "dataContractProperties"
+  },
+  "name": "DataContractProperties",
+  "namespace": "com.linkedin.pegasus2avro.datacontract",
+  "fields": [
+    {
+      "Relationship": {
+        "entityTypes": [
+          "dataset"
+        ],
+        "name": "ContractFor"
+      },
+      "java": {
+        "class": "com.linkedin.pegasus2avro.common.urn.Urn"
+      },
+      "type": "string",
+      "name": "entity",
+      "doc": "The entity that this contract is associated with. Currently, we only support Dataset contracts, but\nin the future we may also support Data Product level contracts.",
+      "Urn": "Urn",
+      "entityTypes": [
+        "dataset"
+      ]
+    },
+    {
+      "type": [
+        "null",
+        {
+          "type": "array",
+          "items": {
+            "type": "record",
+            "name": "SlaContract",
+            "namespace": "com.linkedin.pegasus2avro.datacontract",
+            "fields": [
+              {
+                "Relationship": {
+                  "entityTypes": [
+                    "assertion"
+                  ],
+                  "name": "IncludesSchemaAssertion"
+                },
+                "java": {
+                  "class": "com.linkedin.pegasus2avro.common.urn.Urn"
+                },
+                "type": "string",
+                "name": "assertion",
+                "doc": "The assertion representing the SLA contract.",
+                "Urn": "Urn",
+                "entityTypes": [
+                  "assertion"
+                ]
+              }
+            ],
+            "doc": "A contract pertaining to the operational SLAs of a physical data asset"
+          }
+        }
+      ],
+      "name": "sla",
+      "default": null,
+      "doc": "An optional set of SLA contracts. If this is a dataset contract, there will only be one."
+    },
+    {
+      "Relationship": {
+        "/*/assertion": {
+          "entityTypes": [
+            "assertion"
+          ],
+          "name": "IncludesSchemaAssertion"
+        }
+      },
+      "type": [
+        "null",
+        {
+          "type": "array",
+          "items": {
+            "type": "record",
+            "name": "SchemaContract",
+            "namespace": "com.linkedin.pegasus2avro.datacontract",
+            "fields": [
+              {
+                "Relationship": {
+                  "entityTypes": [
+                    "assertion"
+                  ],
+                  "name": "IncludesSchemaAssertion"
+                },
+                "java": {
+                  "class": "com.linkedin.pegasus2avro.common.urn.Urn"
+                },
+                "type": "string",
+                "name": "assertion",
+                "doc": "The assertion representing the schema contract.",
+                "Urn": "Urn",
+                "entityTypes": [
+                  "assertion"
+                ]
+              }
+            ],
+            "doc": "Expectations for a logical schema"
+          }
+        }
+      ],
+      "name": "schema",
+      "default": null,
+      "doc": "An optional set of schema contracts. If this is a dataset contract, there will only be one."
+    },
+    {
+      "Relationship": {
+        "/*/assertion": {
+          "entityTypes": [
+            "assertion"
+          ],
+          "name": "IncludesFreshnessAssertion"
+        }
+      },
+      "type": [
+        "null",
+        {
+          "type": "array",
+          "items": {
+            "type": "record",
+            "name": "FreshnessContract",
+            "namespace": "com.linkedin.pegasus2avro.datacontract",
+            "fields": [
+              {
+                "java": {
+                  "class": "com.linkedin.pegasus2avro.common.urn.Urn"
+                },
+                "type": "string",
+                "name": "assertion",
+                "doc": "The assertion representing the SLA contract.",
+                "Urn": "Urn"
+              }
+            ],
+            "doc": "A contract pertaining to the operational SLAs of a physical data asset"
+          }
+        }
+      ],
+      "name": "freshness",
+      "default": null,
+      "doc": "An optional set of FRESHNESS contracts. If this is a dataset contract, there will only be one."
+    },
+    {
+      "Relationship": {
+        "/*/assertion": {
+          "entityTypes": [
+            "assertion"
+          ],
+          "name": "IncludesDataQualityAssertion"
+        }
+      },
+      "type": [
+        "null",
+        {
+          "type": "array",
+          "items": {
+            "type": "record",
+            "name": "DataQualityContract",
+            "namespace": "com.linkedin.pegasus2avro.datacontract",
+            "fields": [
+              {
+                "Relationship": {
+                  "entityTypes": [
+                    "assertion"
+                  ],
+                  "name": "IncludesDataQualityAssertion"
+                },
+                "java": {
+                  "class": "com.linkedin.pegasus2avro.common.urn.Urn"
+                },
+                "type": "string",
+                "name": "assertion",
+                "doc": "The assertion representing the Data Quality contract.\nE.g. a table or column-level assertion.",
+                "Urn": "Urn",
+                "entityTypes": [
+                  "assertion"
+                ]
+              }
+            ],
+            "doc": "A data quality contract pertaining to a physical data asset\nData Quality contracts are used to make assertions about data quality metrics for a physical data asset"
+          }
+        }
+      ],
+      "name": "dataQuality",
+      "default": null,
+      "doc": "An optional set of Data Quality contracts, e.g. table and column level contract constraints."
+    },
+    {
+      "type": [
+        "null",
+        "string"
+      ],
+      "name": "rawContract",
+      "default": null,
+      "doc": "YAML-formatted contract definition"
+    }
+  ],
+  "doc": "Information about a data contract"
+}

acryl_datahub_cloud/metadata/schemas/DataContractStatus.avsc

@@ -0,0 +1,44 @@
+{
+  "type": "record",
+  "Aspect": {
+    "name": "dataContractStatus"
+  },
+  "name": "DataContractStatus",
+  "namespace": "com.linkedin.pegasus2avro.datacontract",
+  "fields": [
+    {
+      "Searchable": {
+        "/*": {
+          "fieldType": "TEXT",
+          "queryByDefault": true
+        }
+      },
+      "type": {
+        "type": "map",
+        "values": "string"
+      },
+      "name": "customProperties",
+      "default": {},
+      "doc": "Custom property bag."
+    },
+    {
+      "Searchable": {},
+      "type": {
+        "type": "enum",
+        "symbolDocs": {
+          "ACTIVE": "The data contract is active.",
+          "PENDING": "The data contract is pending implementation."
+        },
+        "name": "DataContractState",
+        "namespace": "com.linkedin.pegasus2avro.datacontract",
+        "symbols": [
+          "ACTIVE",
+          "PENDING"
+        ]
+      },
+      "name": "state",
+      "doc": "The latest state of the data contract"
+    }
+  ],
+  "doc": "Information about the status of a data contract"
+}