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

acryl_datahub_cloud/metadata/schemas/AssertionDryRunEvent.avsc

@@ -0,0 +1,309 @@
+{
+  "type": "record",
+  "Aspect": {
+    "name": "assertionDryRunEvent",
+    "type": "timeseries"
+  },
+  "name": "AssertionDryRunEvent",
+  "namespace": "com.linkedin.pegasus2avro.assertion",
+  "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": {},
+      "java": {
+        "class": "com.linkedin.pegasus2avro.common.urn.Urn"
+      },
+      "type": "string",
+      "name": "asserteeUrn",
+      "Urn": "Urn"
+    },
+    {
+      "TimeseriesField": {},
+      "type": {
+        "type": "enum",
+        "symbolDocs": {
+          "COMPLETE": "The Assertion Run has completed"
+        },
+        "name": "AssertionRunStatus",
+        "namespace": "com.linkedin.pegasus2avro.assertion",
+        "symbols": [
+          "COMPLETE"
+        ],
+        "doc": "The lifecycle status of an assertion run."
+      },
+      "name": "status",
+      "doc": "The status of the assertion dry run as per this timeseries event."
+    },
+    {
+      "type": [
+        "null",
+        {
+          "type": "record",
+          "name": "AssertionDryRunResult",
+          "namespace": "com.linkedin.pegasus2avro.assertion",
+          "fields": [
+            {
+              "TimeseriesField": {},
+              "type": {
+                "type": "enum",
+                "symbolDocs": {
+                  "ERROR": " The Assertion encountered an Error",
+                  "FAILURE": " The Assertion Failed",
+                  "INIT": " The Assertion has not yet been fully evaluated",
+                  "SUCCESS": " The Assertion Succeeded"
+                },
+                "name": "AssertionResultType",
+                "namespace": "com.linkedin.pegasus2avro.assertion",
+                "symbols": [
+                  "INIT",
+                  "SUCCESS",
+                  "FAILURE",
+                  "ERROR"
+                ],
+                "doc": " The final result of evaluating an assertion, e.g. SUCCESS, FAILURE, or ERROR."
+              },
+              "name": "type",
+              "doc": " The final result, e.g. either SUCCESS, FAILURE, or ERROR."
+            },
+            {
+              "type": [
+                "null",
+                "long"
+              ],
+              "name": "rowCount",
+              "default": null,
+              "doc": "Number of rows for evaluated batch"
+            },
+            {
+              "type": [
+                "null",
+                "long"
+              ],
+              "name": "missingCount",
+              "default": null,
+              "doc": "Number of rows with missing value for evaluated batch"
+            },
+            {
+              "type": [
+                "null",
+                "long"
+              ],
+              "name": "unexpectedCount",
+              "default": null,
+              "doc": "Number of rows with unexpected value for evaluated batch"
+            },
+            {
+              "type": [
+                "null",
+                {
+                  "type": "map",
+                  "values": "string"
+                }
+              ],
+              "name": "nativeResults",
+              "default": null,
+              "doc": "Other results of evaluation"
+            },
+            {
+              "type": [
+                "null",
+                {
+                  "type": "record",
+                  "name": "AssertionResultError",
+                  "namespace": "com.linkedin.pegasus2avro.assertion",
+                  "fields": [
+                    {
+                      "type": {
+                        "type": "enum",
+                        "symbolDocs": {
+                          "CUSTOM_SQL_ERROR": " Error while executing a custom SQL assertion",
+                          "FIELD_ASSERTION_ERROR": " Error while executing a field assertion",
+                          "INSUFFICIENT_DATA": " Insufficient data to evaluate the assertion",
+                          "INVALID_PARAMETERS": " Invalid parameters were detected",
+                          "INVALID_SOURCE_TYPE": " Event type not supported by the specified source",
+                          "SOURCE_CONNECTION_ERROR": " Source is unreachable",
+                          "SOURCE_QUERY_FAILED": " Source query failed to execute",
+                          "UNKNOWN_ERROR": " Unknown error",
+                          "UNSUPPORTED_PLATFORM": " Unsupported platform"
+                        },
+                        "name": "AssertionResultErrorType",
+                        "namespace": "com.linkedin.pegasus2avro.assertion",
+                        "symbols": [
+                          "SOURCE_CONNECTION_ERROR",
+                          "SOURCE_QUERY_FAILED",
+                          "INSUFFICIENT_DATA",
+                          "INVALID_PARAMETERS",
+                          "INVALID_SOURCE_TYPE",
+                          "UNSUPPORTED_PLATFORM",
+                          "CUSTOM_SQL_ERROR",
+                          "FIELD_ASSERTION_ERROR",
+                          "UNKNOWN_ERROR"
+                        ]
+                      },
+                      "name": "type",
+                      "doc": " The type of error encountered"
+                    },
+                    {
+                      "type": [
+                        "null",
+                        {
+                          "type": "map",
+                          "values": "string"
+                        }
+                      ],
+                      "name": "properties",
+                      "default": null,
+                      "doc": " Additional metadata depending on the type of error"
+                    }
+                  ],
+                  "doc": " An error encountered when evaluating an AssertionResult"
+                }
+              ],
+              "name": "error",
+              "default": null,
+              "doc": " The error object if AssertionResultType is an Error"
+            }
+          ],
+          "doc": "The result of dry running an assertion.\n\nA subset of AssertionResult without the @Searchables."
+        }
+      ],
+      "name": "result",
+      "default": null,
+      "doc": "Results of assertion, present if the status is COMPLETE"
+    },
+    {
+      "type": [
+        "null",
+        {
+          "type": "map",
+          "values": "string"
+        }
+      ],
+      "name": "runtimeContext",
+      "default": null,
+      "doc": "Runtime parameters of evaluation"
+    }
+  ],
+  "doc": "An event produced when assertion is executed in dry-run mode, e.g.\nwhen an assertion is being \"dark-tested\".\n\nThis is used primarily in Acryl Observe to test assertion inference before production release.\n\nA subset of AssertionRunEvent without the @Searchables."
+}

acryl_datahub_cloud/metadata/schemas/AssertionInferenceDetails.avsc

@@ -0,0 +1,105 @@
+{
+  "type": "record",
+  "Aspect": {
+    "name": "assertionInferenceDetails"
+  },
+  "name": "AssertionInferenceDetails",
+  "namespace": "com.linkedin.pegasus2avro.assertion",
+  "fields": [
+    {
+      "type": [
+        "null",
+        "string"
+      ],
+      "name": "modelId",
+      "default": null,
+      "doc": "The id of the assertion prediction model."
+    },
+    {
+      "type": [
+        "null",
+        "string"
+      ],
+      "name": "modelVersion",
+      "default": null,
+      "doc": "The version of the model used to make the assertion."
+    },
+    {
+      "type": [
+        "null",
+        "float"
+      ],
+      "name": "confidence",
+      "default": null,
+      "doc": "The confidence of the inference, if available."
+    },
+    {
+      "type": [
+        "null",
+        {
+          "type": "map",
+          "values": "string"
+        }
+      ],
+      "name": "parameters",
+      "default": null,
+      "doc": "The parameters used to make the inference."
+    },
+    {
+      "type": [
+        "null",
+        "long"
+      ],
+      "name": "generatedAt",
+      "default": null,
+      "doc": "The time the inferred assertion was generated at in milliseconds."
+    },
+    {
+      "type": [
+        "null",
+        {
+          "type": "record",
+          "name": "AssertionAdjustmentSettings",
+          "namespace": "com.linkedin.pegasus2avro.assertion",
+          "fields": [
+            {
+              "type": {
+                "type": "enum",
+                "name": "AdjustmentAlgorithm",
+                "namespace": "com.linkedin.pegasus2avro.assertion",
+                "symbols": [
+                  "CUSTOM"
+                ],
+                "doc": "The adjustment algorithm to use when determining the threshold for an assertion."
+              },
+              "name": "algorithm",
+              "doc": "The algorithm to use to adjust assertion values to power the sensitivity control feature\n\nNote that for algorithm \"STDDEV\" which leverages the standard deviation of trailing values,\nthe \"stdDev\" field must be provided inside of AssertionEvaluationContext.pdl by the offline pipeline."
+            },
+            {
+              "type": "string",
+              "name": "algorithmName",
+              "doc": "The name of the algorithm to use to adjust assertion values to power the sensitivity control feature"
+            },
+            {
+              "type": [
+                "null",
+                {
+                  "type": "map",
+                  "values": "string"
+                }
+              ],
+              "name": "context",
+              "default": null,
+              "doc": "The context to use to adjust assertion values"
+            }
+          ],
+          "doc": "A set of settings that can be used to adjust assertion values\nThis is mainly applied against inferred assertions"
+        }
+      ],
+      "name": "adjustmentSettings",
+      "default": null,
+      "doc": "The settings used to adjust the assertion."
+    }
+  ],
+  "doc": "The details of an assertion that was inferred."
+}