Google Cloud Dataflow Pipeline Description

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://github.com/api-search/google-cloud-dataflow/json-schema/google-cloud-dataflow-pipeline-schema.json",
  "title": "Google Cloud Dataflow Pipeline Description",
  "description": "A descriptive representation of a Dataflow pipeline, providing structural information about the stages and transforms in the pipeline graph.",
  "type": "object",
  "properties": {
    "originalPipelineTransform": {
      "type": "array",
      "description": "Description of each transform in the pipeline as provided by the user.",
      "items": {
        "$ref": "#/$defs/TransformSummary"
      }
    },
    "executionPipelineStage": {
      "type": "array",
      "description": "Description of each stage of execution after the pipeline has been optimized by the service.",
      "items": {
        "$ref": "#/$defs/ExecutionStageSummary"
      }
    },
    "displayData": {
      "type": "array",
      "description": "Pipeline level display data.",
      "items": {
        "$ref": "#/$defs/DisplayData"
      }
    }
  },
  "$defs": {
    "TransformSummary": {
      "type": "object",
      "description": "Description of a transform executed as part of a Dataflow job.",
      "properties": {
        "kind": {
          "type": "string",
          "description": "The type of transform.",
          "enum": [
            "UNKNOWN_KIND",
            "PAR_DO_KIND",
            "GROUP_BY_KEY_KIND",
            "FLATTEN_KIND",
            "READ_KIND",
            "WRITE_KIND",
            "CONSTANT_KIND",
            "SINGLETON_KIND",
            "SHUFFLE_KIND"
          ]
        },
        "id": {
          "type": "string",
          "description": "SDK-generated unique identifier of the transform."
        },
        "name": {
          "type": "string",
          "description": "User-provided name of the transform."
        },
        "displayData": {
          "type": "array",
          "description": "Transform-specific display data.",
          "items": {
            "$ref": "#/$defs/DisplayData"
          }
        },
        "outputCollectionName": {
          "type": "array",
          "description": "User names for the output collections of this transform.",
          "items": {
            "type": "string"
          }
        },
        "inputCollectionName": {
          "type": "array",
          "description": "User names for the input collections of this transform.",
          "items": {
            "type": "string"
          }
        }
      }
    },
    "ExecutionStageSummary": {
      "type": "object",
      "description": "Description of a stage of execution after pipeline optimization.",
      "properties": {
        "name": {
          "type": "string",
          "description": "Dataflow service generated name for this stage."
        },
        "id": {
          "type": "string",
          "description": "Dataflow service generated unique ID for this stage."
        },
        "kind": {
          "type": "string",
          "description": "The type of execution stage.",
          "enum": [
            "UNKNOWN_KIND",
            "PAR_DO_KIND",
            "GROUP_BY_KEY_KIND",
            "FLATTEN_KIND",
            "READ_KIND",
            "WRITE_KIND",
            "CONSTANT_KIND",
            "SINGLETON_KIND",
            "SHUFFLE_KIND"
          ]
        },
        "inputSource": {
          "type": "array",
          "description": "Input sources for this stage.",
          "items": {
            "$ref": "#/$defs/StageSource"
          }
        },
        "outputSource": {
          "type": "array",
          "description": "Output sources for this stage.",
          "items": {
            "$ref": "#/$defs/StageSource"
          }
        },
        "componentTransform": {
          "type": "array",
          "description": "Transforms that comprise this execution stage.",
          "items": {
            "type": "object",
            "properties": {
              "userName": {
                "type": "string",
                "description": "Human-readable name for this transform."
              },
              "name": {
                "type": "string",
                "description": "Dataflow service generated name for this transform."
              },
              "originalTransform": {
                "type": "string",
                "description": "User name for the original user transform."
              }
            }
          }
        },
        "componentSource": {
          "type": "array",
          "description": "Collections produced and consumed by component transforms.",
          "items": {
            "type": "object",
            "properties": {
              "userName": {
                "type": "string",
                "description": "Human-readable name for this source."
              },
              "name": {
                "type": "string",
                "description": "Dataflow service generated name for this source."
              },
              "originalTransformOrCollection": {
                "type": "string",
                "description": "User name for the original transform or collection."
              }
            }
          }
        },
        "prerequisiteStage": {
          "type": "array",
          "description": "Other stages that must complete before this stage can run.",
          "items": {
            "type": "string"
          }
        }
      }
    },
    "StageSource": {
      "type": "object",
      "description": "Describes a stream of data that flows in or out of a stage.",
      "properties": {
        "userName": {
          "type": "string",
          "description": "Human-readable name for this source."
        },
        "name": {
          "type": "string",
          "description": "Dataflow service generated name for this source."
        },
        "originalTransformOrCollection": {
          "type": "string",
          "description": "User name for the original transform or collection."
        },
        "sizeBytes": {
          "type": "string",
          "format": "int64",
          "description": "Size of the source in bytes, if known."
        }
      }
    },
    "DisplayData": {
      "type": "object",
      "description": "Data provided with a pipeline or transform for descriptive information.",
      "properties": {
        "key": {
          "type": "string",
          "description": "The key identifying the display data."
        },
        "namespace": {
          "type": "string",
          "description": "The namespace for the key, usually a class name."
        },
        "strValue": {
          "type": "string",
          "description": "Contains value if the data is of string type."
        },
        "int64Value": {
          "type": "string",
          "format": "int64",
          "description": "Contains value if the data is of int64 type."
        },
        "floatValue": {
          "type": "number",
          "format": "float",
          "description": "Contains value if the data is of float type."
        },
        "javaClassValue": {
          "type": "string",
          "description": "Contains value if the data is of java class type."
        },
        "timestampValue": {
          "type": "string",
          "format": "date-time",
          "description": "Contains value if the data is of timestamp type."
        },
        "durationValue": {
          "type": "string",
          "description": "Contains value if the data is of duration type."
        },
        "boolValue": {
          "type": "boolean",
          "description": "Contains value if the data is of bool type."
        },
        "shortStrValue": {
          "type": "string",
          "description": "A possible additional shorter value to display."
        },
        "url": {
          "type": "string",
          "description": "An optional full URL."
        },
        "label": {
          "type": "string",
          "description": "An optional label to display with the value."
        }
      }
    }
  }
}