{
    "$schema": "http://json-schema.org/draft-07/schema#",
    "$id": "https://raw.githubusercontent.com/earthpulse/ml-dataset/main/json-schema/schema.json#",
    "title": "ML Dataset Extension",
    "description": "STAC ML Dataset Extension for STAC Catalogs.",
    "oneOf": [
        {
            "$comment": "This is the schema for STAC Catalogs.",
            "allOf": [
                {
                    "type": "object",
                    "required": [
                        "type",
                        "ml-dataset:name",
                        "ml-dataset:tasks",
                        "ml-dataset:inputs-type",
                        "ml-dataset:annotations-type",
                        "ml-dataset:version"
                    ],
                    "properties": {
                        "type": {
                            "const": "Catalog"
                        },
                        "ml-dataset:name": {
                            "type": "string",
                            "title": "ML Dataset Name"
                        },
                        "ml-dataset:tasks": {
                            "type": "array",
                            "items": {
                                "type": "string",
                                "enum": [
                                    "image classification",
                                    "object detection",
                                    "segmentation"
                                ]
                            },
                            "title": "ML Dataset tasks"
                        },
                        "ml-dataset:type": {
                            "type": "string",
                            "enum": [
                                "training",
                                "test",
                                "validation",
                                "reference",
                                "benchmark",
                                "legacy"
                            ],
                            "title": "ML Dataset type"
                        },
                        "ml-dataset:inputs-type": {
                            "type": "array",
                            "items": {
                                "type": "string",
                                "enum": [
                                    "text",
                                    "image",
                                    "satellite imagery",
                                    "video"
                                ]
                            },
                            "title": "ML Dataset Inputs type"
                        },
                        "ml-dataset:annotations-type": {
                            "type": "string",
                            "enum": [
                                "raster",
                                "vector"
                            ],
                            "title": "ML Dataset Annotations type"
                        },
                        "ml-dataset:quality-metrics": {
                            "type": "array",
                            "items": {
                                "type": "object"
                            },
                            "additionalItems": true,
                            "title": "ML Dataset quality metrics"
                        },
                        "ml-dataset:version": {
                            "type": "string",
                            "title": "ML Dataset version"
                        },
                        "ml-dataset:splits": {
                            "type": "array",
                            "items": {
                                "type": "string"
                            },
                            "title": "ML Dataset splits"
                        },
                        "links": {
                            "type": "array",
                            "items": {
                                "$ref": "#/definitions/fields"
                            }
                        }
                    }
                },
                {
                    "$ref": "#/definitions/stac_extensions"
                }
            ]
        },
        {
            "$comment": "This is the schema for STAC Collections.",
            "allOf": [
                {
                    "type": "object",
                    "required": [
                        "type",
                        "ml-dataset:split-items"
                    ],
                    "properties": {
                        "type": {
                            "const": "Collection"
                        },
                        "ml-dataset:split-items": {
                            "type": "array",
                            "items": {
                                "type": "object"
                            },
                            "title": "ML Dataset Items that conform the splits"
                        }
                    }
                },
                {
                    "$ref": "#/definitions/stac_extensions"
                }
            ]
        },
        {
            "$comment": "This is the schema for STAC Items.",
            "allOf": [
              {
                "type": "object",
                "required": [
                  "type",
                  "properties",
                  "links",
                  "assets"
                ],
                "properties": {
                  "type": {
                    "const": "Feature"
                  },
                  "properties": {
                    "allOf": [
                      {
                        "$comment": "Require fields here for item properties.",
                        "required": [
                          " ml-dataset:split"
                        ]
                      },
                      {
                        "$ref": "#/definitions/fields"
                      }
                    ]
                  },
                  "assets": {
                    "type": "object",
                    "additionalProperties": {
                      "$ref": "#/definitions/fields"
                    }
                  }
                }
              },
              {
                "$ref": "#/definitions/stac_extensions"
              }
            ]
          }
    ],
    "definitions": {
        "stac_extensions": {
            "type": "object",
            "required": [
                "stac_extensions"
            ],
            "properties": {
                "stac_extensions": {
                    "type": "array",
                    "contains": {
                        "const": "https://raw.githubusercontent.com/earthpulse/ml-dataset/main/json-schema/schema.json"
                    }
                }
            }
        },
        "fields": {
            "$comment": "Add your new fields here. Don't require them here, do that above in the item schema.",
            "type": "object",
            "properties": {
                "ml-dataset:name": {
                    "type": "string",
                    "title": "ML Dataset Name"
                },
                "ml-dataset:tasks": {
                    "type": "array",
                    "items": {
                        "type": "string",
                        "enum": [
                            "image classification",
                            "object detection",
                            "segmentation"
                        ]
                    },
                    "title": "ML Dataset tasks"
                },
                "ml-dataset:type": {
                    "type": "string",
                    "enum": [
                        "training",
                        "test",
                        "validation",
                        "reference",
                        "benchmark",
                        "legacy"
                    ],
                    "title": "ML Dataset type"
                },
                "ml-dataset:inputs-type": {
                    "type": "array",
                    "items": {
                        "type": "string",
                        "enum": [
                            "text",
                            "image",
                            "satellite imagery",
                            "video"
                        ]
                    },
                    "title": "ML Dataset Inputs type"
                },
                "ml-dataset:annotations-type": {
                    "type": "string",
                    "enum": [
                        "raster",
                        "vector"
                    ],
                    "title": "ML Dataset Annotations type"
                },
                "ml-dataset:quality-metrics": {
                    "type": "array",
                    "items": {
                        "type": "object"
                    },
                    "additionalItems": true,
                    "title": "ML Dataset quality metrics"
                },
                "ml-dataset:version": {
                    "type": "string",
                    "title": "ML Dataset version"
                },
                "ml-dataset:splits": {
                    "type": "array",
                    "items": {
                        "type": "string"
                    },
                    "title": "ML Dataset splits"
                },
                "ml-dataset:split-items": {
                    "type": "array",
                    "items": {
                        "type": "object"
                    },
                    "title": "ML Dataset Items that conform the split"
                },
                "ml-dataset:split": {
                    "type": "string",
                    "title": "ML Dataset name of the split the Item is included"
                }
            },
            "patternProperties": {
                "^(?!file:)": {
                    "$comment": "Above, change `template` to the prefix of this extension"
                }
            },
            "additionalProperties": false
        }
    }
}