{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "$id": "https://pdf2zh/scanned/parsed_document.schema.json",
  "title": "ParsedDocument",
  "description": "Stage A output - intermediate JSON representation of a scanned PDF.",
  "type": "object",
  "required": [
    "pdf_path",
    "pages",
    "chapters",
    "glossary"
  ],
  "additionalProperties": false,
  "properties": {
    "pdf_path": {
      "type": "string",
      "minLength": 1
    },
    "pages": {
      "type": "array",
      "items": {
        "$ref": "#/$defs/PageData"
      }
    },
    "chapters": {
      "type": "array",
      "items": {
        "$ref": "#/$defs/ChapterInfo"
      }
    },
    "glossary": {
      "type": "object",
      "additionalProperties": {
        "type": "string"
      }
    }
  },
  "$defs": {
    "BboxPdf": {
      "type": "array",
      "description": "(x0, y0, x1, y1) in PDF points, top-left origin. x0<x1, y0<y1.",
      "items": {
        "type": "number"
      },
      "minItems": 4,
      "maxItems": 4
    },
    "BboxImage": {
      "type": "array",
      "description": "(x0, y0, x1, y1) in image pixels, top-left origin. x0<x1, y0<y1.",
      "items": {
        "type": "number"
      },
      "minItems": 4,
      "maxItems": 4
    },
    "ElementCategory": {
      "type": "string",
      "enum": [
        "BYPASS",
        "FLOWING_TEXT",
        "IN_PLACE",
        "TABLE",
        "EQUATION"
      ]
    },
    "CellData": {
      "type": "object",
      "required": [
        "bbox_pdf",
        "bbox_text",
        "source_text",
        "translated_text"
      ],
      "additionalProperties": false,
      "properties": {
        "bbox_pdf": {
          "$ref": "#/$defs/BboxPdf"
        },
        "bbox_text": {
          "$ref": "#/$defs/BboxPdf"
        },
        "source_text": {
          "type": "string"
        },
        "translated_text": {
          "type": "string"
        }
      }
    },
    "ElementData": {
      "type": "object",
      "required": [
        "label",
        "category",
        "bbox_pdf",
        "source_text",
        "translated_text",
        "cells"
      ],
      "additionalProperties": false,
      "properties": {
        "label": {
          "type": "string",
          "minLength": 1
        },
        "category": {
          "$ref": "#/$defs/ElementCategory"
        },
        "bbox_pdf": {
          "$ref": "#/$defs/BboxPdf"
        },
        "source_text": {
          "type": "string"
        },
        "translated_text": {
          "type": "string"
        },
        "cells": {
          "type": "array",
          "items": {
            "$ref": "#/$defs/CellData"
          }
        }
      }
    },
    "PageData": {
      "type": "object",
      "required": [
        "page_index",
        "page_width",
        "page_height",
        "elements",
        "raw_text",
        "chapter_id"
      ],
      "additionalProperties": false,
      "properties": {
        "page_index": {
          "type": "integer",
          "minimum": 0
        },
        "page_width": {
          "type": "number",
          "exclusiveMinimum": 0
        },
        "page_height": {
          "type": "number",
          "exclusiveMinimum": 0
        },
        "elements": {
          "type": "array",
          "items": {
            "$ref": "#/$defs/ElementData"
          }
        },
        "raw_text": {
          "type": "string"
        },
        "chapter_id": {
          "type": "string"
        }
      }
    },
    "ChapterInfo": {
      "type": "object",
      "required": [
        "chapter_id",
        "title",
        "start_page",
        "end_page",
        "summary",
        "glossary"
      ],
      "additionalProperties": false,
      "properties": {
        "chapter_id": {
          "type": "string",
          "minLength": 1
        },
        "title": {
          "type": "string"
        },
        "start_page": {
          "type": "integer",
          "minimum": 0
        },
        "end_page": {
          "type": "integer",
          "minimum": 0
        },
        "summary": {
          "type": "string"
        },
        "glossary": {
          "type": "object",
          "additionalProperties": {
            "type": "string"
          }
        }
      }
    }
  }
}