> ## Documentation Index
> Fetch the complete documentation index at: https://docs.trueparser.com/llms.txt
> Use this file to discover all available pages before exploring further.

# PDF

> Public result contract for PDF outputs.

# PDF

Use this contract for PDF results. It is designed for fixed-layout document workflows where teams need search, OCR-aware extraction, policy review, contract analysis, and RAG over page-faithful content.

## Supported modes

* **Basic Single Column**: Standard extraction for single-column PDF layouts.
* **Basic Multi Column**: Standard extraction for multi-column PDF layouts.
* **Advanced**: OCR for scanned PDFs and OCR-aware extraction for complex digital layouts. Advanced runs are limited to 100 pages per run in beta.

## Beta limits

* Maximum file size: 25 MB
* Advanced mode page limit: 100 pages per run

## Top-level envelope

```json theme={null}
{
  "schema_version": "1.0",
  "document": {},
  "warnings": [],
  "content": []
}
```

## Document fields

| Field                                   | Notes                         |
| --------------------------------------- | ----------------------------- |
| `source_file`                           | Source file name.             |
| `format`                                | Always `pdf`.                 |
| `format_family`                         | Always `pdf`.                 |
| `title`, `author`, `subject`, `company` | Document metadata fields.     |
| `created_at`, `modified_at`             | Timestamps when available.    |
| `page_count`                            | Page count when available.    |
| `source_mode`                           | Public extraction mode label. |
| `source_engine`                         | Engine identity.              |

## Universal content/block shape

Every public content record uses the same base shape.

| Field         | Notes                             |
| ------------- | --------------------------------- |
| `id`          | Stable record id.                 |
| `type`        | Public record type.               |
| `path`        | Public structural path.           |
| `parent_id`   | Parent record id.                 |
| `depth`       | Structural depth.                 |
| `page_number` | Page number.                      |
| `order`       | Deterministic order.              |
| `bbox`        | Required bounding box.            |
| `source_ref`  | Provenance object.                |
| `is_inferred` | Inference marker.                 |
| `chunk_hint`  | Present only when you request it. |
| `text`        | Searchable text projection.       |
| `attributes`  | PDF-specific structured data.     |

Common record types include `paragraph`, `heading`, `list`, `table`, `image`, and `header_footer`.

## Warnings

* `warnings` is always present.
* Use plain strings.
* Keep warnings human-readable.
* Use warnings for client-facing notes, not for hidden parser behavior.

## What clients can rely on

* Page order stays stable.
* Section structure stays explicit.
* `bbox` and `source_ref` stay attached to each block.
* `attributes` carries PDF-specific details.
* The public contract does not expose internal transport or worker details.

## Example

```json theme={null}
{
  "schema_version": "1.0",
  "document": {
    "source_file": "policy.pdf",
    "format": "pdf",
    "format_family": "pdf",
    "title": "Insurance Policy",
    "page_count": 2,
    "source_mode": "basic",
    "source_engine": "TrueParserPdf.Basic"
  },
  "warnings": [],
  "content": [
    {
      "id": "p1-o1-paragraph",
      "type": "heading",
      "path": ["EXECUTIVE SUMMARY"],
      "parent_id": null,
      "depth": 1,
      "page_number": 1,
      "order": 1,
      "bbox": { "x": 72, "y": 50, "width": 200, "height": 16 },
      "source_ref": {
        "page": 1,
        "block_id": "p1-o1-paragraph",
        "source_mode": "basic",
        "source_engine": "TrueParserPdf.Basic"
      },
      "is_inferred": true,
      "text": "EXECUTIVE SUMMARY",
      "attributes": {
        "level": 1
      }
    }
  ]
}
```