Biblicus

Make your documents usable by your assistant, then decide later how you will search and retrieve them.

Mock Layout Detector

The mock-layout-detector extractor emits deterministic layout metadata for images. It is intended for tests and documentation examples that demonstrate layout-aware OCR pipelines.

This extractor does not perform real layout detection. It simply returns predefined regions based on the layout_type configuration.

Usage

biblicus extract build \
  --corpus my-corpus \
  --stage "mock-layout-detector:layout_type=two-column" \
  --stage "ocr-tesseract:use_layout_metadata=true"

Configuration

Output Metadata

Example metadata for two-column:

{
  "layout_detector": "mock-layout-detector",
  "layout_type": "two-column",
  "regions": [
    {"id": 1, "type": "header", "bbox": [50, 50, 700, 100], "order": 1},
    {"id": 2, "type": "text", "bbox": [50, 120, 350, 800], "order": 2},
    {"id": 3, "type": "text", "bbox": [380, 120, 350, 800], "order": 3}
  ],
  "num_regions": 3
}