# Paper-ingestion engines — Python dependencies (local, keyless).
#
# Every default engine is license-unrestricted: no copyleft, and no revenue or
# user threshold.
#   pypdfium2   Apache-2.0 / BSD-3-Clause (PDFium: BSD)   fast path + page probe
#   docling     MIT; models docling-layout-heron (Apache-2.0), TableFormer,
#               Granite-Docling-258M (Apache-2.0) / CodeFormulaV2
#               (CDLA-Permissive-2.0), RapidOCR (Apache-2.0)   heavy pages
#   mlx-vlm     MIT, Apple Silicon only: runs the formula model on the GPU
#               (7 s per page instead of about 4 minutes on the CPU)
#
# Install into the project environment with ./setup.sh (micromamba), or:
#   pip install -r requirements.txt
#
# Marker (engine: marker) is no longer installed by default: its model weights
# use a modified OpenRAIL-M license, free only below USD 5M in funding or
# revenue. Opt in with `pip install marker-pdf==2.0.0`.

pypdfium2>=4.30
docling==2.141.0
mlx-vlm==0.7.10; sys_platform == "darwin" and platform_machine == "arm64"
PyYAML>=6.0
