mkdocs.yml

site_name: GlossAPI
site_description: Academic document processing pipeline (Docling + RapidOCR + Rust)
repo_url: https://github.com/eellak/glossAPI
theme:
  name: material
  features:
    - navigation.tabs
    - navigation.sections
    - content.code.copy
nav:
  - Home: index.md
  - Onboarding:
      - Onboarding Guide: getting_started.md
      - Quickstart Recipes: quickstart.md
      - Lightweight PDF Corpus: lightweight_corpus.md
  - Architecture:
      - Architecture Overview: architecture/index.md
      - Core Design Principles: architecture/core_design_principles.md
      - Docling Throughput and Batching: architecture/docling_throughput_and_batching.md
      - Failure Recovery and Skiplist: architecture/docling_failure_recovery_and_skiplist.md
      - Greek Text Validation: architecture/greek_text_validation.md
      - Metadata, Artifacts, and Run Diagnostics: architecture/metadata_artifacts_and_run_diagnostics.md
      - Artifact Layout and Stage Handoffs: architecture/artifact_layout_and_stage_handoffs.md
      - Resumability, Recovery, and Retention: architecture/resumability_recovery_and_retention.md
  - Pipeline:
      - Pipeline Overview: pipeline.md
      - OCR & Math Enrichment: ocr_and_math_enhancement.md
      - Multi-GPU & Scaling: multi_gpu.md
      - Stage Reference:
          - Stage Overview: stages/index.md
          - Download: stages/download.md
          - Extract: stages/extract.md
          - Clean: stages/clean.md
          - OCR: stages/ocr.md
          - Section: stages/section.md
          - Annotate: stages/annotate.md
          - Export: stages/export.md
  - Configuration & Ops:
      - Configuration: configuration.md
      - AWS Job Distribution: aws_job_distribution.md
      - Troubleshooting: troubleshooting.md
  - Reference:
      - Code Map: code_map.md
      - Corpus API: api/corpus.md
      - Legacy Corpus API Notes: api_corpus_tmp.md
      - Math Enrichment Runtime: math_enrichment_runtime.md
docs_dir: docs
markdown_extensions:
  - admonition
  - toc:
      permalink: true
  - tables
plugins:
  - search