# Filing parsing configuration
parsing:
  html:
    parser: lxml
    remove_tags:
      - script
      - style
      - meta
      - link
    preserve_tables: true
    preserve_lists: true
    normalize_whitespace: true
    max_file_size_mb: 50

  sections:
    form_10k:
      - name: "business"
        item: "1"
        aliases: ["item 1", "item 1.", "item1"]
      - name: "risk_factors"
        item: "1A"
        aliases: ["item 1a", "item 1a.", "item1a"]
      - name: "unresolved_staff_comments"
        item: "1B"
        aliases: ["item 1b", "item 1b."]
      - name: "properties"
        item: "2"
        aliases: ["item 2", "item 2."]
      - name: "legal_proceedings"
        item: "3"
        aliases: ["item 3", "item 3."]
      - name: "mine_safety"
        item: "4"
        aliases: ["item 4", "item 4."]
      - name: "market_for_equity"
        item: "5"
        aliases: ["item 5", "item 5."]
      - name: "selected_financial_data"
        item: "6"
        aliases: ["item 6", "item 6."]
      - name: "md_and_a"
        item: "7"
        aliases: ["item 7", "item 7."]
      - name: "quantitative_disclosures"
        item: "7A"
        aliases: ["item 7a", "item 7a."]
      - name: "financial_statements"
        item: "8"
        aliases: ["item 8", "item 8."]
      - name: "changes_in_accounting"
        item: "9"
        aliases: ["item 9", "item 9."]
      - name: "controls_and_procedures"
        item: "9A"
        aliases: ["item 9a", "item 9a."]
      - name: "other_information"
        item: "9B"
        aliases: ["item 9b", "item 9b."]

    form_10q:
      - name: "financial_statements"
        item: "1"
        part: 1
      - name: "md_and_a"
        item: "2"
        part: 1
      - name: "quantitative_disclosures"
        item: "3"
        part: 1
      - name: "controls_and_procedures"
        item: "4"
        part: 1
      - name: "legal_proceedings"
        item: "1"
        part: 2
      - name: "risk_factors"
        item: "1A"
        part: 2
      - name: "unregistered_sales"
        item: "2"
        part: 2

  chunking:
    strategy: sentence_window
    chunk_size: 512
    chunk_overlap: 64
    min_chunk_size: 50
    max_chunk_size: 1024
    sentence_boundary: true

  parallel:
    max_workers: 4
    batch_size: 100