# Contract extraction profile
# Matches legal contracts and agreements with parties, effective date, term, governing law, and signatures
name: contract
description: Legal contracts and agreements with parties, effective date, term, governing law, and signatures
priority: 20

match:
  all:
    - any:
        - text_contains:
            patterns: ["AGREEMENT", "CONTRACT", "WHEREAS", "NOW THEREFORE", "In witness whereof"]
        - heading_matches:
            pattern: "^(Agreement|Contract|Memorandum of Understanding)"
    - structural:
        has_table: false
        has_form_field: false
        has_math: false
        page_count:
          min: 2
          max: 200
  none:
    - text_contains:
        patterns: ["Invoice #", "Receipt"]

extraction:
  reading_order: xy_cut
  table_detection: off
  readability_threshold: 0.5
  include_invisible: false
  include_headers_footers: false
  force_ocr: false
  min_block_chars: 0

fields:
  parties:
    type: string
    extraction:
      near: ["between", "party of the first part", "BY AND BETWEEN"]
      pick: nearest_below
      parse: string

  effective_date:
    type: date
    extraction:
      near: ["Effective Date", "Date of Agreement", "as of"]
      parse: date

  term:
    type: string
    extraction:
      near: ["Term", "Initial Term", "expires on", "shall remain in effect"]
      regex: "\\d+\\s+(years?|months?)|expires?\\s+\\d{4}"
      parse: string

  governing_law:
    type: string
    extraction:
      near: ["Governing Law", "governed by the laws of"]
      pick: nearest_right
      parse: string

  signatures:
    type: array
    extraction:
      region: bottom_quarter
      fallback: []