80dbf0f703
- Add profile source modules: apply_profile, extraction, extraction_loader, field_extractor, match_eval - Add profiles CLI subcommand (profiles_cmd.rs) - Update all 9 built-in profile YAMLs (invoice, receipt, contract, scientific_paper, slide_deck, form, bank_statement, legal_filing, book_chapter) - Add 50 invoice fixture PDFs - Add 2 receipt fixture PDFs Part of: pdftract-3a310 (Phase 7.10 coordinator)
62 lines
1.5 KiB
YAML
62 lines
1.5 KiB
YAML
# Legal Filing extraction profile
|
|
# Matches court filings: motions, briefs, orders, docket entries
|
|
name: legal_filing
|
|
description: Court filings: motions, briefs, orders, docket entries
|
|
priority: 40
|
|
|
|
match:
|
|
all:
|
|
- any:
|
|
- text_contains:
|
|
patterns: ["UNITED STATES DISTRICT COURT", "IN THE COURT OF", "IN THE MATTER OF", "Case No.", "Civil Action No.", "Plaintiff", "Defendant", "Petitioner", "Respondent", "COMPLAINT", "MOTION TO", "ORDER GRANTING", "OPINION"]
|
|
- heading_matches:
|
|
pattern: "^(COMPLAINT|MOTION|ORDER|OPINION|BRIEF)"
|
|
- structural:
|
|
has_table: false
|
|
has_form_field: false
|
|
has_math: false
|
|
page_count:
|
|
min: 1
|
|
max: 500
|
|
|
|
extraction:
|
|
reading_order: xy_cut
|
|
table_detection: default
|
|
readability_threshold: 0.5
|
|
include_invisible: false
|
|
include_headers_footers: true
|
|
force_ocr: false
|
|
min_block_chars: 0
|
|
|
|
fields:
|
|
case_number:
|
|
type: string
|
|
extraction:
|
|
near: ["Case No.", "Civil Action No.", "Docket No.", "Cause No."]
|
|
regex: "[\\w-]+:?\\s*\\d+[\\w-]*"
|
|
parse: string
|
|
|
|
court:
|
|
type: string
|
|
extraction:
|
|
region: top_quarter
|
|
pick: largest_font
|
|
parse: string
|
|
|
|
parties:
|
|
type: array
|
|
extraction:
|
|
near: ["Plaintiff", "Defendant", "Petitioner", "Respondent", "v."]
|
|
fallback: []
|
|
|
|
filing_date:
|
|
type: date
|
|
extraction:
|
|
near: ["Filed", "Date Filed", "Dated"]
|
|
parse: date
|
|
|
|
docket_entries:
|
|
type: array
|
|
extraction:
|
|
region: bottom_half
|
|
fallback: []
|