8b63217dbf
Implements the legal_filing document profile for court filings (motions, briefs, orders, docket entries) with: - Profile YAML at profiles/builtin/legal_filing/profile.yaml - Fields: case_number, court, parties, filing_date, docket_entries - Match predicates for court name, case numbers, party markers - Extraction: xy_cut reading order, include_headers_footers=true - 5 synthetic PDF fixtures at tests/fixtures/profiles/legal_filing/ - federal_complaint: Federal district court complaint - state_motion: State superior court motion to dismiss - appellate_brief: Federal appellate brief - court_order: Federal district court order - docket_sheet: Docket sheet with entries - 5 expected output JSON files with profile_fields - Regression tests at crates/pdftract-cli/tests/test_legal_filing.rs - 14/14 tests pass - Verifies profile schema, fixture structure, match predicates Acceptance criteria (from bead pdftract-260a3): - ✅ profiles/builtin/legal_filing.yaml validates - ✅ 5+ public-domain fixtures with expected outputs - ✅ tests/test_legal_filing.rs passes - ✅ Per-field accuracy thresholds defined (integration tests pending Phase 7.10) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| appellate_brief-expected.json | ||
| appellate_brief.pdf | ||
| court_order-expected.json | ||
| court_order.pdf | ||
| docket_sheet-expected.json | ||
| docket_sheet.pdf | ||
| federal_complaint-expected.json | ||
| federal_complaint.pdf | ||
| PROVENANCE.md | ||
| README.md | ||
| state_motion-expected.json | ||
| state_motion.pdf | ||
Legal Filing Profile Fixtures
This directory contains test fixtures for the legal filing document profile.
Fixture Types
- federal_complaint.pdf (3 pages) - Federal district court complaint with case number, court, parties, filing date, and verification
- state_motion.pdf (2 pages) - State superior court motion to dismiss with notice of motion and legal argument
- appellate_brief.pdf (3 pages) - Federal appellate brief with jurisdiction statement, issue summary, and argument
- court_order.pdf (2 pages) - Court order granting motion with background and analysis
- docket_sheet.pdf (2 pages) - Docket sheet with docket entries showing case history
Expected Output Format
Each fixture has a corresponding *-expected.json file with the following structure:
{
"metadata": {
"document_type": "legal_filing",
"document_type_confidence": 0.XX,
"document_type_reasons": [...],
"profile_name": "legal_filing",
"profile_version": "1.0.0",
"profile_fields": {
"case_number": "string",
"court": "string",
"parties": ["Party One", "Party Two"],
"filing_date": "YYYY-MM-DD",
"docket_entries": [...]
}
}
}
Provenance
All fixtures are synthetic PDFs generated by tests/fixtures/generate_legal_filing_fixtures.rs. They are created programmatically as minimal valid PDFs for testing purposes. No real court filings or PII are included.
See PROVENANCE.md for detailed generation information.
Field Accuracy Notes
- case_number: Regex-based extraction; handles federal (1:24-cv-00123), state (CGC-24-123456), and appellate (24-1234) formats
- court: Extracted from top_quarter region with largest_font heuristics; may fail for graphical court headers
- parties: Captured verbatim block; multi-party cases may have incomplete extraction
- filing_date: Date parsing with flexible format detection
- docket_entries: BEST-EFFORT structured extraction; only present for docket_sheet fixture
Acceptance Criteria
- Per-field accuracy: >= 90% across the 5-fixture corpus
- parties and docket_entries relaxed to >= 80% due to complexity