sahilfarib/Legal-Document-Intelligence

1"""2tests/test_validator.py — Tests for post-generation citation validator3"""4import sys5from pathlib import Path6import pytest7 8sys.path.insert(0, str(Path(__file__).parent.parent))9 10from models import EvidenceChunk11from generation.validator import CitationValidator, ValidationReport12 13 14@pytest.fixture15def validator():16    return CitationValidator()17 18 19@pytest.fixture20def evidence():21    return [22        EvidenceChunk(chunk_id="aabb001100000000", text="Contract text.", source_file="f.pdf", score=0.9),23        EvidenceChunk(chunk_id="ccdd002200000000", text="Payment terms.", source_file="f.pdf", score=0.8),24    ]25 26 27class TestValidator:28    def test_fully_cited_draft(self, validator, evidence):29        draft = (30            "## 1. Document Overview\n"31            "The contract was signed in January. [CHUNK-aabb0011]\n"32            "## 2. Parties and Roles\n"33            "Payment is $25,000 monthly. [CHUNK-ccdd0022]\n"34            "## 3. Key Facts and Dates\n"35            "The term is 12 months. [CHUNK-aabb0011]\n"36            "## 4. Obligations and Terms\n"37            "Late fees apply at 1.5%. [CHUNK-ccdd0022]\n"38            "## 5. Red Flags and Issues\n"39            "Limitation clause may be unenforceable. [CHUNK-aabb0011]\n"40            "## 6. Analyst Notes\n"41            "Recommend further review of liability cap. [CHUNK-ccdd0022]\n"42        )43        report = validator.validate(draft, evidence)44        assert report.uncited_ratio < 0.1045        assert report.is_clean()46        assert report.severity() == "green"47 48    def test_uncited_sentences_detected(self, validator, evidence):49        draft = (50            "## 1. Document Overview\n"51            "The contract was signed in January. [CHUNK-aabb0011]\n"52            "This sentence has no citation and is long enough to count.\n"53            "Another uncited claim that should be flagged by the validator.\n"54        )55        report = validator.validate(draft, evidence)56        assert len(report.uncited_sentences) == 257        assert report.uncited_ratio > 058 59    def test_invalid_citation_detected(self, validator, evidence):60        draft = "The contract mentions something. [CHUNK-deadbeef]\n"61        report = validator.validate(draft, evidence)62        assert len(report.invalid_citations) == 163 64    def test_not_evidenced_marker(self, validator, evidence):65        draft = "The jurisdiction is not stated. [NOT EVIDENCED]\n"66        report = validator.validate(draft, evidence)67        assert report.not_evidenced_count == 168        assert len(report.uncited_sentences) == 069 70    def test_section_completeness(self, validator, evidence):71        draft = "## 1. Document Overview\nSome text. [CHUNK-aabb0011]\n"72        report = validator.validate(draft, evidence)73        assert len(report.missing_sections) > 074        assert report.section_completeness < 1.075 76    def test_empty_draft(self, validator, evidence):77        report = validator.validate("", evidence)78        assert report.total_sentences == 079 80 81class TestValidationReport:82    def test_severity_green(self):83        r = ValidationReport(uncited_ratio=0.05)84        assert r.severity() == "green"85 86    def test_severity_yellow(self):87        r = ValidationReport(uncited_ratio=0.15)88        assert r.severity() == "yellow"89 90    def test_severity_red(self):91        r = ValidationReport(uncited_ratio=0.30)92        assert r.severity() == "red"93 94    def test_to_dict(self):95        r = ValidationReport(total_sentences=10, cited_sentences=8)96        d = r.to_dict()97        assert "total_sentences" in d98        assert "severity" in d99