tests/test_validator.py

"""
tests/test_validator.py — Tests for post-generation citation validator
"""
import sys
from pathlib import Path
import pytest

sys.path.insert(0, str(Path(__file__).parent.parent))

from models import EvidenceChunk
from generation.validator import CitationValidator, ValidationReport


@pytest.fixture
def validator():
    return CitationValidator()


@pytest.fixture
def evidence():
    return [
        EvidenceChunk(chunk_id="aabb001100000000", text="Contract text.", source_file="f.pdf", score=0.9),
        EvidenceChunk(chunk_id="ccdd002200000000", text="Payment terms.", source_file="f.pdf", score=0.8),
    ]


class TestValidator:
    def test_fully_cited_draft(self, validator, evidence):
        draft = (
            "## 1. Document Overview\n"
            "The contract was signed in January. [CHUNK-aabb0011]\n"
            "## 2. Parties and Roles\n"
            "Payment is $25,000 monthly. [CHUNK-ccdd0022]\n"
            "## 3. Key Facts and Dates\n"
            "The term is 12 months. [CHUNK-aabb0011]\n"
            "## 4. Obligations and Terms\n"
            "Late fees apply at 1.5%. [CHUNK-ccdd0022]\n"
            "## 5. Red Flags and Issues\n"
            "Limitation clause may be unenforceable. [CHUNK-aabb0011]\n"
            "## 6. Analyst Notes\n"
            "Recommend further review of liability cap. [CHUNK-ccdd0022]\n"
        )
        report = validator.validate(draft, evidence)
        assert report.uncited_ratio < 0.10
        assert report.is_clean()
        assert report.severity() == "green"

    def test_uncited_sentences_detected(self, validator, evidence):
        draft = (
            "## 1. Document Overview\n"
            "The contract was signed in January. [CHUNK-aabb0011]\n"
            "This sentence has no citation and is long enough to count.\n"
            "Another uncited claim that should be flagged by the validator.\n"
        )
        report = validator.validate(draft, evidence)
        assert len(report.uncited_sentences) == 2
        assert report.uncited_ratio > 0

    def test_invalid_citation_detected(self, validator, evidence):
        draft = "The contract mentions something. [CHUNK-deadbeef]\n"
        report = validator.validate(draft, evidence)
        assert len(report.invalid_citations) == 1

    def test_not_evidenced_marker(self, validator, evidence):
        draft = "The jurisdiction is not stated. [NOT EVIDENCED]\n"
        report = validator.validate(draft, evidence)
        assert report.not_evidenced_count == 1
        assert len(report.uncited_sentences) == 0

    def test_section_completeness(self, validator, evidence):
        draft = "## 1. Document Overview\nSome text. [CHUNK-aabb0011]\n"
        report = validator.validate(draft, evidence)
        assert len(report.missing_sections) > 0
        assert report.section_completeness < 1.0

    def test_empty_draft(self, validator, evidence):
        report = validator.validate("", evidence)
        assert report.total_sentences == 0


class TestValidationReport:
    def test_severity_green(self):
        r = ValidationReport(uncited_ratio=0.05)
        assert r.severity() == "green"

    def test_severity_yellow(self):
        r = ValidationReport(uncited_ratio=0.15)
        assert r.severity() == "yellow"

    def test_severity_red(self):
        r = ValidationReport(uncited_ratio=0.30)
        assert r.severity() == "red"

    def test_to_dict(self):
        r = ValidationReport(total_sentences=10, cited_sentences=8)
        d = r.to_dict()
        assert "total_sentences" in d
        assert "severity" in d