112 lines
4.7 KiB
Python
112 lines
4.7 KiB
Python
"""Tests for app/domain/reports/citation_validator.py"""
|
|
|
|
from app.domains.reports.citation_validator import validate_section
|
|
|
|
|
|
class TestCitationValidator:
|
|
"""Tests for validate_section function."""
|
|
|
|
def test_valid_citation(self):
|
|
"""Test that valid citations pass validation."""
|
|
result = validate_section(
|
|
"This token has a risk score of 75/100 [1]. It is flagged [2].",
|
|
["Risk score 75/100 detected", "Token flagged as suspicious"],
|
|
on_unciteable="strip",
|
|
)
|
|
assert result["validation_rate"] == 1.0
|
|
assert result["unciteable_count"] == 0
|
|
assert "75/100" in result["validated_text"]
|
|
|
|
def test_invalid_citation(self):
|
|
"""Test that invalid citations are stripped."""
|
|
result = validate_section(
|
|
"This claim [99] is invalid [1].", ["Only source 1 available"], on_unciteable="strip"
|
|
)
|
|
assert result["validation_rate"] == 0.0
|
|
assert result["unciteable_count"] == 1
|
|
assert "[Data not available]" in result["validated_text"]
|
|
|
|
def test_no_citations(self):
|
|
"""Test that text without citations is marked unciteable."""
|
|
result = validate_section(
|
|
"This has no citations but should match source.",
|
|
["Source text for validation"],
|
|
on_unciteable="strip",
|
|
)
|
|
assert result["validation_rate"] == 0.0
|
|
assert result["unciteable_count"] == 1
|
|
|
|
def test_multiple_citations_same_sentence(self):
|
|
"""Test that multiple citations in one sentence are parsed.
|
|
|
|
Per T05 spec: validator must (1) parse [N,M,K] correctly AND
|
|
(2) verify claim is supported by source content. With the strict
|
|
default threshold, the literal claim 'This is supported by'
|
|
won't match 'Source one content' terms - so it should fail
|
|
the support check (unciteable_count > 0).
|
|
"""
|
|
result = validate_section(
|
|
"This is supported by [1,2,3].",
|
|
["Source one content", "Source two content", "Source three content"],
|
|
on_unciteable="strip",
|
|
)
|
|
# Parsing succeeded (multi-citation [1,2,3] all valid indices)
|
|
# But claim doesn't actually overlap with any source content
|
|
assert result["unciteable_count"] == 1 # claim unsupported
|
|
|
|
def test_citation_range(self):
|
|
"""Test that citation ranges [1-3] are parsed correctly.
|
|
|
|
Per T05 spec: validator parses [1-N] and checks support.
|
|
The literal claim 'This is supported by' doesn't overlap
|
|
with 'Source one content' terms, so should fail support.
|
|
"""
|
|
result = validate_section(
|
|
"This is supported by [1-2].",
|
|
["Source one content", "Source two content", "Source three content"],
|
|
on_unciteable="strip",
|
|
)
|
|
# Parsed correctly (range expanded)
|
|
# But claim doesn't match source content
|
|
assert result["unciteable_count"] == 1
|
|
|
|
def test_empty_sources(self):
|
|
"""Test that empty sources list marks all as unciteable."""
|
|
result = validate_section("Some text [1] with citations.", [], on_unciteable="strip")
|
|
assert result["validation_rate"] == 0.0
|
|
assert "Data not available" in result["validated_text"]
|
|
|
|
def test_keep_unciteable(self):
|
|
"""Test that on_unciteable='keep' preserves unciteable text."""
|
|
result = validate_section(
|
|
"Some text [99] invalid.", ["Source one content"], on_unciteable="keep"
|
|
)
|
|
assert result["unciteable_count"] == 1
|
|
assert "Some text [99] invalid." in result["validated_text"]
|
|
|
|
def test_strip_unciteable(self):
|
|
"""Test that on_unciteable='strip' removes unciteable text."""
|
|
result = validate_section(
|
|
"Some text [99] invalid.", ["Source one content"], on_unciteable="strip"
|
|
)
|
|
assert result["unciteable_count"] == 1
|
|
assert "Data not available" in result["validated_text"]
|
|
|
|
def test_validation_report_structure(self):
|
|
"""Test that the validation report has the correct structure."""
|
|
result = validate_section("Test [1].", ["Source one content"])
|
|
assert "validated_text" in result
|
|
assert "citations" in result
|
|
assert "unciteable_count" in result
|
|
assert "validation_rate" in result
|
|
assert isinstance(result["citations"], list)
|
|
|
|
def test_citation_details(self):
|
|
"""Test that citations include claim, source_idx, source_text, supported."""
|
|
result = validate_section("Test [1].", ["Source one content"])
|
|
if result["citations"]:
|
|
c = result["citations"][0]
|
|
assert "claim" in c
|
|
assert "source_idx" in c
|
|
assert "source_text" in c
|
|
assert "supported" in c
|