Implemented a second-stage validator for grammar corrections that allows homophone-related changes

This commit is contained in:
2026-04-23 11:25:35 -05:00
parent dee6ae4067
commit c4e2db75f1
13 changed files with 638 additions and 65 deletions

View File

@@ -8,6 +8,7 @@ from audita.config import (
DEFAULT_GLOSSARY_MAX_LLM_PASSES,
DEFAULT_GRAMMAR_CONFIDENCE_THRESHOLD,
DEFAULT_GRAMMAR_MAX_LLM_PASSES,
DEFAULT_GRAMMAR_SPOKEN_FORM_VALIDATION_CONFIDENCE_THRESHOLD,
DEFAULT_GRAMMAR_VALIDATION_CONFIDENCE_THRESHOLD,
DEFAULT_GRAMMAR_VALIDATION_ENABLED,
DEFAULT_MAX_RETRIES,
@@ -36,6 +37,11 @@ def test_config_uses_defaults_with_api_key():
assert config.grammar_validation_enabled is True
assert config.grammar_validation_confidence_threshold == DEFAULT_GRAMMAR_VALIDATION_CONFIDENCE_THRESHOLD
assert config.grammar_validation_confidence_threshold == 0.8
assert (
config.grammar_spoken_form_validation_confidence_threshold
== DEFAULT_GRAMMAR_SPOKEN_FORM_VALIDATION_CONFIDENCE_THRESHOLD
)
assert config.grammar_spoken_form_validation_confidence_threshold == 0.8
assert config.normalize_max_segment_gap == DEFAULT_NORMALIZE_MAX_SEGMENT_GAP
assert config.normalize_max_segment_gap == 5.0
assert config.normalize_ellipsis_gap == DEFAULT_NORMALIZE_ELLIPSIS_GAP
@@ -58,6 +64,7 @@ def test_config_env_overrides_defaults():
"AUDITA_GRAMMAR_MAX_LLM_PASSES": "4",
"AUDITA_GRAMMAR_VALIDATION_ENABLED": "false",
"AUDITA_GRAMMAR_VALIDATION_CONFIDENCE_THRESHOLD": "0.91",
"AUDITA_GRAMMAR_SPOKEN_FORM_VALIDATION_CONFIDENCE_THRESHOLD": "0.87",
"AUDITA_NORMALIZE_MAX_SEGMENT_GAP": "4.5",
"AUDITA_NORMALIZE_ELLIPSIS_GAP": "1.5",
"AUDITA_NORMALIZE_MAX_SEGMENT_DURATION": "45.0",
@@ -74,6 +81,7 @@ def test_config_env_overrides_defaults():
assert config.grammar_max_llm_passes == 4
assert config.grammar_validation_enabled is False
assert config.grammar_validation_confidence_threshold == 0.91
assert config.grammar_spoken_form_validation_confidence_threshold == 0.87
assert config.normalize_max_segment_gap == 4.5
assert config.normalize_ellipsis_gap == 1.5
assert config.normalize_max_segment_duration == 45.0
@@ -91,6 +99,7 @@ def test_config_cli_overrides_env():
"AUDITA_GRAMMAR_MAX_LLM_PASSES": "6",
"AUDITA_GRAMMAR_VALIDATION_ENABLED": "false",
"AUDITA_GRAMMAR_VALIDATION_CONFIDENCE_THRESHOLD": "0.91",
"AUDITA_GRAMMAR_SPOKEN_FORM_VALIDATION_CONFIDENCE_THRESHOLD": "0.87",
"AUDITA_NORMALIZE_MAX_SEGMENT_GAP": "4.5",
"AUDITA_NORMALIZE_ELLIPSIS_GAP": "1.5",
"AUDITA_NORMALIZE_MAX_SEGMENT_DURATION": "45.0",
@@ -106,6 +115,7 @@ def test_config_cli_overrides_env():
grammar_max_llm_passes=3,
grammar_validation_enabled=True,
grammar_validation_confidence_threshold=0.75,
grammar_spoken_form_validation_confidence_threshold=0.72,
normalize_max_segment_gap=3.0,
normalize_ellipsis_gap=1.0,
normalize_max_segment_duration=30.0,
@@ -122,6 +132,7 @@ def test_config_cli_overrides_env():
assert config.grammar_max_llm_passes == 3
assert config.grammar_validation_enabled is True
assert config.grammar_validation_confidence_threshold == 0.75
assert config.grammar_spoken_form_validation_confidence_threshold == 0.72
assert config.normalize_max_segment_gap == 3.0
assert config.normalize_ellipsis_gap == 1.0
assert config.normalize_max_segment_duration == 30.0
@@ -168,6 +179,13 @@ def test_config_rejects_invalid_stage_thresholds():
AuditaConfig.from_sources(
env={"OPENROUTER_API_KEY": "key", "AUDITA_GRAMMAR_VALIDATION_CONFIDENCE_THRESHOLD": "1.1"}
)
with pytest.raises(AuditaConfigError):
AuditaConfig.from_sources(
env={
"OPENROUTER_API_KEY": "key",
"AUDITA_GRAMMAR_SPOKEN_FORM_VALIDATION_CONFIDENCE_THRESHOLD": "-0.1",
}
)
def test_config_rejects_invalid_grammar_validation_enabled():