Added support for multiple LLM passes to correct the transcript based upon the provided glossary

This commit is contained in:
2026-04-21 13:12:08 -05:00
parent 4296e3576e
commit 7e1a3f721a
12 changed files with 306 additions and 55 deletions

View File

@@ -3,7 +3,7 @@ from pathlib import Path
import pytest
from audita.config import AuditaConfig, ConfigOverrides
from audita.config import DEFAULT_MAX_RETRIES, DEFAULT_MAX_SECTION_TOKENS, DEFAULT_WORK_DIR
from audita.config import DEFAULT_GLOSSARY_MAX_LLM_PASSES, DEFAULT_MAX_RETRIES, DEFAULT_MAX_SECTION_TOKENS, DEFAULT_WORK_DIR
from audita.errors import AuditaConfigError
@@ -12,6 +12,7 @@ def test_config_uses_defaults_with_api_key():
assert config.max_section_tokens == DEFAULT_MAX_SECTION_TOKENS
assert config.max_retries == DEFAULT_MAX_RETRIES
assert config.glossary_max_llm_passes == DEFAULT_GLOSSARY_MAX_LLM_PASSES
assert config.work_dir == Path(DEFAULT_WORK_DIR)
@@ -22,6 +23,7 @@ def test_config_env_overrides_defaults():
"AUDITA_MAX_SECTION_TOKENS": "42",
"AUDITA_CONFIDENCE_THRESHOLD": "0.9",
"AUDITA_MAX_RETRIES": "5",
"AUDITA_GLOSSARY_MAX_LLM_PASSES": "7",
"AUDITA_WORK_DIR": "/tmp/custom-audita",
}
)
@@ -29,6 +31,7 @@ def test_config_env_overrides_defaults():
assert config.max_section_tokens == 42
assert config.confidence_threshold == 0.9
assert config.max_retries == 5
assert config.glossary_max_llm_passes == 7
assert config.work_dir == Path("/tmp/custom-audita")
@@ -38,17 +41,20 @@ def test_config_cli_overrides_env():
"OPENROUTER_API_KEY": "key",
"AUDITA_MAX_SECTION_TOKENS": "42",
"AUDITA_MAX_RETRIES": "5",
"AUDITA_GLOSSARY_MAX_LLM_PASSES": "7",
"AUDITA_WORK_DIR": "/tmp/env-audita",
},
overrides=ConfigOverrides(
max_section_tokens=100,
max_retries=3,
glossary_max_llm_passes=2,
work_dir=Path("/tmp/cli-audita"),
),
)
assert config.max_section_tokens == 100
assert config.max_retries == 3
assert config.glossary_max_llm_passes == 2
assert config.work_dir == Path("/tmp/cli-audita")
@@ -63,3 +69,9 @@ def test_config_rejects_bad_env_int():
env={"OPENROUTER_API_KEY": "key", "AUDITA_MAX_SECTION_TOKENS": "many"}
)
def test_config_rejects_invalid_glossary_pass_count():
with pytest.raises(AuditaConfigError):
AuditaConfig.from_sources(
env={"OPENROUTER_API_KEY": "key", "AUDITA_GLOSSARY_MAX_LLM_PASSES": "0"}
)