KMP-IMPACT

AnalysisConfig

The CLI surface flattened into a configuration dataclass. Built once at the top of pipeline.run_pipeline(). Defined in src/kmp_impact_analyzer/config.py.

Fields

Field	Type	Default	Notes
repo_path	str	""	Path or URL to the KMP project repository.
dependency_group	str	""	Maven group or coordinate substring (e.g. io.ktor).
before_version	str	""	Version before the change.
after_version	str	""	Version after the change.
output_dir	str	"output"	Where shadow copies, phase outputs, and the report are written.
skip_dynamic	bool	False	Skip Phase 3 entirely.
init_script_path	str	""	Path to the Gradle init script used by the static phase. Empty means use the bundled gradle-init/impact-analyzer-init.gradle.kts.
droidbot_timeout	int	120	DroidBot per-event timeout in seconds.
droidbot_policy	str	"dfs_greedy"	DroidBot exploration policy.
before_apk	str	""	Optional pre-built APK to reuse for the BEFORE state.
after_apk	str	""	Optional pre-built APK to reuse for the AFTER state.
droidbot_before_output	str	""	Optional pre-generated DroidBot output directory for the BEFORE APK.
droidbot_after_output	str	""	Optional pre-generated DroidBot output directory for the AFTER APK.
extra_seed_packages	list[str]	[]	Additional Kotlin package roots to treat as impact seeds, on top of the auto-resolved ones.

Construction helpers

from pathlib import Path
from kmp_impact_analyzer.config import AnalysisConfig

# From CLI keyword arguments (extra keys are ignored)
config = AnalysisConfig.from_cli(
    repo_path="/path/to/kmp/project",
    dependency_group="io.ktor",
    before_version="2.3.8",
    after_version="2.3.11",
    output_dir="output",
    skip_dynamic=True,
)

# From a YAML file (scenario)
config = AnalysisConfig.from_yaml(Path("scenarios/ktor/scenario.yml"))

from_yaml reads every key from the YAML file and silently discards keys that are not fields of the dataclass. This is what makes scenario.yml forward-compatible with new fields.

Consumed by

• kmp_impact_analyzer.pipeline.run_pipeline()
• The four CLI subcommands map their flags to this object before invoking the pipeline.

See also

• analyze — the most common way to build this config.
• run-scenario — builds it from scenario.yml.

This site is open source. Improve this page.