[build-system] requires = ["hatchling"] build-backend = "hatchling.build" [project] name = "truthound" version = "3.3.2" description = "Zero-Configuration Data Quality Framework Powered by Polars" readme = "README.md" license = "Apache-2.0" requires-python = ">=3.11" authors = [ { name = "seadonggyun4", email = "seadonggyun4@gmail.com" } ] keywords = [ "data-quality", "data-validation", "polars", "pii-detection", "data-masking", "mcp", "csv", "parquet", "data-profiling", ] classifiers = [ "Development Status :: 5 - Production/Stable", "Intended Audience :: Developers", "Intended Audience :: Science/Research", "License :: OSI Approved :: Apache Software License", "Operating System :: OS Independent", "Programming Language :: Python :: 3", "Programming Language :: Python :: 3.11", "Programming Language :: Python :: 3.12", "Topic :: Scientific/Engineering", "Topic :: Software Development :: Libraries :: Python Modules", "Typing :: Typed", ] dependencies = [ "polars>=1.0.0,<2.0.0", "pyyaml>=6.0.0", "rich>=13.0.0", "typer>=0.12.0", ] [project.optional-dependencies] # Graphviz images also require the system dot executable on PATH. lineage-visualization = ["graphviz>=0.21,<1.0.0"] profiler-visualization = ["plotly>=5.24.0,<7.0.0"] # Local MCP transport; Core remains usable without the SDK. mcp = ["mcp>=1.30.0,<2.0.0", "anyio>=4.6,<5.0", "polars>=1.0.0,<2.0.0", "pydantic>=2.11.0,<3.0.0", "psutil>=5.9.0,<8.0.0", "filelock>=3.16.0,<4.0.0"] # Native process-locking providers; no-op locking cannot fence durable store recovery. stores-concurrency = ["filelock>=3.16.0,<4.0.0", "portalocker>=3.0.0,<4.0.0"] stores-streaming = [ "filelock>=3.16.0,<4.0.0", "zstandard>=0.22.0,<1.0.0", "lz4>=4.3.0,<5.0.0", "python-snappy>=0.7.0,<1.0.0", ] # Native local/distributed execution; client/cluster configuration remains caller-owned. execution-dask = ["dask[dataframe,distributed]>=2024.1.0,<2027.0.0", "psutil>=5.9.0,<8.0.0"] execution-ray = ["ray[data]>=2.48.0,!=2.59.0,<3.0.0", "psutil>=5.9.0,<8.0.0"] execution-spark = ["pyspark[sql]>=4.0.0,<5.0.0", "psutil>=5.9.0,<8.0.0"] # Optional AI artifact and provider contract surface ai = [ "pydantic>=2.11.0,<3.0.0", "openai>=1.35.0,<2.0.0", ] # Report generation reports = [ "jinja2>=3.0.0", ] # Statistical drift detection drift = [ "scipy>=1.10.0", "river>=0.26.1,<0.27.0", ] # Anomaly detection with ML anomaly = [ "scipy>=1.10.0", "scikit-learn>=1.7.0", ] # Cloud storage backends geospatial = ["shapely>=2.0.0,<3.0.0"] s3 = [ "boto3>=1.42.26,<2.0.0", ] gcs = [ "google-cloud-storage>=2.0.0", ] azure = [ "azure-storage-blob>=12.0.0", "azure-identity>=1.15.0,<2.0.0", ] # Database storage backend database = [ "sqlalchemy>=2.0.0", ] # SQL datasource providers postgresql = [ "psycopg2-binary>=2.9.9", ] mysql = [ "pymysql>=1.1.0", ] snowflake = [ "snowflake-connector-python>=3.12.0", ] bigquery = [ "google-cloud-bigquery>=3.25.0", "db-dtypes>=1.3.0", ] redshift = [ "redshift-connector>=2.1.3", ] databricks = [ "databricks-sql-connector>=3.4.0", ] oracle = [ "oracledb>=2.4.0", ] sqlserver = [ "pymssql>=2.3.1", ] enterprise = [ "oracledb>=2.4.0", "pymssql>=2.3.1", ] sql-connectors = [ "psycopg2-binary>=2.9.9", "pymysql>=1.1.0", "duckdb>=1.0.0", "snowflake-connector-python>=3.12.0", "google-cloud-bigquery>=3.25.0", "db-dtypes>=1.3.0", "redshift-connector>=2.1.3", "databricks-sql-connector>=3.4.0", "oracledb>=2.4.0", "pymssql>=2.3.1", ] # All storage backends stores = [ "boto3>=1.42.26,<2.0.0", "google-cloud-storage>=2.0.0", "azure-storage-blob>=12.0.0", "azure-identity>=1.15.0,<2.0.0", "sqlalchemy>=2.0.0", ] # DuckDB database duckdb = [ "duckdb>=1.0.0", ] # NoSQL databases mongodb = [ "motor>=3.0.0", ] elasticsearch = [ "elasticsearch[async]>=8.0.0", ] nosql = [ "motor>=3.0.0", "elasticsearch[async]>=8.0.0", ] # Streaming platforms kafka = [ "aiokafka>=0.9.0", ] streaming = [ "aiokafka>=0.9.0", "pyarrow>=14.0.0", ] # All async datasources async-datasources = [ "motor>=3.0.0", "elasticsearch[async]>=8.0.0", "aiokafka>=0.9.0", ] # Native APScheduler 3.x adapter (the 4.x API is incompatible). checkpoint-scheduler = ["apscheduler>=3.10.0,<4.0.0"] # Native Streamlit process lifecycle; separate from the Reflex dashboard extra. dashboard-streamlit = ["streamlit>=1.40.0,<2.0.0"] # Interactive dashboard (Phase 8) dashboard = [ "reflex>=0.8.25,<0.9.0", "sqlalchemy[asyncio]>=2.0.0,<3.0.0", "uvicorn[standard]>=0.35.0,<1.0.0", ] # PDF export support pdf = [ "weasyprint>=60.0", ] # Benchmark parity and release-gate dependencies benchmarks = [ "pandas>=2.0.0,<3.0.0", "great-expectations>=1.15.0,<2.0.0", "psutil>=5.9.0,<8.0.0", "duckdb>=1.0.0,<2.0.0", "sqlalchemy>=2.0.0,<3.0.0", ] # Performance optimization perf = [ "xxhash>=3.4.0", ] # Full installation with all optional dependencies all = [ "mcp>=1.30.0,<2.0.0", "anyio>=4.6,<5.0", "graphviz>=0.21,<1.0.0", "plotly>=5.24.0,<7.0.0", "filelock>=3.16.0,<4.0.0", "portalocker>=3.0.0,<4.0.0", "dask[dataframe,distributed]>=2024.1.0,<2027.0.0", "ray[data]>=2.48.0,!=2.59.0,<3.0.0", "pyspark[sql]>=4.0.0,<5.0.0", "zstandard>=0.22.0,<1.0.0", "lz4>=4.3.0,<5.0.0", "python-snappy>=0.7.0,<1.0.0", "shapely>=2.0.0,<3.0.0", "pydantic>=2.11.0,<3.0.0", "openai>=1.35.0,<2.0.0", "jinja2>=3.0.0", "pyarrow>=14.0.0", "pandas>=2.0.0,<3.0.0", "scipy>=1.10.0", "river>=0.26.1,<0.27.0", "scikit-learn>=1.7.0", "boto3>=1.42.26,<2.0.0", "google-cloud-storage>=2.0.0", "azure-storage-blob>=12.0.0", "azure-identity>=1.15.0,<2.0.0", "sqlalchemy>=2.0.0", "duckdb>=1.0.0,<2.0.0", "psycopg2-binary>=2.9.9", "pymysql>=1.1.0", "snowflake-connector-python>=3.12.0", "google-cloud-bigquery>=3.25.0", "db-dtypes>=1.3.0", "redshift-connector>=2.1.3", "databricks-sql-connector>=3.4.0", "oracledb>=2.4.0", "pymssql>=2.3.1", "reflex>=0.8.25,<0.9.0", "sqlalchemy[asyncio]>=2.0.0,<3.0.0", "uvicorn[standard]>=0.35.0,<1.0.0", "streamlit>=1.40.0,<2.0.0", "apscheduler>=3.10.0,<4.0.0", "weasyprint>=60.0", "motor>=3.0.0", "elasticsearch[async]>=8.0.0", "aiokafka>=0.9.0", "great-expectations>=1.15.0,<2.0.0", "psutil>=5.9.0,<8.0.0", "xxhash>=3.4.0", "sqlalchemy>=2.0.0,<3.0.0", ] # Development dependencies dev = [ # Shared native MCP fixtures must collect and execute in the normal dev lane. "mcp>=1.30.0,<2.0.0", "anyio>=4.6,<5.0", "graphviz>=0.21,<1.0.0", "weasyprint>=60.0", "plotly>=5.24.0,<7.0.0", # Offline S3 service-model contracts execute actual botocore validation. "boto3>=1.42.26,<2.0.0", "google-cloud-storage>=3.7.0,<4.0.0", "azure-storage-blob>=12.28.0,<13.0.0", "azure-identity>=1.15.0,<2.0.0", "zstandard>=0.22.0,<1.0.0", "python-snappy>=0.7.0,<1.0.0", # Native cross-platform store locking must execute actual providers. "filelock>=3.16.0,<4.0.0", "portalocker>=3.0.0,<4.0.0", # Native engine contracts execute actual local SDK partitions and a Java 17+ JVM. "ray[data]>=2.48.0,!=2.59.0,<3.0.0", "pyspark[sql]>=4.0.0,<5.0.0", "dask[dataframe,distributed]>=2024.1.0,<2027.0.0", # Durable restart tests must exercise the actual supported backend. "apscheduler>=3.10.0,<4.0.0", # Dashboard lifecycle contracts run a real loopback Streamlit process. "streamlit>=1.40.0,<2.0.0", "reflex>=0.8.25,<0.9.0", "sqlalchemy[asyncio]>=2.0.0,<3.0.0", "uvicorn[standard]>=0.35.0,<1.0.0", "duckdb>=1.4.0,<2.0.0", "shapely>=2.0.0,<3.0.0", # Full encryption roundtrip contracts exercise the default LZ4 pipeline. "lz4>=4.3.0,<5.0.0", # Default file-key-store KDF contracts must execute, not skip via transitives. "argon2-cffi>=23.1.0,<26.0.0", "cryptography>=41.0.0,<51.0.0", # Lifecycle contracts observe real child PIDs and reaping, not mocked tasks. "psutil>=5.9.0,<8.0.0", "pydantic>=2.11.0,<3.0.0", "openai>=1.35.0,<2.0.0", "pytest>=9.0.0,<10.0.0", "pytest-timeout>=2.4.0,<3.0.0", "jsonschema>=4.23.0,<5.0.0", "pytest-cov>=4.0.0", "pytest-asyncio>=0.23.0", "ruff>=0.4.0", "mypy>=1.10.0", "jinja2>=3.0.0", "pyarrow>=14.0.0", "pandas>=2.0.0,<3.0.0", "scipy>=1.10.0", "river>=0.26.1,<0.27.0", "scikit-learn>=1.7.0", ] docs = [ "mkdocs>=1.6.0", "mkdocs-material>=9.5.0", "mkdocs-static-i18n>=1.3.0", "mkdocstrings[python]>=0.24.0", "pymdown-extensions>=10.8.0", ] # CI-only prerequisites; these are not published user-facing package extras. [dependency-groups] nightly-local = [ "aiokafka>=0.12.0,<1.0.0", "aiobotocore>=2.18.0,<4.0.0", "testcontainers[kafka,localstack]>=4.9.0,<5.0.0", "kafka-python>=2.0.2,<3.0.0", "boto3>=1.42.26,<2.0.0", "azure-storage-blob>=12.19.0,<13.0.0", "redis>=5.0.0,<7.0.0", "elasticsearch>=8.0.0,<9.0.0", "opentelemetry-sdk>=1.20.0,<2.0.0", "zstandard>=0.22.0,<1.0.0", "pypdf==6.18.0", ] # Internal Depot runtime contract. # This table is intentionally kept outside `project.optional-dependencies` so it # does not become public package metadata or a user-facing install surface. # It defines the baseline optional Truthound features that private # Truthound Depot deployments must provision through their own private runtime # artifact. The table key keeps its legacy `dashboard_runtime` name until the # private runtime compatibility surface is renamed. # # Scope rules: # - Included extras are baseline optional-core features that the Depot console # product must be able to assume across private operator deployments. # - Excluded extras stay out of the baseline runtime because they are connector- # specific, environment-specific, or otherwise optional for Depot. # - When Depot starts depending on a new baseline optional-core feature, update # this manifest first, then sync the sibling Depot contract and private # runtime artifact generation. [tool.truthound.internal_runtime.dashboard_runtime] included_extras = ["ai", "drift", "anomaly", "reports"] excluded_extras = [ "profiler-visualization", "dashboard", "pdf", "benchmarks", "perf", "stores", "s3", "gcs", "azure", "database", "duckdb", "mongodb", "elasticsearch", "nosql", "kafka", "streaming", "async-datasources", "dev", "docs", "all", ] rationale = "Baseline dashboard runtime only; no connector extras or public packaging exposure." version_policy = "Tracks the Truthound patch version consumed by the current dashboard release line." [project.scripts] truthound-mcp = "truthound.mcp.__main__:main" truthound = "truthound.cli:app" [project.urls] Documentation = "https://truthound.netlify.app/" "Local MCP" = "https://truthound.netlify.app/guides/mcp/" "Claude marketplace" = "https://github.com/seadonggyun4/styxis-plugin" Homepage = "https://github.com/seadonggyun4/Truthound" Repository = "https://github.com/seadonggyun4/Truthound" Issues = "https://github.com/seadonggyun4/Truthound/issues" [tool.hatch.build.targets.wheel] packages = ["src/truthound"] exclude = ["/internal-development/**", "/docs-private/**"] [tool.hatch.build.targets.wheel.force-include] "src/truthound/mcp/validator_contracts.json" = "truthound/mcp/validator_contracts.json" "src/truthound/ai/data/prompt_lexicon.ko.json" = "truthound/ai/data/prompt_lexicon.ko.json" "src/truthound/ai/data/validation_dsl_coverage.v1.json" = "truthound/ai/data/validation_dsl_coverage.v1.json" "src/truthound/schema_audit/schemas/document-v1.json" = "truthound/schema_audit/schemas/document-v1.json" [tool.hatch.build.targets.sdist] # Generated verification output is not source or a runtime package resource. exclude = ["/dist/**", "/verification/**", "/internal-development/**", "/docs-private/**", "**/__pycache__/**", "**/*.py[cod]"] [tool.hatch.build.targets.sdist.force-include] "src/truthound/mcp/validator_contracts.json" = "src/truthound/mcp/validator_contracts.json" "src/truthound/ai/data/prompt_lexicon.ko.json" = "src/truthound/ai/data/prompt_lexicon.ko.json" "src/truthound/ai/data/validation_dsl_coverage.v1.json" = "src/truthound/ai/data/validation_dsl_coverage.v1.json" "src/truthound/schema_audit/schemas/document-v1.json" = "src/truthound/schema_audit/schemas/document-v1.json" [tool.hatch.envs.default] dependencies = [ "pytest>=8.0.0", "pytest-cov>=4.0.0", "ruff>=0.4.0", "mypy>=1.10.0", "pandas>=2.0.0", ] [tool.hatch.envs.default.scripts] test = "pytest {args:tests}" test-cov = "pytest --cov=truthound --cov-report=term-missing {args:tests}" lint = "ruff check src tests" format = "ruff format src tests" typecheck = "mypy src" [tool.ruff] target-version = "py311" line-length = 100 src = ["src", "tests"] [tool.ruff.lint] select = [ "E", # pycodestyle errors "W", # pycodestyle warnings "F", # Pyflakes "I", # isort "UP", # pyupgrade "B", # flake8-bugbear "SIM", # flake8-simplify "TCH", # flake8-type-checking ] ignore = [ "E501", # line too long (handled by formatter) ] [tool.ruff.lint.isort] known-first-party = ["truthound"] [tool.mypy] python_version = "3.11" strict = true warn_return_any = true warn_unused_ignores = true [tool.pytest.ini_options] testpaths = ["tests"] pythonpath = ["src"] addopts = "--import-mode=importlib" markers = [ "live_account: optional live-account or paid API verification, reported separately from PASS", "optional_consumer: existing optional sibling-consumer verification, not core PASS", "contract: deterministic public-contract regression coverage", "fault: deterministic failure injection, timeout, corruption, and concurrency coverage", "integration: tests requiring external services or backend integration", "soak: long-running or probabilistic chaos coverage intended for nightly execution", "manual_verification: manual verification artifacts that must not run under pytest by default", "slow: slow tests that are skipped by the fast default lane", "e2e: short end-to-end scenarios allowed in the default PR lane", "scale_100m: large-scale tests intended for nightly execution", "expensive: tests with elevated runtime or infrastructure cost", "stress: load-oriented stress tests intended for nightly execution", ]