pyproject.toml

[build-system]
requires = [
  "hatchling>=1.8.0",
]
build-backend = "hatchling.build"

[project]
name = "haystack-ai"
dynamic = [
  "version",
]
description = "LLM framework to build customizable, production-ready LLM applications. Connect components (models, vector DBs, file converters) to pipelines or agents that can interact with your data."
readme = "README.md"
license = "Apache-2.0"
requires-python = ">=3.8"
authors = [
  { name = "deepset.ai", email = "malte.pietsch@deepset.ai" },
]
keywords = [
  "BERT",
  "QA",
  "Question-Answering",
  "Reader",
  "Retriever",
  "albert",
  "language-model",
  "mrc",
  "roberta",
  "search",
  "semantic-search",
  "squad",
  "transfer-learning",
  "transformer",
]
classifiers = [
  "Development Status :: 5 - Production/Stable",
  "Intended Audience :: Science/Research",
  "License :: Freely Distributable",
  "License :: OSI Approved :: Apache Software License",
  "Operating System :: OS Independent",
  "Programming Language :: Python",
  "Programming Language :: Python :: 3",
  "Programming Language :: Python :: 3.8",
  "Programming Language :: Python :: 3.9",
  "Programming Language :: Python :: 3.10",
  "Topic :: Scientific/Engineering :: Artificial Intelligence",
]
dependencies = [
  "pandas",
  "haystack-bm25",
  "tqdm",
  "tenacity",
  "lazy-imports",
  "openai>=1.1.0",
  "Jinja2",
  "posthog",  # telemetry
  "pyyaml",
  "more-itertools",  # TextDocumentSplitter
  "networkx", # Pipeline graphs
  "typing_extensions>=4.7", # typing support for Python 3.8
  "boilerpy3", # Fulltext extraction from HTML pages
  "requests",
  "numpy",
  "python-dateutil",
]

[tool.hatch.envs.default]
dependencies = [
  "pre-commit",
  # Type check
  "mypy",
  # Test
  "pytest",
  "pytest-cov",
  "pytest-custom_exit_code",  # used in the CI
  "pytest-asyncio",
  "pytest-rerunfailures",
  "responses",
  "tox",
  "coverage",
  "python-multipart",
  "psutil",
  # Linting
  "pylint",
  "ruff",
  # Documentation
  "toml",
  "reno",
  # dulwich is a reno dependency, they pin it at >=0.15.0 so pip takes ton of time to resolve the dependency tree.
  # We pin it here to avoid taking too much time.
  # https://opendev.org/openstack/reno/src/branch/master/requirements.txt#L7
  "dulwich>=0.21.0,<1.0.0",
  # Version specified following Black stability policy:
  # https://black.readthedocs.io/en/stable/the_black_code_style/index.html#stability-policy
  "black[jupyter]~=23.0",
]

[tool.hatch.envs.default.scripts]
format = "black ."
format-check = "black --check ."

[tool.hatch.envs.test]
extra-dependencies = [
  "transformers[torch,sentencepiece]==4.38.2",  # ExtractiveReader, TransformersSimilarityRanker, LocalWhisperTranscriber, HFGenerators...
  "huggingface_hub>=0.22.0", # TGI Generators and TEI Embedders
  "spacy>=3.7,<3.8",  # NamedEntityExtractor
  "spacy-curated-transformers>=0.2,<=0.3",  # NamedEntityExtractor
  "en-core-web-trf @ https://github.com/explosion/spacy-models/releases/download/en_core_web_trf-3.7.3/en_core_web_trf-3.7.3-py3-none-any.whl",  # NamedEntityExtractor

  # Converters
  "pypdf",  # PyPDFConverter
  "markdown-it-py",  # MarkdownToDocument
  "mdit_plain",  # MarkdownToDocument
  "tika",  # TikaDocumentConverter
  "azure-ai-formrecognizer>=3.2.0b2",  # AzureOCRDocumentConverter
  "langdetect",  # TextLanguageRouter and DocumentLanguageClassifier
  "sentence-transformers>=2.2.0",  # SentenceTransformersTextEmbedder and SentenceTransformersDocumentEmbedder
  "openai-whisper>=20231106",  # LocalWhisperTranscriber
  "chroma-haystack",  # pipeline predefined templates

  # OpenAPI
  "jsonref",  # OpenAPIServiceConnector, OpenAPIServiceToFunctions
  "openapi3",

  # Validation
  "jsonschema",

  # Tracing
  "opentelemetry-sdk",
  "ddtrace",

  # Structured logging
  "structlog",

  # Looking for missing imports
  "isort",
  "pyproject-parser",
]

[tool.hatch.envs.test.scripts]
e2e = "pytest e2e"
unit = 'pytest --cov-report xml:coverage.xml --cov="haystack" -m "not integration" {args:test}'
integration = 'pytest --maxfail=5 -m "integration" {args:test}'
integration-mac = 'pytest --maxfail=5 -m "integration" -k "not tika" {args:test}'
integration-windows = 'pytest --maxfail=5 -m "integration" -k "not tika" {args:test}'
types = "mypy --install-types --non-interactive --cache-dir=.mypy_cache/ {args:haystack}"
lint = [
  "ruff check {args:haystack}",
  "pylint -ry -j 0 {args:haystack}"
]
lint-fix = [
  "black .",
  "ruff check {args:haystack} --fix",
]

[tool.hatch.envs.readme]
detached = true  # To avoid installing the dependencies from the default environment
dependencies = [
  "haystack-pydoc-tools",
]

[tool.hatch.envs.readme.scripts]
sync = "./.github/utils/pydoc-markdown.sh"
delete-outdated = "python ./.github/utils/delete_outdated_docs.py {args}"

[tool.hatch.envs.snippets]
extra-dependencies = [
  "torch",
  "pydantic",
]

[project.urls]
"CI: GitHub" = "https://github.com/deepset-ai/haystack/actions"
"Docs: RTD" = "https://haystack.deepset.ai/overview/intro"
"GitHub: issues" = "https://github.com/deepset-ai/haystack/issues"
"GitHub: repo" = "https://github.com/deepset-ai/haystack"
Homepage = "https://github.com/deepset-ai/haystack"

[tool.hatch.version]
path = "VERSION.txt"
pattern = "(?P<version>.+)"

[tool.hatch.metadata]
allow-direct-references = true

[tool.hatch.build.targets.sdist]
include = [
  "/haystack",
  "/VERSION.txt",
]

[tool.hatch.build.targets.wheel]
packages = [
  "haystack",
]

[tool.black]
line-length = 120
skip_magic_trailing_comma = true  # For compatibility with pydoc>=4.6, check if still needed.

[tool.codespell]
ignore-words-list = "ans,astroid,nd,ned,nin,ue,rouge,ist"
quiet-level = 3
skip = "test/nodes/*,test/others/*,test/samples/*"

[tool.pylint.'MESSAGES CONTROL']
max-line-length=120
disable = [

  # To keep
  "fixme",
  "c-extension-no-member",

  # To review:
  "missing-docstring",
  "unused-argument",
  "no-member",
  "line-too-long",
  "protected-access",
  "too-few-public-methods",
  "raise-missing-from",
  "invalid-name",
  "duplicate-code",
  "arguments-differ",
  "consider-using-f-string",
  "no-else-return",
  "attribute-defined-outside-init",
  "super-with-arguments",
  "redefined-builtin",
  "abstract-method",
  "unspecified-encoding",
  "unidiomatic-typecheck",
  "no-name-in-module",
  "consider-using-with",
  "redefined-outer-name",
  "arguments-renamed",
  "unnecessary-pass",
  "broad-except",
  "unnecessary-comprehension",
  "subprocess-run-check",
  "singleton-comparison",
  "consider-iterating-dictionary",
  "undefined-loop-variable",
  "consider-using-in",
  "bare-except",
  "unexpected-keyword-arg",
  "simplifiable-if-expression",
  "use-list-literal",
  "broad-exception-raised",

  # To review later
  "cyclic-import",
  "import-outside-toplevel",
  "deprecated-method",
]
[tool.pylint.'DESIGN']
max-args = 38  # Default is 5
max-attributes = 28  # Default is 7
max-branches = 34  # Default is 12
max-locals = 45  # Default is 15
max-module-lines = 2468  # Default is 1000
max-nested-blocks = 9  # Default is 5
max-statements = 206  # Default is 50
[tool.pylint.'SIMILARITIES']
min-similarity-lines=6

[tool.pytest.ini_options]
minversion = "6.0"
addopts = "--strict-markers"
markers = [
  "unit: unit tests",
  "integration: integration tests",

  "generator: generator tests",
  "summarizer: summarizer tests",
  "embedding_dim: uses a document store with non-default embedding dimension (e.g @pytest.mark.embedding_dim(128))",

  "tika: requires Tika container",
  "parsr: requires Parsr container",
  "ocr: requires Tesseract",

  "elasticsearch: requires Elasticsearch container",
  "weaviate: requires Weaviate container",
  "pinecone: requires Pinecone credentials",
  "faiss: uses FAISS",
  "opensearch",
  "document_store",
]
log_cli = true

[tool.mypy]
warn_return_any = false
warn_unused_configs = true
ignore_missing_imports = true

[tool.ruff]
line-length = 301
target-version = "py38"

[tool.ruff.lint]
select = [
  "AIR",    # Airflow
  "ASYNC",  # flake8-async
  "C4",     # flake8-comprehensions
  "C90",    # McCabe cyclomatic complexity
  "DJ",     # flake8-django
  "E501",   # Long lines
  "EXE",    # flake8-executable
  "F",      # Pyflakes
  "INT",    # flake8-gettext
  "PERF",   # Perflint
  "PL",     # Pylint
  "Q",      # flake8-quotes
  "SIM",    # flake8-simplify
  "SLOT",   # flake8-slots
  "T10",    # flake8-debugger
  "W",      # pycodestyle
  "YTT",    # flake8-2020
  "I"       # isort
  # "E",    # pycodestyle
  # "NPY",  # NumPy-specific rules
  # "PD",   # pandas-vet
  # "PT",   # flake8-pytest-style
  # "UP",   # pyupgrade
]

ignore = [
  "F401",     # unused-import
  "PERF203",	# `try`-`except` within a loop incurs performance overhead
  "PERF401",	# Use a list comprehension to create a transformed list
  "PLR1714",  # repeated-equality-comparison
  "PLR5501",  # collapsible-else-if
  "PLW0603",  # global-statement
  "PLW1510",  # subprocess-run-without-check
  "PLW2901",  # redefined-loop-name
  "SIM108",   # if-else-block-instead-of-if-exp
  "SIM115",   # open-file-with-context-handler
  "SIM118",   # in-dict-keys
]

[tool.ruff.lint.mccabe]
max-complexity = 28

[tool.ruff.lint.per-file-ignores]
"examples/basic_qa_pipeline.py" = ["C416"]
"haystack/preview/testing/document_store.py" = ["C416", "F821"]
"haystack/telemetry.py" = ["F821"]

[tool.ruff.lint.pylint]
allow-magic-value-types = ["float", "int", "str"]
max-args = 14  # Default is 5
max-branches = 21  # Default is 12
max-public-methods = 20  # Default is 20
max-returns = 7  # Default is 6
max-statements = 60  # Default is 50

[tool.coverage.run]
omit = [
    "haystack/testing/*",
]