Merge pull request #326690 from natsukium/graphrag/init

{

  lib,

  buildPythonPackage,

  fetchFromGitHub,

  setuptools,

  azure-common,

  azure-core,

  isodate,

  pytestCheckHook,

}:

buildPythonPackage rec {

  pname = "azure-search-documents";

  version = "11.4.0";

  pyproject = true;

  src = fetchFromGitHub {

    owner = "Azure";

    repo = "azure-sdk-for-python";

    rev = "azure-search-documents_${version}";

    hash = "sha256-0J9AXDH7TOkcKDwFbICiMatLAwiFq3Jtoji8fJSOg8k=";

  };

  sourceRoot = "${src.name}/sdk/search/azure-search-documents";

  build-system = [ setuptools ];

  dependencies = [

    azure-common

    azure-core

    isodate

  ];

  pythonImportsCheck = [ "azure.search.documents" ];

  # require devtools_testutils which is a internal package for azure-sdk

  doCheck = false;

  meta = {

    description = "Microsoft Azure Cognitive Search Client Library for Python";

    homepage = "https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/search/azure-search-documents";

    changelog = "https://github.com/Azure/azure-sdk-for-python/blob/${src.rev}/sdk/search/azure-search-documents/CHANGELOG.md";

    license = lib.licenses.mit;

    maintainers = with lib.maintainers; [ natsukium ];

  };

}

{

  lib,

  buildPythonPackage,

  fetchPypi,

  pythonOlder,

  pythonRelaxDepsHook,

  poetry-core,

  dacite,

  diskcache,

  jsonschema,

  pandas,

  pyarrow,

}:

buildPythonPackage rec {

  pname = "datashaper";

  version = "0.0.49";

  pyproject = true;

  disabled = pythonOlder "3.10";

  src = fetchPypi {

    inherit pname version;

    hash = "sha256-Bb+6WWRHSmK91SWew/oBc9AeNlIItqSv9OoOYwlqdTM=";

  };

  build-system = [ poetry-core ];

  nativeBuildInputs = [ pythonRelaxDepsHook ];

  pythonRelaxDeps = [ "pyarrow" ];

  dependencies = [

    dacite

    diskcache

    jsonschema

    pandas

    pyarrow

  ];

  pythonImportsCheck = [ "datashaper" ];

  # pypi tarball has no tests

  doCheck = false;

  meta = {

    description = "Collection of utilities for doing lightweight data wrangling";

    homepage = "https://github.com/microsoft/datashaper/tree/main/python/datashaper";

    license = lib.licenses.mit;

    maintainers = with lib.maintainers; [ natsukium ];

  };

}

{

  lib,

  buildPythonPackage,

  fetchFromGitHub,

  poetry-core,

  poetry-dynamic-versioning,

  aiofiles,

  aiolimiter,

  azure-identity,

  azure-search-documents,

  azure-storage-blob,

  datashaper,

  devtools,

  environs,

  fastparquet,

  graspologic,

  lancedb,

  networkx,

  nltk,

  numba,

  numpy,

  openai,

  pyaml-env,

  pydantic,

  python-dotenv,

  pyyaml,

  rich,

  scipy,

  swifter,

  tenacity,

  textual,

  tiktoken,

  typing-extensions,

  uvloop,

  nbformat,

  pytest-asyncio,

  pytestCheckHook,

}:

buildPythonPackage rec {

  pname = "graphrag";

  version = "0.1.1";

  pyproject = true;

  src = fetchFromGitHub {

    owner = "microsoft";

    repo = "graphrag";

    rev = "refs/tags/v${version}";

    hash = "sha256-hIAQOIqm9S9AtssE6UxcXfaIbSt3+506ueMrlathNaQ=";

  };

  build-system = [

    poetry-core

    poetry-dynamic-versioning

  ];

  pythonRelaxDeps = [

    "aiofiles"

    "azure-identity"

    "scipy"

    "tiktoken"

  ];

  dependencies = [

    aiofiles

    aiolimiter

    azure-identity

    azure-search-documents

    azure-storage-blob

    datashaper

    devtools

    environs

    fastparquet

    graspologic

    lancedb

    networkx

    nltk

    numba

    numpy

    openai

    pyaml-env

    pydantic

    python-dotenv

    pyyaml

    rich

    scipy

    swifter

    tenacity

    textual

    tiktoken

    typing-extensions

    uvloop

  ];

  env.NUMBA_CACHE_DIR = "$TMPDIR";

  pythonImportsCheck = [ "graphrag" ];

  nativeCheckInputs = [

    nbformat

    pytest-asyncio

    pytestCheckHook

  ];

  pytestFlagsArray = [ "tests/unit" ];

  disabledTests = [

    # touch the network

    "test_child"

    "test_dotprefix"

    "test_find"

    "test_run_extract_entities_multiple_documents"

    "test_run_extract_entities_single_document"

  ];

  meta = {

    description = "Modular graph-based Retrieval-Augmented Generation (RAG) system";

    homepage = "https://github.com/microsoft/graphrag";

    license = lib.licenses.mit;

    maintainers = with lib.maintainers; [ natsukium ];

  };

}

{

  lib,

  buildPythonPackage,

  fetchFromGitHub,

  setuptools,

  pyyaml,

  pytestCheckHook,

}:

buildPythonPackage rec {

  pname = "pyaml-env";

  version = "1.2.1";

  pyproject = true;

  src = fetchFromGitHub {

    owner = "mkaranasou";

    repo = "pyaml_env";

    rev = "refs/tags/v${version}";

    hash = "sha256-xSu+dksSVugShJwOqedXBrXIKaH0G5JAsynauOuP3OA=";

  };

  build-system = [ setuptools ];

  dependencies = [ pyyaml ];

  pythonImportsCheck = [ "pyaml_env" ];

  nativeCheckInputs = [ pytestCheckHook ];

  meta = {

    description = "Parse YAML configuration with environment variables in Python";

    homepage = "https://github.com/mkaranasou/pyaml_env";

    changelog = "https://github.com/mkaranasou/pyaml_env/releases/tag/v${version}";

    license = lib.licenses.mit;

    maintainers = with lib.maintainers; [ natsukium ];

  };

}

{

  lib,

  buildPythonPackage,

  fetchFromGitHub,

  setuptools,

  dask,

  pandas,

  psutil,

  tqdm,

  ipywidgets,

  ray,

}:

buildPythonPackage rec {

  pname = "swifter";

  version = "1.4.0";

  pyproject = true;

  src = fetchFromGitHub {

    owner = "jmcarpenter2";

    repo = "swifter";

    rev = "refs/tags/${version}";

    hash = "sha256-lgdf8E9GGjeLY4ERzxqtjQuYVtdtIZt2HFLSiNBbtX4=";

  };

  build-system = [ setuptools ];

  dependencies = [

    pandas

    psutil

    dask

    tqdm

  ] ++ dask.optional-dependencies.dataframe;

  optional-dependencies = {

    groupby = [ ray ];

    notebook = [ ipywidgets ];

  };

  pythonImportsCheck = [ "swifter" ];

  # tests may hang due to ignoring cpu core limit

  # https://github.com/jmcarpenter2/swifter/issues/221

  doCheck = false;

  meta = {

    description = "Package which efficiently applies any function to a pandas dataframe or series in the fastest available manner";

    homepage = "https://github.com/jmcarpenter2/swifter";

    license = lib.licenses.mit;

    maintainers = with lib.maintainers; [ natsukium ];

  };

}