Merge pull request #311753 from r-ryantm/auto-update/python311Packages.argilla

{ lib

, buildPythonPackage

, fetchFromGitHub

, pythonOlder

, pythonRelaxDepsHook

, deprecated

, rich

, backoff

, packaging

, pydantic

, typer

, tqdm

, wrapt

, numpy

, httpx

, pandas

, monotonic

# optional-dependencies

, fastapi

, opensearch-py

, elasticsearch8

, uvicorn

, smart-open

, brotli-asgi

, alembic

, sqlalchemy

, greenlet

, aiosqlite

, luqum

, scikit-learn

, aiofiles

, pyyaml

, python-multipart

, python-jose

, passlib

, psutil

# , segment-analytics-python

, asyncpg

, psycopg2

, schedule

, prodict

, cleanlab

, datasets

, huggingface-hub

{

  lib,

  aiofiles,

  aiosqlite,

  alembic,

  asyncpg,

  backoff,

  brotli-asgi,

  buildPythonPackage,

  cleanlab,

  datasets,

  deprecated,

  elasticsearch8,

  evaluate,

  factory-boy,

  faiss,

  fastapi,

  fetchFromGitHub,

  flyingsquid,

  greenlet,

  httpx,

  huggingface-hub,

  luqum,

  monotonic,

  numpy,

  openai,

  opensearch-py,

  packaging,

  pandas,

  passlib,

  peft,

  pgmpy,

  plotly,

  prodict,

  psutil,

  psycopg2,

  pydantic,

  pytest-asyncio,

  pytest-mock,

  pytestCheckHook,

  python-jose,

  python-multipart,

  pythonOlder,

  pythonRelaxDepsHook,

  pyyaml,

  rich,

  schedule,

  scikit-learn,

  sentence-transformers,

  seqeval,

  setuptools,

  smart-open,

  snorkel,

  spacy,

  spacy-transformers,

  sqlalchemy,

  tqdm,

  transformers,

  typer,

  uvicorn,

  wrapt,

# , flair

, faiss

, flyingsquid

, pgmpy

, plotly

, snorkel

, spacy

, transformers

, evaluate

, seqeval

# , setfit

# , spacy-huggingface-hub

# , span_marker

, openai

, peft

# test dependencies

, pytestCheckHook

, pytest-cov

, pytest-mock

, pytest-asyncio

, factory-boy

# , trl

}:

let

buildPythonPackage rec {

  pname = "argilla";

  version = "1.27.0";

  optional-dependencies = {

    server = [

      fastapi

      opensearch-py

      elasticsearch8

      uvicorn

      smart-open

      brotli-asgi

      alembic

      sqlalchemy

      greenlet

      aiosqlite

      luqum

      scikit-learn

      aiofiles

      pyyaml

      python-multipart

      python-jose

      passlib

      psutil

      # segment-analytics-python

    ] ++

      elasticsearch8.optional-dependencies.async ++

      uvicorn.optional-dependencies.standard ++

      python-jose.optional-dependencies.cryptography ++

      passlib.optional-dependencies.bcrypt;

    postgresql = [ asyncpg psycopg2 ];

    listeners = [ schedule prodict ];

    integrations = [

      pyyaml

      cleanlab

      datasets

      huggingface-hub

      # flair

      faiss

      flyingsquid

      pgmpy

      plotly

      snorkel

      spacy

      transformers

      evaluate

      seqeval

      # setfit

      # span_marker

      openai

      peft

    ] ++ transformers.optional-dependencies.torch;

  };

in

buildPythonPackage {

  inherit pname version;

  format = "setuptools";

  version = "1.28.0";

  pyproject = true;

  disabled = pythonOlder "3.8";

  src = fetchFromGitHub {

    owner = "argilla-io";

    repo = pname;

    repo = "argilla";

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

    hash = "sha256-CBVP/+XFKnJBMcxsDd7lgQ1JFX7zFlHmdBwkAMmq85g=";

    hash = "sha256-gQpJ2umi3IE5BhRu3bM7ONPIP0hb2YG37jGvDKQHZWA=";

  };

  pythonRelaxDeps = [

    "typer"

    "rich"

    "httpx"

    "numpy"

    "rich"

    "typer"

    "wrapt"

  ];

  nativeBuildInputs = [

    pythonRelaxDepsHook

  ];

  build-system = [ setuptools ];

  propagatedBuildInputs = [

  nativeBuildInputs = [ pythonRelaxDepsHook ];

  dependencies = [

    httpx

    deprecated

    packaging

    typer

  ];

  # still quite a bit of optional dependencies missing

  passthru.optional-dependencies = {

    server =

      [

        aiofiles

        aiosqlite

        alembic

        brotli-asgi

        elasticsearch8

        fastapi

        greenlet

        luqum

        opensearch-py

        passlib

        psutil

        python-jose

        python-multipart

        pyyaml

        scikit-learn

        smart-open

        sqlalchemy

        uvicorn

      ]

      ++ elasticsearch8.optional-dependencies.async

      ++ uvicorn.optional-dependencies.standard

      ++ python-jose.optional-dependencies.cryptography

      ++ passlib.optional-dependencies.bcrypt;

    postgresql = [

      asyncpg

      psycopg2

    ];

    listeners = [

      schedule

      prodict

    ];

    integrations = [

      cleanlab

      datasets

      evaluate

      faiss

      flyingsquid

      huggingface-hub

      openai

      peft

      pgmpy

      plotly

      pyyaml

      sentence-transformers

      seqeval

      snorkel

      spacy

      spacy-transformers

      transformers

      # flair

      # setfit

      # span_marker

      # trl

      # spacy-huggingface-hub

    ] ++ transformers.optional-dependencies.torch;

  };

  # Still quite a bit of optional dependencies missing

  doCheck = false;

  preCheck = ''

  nativeCheckInputs = [

    pytestCheckHook

    pytest-cov

    pytest-mock

    pytest-asyncio

    factory-boy

  ]

    ++ optional-dependencies.server

    ++ optional-dependencies.postgresql

    ++ optional-dependencies.listeners

    ++ optional-dependencies.integrations;

  pytestFlagsArray = [ "--ignore=tests/server/datasets/test_dao.py" ];

  ] ++ lib.flatten (builtins.attrValues passthru.optional-dependencies);

  passthru.optional-dependencies = optional-dependencies;

  disabledTestPaths = [ "tests/server/datasets/test_dao.py" ];

  meta = with lib; {

    description = "Argilla: the open-source data curation platform for LLMs";

    mainProgram = "argilla";

    description = "Open-source data curation platform for LLMs";

    homepage = "https://github.com/argilla-io/argilla";

    changelog = "https://github.com/argilla-io/argilla/releases/tag/v${version}";

    license = licenses.asl20;

    maintainers = with maintainers; [ happysalada ];

    mainProgram = "argilla";

  };

}