nixpkgs-extension/pkgs/by-category/pythonPackages/prefab-specs.nix

{
  lib,
  hatchling,
  poetry-core,

  huggingface-hub,
  tqdm,
  aiohttp,
  configparser,
  google-api-core,
  google-genai,
  json-repair,
  networkx,
  numpy,
  pandas,
  pydantic,
  pypinyin,
  python-dotenv,
  tenacity,
  tiktoken,
  xlsxwriter,
  boto3,
  click,
  loguru,
  pdfminer-six,
  requests,
  httpx,
  pillow,
  pypdfium2,
  pypdf,
  reportlab,
  modelscope,
  opencv-python,
  scikit-image,
  openai,
  beautifulsoup4,
  magika,
  av,
  pdm-backend,
  fasttext-predict,
  pydantic-settings,
  aiofiles,
  hatch-fancy-pypi-readme,
  colorlog,
  torch,
  torchvision,
  wcwidth,
  matplotlib,
  pyyaml,
  scipy,
  psutil,
  py-cpuinfo,
  seaborn,
  albumentations,
  transformers,
  accelerate,
  ultralytics,
  dill,
  ftfy,
  shapely,
  pyclipper,
  omegaconf,
  onnxruntime,
  fastapi,
  python-multipart,
  uvicorn,
  gradio,
  gradio-pdf,
  ultralytics-thop,
}:

self:
let
  inherit (self) # locally defined packages
    lightrag-hku
    mineru
    nano-vectordb
    pipmaster
    ascii-colors
    pdftext
    fast-langdetect
    mineru-vl-utils
    qwen-vl-utils
    httpx-retries
    robust-downloader
    doclayout-yolo
    # decord
    ;
in
{
  raganything = {
    url = "https://github.com/HKUDS/RAG-Anything/tree/v1.2.9";
    hash = "sha256-yepiLYzPD6UcJRbAbovg/BwFE8nh903o/tHypiHGKSw=";
    propagatedBuildDeps = [
      huggingface-hub
      lightrag-hku
      # mineru
      (mineru.optionalDeps [ "core" ])
      tqdm
    ];
    pythonImportsCheck = [ "raganything" ];
    meta = {
      description = "RAGAnything: All-in-One RAG System";
      license = lib.licenses.mit;
    };
  };
  lightrag-hku = {
    url = "https://github.com/HKUDS/LightRAG/tree/v1.4.9.11";
    hash = "sha256-TX/HSF2ZqoDo2SRlFzM+bkUxQXiUpnrl6kwI+lljjdo=";
    propagatedBuildDeps = [
      aiohttp
      configparser
      google-api-core
      google-genai
      json-repair
      nano-vectordb
      networkx
      numpy
      pandas
      pipmaster
      pydantic
      pypinyin
      python-dotenv
      tenacity
      tiktoken
      xlsxwriter
    ];
    meta = {
      description = "LightRAG: Simple and Fast Retrieval-Augmented Generation";
      license = lib.licenses.mit;
    };
  };
  pipmaster = {
    url = "https://github.com/ParisNeo/pipmaster/tree/820acdbc9d541443942bd8afd6ba968036bd8452";
    hash = "sha256-H1R/hXPyjn1r6Dal0QMEQeBR5xUlgacxFsuDUZVwV+s=";
    force.version = "1.1.0";
    propagatedBuildDeps = [
      ascii-colors
    ];
    meta = with lib; {
      description = "A versatile Python package manager utility for simplifying package installation, updates, checks, and environment management.";
      license = licenses.asl20;
    };
  };
  nano-vectordb = {
    url = "https://github.com/gusye1234/nano-vectordb/tree/03f821348f04a93a9c36bb575faae05b61f4c02e";
    hash = "sha256-jLco+1TAncF8Ep+VGd3DhsoiZTW7G/H8fAEwnNsovTY=";
    force.version = "0.0.4.3";
    propagatedBuildDeps = [
      numpy
    ];
    meta = {
      description = "A simple, easy-to-hack Vector Database implementation";
      license = lib.licenses.mit;
    };
  };
  ascii-colors = {
    url = "https://github.com/ParisNeo/ascii_colors/tree/817a21485136635e3da89ff08444183254b25aab";
    hash = "sha256-zK4vM2sOfEFDJBpFjcJmQXljg3pgprP+VFuuk5JNWnE=";
    force.version = "0.11.12";
    propagatedBuildDeps = [
      wcwidth
    ];
    meta = {
      description = "A python library for displaying stuff on the console in a pretty way";
      license = lib.licenses.asl20;
    };
  };
  mineru = {
    url = "https://github.com/opendatalab/MinerU/tree/mineru-2.7.6-released";
    hash = "sha256-A/nFNQYGEfmWUdpX8N1lbex3mdiF8+eN8s0UKQFc42E=";
    propagatedBuildDeps = [
      boto3
      click
      loguru
      numpy
      pdfminer-six
      tqdm
      requests
      httpx
      pillow
      pypdfium2
      pypdf
      reportlab
      pdftext
      modelscope
      huggingface-hub
      json-repair
      opencv-python
      fast-langdetect
      scikit-image
      openai
      beautifulsoup4
      magika
      mineru-vl-utils
      qwen-vl-utils
    ];
    optional-dependencies =
      let
        _gradio = gradio;
      in
      rec {
        vlm = [
          torch
          transformers
          accelerate
        ];
        # vllm = [
        #   vllm
        # ];
        # lmdeploy = [
        #   lmdeploy
        # ];
        # mlx = [
        #   mlx-vlm
        # ];
        pipeline = [
          matplotlib
          ultralytics
          doclayout-yolo
          dill
          pyyaml
          ftfy
          shapely
          pyclipper
          omegaconf
          torch
          torchvision
          transformers
          onnxruntime
        ];
        api = [
          fastapi
          python-multipart
          uvicorn
        ];
        gradio = [
          _gradio
          gradio-pdf
        ];
        core = vlm ++ pipeline ++ api ++ gradio;
        # all = core ++ mlx ++ vllm ++ lmdeploy;
      };
    pythonRelaxDeps = [
      "fast-langdetect"
    ];
    meta = {
      description = "Transforms complex documents like PDFs into LLM-ready markdown/JSON for your Agentic workflows.";
      license = lib.licenses.gpl3;
    };
  };
  doclayout-yolo = {
    url = "https://pypi.org/project/doclayout-yolo/0.0.4/";
    hash = "sha256-gDEdEL7QPPiExb/MYkv+D9z/bnhLp2eZNHsCkNedy00=";
    propagatedBuildDeps = [
      matplotlib
      opencv-python
      pillow
      pyyaml
      requests
      scipy
      torch
      torchvision
      tqdm
      psutil
      py-cpuinfo
      pandas
      seaborn
      albumentations
      huggingface-hub
    ]
    ++ [
      ultralytics-thop # thop
    ];
    postInstall = ''
      rm -f "$out/bin/yolo" # collision with $'{pkgs.python3Packages.ultralytics}/bin/.yolo-wrapped
    '';
    pythonRelaxDeps = [
      "thop"
    ];
    pythonRemoveDeps = [
      "thop"
    ];
    meta = {
      description = "DocLayout-YOLO: Enhancing Document Layout Analysis through Diverse Synthetic Data and Global-to-Local Adaptive Perception";
      license = lib.licenses.agpl3Only;
    };
  };
  # thop = { url = "use ultralytics-thop instead"; };
  qwen-vl-utils = {
    url = "https://github.com/QwenLM/Qwen3-VL/tree/fe12058/qwen-vl-utils";
    hash = "sha256-Vha/Tc4q2v5RCL31hB9U4ZrfIFxfZjgwM6PYlvDeoAQ=";
    force = {
      version = "0.0.14";
    };
    propagatedBuildDeps = [
      hatchling
      av
      pillow
      requests
      torch
      torchvision
    ];
    pythonImportsCheck = [ "qwen_vl_utils" ];
    optional-dependencies = {
      # decord' = [
      #   decord
      # ];
    };
    meta = {
      description = "Qwen-VL Utils contains a set of helper functions for processing and integrating visual language information with Qwen-VL Series Model.";
      license = lib.licenses.asl20;
    };
  };
  # decord = {
  #   url = "https://github.com/dmlc/decord/tree/v0.6.0/python";
  #   hash = "sha256-7YedKE0FAuZOCiQlgocZCMfkXvykxdf10ES7KwVZ1hc=";
  #   meta = {
  #     description = "An efficient video loader for deep learning with smart shuffling that's super easy to digest";
  #     license = lib.licenses.asl20;
  #   };
  # };
  mineru-vl-utils = {
    url = "https://github.com/opendatalab/mineru-vl-utils/tree/mineru_vl_utils-0.1.22-released";
    hash = "sha256-hpTW/1nwXPxfld4nx0XHZBMerWj+UL1vzDhYpwjezRU=";
    propagatedBuildDeps = [
      httpx
      httpx-retries
      aiofiles
      pillow
      pydantic
      loguru
    ];
    optional-dependencies = {
      transformers = [
        torch
        transformers
        accelerate
        torchvision
      ];
      # vllm = [
      #   vllm
      # ];
      # mlx = [
      #   mlx-vlm
      # ];
      # lmdeploy = [
      #   lmdeploy
      #   qwen-vl-utils
      # ];
    };
    meta = {
      description = "A Python package for interacting with the MinerU Vision-Language Model.";
      license = lib.licenses.gpl3;
    };
  };
  httpx-retries = {
    url = "https://github.com/will-ockmore/httpx-retries/tree/0.4.5";
    hash = "sha256-zJ3ExSEWxlHFluSdYA8/XZ3zb4KBelU+IOFyUu4ezvo=";
    propagatedBuildDeps = [
      hatchling
      hatch-fancy-pypi-readme
      httpx
    ];
    meta = {
      description = "A retry layer for HTTPX.";
      license = lib.licenses.mit;
    };
  };
  fast-langdetect = {
    url = "https://github.com/LlmKira/fast-langdetect/tree/pypi_1.0.0";
    hash = "sha256-pj46gHG9cjkSjnYc88bSctL/1LAUe0jkBuM/GZWMsUI=";
    propagatedBuildDeps = [
      pdm-backend
      robust-downloader
      requests
      fasttext-predict
    ];
    meta = {
      description = "⚡️ 80x faster Fasttext language detection out of the box | Split text by language ";
      license = lib.licenses.mit;
    };
  };
  robust-downloader = {
    url = "https://github.com/fedebotu/robust-downloader/tree/0.0.2";
    hash = "sha256-UmzfEIPiMtUkOG6sIMYgLxc8YwL5wRgMBRlywqKomv0=";
    propagatedBuildDeps = [
      tqdm
      colorlog
      requests
    ];
    meta = {
      description = "Minimal Python downloader with robustness in mind - resumable downloads, retries, and more";
      license = lib.licenses.asl20;
    };
  };
  pdftext = {
    url = "https://github.com/datalab-to/pdftext/tree/v0.6.3";
    hash = "sha256-EGVjzjDWtdcEPX//cOm5+xm9FvX0aP+h6fsD25hC8gA=";
    propagatedBuildDeps = [
      poetry-core
      click
      pydantic
      pydantic-settings
      pypdfium2
    ];
    pythonRelaxDeps = [
      "pypdfium2"
    ];
    meta = {
      description = "Extract structured text from pdfs quickly";
      license = lib.licenses.asl20;
    };
  };
}