pkgs/development/python-modules/llama-index-core/default.nix

   1 {
   2   lib,
   3   aiohttp,
   4   buildPythonPackage,
   5   dataclasses-json,
   6   deprecated,
   7   dirtyjson,
   8   fetchFromGitHub,
   9   fsspec,
  10   jsonpath-ng,
  11   llamaindex-py-client,
  12   nest-asyncio,
  13   networkx,
  14   nltk,
  15   nltk-data,
  16   numpy,
  17   openai,
  18   pandas,
  19   pillow,
  20   poetry-core,
  21   pytest-asyncio,
  22   pytest-mock,
  23   pytestCheckHook,
  24   pythonOlder,
  25   pyvis,
  26   pyyaml,
  27   requests,
  28   spacy,
  29   sqlalchemy,
  30   tenacity,
  31   tiktoken,
  32   tree-sitter,
  33   typing-inspect,
  34 }:
  35
  36 buildPythonPackage rec {
  37   pname = "llama-index-core";
  38   version = "0.11.20";
  39   pyproject = true;
  40
  41   disabled = pythonOlder "3.8";
  42
  43   src = fetchFromGitHub {
  44     owner = "run-llama";
  45     repo = "llama_index";
  46     rev = "refs/tags/v${version}";
  47     hash = "sha256-r4xedtxoYv6CcxtDrgwau9LY3kOBg3jXlQm1g59L7x4=";
  48   };
  49
  50   sourceRoot = "${src.name}/${pname}";
  51
  52   # When `llama-index` is imported, it uses `nltk` to look for the following files and tries to
  53   # download them if they aren't present.
  54   # https://github.com/run-llama/llama_index/blob/6efa53cebd5c8ccf363582c932fffde44d61332e/llama-index-core/llama_index/core/utils.py#L59-L67
  55   # Setting `NLTK_DATA` to a writable path can also solve this problem, but it needs to be done in
  56   # every package that depends on `llama-index-core` for `pythonImportsCheck` not to fail, so this
  57   # solution seems more elegant.
  58   postPatch = ''
  59     mkdir -p llama_index/core/_static/nltk_cache/corpora/stopwords/
  60     cp -r ${nltk-data.stopwords}/corpora/stopwords/* llama_index/core/_static/nltk_cache/corpora/stopwords/
  61
  62     mkdir -p llama_index/core/_static/nltk_cache/tokenizers/punkt/
  63     cp -r ${nltk-data.punkt}/tokenizers/punkt/* llama_index/core/_static/nltk_cache/tokenizers/punkt/
  64   '';
  65
  66   pythonRelaxDeps = [ "tenacity" ];
  67
  68   build-system = [ poetry-core ];
  69
  70   dependencies = [
  71     aiohttp
  72     dataclasses-json
  73     deprecated
  74     dirtyjson
  75     fsspec
  76     jsonpath-ng
  77     llamaindex-py-client
  78     nest-asyncio
  79     networkx
  80     nltk
  81     numpy
  82     openai
  83     pandas
  84     pillow
  85     pyvis
  86     pyyaml
  87     requests
  88     spacy
  89     sqlalchemy
  90     tenacity
  91     tiktoken
  92     typing-inspect
  93   ];
  94
  95   nativeCheckInputs = [
  96     tree-sitter
  97     pytest-asyncio
  98     pytest-mock
  99     pytestCheckHook
 100   ];
 101
 102   pythonImportsCheck = [ "llama_index" ];
 103
 104   disabledTestPaths = [
 105     # Tests require network access
 106     "tests/agent/"
 107     "tests/callbacks/"
 108     "tests/chat_engine/"
 109     "tests/evaluation/"
 110     "tests/indices/"
 111     "tests/ingestion/"
 112     "tests/memory/"
 113     "tests/node_parser/"
 114     "tests/objects/"
 115     "tests/playground/"
 116     "tests/postprocessor/"
 117     "tests/query_engine/"
 118     "tests/question_gen/"
 119     "tests/response_synthesizers/"
 120     "tests/retrievers/"
 121     "tests/selectors/"
 122     "tests/test_utils.py"
 123     "tests/text_splitter/"
 124     "tests/token_predictor/"
 125     "tests/tools/"
 126   ];
 127
 128   disabledTests = [
 129     # Tests require network access
 130     "test_from_namespaced_persist_dir"
 131     "test_from_persist_dir"
 132     # asyncio.exceptions.InvalidStateError: invalid state
 133     "test_workflow_context_to_dict_mid_run"
 134   ];
 135
 136   meta = with lib; {
 137     description = "Data framework for your LLM applications";
 138     homepage = "https://github.com/run-llama/llama_index/";
 139     changelog = "https://github.com/run-llama/llama_index/blob/${version}/CHANGELOG.md";
 140     license = licenses.mit;
 141     maintainers = with maintainers; [ fab ];
 142   };
 143 }