pkgs/servers/search/khoj/default.nix

   1 { lib
   2 , stdenv
   3 , fetchFromGitHub
   4 , python3
   5 , postgresql
   6 , postgresqlTestHook
   7 }:
   8
   9 python3.pkgs.buildPythonApplication rec {
  10   pname = "khoj";
  11   version = "1.0.1";
  12   pyproject = true;
  13
  14   src = fetchFromGitHub {
  15     owner = "debanjum";
  16     repo = "khoj";
  17     rev = "refs/tags/${version}";
  18     hash = "sha256-lvOeYTrvW5MfhuJ3lj9n9TRlvpRwVP2vFeaEeJdqIec=";
  19   };
  20
  21   env = {
  22     DJANGO_SETTINGS_MODULE = "khoj.app.settings";
  23     postgresqlEnableTCP = 1;
  24   };
  25
  26   nativeBuildInputs = with python3.pkgs; [
  27     hatch-vcs
  28     hatchling
  29   ];
  30
  31   propagatedBuildInputs = with python3.pkgs; [
  32     aiohttp
  33     anyio
  34     authlib
  35     beautifulsoup4
  36     dateparser
  37     defusedxml
  38     django
  39     fastapi
  40     google-auth
  41     # gpt4all
  42     gunicorn
  43     httpx
  44     itsdangerous
  45     jinja2
  46     langchain
  47     lxml
  48     openai
  49     openai-whisper
  50     pgvector
  51     pillow
  52     psycopg2
  53     pydantic
  54     pymupdf
  55     python-multipart
  56     pyyaml
  57     # rapidocr-onnxruntime
  58     requests
  59     rich
  60     schedule
  61     sentence-transformers
  62     stripe
  63     tenacity
  64     tiktoken
  65     torch
  66     transformers
  67     tzdata
  68     uvicorn
  69   ];
  70
  71   nativeCheckInputs = with python3.pkgs; [
  72     freezegun
  73     factory-boy
  74     pytest-xdist
  75     trio
  76     psutil
  77     pytest-django
  78     pytestCheckHook
  79   ] ++ [
  80     (postgresql.withPackages (p: with p; [ pgvector ]))
  81     postgresqlTestHook
  82   ];
  83
  84   preCheck = ''
  85     export HOME=$(mktemp -d)
  86   '';
  87
  88   pythonImportsCheck = [
  89     "khoj"
  90   ];
  91
  92   disabledTests = [
  93     # Tests require network access
  94     "test_different_user_data_not_accessed"
  95     "test_get_api_config_types"
  96     "test_get_configured_types_via_api"
  97     "test_image_metadata"
  98     "test_image_search"
  99     "test_image_search_by_filepath"
 100     "test_image_search_query_truncated"
 101     "test_index_update"
 102     "test_index_update_with_no_auth_key"
 103     "test_notes_search"
 104     "test_notes_search_with_exclude_filter"
 105     "test_notes_search_with_include_filter"
 106     "test_parse_html_plaintext_file"
 107     "test_regenerate_index_with_new_entry"
 108     "test_regenerate_with_github_fails_without_pat"
 109     "test_regenerate_with_invalid_content_type"
 110     "test_regenerate_with_valid_content_type"
 111     "test_search_for_user2_returns_empty"
 112     "test_search_with_invalid_auth_key"
 113     "test_search_with_invalid_content_type"
 114     "test_search_with_no_auth_key"
 115     "test_search_with_valid_content_type"
 116     "test_text_index_same_if_content_unchanged"
 117     "test_text_indexer_deletes_embedding_before_regenerate"
 118     "test_text_search"
 119     "test_text_search_setup_batch_processes"
 120     "test_update_with_invalid_content_type"
 121     "test_user_no_data_returns_empty"
 122
 123     # Tests require rapidocr-onnxruntime
 124     "test_multi_page_pdf_to_jsonl"
 125     "test_single_page_pdf_to_jsonl"
 126     "test_ocr_page_pdf_to_jsonl"
 127   ];
 128
 129   disabledTestPaths = [
 130     # Tests require network access
 131     "tests/test_conversation_utils.py"
 132   ];
 133
 134   meta = with lib; {
 135     description = "Natural Language Search Assistant for your Org-Mode and Markdown notes, Beancount transactions and Photos";
 136     homepage = "https://github.com/debanjum/khoj";
 137     changelog = "https://github.com/debanjum/khoj/releases/tag/${version}";
 138     license = licenses.agpl3Plus;
 139     maintainers = with maintainers; [ dit7ya ];
 140     broken = true; # last successful build 2024-01-10
 141   };
 142 }