khoj: 0.3.0 -> 1.0.1

Diff: https://github.com/debanjum/khoj/compare/refs/tags/0.3.0...1.0.1

Changelog: https://github.com/debanjum/khoj/releases/tag/1.0.1

pkgs/servers/search/khoj/default.nix

@@ -2,74 +2,84 @@
 , stdenv
 , fetchFromGitHub
 , python3
-, qt6
+, postgresql
+, postgresqlTestHook
 }:
 
 python3.pkgs.buildPythonApplication rec {
   pname = "khoj";
-  version = "0.3.0";
-  format = "pyproject";
+  version = "1.0.1";
+  pyproject = true;
 
   src = fetchFromGitHub {
     owner = "debanjum";
     repo = "khoj";
     rev = "refs/tags/${version}";
-    hash = "sha256-9kKK0DXpLfPB2LMnYcC6BKgZaoRsNHBZVe4thI7b9tk=";
+    hash = "sha256-lvOeYTrvW5MfhuJ3lj9n9TRlvpRwVP2vFeaEeJdqIec=";
   };
 
-  SETUPTOOLS_SCM_PRETEND_VERSION = version;
-
-  postPatch = ''
-    substituteInPlace pyproject.toml \
-      --replace "dateparser == 1.1.1" "dateparser" \
-      --replace "defusedxml == 0.7.1" ""defusedxml"" \
-      --replace "fastapi == 0.77.1" "fastapi" \
-      --replace "jinja2 == 3.1.2" "jinja2" \
-      --replace "openai == 0.20.0" "openai" \
-      --replace "pillow == 9.3.0" "pillow" \
-      --replace "pydantic == 1.9.1" "pydantic" \
-      --replace "pyyaml == 6.0" "pyyaml" \
-      --replace "pyqt6 == 6.3.1" "pyqt6" \
-      --replace "rich >= 13.3.1" "rich" \
-      --replace "schedule == 1.1.0" "schedule" \
-      --replace "sentence-transformers == 2.2.2" "sentence-transformers" \
-      --replace "torch == 1.13.1" "torch" \
-      --replace "uvicorn == 0.17.6" "uvicorn"
-  '';
+  env = {
+    SETUPTOOLS_SCM_PRETEND_VERSION = version;
+    DJANGO_SETTINGS_MODULE = "khoj.app.settings";
+    postgresqlEnableTCP = 1;
+  };
 
   nativeBuildInputs = with python3.pkgs; [
     hatch-vcs
     hatchling
-  ] ++ (with qt6; [
-    wrapQtAppsHook
-  ]);
-
-  buildInputs = lib.optionals stdenv.isLinux [
-    qt6.qtwayland
-  ] ++ lib.optionals stdenv.isDarwin [
-    qt6.qtbase
   ];
 
   propagatedBuildInputs = with python3.pkgs; [
+    aiohttp
+    anyio
+    authlib
+    beautifulsoup4
     dateparser
     defusedxml
+    django
     fastapi
+    google-auth
+    # gpt4all
+    gunicorn
+    httpx
+    itsdangerous
     jinja2
-    numpy
+    langchain
+    lxml
     openai
+    openai-whisper
+    pgvector
     pillow
+    psycopg2
     pydantic
-    pyqt6
+    pymupdf
+    python-multipart
     pyyaml
+    # rapidocr-onnxruntime
+    requests
     rich
     schedule
     sentence-transformers
+    stripe
+    tenacity
+    tiktoken
     torch
+    transformers
+    tzdata
     uvicorn
   ];
 
   nativeCheckInputs = with python3.pkgs; [
+    freezegun
+    factory-boy
+    pytest-xdist
+    trio
+    psutil
+    pytest-django
     pytestCheckHook
+  ] ++ [
+    (postgresql.withPackages (p: with p; [ pgvector ]))
+    postgresqlTestHook
   ];
 
   preCheck = ''
@@ -82,32 +92,51 @@ python3.pkgs.buildPythonApplication rec {
 
   disabledTests = [
     # Tests require network access
-    "test_search_with_valid_content_type"
-    "test_update_with_valid_content_type"
-    "test_regenerate_with_valid_content_type"
-    "test_image_search"
-    "test_notes_search"
-    "test_notes_search_with_only_filters"
-    "test_notes_search_with_include_filter"
-    "test_notes_search_with_exclude_filter"
+    "test_different_user_data_not_accessed"
+    "test_get_api_config_types"
+    "test_get_configured_types_via_api"
     "test_image_metadata"
     "test_image_search"
-    "test_image_search_query_truncated"
     "test_image_search_by_filepath"
-    "test_asymmetric_setup_with_missing_file_raises_error"
-    "test_asymmetric_setup_with_empty_file_raises_error"
-    "test_asymmetric_reload"
-    "test_asymmetric_setup"
-    "test_asymmetric_search"
-    "test_entry_chunking_by_max_tokens"
-    "test_incremental_update"
+    "test_image_search_query_truncated"
+    "test_index_update"
+    "test_index_update_with_no_auth_key"
+    "test_notes_search"
+    "test_notes_search_with_exclude_filter"
+    "test_notes_search_with_include_filter"
+    "test_parse_html_plaintext_file"
+    "test_regenerate_index_with_new_entry"
+    "test_regenerate_with_github_fails_without_pat"
+    "test_regenerate_with_invalid_content_type"
+    "test_regenerate_with_valid_content_type"
+    "test_search_for_user2_returns_empty"
+    "test_search_with_invalid_auth_key"
+    "test_search_with_invalid_content_type"
+    "test_search_with_no_auth_key"
+    "test_search_with_valid_content_type"
+    "test_text_index_same_if_content_unchanged"
+    "test_text_indexer_deletes_embedding_before_regenerate"
+    "test_text_search"
+    "test_text_search_setup_batch_processes"
+    "test_update_with_invalid_content_type"
+    "test_user_no_data_returns_empty"
+
+    # Tests require rapidocr-onnxruntime
+    "test_multi_page_pdf_to_jsonl"
+    "test_single_page_pdf_to_jsonl"
+    "test_ocr_page_pdf_to_jsonl"
+  ];
+
+  disabledTestPaths = [
+    # Tests require network access
+    "tests/test_conversation_utils.py"
   ];
 
   meta = with lib; {
     description = "Natural Language Search Assistant for your Org-Mode and Markdown notes, Beancount transactions and Photos";
     homepage = "https://github.com/debanjum/khoj";
     changelog = "https://github.com/debanjum/khoj/releases/tag/${version}";
-    license = licenses.gpl3Only;
+    license = licenses.agpl3Plus;
     maintainers = with maintainers; [ dit7ya ];
     # src/tcmalloc.cc:333] Attempt to free invalid pointer
     broken = stdenv.isDarwin;