python3Packages.pymupdf4llm: init at 0.0.17 (#409244)

maintainers/maintainer-list.nix

@@ -21768,6 +21768,12 @@
     githubId = 7309170;
     name = "Ryota Kameoka";
   };
+  ryota2357 = {
+    email = "contact@ryota2357.com";
+    github = "ryota2357";
+    githubId = 61523777;
+    name = "Ryota Otsuki";
+  };
   rypervenche = {
     email = "git@ryper.org";
     github = "rypervenche";

pkgs/development/python-modules/pymupdf4llm/default.nix

@@ -0,0 +1,56 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  setuptools,
+  pymupdf,
+}:
+
+buildPythonPackage rec {
+  pname = "pymupdf4llm";
+  version = "0.0.17";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "pymupdf";
+    repo = "RAG";
+    tag = "v${version}";
+    hash = "sha256-+RLK+UorkU8eVQJGrc7pVNZPtIpxMgA9mBKA6GeWUa0=";
+  };
+
+  sourceRoot = "${src.name}/pymupdf4llm";
+
+  build-system = [ setuptools ];
+
+  dependencies = [ pymupdf ];
+
+  checkPhase = ''
+    runHook preCheck
+
+    python3 - <<'EOF'
+    import fitz
+    import pymupdf4llm
+
+    doc = fitz.open()
+    page = doc.new_page()
+    page.insert_text((72, 72), "Hello, Nix!")
+    doc.save("input.pdf")
+
+    md = pymupdf4llm.to_markdown("input.pdf")
+    assert isinstance(md, str), "Returned value is not a string"
+    assert "Hello, Nix!" in md, "Returned value does not contain the expected text"
+    EOF
+
+    runHook postCheck
+  '';
+
+  pythonImportsCheck = [ "pymupdf4llm" ];
+
+  meta = {
+    description = "PyMuPDF Utilities for LLM/RAG - converts PDF pages to Markdown format for Retrieval-Augmented Generation";
+    homepage = "https://github.com/pymupdf/RAG";
+    changelog = "https://github.com/pymupdf/RAG/blob/${src.tag}/CHANGES.md";
+    license = lib.licenses.agpl3Only;
+    maintainers = with lib.maintainers; [ ryota2357 ];
+  };
+}

pkgs/top-level/python-packages.nix

@@ -12918,6 +12918,8 @@ self: super: with self; {
 
   pymupdf-fonts = callPackage ../development/python-modules/pymupdf-fonts { };
 
+  pymupdf4llm = callPackage ../development/python-modules/pymupdf4llm { };
+
   pymvglive = callPackage ../development/python-modules/pymvglive { };
 
   pymysensors = callPackage ../development/python-modules/pymysensors { };