1{
2 lib,
3 buildPythonPackage,
4 fetchFromGitHub,
5 setuptools,
6 pymupdf,
7}:
8
9buildPythonPackage rec {
10 pname = "pymupdf4llm";
11 version = "0.0.27";
12 pyproject = true;
13
14 src = fetchFromGitHub {
15 owner = "pymupdf";
16 repo = "RAG";
17 tag = "v${version}";
18 hash = "sha256-rezdDsjNCDetvrX3uvykYuL/y40MZnr0fFMvQY3JRr0=";
19 };
20
21 sourceRoot = "${src.name}/pymupdf4llm";
22
23 build-system = [ setuptools ];
24
25 dependencies = [ pymupdf ];
26
27 checkPhase = ''
28 runHook preCheck
29
30 python3 - <<'EOF'
31 import fitz
32 import pymupdf4llm
33
34 doc = fitz.open()
35 page = doc.new_page()
36 page.insert_text((72, 72), "Hello, Nix!")
37 doc.save("input.pdf")
38
39 md = pymupdf4llm.to_markdown("input.pdf")
40 assert isinstance(md, str), "Returned value is not a string"
41 assert "Hello, Nix!" in md, "Returned value does not contain the expected text"
42 EOF
43
44 runHook postCheck
45 '';
46
47 pythonImportsCheck = [ "pymupdf4llm" ];
48
49 meta = {
50 description = "PyMuPDF Utilities for LLM/RAG - converts PDF pages to Markdown format for Retrieval-Augmented Generation";
51 homepage = "https://github.com/pymupdf/RAG";
52 changelog = "https://github.com/pymupdf/RAG/blob/${src.tag}/CHANGES.md";
53 license = lib.licenses.agpl3Only;
54 maintainers = with lib.maintainers; [ ryota2357 ];
55 };
56}