pythonPackages.pdfx: init at 1.3.1 (#42815)

2 files changed, 35 insertions, 0 deletions

diff --git a/pkgs/development/python-modules/pdfx/default.nix b/pkgs/development/python-modules/pdfx/default.nix
new file mode 100644
index 000000000000..d2397fb5c82b
--- /dev/null
+++ b/pkgs/development/python-modules/pdfx/default.nix
@@ -0,0 +1,33 @@
+{ stdenv, buildPythonPackage, fetchFromGitHub, pdfminer, chardet, pytest }:
+
+buildPythonPackage rec {
+  pname = "pdfx";
+  version = "1.3.1";
+
+  src = fetchFromGitHub {
+    owner = "metachris";
+    repo = "pdfx";
+    rev = "v${version}";
+    sha256 = "1183k4h5qdf8y0imbir9ja3yzzsvdmqgbv3bi6dnkgr1wy2xfr0v";
+  };
+
+  # Remove after https://github.com/metachris/pdfx/pull/28
+  prePatch = ''
+    sed -i -e "s|pdfminer2|pdfminer.six|" setup.py
+  '';
+
+  propagatedBuildInputs = [ pdfminer chardet ];
+
+  checkInputs = [ pytest ];
+
+  checkPhase = ''
+    py.test
+  '';
+
+  meta = with stdenv.lib; {
+    inherit (src.meta) homepage;
+    description = "Extract references (pdf, url, doi, arxiv) and metadata from a PDF";
+    license = licenses.asl20;
+    maintainers = with maintainers; [ marsam ];
+  };
+}
diff --git a/pkgs/top-level/python-packages.nix b/pkgs/top-level/python-packages.nix
index 6ac09121fac2..b2b1b295896d 100644
--- a/pkgs/top-level/python-packages.nix
+++ b/pkgs/top-level/python-packages.nix
@@ -335,6 +335,8 @@ in {
 
   pdfminer = callPackage ../development/python-modules/pdfminer_six { };
 
+  pdfx = callPackage ../development/python-modules/pdfx { };
+
   plantuml = callPackage ../tools/misc/plantuml { };
 
   Pmw = callPackage ../development/python-modules/Pmw { };