Merge pull request #1348 from TikhonJelvis/pytesseract-executable-dep…

…endency

Add dependency on native Tesseract OCR executable for pytesseract

overrides/default.nix

@@ -2141,6 +2141,34 @@ lib.composeManyExtensions [
         buildInputs = (old.buildInputs or [ ]) ++ [ pkgs.taglib ];
       });
 
+      pytesseract =
+        let
+          pytesseract-cmd-patch = pkgs.writeText "pytesseract-cmd.patch" ''
+            --- a/pytesseract/pytesseract.py
+            +++ b/pytesseract/pytesseract.py
+            @@ -27,7 +27,7 @@
+             from PIL import Image
+
+
+            -tesseract_cmd = 'tesseract'
+            +tesseract_cmd = '${pkgs.tesseract4}/bin/tesseract'
+
+             numpy_installed = find_loader('numpy') is not None
+             if numpy_installed:
+          '';
+        in
+        super.pytesseract.overridePythonAttrs (old: {
+          buildInputs = (old.buildInputs or [ ]) ++ [ pkgs.tesseract4 ];
+          patches = (old.patches or [ ]) ++ lib.optionals (!(old.src.isWheel or false)) [ pytesseract-cmd-patch ];
+
+          # apply patch in postInstall if the source is a wheel
+          postInstall = lib.optionalString (old.src.isWheel or false) ''
+            pushd "$out/${self.python.sitePackages}"
+            patch -p1 < "${pytesseract-cmd-patch}"
+            popd
+          '';
+        });
+
       pytezos = super.pytezos.overridePythonAttrs (old: {
         buildInputs = (old.buildInputs or [ ]) ++ [ pkgs.libsodium ];
       });

tests/default.nix

@@ -162,6 +162,7 @@ in
   rpds-py-no-wheel = callTest ./rpds-py-no-wheel { };
   contourpy-wheel = callTest ./contourpy-wheel { };
   contourpy-no-wheel = callTest ./contourpy-no-wheel { };
+  pytesseract = callTest ./pytesseract { };
 } // lib.optionalAttrs (!stdenv.isDarwin) {
   # pyqt5 = (callTest ./pyqt5 { });
 

tests/pytesseract/default.nix

@@ -0,0 +1,13 @@
+{ poetry2nix, runCommand, python3 }:
+let
+  env = poetry2nix.mkPoetryEnv {
+    python = python3;
+    pyproject = ./pyproject.toml;
+    poetrylock = ./poetry.lock;
+    preferWheels = true;
+  };
+  py = env.python;
+in
+runCommand "pytesseract-test" { } ''
+  ${env}/bin/python -c 'import pytesseract; print(pytesseract.get_tesseract_version())' > $out
+''

tests/pytesseract/pyproject.toml

@@ -0,0 +1,9 @@
+[tool.poetry]
+name = "pytesseract-test"
+version = "0.1.0"
+description = ""
+authors = ["Your Name <[email protected]>"]
+
+[tool.poetry.dependencies]
+python = "^3.10"
+pytesseract = "*"