From 7d2cee008b636b2bb754bfbe43f09d8fbc460296 Mon Sep 17 00:00:00 2001
From: Erik Arvstedt <erik.arvstedt@gmail.com>
Date: Mon, 2 May 2022 12:38:51 +0200
Subject: [PATCH] tesseract: add wrapper test

---
 .../graphics/tesseract/wrapper.nix            | 32 +++++++++++++++++--
 1 file changed, 29 insertions(+), 3 deletions(-)

diff --git a/pkgs/applications/graphics/tesseract/wrapper.nix b/pkgs/applications/graphics/tesseract/wrapper.nix
index 1dfc4393c46b..59686e3251f2 100644
--- a/pkgs/applications/graphics/tesseract/wrapper.nix
+++ b/pkgs/applications/graphics/tesseract/wrapper.nix
@@ -1,4 +1,5 @@
 { lib, makeWrapper, tesseractBase, languages
+, runCommand, imagemagick
 
 # A list of languages like [ "eng" "spa" … ] or `null` for all available languages
 , enableLanguages ? null
@@ -12,8 +13,6 @@
 }:
 
 let
-  passthru = { inherit tesseractBase languages tessdata; };
-
   tesseractWithData = tesseractBase.overrideAttrs (_: {
     inherit tesseractBase tessdata;
 
@@ -47,9 +46,36 @@ let
          exit 1
       fi
     '';
+
   });
 
-  tesseract = (if enableLanguages == [] then tesseractBase else tesseractWithData) // passthru;
+  passthru = { inherit tesseractBase languages tessdata; };
+
+  # Only run test when all languages are available
+  test = lib.optionalAttrs (enableLanguages == null) {
+    tests.default = runCommand "tesseract-test-ocr" {
+      buildInputs = [
+        tesseractWithData
+        imagemagick
+      ];
+    } ''
+      text="hello nix"
+
+      convert -size 400x40 xc:white -font 'DejaVu-Sans' -pointsize 20 \
+        -fill black -annotate +5+20 "$text" /tmp/test-img.png 2>/dev/null
+      ocrResult=$(tesseract /tmp/test-img.png - | tr -d "\f")
+
+      if [[ $ocrResult != $text ]]; then
+        echo "OCR test failed"
+        echo "expected: '$text'"
+        echo "actual: '$ocrResult'"
+        exit 1
+      fi
+      touch $out
+    '';
+  };
+
+  tesseract = (if enableLanguages == [] then tesseractBase else tesseractWithData) // passthru // test;
 in
   if enableLanguagesHash == null then
     tesseract