pythonPackages.pyocr: 0.4.7 -> 0.5.3

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

@@ -1,47 +1,39 @@
-{ lib, fetchFromGitHub, buildPythonPackage, pillow, six
-, tesseract, cuneiform, isPy3k
+{ lib, fetchFromGitLab, buildPythonPackage, pillow, six
+, tesseract, cuneiform, isPy3k, substituteAll, pytest, tox
 }:
 
 buildPythonPackage rec {
   pname = "pyocr";
-  version = "0.4.7";
+  version = "0.5.3";
   name = pname + "-" + version;
   disabled = !isPy3k;
 
   # Don't fetch from PYPI because it doesn't contain tests.
-  src = fetchFromGitHub {
-    owner = "jflesch";
+  src = fetchFromGitLab {
+    domain = "gitlab.gnome.org";
+    group = "World";
+    owner = "OpenPaperwork";
     repo = "pyocr";
     rev = version;
-    sha256 = "1iw73r8yrgjf8g00yzpz62ymqbf89cqhyhl9g430srmsrq7mn2yd";
+    sha256 = "1nihf0qmbpg3yj3yp11jp6hp5z5dqf39nz6j9lqbvgi1nqbs7x15";
   };
 
-  NIX_CUNEIFORM_CMD = "${cuneiform}/bin/cuneiform";
-  NIX_CUNEIFORM_DATA = "${cuneiform}/share/cuneiform";
-  NIX_LIBTESSERACT_PATH = "${tesseract}/lib/libtesseract.so";
-  NIX_TESSDATA_PREFIX = "${tesseract}/share/tessdata";
-  NIX_TESSERACT_CMD = "${tesseract}/bin/tesseract";
-
-  patches = [ ./paths.patch ];
+  patches = [ (substituteAll {
+    src = ./paths.patch;
+    inherit cuneiform tesseract;
+  })
+  ];
 
   postPatch = ''
-    substituteInPlace src/pyocr/cuneiform.py \
-      --subst-var NIX_CUNEIFORM_CMD \
-      --subst-var NIX_CUNEIFORM_CMD
-
-    substituteInPlace src/pyocr/tesseract.py \
-      --subst-var NIX_TESSERACT_CMD
-
-    substituteInPlace src/pyocr/libtesseract/tesseract_raw.py \
-      --subst-var NIX_TESSDATA_PREFIX \
-      --subst-var NIX_LIBTESSERACT_PATH
+    echo 'version = "${version}"' > src/pyocr/_version.py
 
     # Disable specific tests that are probably failing because of this issue:
     # https://github.com/jflesch/pyocr/issues/52
     for test in $disabledTests; do
       file="''${test%%:*}"
       fun="''${test#*:}"
-      echo "$fun = unittest.skip($fun)" >> "tests/tests_$file.py"
+      echo "import pytest" >> "tests/tests_$file.py"
+      echo "$fun = pytest.mark.skip($fun)" >> "tests/tests_$file.py"
     done
   '';
 
@@ -57,14 +49,18 @@ buildPythonPackage rec {
     "libtesseract:TestLineBox.test_japanese"
     "libtesseract:TestTxt.test_japanese"
     "libtesseract:TestWordBox.test_japanese"
+    "libtesseract:TestTxt.test_multi"
+    "tesseract:TestTxt.test_multi"
     "tesseract:TestDigitLineBox.test_digits"
     "tesseract:TestTxt.test_japanese"
   ];
 
   propagatedBuildInputs = [ pillow six ];
+  checkInputs = [ pytest tox ];
+  checkPhase = "pytest";
 
   meta = {
-    homepage = "https://github.com/jflesch/pyocr";
+    inherit (src) homepage;
     description = "A Python wrapper for Tesseract and Cuneiform";
     license = lib.licenses.gpl3Plus;
   };

pkgs/development/python-modules/pyocr/paths.patch

@@ -1,28 +1,28 @@
-diff --git a/src/pyocr/cuneiform.py b/src/pyocr/cuneiform.py
-index a461d92..1f2b914 100644
---- a/src/pyocr/cuneiform.py
-+++ b/src/pyocr/cuneiform.py
+Index: current/src/pyocr/cuneiform.py
+===================================================================
+--- current.orig/src/pyocr/cuneiform.py
++++ current/src/pyocr/cuneiform.py
 @@ -27,13 +27,9 @@ from . import error
  from . import util
  
  
 -# CHANGE THIS IF CUNEIFORM IS NOT IN YOUR PATH, OR IS NAMED DIFFERENTLY
 -CUNEIFORM_CMD = 'cuneiform'
-+CUNEIFORM_CMD = '@NIX_CUNEIFORM_CMD@'
++CUNEIFORM_CMD = '@cuneiform@/bin/cuneiform'
  
 -CUNEIFORM_DATA_POSSIBLE_PATHS = [
 -    "/usr/local/share/cuneiform",
 -    "/usr/share/cuneiform",
 -]
-+CUNEIFORM_DATA_POSSIBLE_PATHS = ['@NIX_CUNEIFORM_DATA@']
++CUNEIFORM_DATA_POSSIBLE_PATHS = ['@cuneiform@/share/cuneiform']
  
  LANGUAGES_LINE_PREFIX = "Supported languages: "
  LANGUAGES_SPLIT_RE = re.compile("[^a-z]")
-diff --git a/src/pyocr/libtesseract/tesseract_raw.py b/src/pyocr/libtesseract/tesseract_raw.py
-index b4e7bda..47505f7 100644
---- a/src/pyocr/libtesseract/tesseract_raw.py
-+++ b/src/pyocr/libtesseract/tesseract_raw.py
-@@ -1,55 +1,13 @@
+Index: current/src/pyocr/libtesseract/tesseract_raw.py
+===================================================================
+--- current.orig/src/pyocr/libtesseract/tesseract_raw.py
++++ current/src/pyocr/libtesseract/tesseract_raw.py
+@@ -1,52 +1,13 @@
  import ctypes
  import logging
  import os
@@ -56,7 +56,13 @@ index b4e7bda..47505f7 100644
 -        # Jflesch> Don't they have the equivalent of LD_LIBRARY_PATH on
 -        # Windows ?
 -        "../vs2010/DLL_Release/libtesseract302.dll",
+-        # prefer the most recent first
+-        "libtesseract305.dll",
+-        "libtesseract304.dll",
+-        "libtesseract303.dll",
 -        "libtesseract302.dll",
+-        "libtesseract400.dll",  # Tesseract 4 is still in alpha stage
+-        "libtesseract.dll",
 -        "C:\\Program Files (x86)\\Tesseract-OCR\\libtesseract-4.dll",
 -        "C:\\Program Files (x86)\\Tesseract-OCR\\libtesseract-3.dll",
 -    ]
@@ -66,27 +72,18 @@ index b4e7bda..47505f7 100644
 -        "libtesseract.so.3",
 -    ]
 -
--
--g_libtesseract = None
--
--for libname in libnames:
--    try:
--        g_libtesseract = ctypes.cdll.LoadLibrary(libname)
--        break
--    except OSError:
--        pass
-+g_libtesseract = ctypes.cdll.LoadLibrary('@NIX_LIBTESSERACT_PATH@')
++libnames = [ "@tesseract@/lib/libtesseract.so" ]
  
+ g_libtesseract = None
  
- class PageSegMode(object):
-@@ -326,12 +284,11 @@ def init(lang=None):
+@@ -346,12 +307,11 @@ def init(lang=None):
      try:
          if lang:
              lang = lang.encode("utf-8")
 -        prefix = None
 -        if TESSDATA_PREFIX:
 -            prefix = TESSDATA_PREFIX.encode("utf-8")
-+        prefix = os.getenv('TESSDATA_PREFIX', '@NIX_TESSDATA_PREFIX@')
++        prefix = os.getenv('TESSDATA_PREFIX', '@tesseract@/share/tessdata')
 +        os.environ['TESSDATA_PREFIX'] = prefix
          g_libtesseract.TessBaseAPIInit3(
              ctypes.c_void_p(handle),
@@ -95,17 +92,17 @@ index b4e7bda..47505f7 100644
              ctypes.c_char_p(lang)
          )
          g_libtesseract.TessBaseAPISetVariable(
-diff --git a/src/pyocr/tesseract.py b/src/pyocr/tesseract.py
-index c935881..7139ffe 100755
---- a/src/pyocr/tesseract.py
-+++ b/src/pyocr/tesseract.py
-@@ -31,8 +31,7 @@ from .builders import DigitBuilder  # backward compatibility
+Index: current/src/pyocr/tesseract.py
+===================================================================
+--- current.orig/src/pyocr/tesseract.py
++++ current/src/pyocr/tesseract.py
+@@ -31,8 +31,7 @@ from .builders import DigitBuilder  # ba
  from .error import TesseractError  # backward compatibility
  from .util import digits_only
  
 -# CHANGE THIS IF TESSERACT IS NOT IN YOUR PATH, OR IS NAMED DIFFERENTLY
 -TESSERACT_CMD = 'tesseract.exe' if os.name == 'nt' else 'tesseract'
-+TESSERACT_CMD = '@NIX_TESSERACT_CMD@'
++TESSERACT_CMD = '@tesseract@/bin/tesseract'
  
  TESSDATA_EXTENSION = ".traineddata"