Extend the ability to process image only from disk to from both disk and memory

ocrolib/common.py

@@ -152,15 +152,18 @@ def isintarray(a):
 def isintegerarray(a):
     return a.dtype in [dtype('int32'),dtype('int64'),dtype('uint32'),dtype('uint64')]
 
-@checks(str,pageno=int,_=GRAYSCALE)
+@checks({str,object},pageno=int,_=GRAYSCALE)
 def read_image_gray(fname,pageno=0):
     """Read an image and returns it as a floating point array.
     The optional page number allows images from files containing multiple
     images to be addressed.  Byte and short arrays are rescaled to
     the range 0...1 (unsigned) or -1...1 (signed)."""
     if type(fname)==tuple: fname,pageno = fname
     assert pageno==0
-    pil = PIL.Image.open(fname)
+    if issubclass(type(fname), PIL.Image.Image):
+        pil = fname
+    else:
+        pil = PIL.Image.open(fname)
     a = pil2array(pil)
     if a.dtype==dtype('uint8'):
         a = a/255.0
@@ -191,13 +194,16 @@ def write_image_gray(fname,image,normalize=0,verbose=0):
     im = array2pil(image)
     im.save(fname)
 
-@checks(str,_=ABINARY2)
+@checks({str,object},_=ABINARY2)
 def read_image_binary(fname,dtype='i',pageno=0):
     """Read an image from disk and return it as a binary image
     of the given dtype."""
     if type(fname)==tuple: fname,pageno = fname
     assert pageno==0
-    pil = PIL.Image.open(fname)
+    if issubclass(type(fname), PIL.Image.Image):
+        pil = fname
+    else:
+        pil = PIL.Image.open(fname)
     a = pil2array(pil)
     if a.ndim==3: a = amax(a,axis=2)
     return array(a>0.5*(amin(a)+amax(a)),dtype)

tests/run-unit

@@ -1,54 +1,6 @@
-#!/usr/bin/env python
+#/bin/bash
 
-import sys
-
-from ocrolib import edist, utils
-
-# Test the levenshtein function and returns 0 if the computed value
-# equals the one it should be, otherwise returns 1 for failed tests.
-def testLevenshtein(a, b, should):
-    if edist.levenshtein(a, b) == should:
-        print 'ok - levenshtein(%s, %s) == %s' % (a,b,should)
-        return 0
-    else:
-        print 'not ok - levenshtein(%s, %s) == %s' % (a,b,should)
-        return 1
-
-
-def testXLevenshtein(a, b, context, should):
-    #print(edist.xlevenshtein(a, b, context))
-    if edist.xlevenshtein(a, b, context) == should:
-        print 'ok - xlevenshtein(%s, %s, %s) == %s' % (a,b,context,should)
-        return 0
-    else:
-        print 'not ok - xlevenshtein(%s, %s, %s) == %s' % (a,b,context,should)
-        return 1
-
-
-failed_tests = 0
-
-print('# 1 Test function "levenshtein" in edist.py')
-failed_tests += testLevenshtein('a', 'a', 0)
-failed_tests += testLevenshtein('', '', 0)
-failed_tests += testLevenshtein('a', '', 1)
-failed_tests += testLevenshtein('', 'a', 1)
-failed_tests += testLevenshtein('aa', 'aaaaaa', 4)
-failed_tests += testLevenshtein('aba', 'bab', 2)
-
-print('\n# 2 Test function "xlevenshtein" in edist.py')
-failed_tests += testXLevenshtein('exccpt', 'except', 1, should=(1.0, [('ccp', 'cep')]))
-failed_tests += testXLevenshtein('exccpt', 'except', 2, should=(1.0, [('xccpt', 'xcept')]))
-failed_tests += testXLevenshtein('exccpt', 'except', 3, should=(1.0, [('exccpt ', 'except ')]))
-failed_tests += testXLevenshtein('exccpt', 'except', 4, should=(1.0, [(' exccpt  ', ' except  ')]))
-failed_tests += testXLevenshtein('', 'test', 1, should=(4.0, []))
-failed_tests += testXLevenshtein('aaaaaaaaaaa', 'a', 1, should=(10.0, [('aaaaaaaaaaa ', 'a__________ ')]))
-failed_tests += testXLevenshtein('123 111 456', '132 111 444', 1, should=(4.0, [('123_ ', '1_32 '), ('456 ', '444 ')]))
-
-print('\n# 3 utils.sumouter / utils.sumprod')
-from pylab import randn
-utils.sumouter(randn(10,3),randn(10,4),out=randn(3,4))
-print('ok - dimensions of sumouter')
-utils.sumprod(randn(11,7),randn(11,7),out=randn(7))
-print('ok - dimensions of sumprod')
-
-sys.exit(failed_tests)
+BASE=$(dirname $0)/..
+export PYTHONPATH="$BASE":"$PYTHONPATH"
+python2 "$BASE"/tests/test_read_image.py
+python2 "$BASE"/tests/test-levenshtein.py

tests/test-levenshtein.py

@@ -0,0 +1,54 @@
+#!/usr/bin/env python
+
+import sys
+
+from ocrolib import edist, utils
+
+# Test the levenshtein function and returns 0 if the computed value
+# equals the one it should be, otherwise returns 1 for failed tests.
+def testLevenshtein(a, b, should):
+    if edist.levenshtein(a, b) == should:
+        print 'ok - levenshtein(%s, %s) == %s' % (a,b,should)
+        return 0
+    else:
+        print 'not ok - levenshtein(%s, %s) == %s' % (a,b,should)
+        return 1
+
+
+def testXLevenshtein(a, b, context, should):
+    #print(edist.xlevenshtein(a, b, context))
+    if edist.xlevenshtein(a, b, context) == should:
+        print 'ok - xlevenshtein(%s, %s, %s) == %s' % (a,b,context,should)
+        return 0
+    else:
+        print 'not ok - xlevenshtein(%s, %s, %s) == %s' % (a,b,context,should)
+        return 1
+
+
+failed_tests = 0
+
+print('# 1 Test function "levenshtein" in edist.py')
+failed_tests += testLevenshtein('a', 'a', 0)
+failed_tests += testLevenshtein('', '', 0)
+failed_tests += testLevenshtein('a', '', 1)
+failed_tests += testLevenshtein('', 'a', 1)
+failed_tests += testLevenshtein('aa', 'aaaaaa', 4)
+failed_tests += testLevenshtein('aba', 'bab', 2)
+
+print('\n# 2 Test function "xlevenshtein" in edist.py')
+failed_tests += testXLevenshtein('exccpt', 'except', 1, should=(1.0, [('ccp', 'cep')]))
+failed_tests += testXLevenshtein('exccpt', 'except', 2, should=(1.0, [('xccpt', 'xcept')]))
+failed_tests += testXLevenshtein('exccpt', 'except', 3, should=(1.0, [('exccpt ', 'except ')]))
+failed_tests += testXLevenshtein('exccpt', 'except', 4, should=(1.0, [(' exccpt  ', ' except  ')]))
+failed_tests += testXLevenshtein('', 'test', 1, should=(4.0, []))
+failed_tests += testXLevenshtein('aaaaaaaaaaa', 'a', 1, should=(10.0, [('aaaaaaaaaaa ', 'a__________ ')]))
+failed_tests += testXLevenshtein('123 111 456', '132 111 444', 1, should=(4.0, [('123_ ', '1_32 '), ('456 ', '444 ')]))
+
+print('\n# 3 utils.sumouter / utils.sumprod')
+from pylab import randn
+utils.sumouter(randn(10,3),randn(10,4),out=randn(3,4))
+print('ok - dimensions of sumouter')
+utils.sumprod(randn(11,7),randn(11,7),out=randn(7))
+print('ok - dimensions of sumprod')
+
+sys.exit(failed_tests)

tests/test_read_image.py

@@ -0,0 +1,35 @@
+import unittest
+import ocrolib
+import PIL
+import numpy
+
+### Original image in disk and memory used to test method read_image_gray()
+img_disk = "tests/testpage.png"
+img_mem = PIL.Image.open(img_disk)
+
+### Binarized image in disk and memory used to test method read_image_binary()
+img_bin_disk = "tests/010030.bin.png"
+img_bin_mem = PIL.Image.open(img_bin_disk)
+
+class OcrolibTestCase(unittest.TestCase):
+	"""
+	Tests for processing image from disk and memory for methods 
+	read_image_gray() and read_image_binary() in common.py under ocrolib
+	"""
+
+	def test_read_image_gray(self):
+		"""
+		Test whether the function read_image_gray() will return same result 
+		when pass a image file name (from disk) and a image object (PIL.Image from memory).
+		The return object of read_image_gray() is a 'ndarray' dedfined by 'numpy', thus we use the 
+		built-in function 'array_equal' to compare two ndarray objects
+		"""
+		self.assertTrue(numpy.array_equal(ocrolib.read_image_gray(img_disk), ocrolib.read_image_gray(img_mem)))
+
+
+	def test_read_image_binary(self):
+		self.assertTrue(numpy.array_equal(ocrolib.read_image_binary(img_bin_disk), ocrolib.read_image_binary(img_bin_mem)))
+
+
+if __name__ == '__main__':
+	unittest.main()