more hOCR tests, which are in the hOCR.m file now.

Author: Kirill Makankov

TestsProject/TestsProject.xcodeproj/project.pbxproj

@@ -31,6 +31,8 @@
 		73BE4C091A598F47002C15F1 /* TesseractOCR.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 73C0A7BC1A59561F00D823D4 /* TesseractOCR.framework */; };
 		73BE4C0A1A598F47002C15F1 /* TesseractOCR.framework in Embed Frameworks */ = {isa = PBXBuildFile; fileRef = 73C0A7BC1A59561F00D823D4 /* TesseractOCR.framework */; settings = {ATTRIBUTES = (CodeSignOnCopy, RemoveHeadersOnCopy, ); }; };
 		73BE4C311A5B404C002C15F1 /* well_scaned_page.hOCR in Resources */ = {isa = PBXBuildFile; fileRef = 73BE4C301A5B404C002C15F1 /* well_scaned_page.hOCR */; };
+		73BE4C331A5BC8A6002C15F1 /* hOCR.m in Sources */ = {isa = PBXBuildFile; fileRef = 73BE4C321A5BC8A6002C15F1 /* hOCR.m */; };
+		73BE4C351A5BD324002C15F1 /* image_sample.hOCR in Resources */ = {isa = PBXBuildFile; fileRef = 73BE4C341A5BD324002C15F1 /* image_sample.hOCR */; };
 		8FA2F9CE23919BEC8C64A5EA /* libPods-TestsProjectTests.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 5CD3C116A45C293ADAC81D1B /* libPods-TestsProjectTests.a */; };
 /* End PBXBuildFile section */
 
@@ -93,6 +95,8 @@
 		732C54781A5288CC000322DA /* Defaults.m */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.objc; path = Defaults.m; sourceTree = "<group>"; };
 		736EFF231A5872CA0031B432 /* tessdata-rus */ = {isa = PBXFileReference; lastKnownFileType = folder; name = "tessdata-rus"; path = "TestsProjectTests/tessdata-rus"; sourceTree = SOURCE_ROOT; };
 		73BE4C301A5B404C002C15F1 /* well_scaned_page.hOCR */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = well_scaned_page.hOCR; sourceTree = "<group>"; };
+		73BE4C321A5BC8A6002C15F1 /* hOCR.m */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.objc; path = hOCR.m; sourceTree = "<group>"; };
+		73BE4C341A5BD324002C15F1 /* image_sample.hOCR */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = image_sample.hOCR; sourceTree = "<group>"; };
 		73C0A7BC1A59561F00D823D4 /* TesseractOCR.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; path = TesseractOCR.framework; sourceTree = BUILT_PRODUCTS_DIR; };
 /* End PBXFileReference section */
 
@@ -169,6 +173,7 @@
 		4115B9771A3EF8E90004EC0A /* TestsProjectTests */ = {
 			isa = PBXGroup;
 			children = (
+				73BE4C321A5BC8A6002C15F1 /* hOCR.m */,
 				732C54751A514DA5000322DA /* InitializationTests.m */,
 				4115B97A1A3EF8E90004EC0A /* RecognitionTests.m */,
 				4115B9781A3EF8E90004EC0A /* Supporting Files */,
@@ -211,6 +216,7 @@
 				41C68DB21A41849100848AE1 /* image_sample.jpg */,
 				41C68DB41A41854600848AE1 /* image_sample_tr.png */,
 				412E9EAC1A45872A007DDAA5 /* image_sample_bl.png */,
+				73BE4C341A5BD324002C15F1 /* image_sample.hOCR */,
 				4141211D1A4B1E3900583ED4 /* image_blank.png */,
 			);
 			name = Images;
@@ -311,6 +317,7 @@
 			files = (
 				412E9EAB1A44316C007DDAA5 /* well_scaned_page.png in Resources */,
 				41C68DB31A41849100848AE1 /* image_sample.jpg in Resources */,
+				73BE4C351A5BD324002C15F1 /* image_sample.hOCR in Resources */,
 				4115B96C1A3EF8E90004EC0A /* Images.xcassets in Resources */,
 				4115B9851A3EF9920004EC0A /* Launch Screen.xib in Resources */,
 				4141211E1A4B1E3900583ED4 /* image_blank.png in Resources */,
@@ -380,6 +387,7 @@
 			buildActionMask = 2147483647;
 			files = (
 				732C54791A5288CC000322DA /* Defaults.m in Sources */,
+				73BE4C331A5BC8A6002C15F1 /* hOCR.m in Sources */,
 				414121231A4C5A5700583ED4 /* G8RecognitionTestsHelper.m in Sources */,
 				4115B97B1A3EF8E90004EC0A /* RecognitionTests.m in Sources */,
 				732C54761A514DA6000322DA /* InitializationTests.m in Sources */,

TestsProject/TestsProject/image_sample.hOCR

@@ -0,0 +1,8 @@
+  <div class='ocr_page' id='page_1' title='image ""; bbox 0 0 1121 156; ppageno 0'>
+   <div class='ocr_carea' id='block_1_1' title="bbox 32 20 1091 148">
+    <p class='ocr_par' dir='ltr' id='par_1_1' title="bbox 32 20 1091 148">
+     <span class='ocr_line' id='line_1_1' title="bbox 32 20 1091 148; baseline -0.008 0"><span class='ocrx_word' id='word_1_1' title='bbox 32 20 1091 148; x_wconf 84' lang='eng'>1234567890</span> 
+     </span>
+    </p>
+   </div>
+  </div>

TestsProject/TestsProjectTests/RecognitionTests.m

@@ -238,17 +238,6 @@
 
         [[[[helper.tesseract recognizedBlocksByIteratorLevel:G8PageIteratorLevelWord] should] haveAtLeast:10] items];
     });
-    
-    it(@"Should hOCR", ^{
-        
-        [helper recognizeImage];
-        NSString *hOCR = [helper.tesseract recognizedHOCRForPageNumber:0];
-        
-        NSString *path = [[NSBundle mainBundle] pathForResource:@"well_scaned_page" ofType:@"hOCR"];
-        NSError *error = nil;
-        [[hOCR should] equal:[NSString stringWithContentsOfFile:path encoding:NSUTF8StringEncoding error:&error]];
-        NSAssert(error == nil, @"error loading hOCR from file %@: %@", path, error);
-    });
 });
 
 SPEC_END

TestsProject/TestsProjectTests/hOCR.m

@@ -0,0 +1,61 @@
+//
+//  hOCR.m
+//  TestsProject
+//
+//  Created by Kirill Makankov on 06.01.15.
+//  Copyright (c) 2015 Daniele Galiotto - www.g8production.com. All rights reserved.
+//
+
+#import <UIKit/UIKit.h>
+#import <Kiwi/Kiwi.h>
+
+#import "G8RecognitionTestsHelper.h"
+
+SPEC_BEGIN(hOCRTests)
+
+__block G8RecognitionTestsHelper *helper;
+
+beforeEach(^{
+    helper = [[G8RecognitionTestsHelper alloc] init];
+});
+
+describe(@"hOCR", ^{
+    
+    it(@"Should sample image", ^{
+        
+        NSString *path = [[NSBundle mainBundle] pathForResource:@"image_sample" ofType:@"hOCR"];
+
+        helper.image = [UIImage imageNamed:@"image_sample.jpg"];
+        helper.charWhitelist = @"0123456789";
+        
+        [helper recognizeImage];
+        NSString *hOCR = [helper.tesseract recognizedHOCRForPageNumber:0];
+        
+        NSError *error = nil;
+        [[hOCR should] equal:[NSString stringWithContentsOfFile:path encoding:NSUTF8StringEncoding error:&error]];
+        NSAssert(error == nil, @"error loading hOCR from file %@: %@", path, error);
+    });
+    
+    it(@"Should well scanced page", ^{
+        
+        NSString *path = [[NSBundle mainBundle] pathForResource:@"well_scaned_page" ofType:@"hOCR"];
+
+        helper.image = [UIImage imageNamed:@"well_scaned_page"];
+        
+        [helper recognizeImage];
+        NSString *hOCR = [helper.tesseract recognizedHOCRForPageNumber:0];
+        
+        NSError *error = nil;
+        [[hOCR should] equal:[NSString stringWithContentsOfFile:path encoding:NSUTF8StringEncoding error:&error]];
+        NSAssert(error == nil, @"error loading hOCR from file %@: %@", path, error);
+    });
+    
+    it(@"Should return nil without prerecognition", ^{
+        G8Tesseract *tesseract = [[G8Tesseract alloc] init];
+
+        NSString *hOCR = [tesseract recognizedHOCRForPageNumber:0];
+        [[hOCR should] beNil];
+    });
+});
+
+SPEC_END