Merge pull request gali8#125 from ws233/master

Added hOCR support, switched to static framework.

Podfile

@@ -1,6 +1,6 @@
 source 'https://github.com/CocoaPods/Specs.git'
 workspace 'Tesseract-OCR-iOS'
-xcodeproj 'TestsProject/TestsProject.xcodeproj/'
+xcodeproj 'TestsProject/TestsProject.xcodeproj/', 'Coverage' => :debug
 
 target 'TestsProjectTests' do
 

Podfile.lock

@@ -7,4 +7,4 @@ DEPENDENCIES:
 SPEC CHECKSUMS:
   Kiwi: 73e1400209055ee9c8ba78c6012b6b642d0fb9f7
 
-COCOAPODS: 0.35.0
+COCOAPODS: 0.36.0.beta.1

Template Framework Project/Template Framework Project.xcodeproj/project.pbxproj

@@ -20,24 +20,9 @@
 		64DDA74E188FD7D10025590D /* Main.storyboard in Resources */ = {isa = PBXBuildFile; fileRef = 6400DDEA180C5DE000443362 /* Main.storyboard */; };
 		64DDA75B188FD9140025590D /* tessdata in Resources */ = {isa = PBXBuildFile; fileRef = 64DDA75A188FD9140025590D /* tessdata */; };
 		64E40AB1180C6D4D00C36DDE /* libstdc++.6.0.9.dylib in Frameworks */ = {isa = PBXBuildFile; fileRef = 64E40AB0180C6D4D00C36DDE /* libstdc++.6.0.9.dylib */; };
-		73BE4C121A598F86002C15F1 /* TesseractOCR.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 73C0A7BE1A59565100D823D4 /* TesseractOCR.framework */; };
-		73BE4C131A598F86002C15F1 /* TesseractOCR.framework in Embed Frameworks */ = {isa = PBXBuildFile; fileRef = 73C0A7BE1A59565100D823D4 /* TesseractOCR.framework */; settings = {ATTRIBUTES = (CodeSignOnCopy, RemoveHeadersOnCopy, ); }; };
+		73BE4C421A5D83AB002C15F1 /* TesseractOCR.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 73BE4C411A5D83AB002C15F1 /* TesseractOCR.framework */; };
 /* End PBXBuildFile section */
 
-/* Begin PBXCopyFilesBuildPhase section */
-		73BE4C141A598F86002C15F1 /* Embed Frameworks */ = {
-			isa = PBXCopyFilesBuildPhase;
-			buildActionMask = 2147483647;
-			dstPath = "";
-			dstSubfolderSpec = 10;
-			files = (
-				73BE4C131A598F86002C15F1 /* TesseractOCR.framework in Embed Frameworks */,
-			);
-			name = "Embed Frameworks";
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-/* End PBXCopyFilesBuildPhase section */
-
 /* Begin PBXFileReference section */
 		428615B61845F340005D5A2E /* image_sample.jpg */ = {isa = PBXFileReference; lastKnownFileType = image.jpeg; path = image_sample.jpg; sourceTree = "<group>"; };
 		6400DDD5180C5DE000443362 /* Template Framework Project.app */ = {isa = PBXFileReference; explicitFileType = wrapper.application; includeInIndex = 0; path = "Template Framework Project.app"; sourceTree = BUILT_PRODUCTS_DIR; };
@@ -57,6 +42,7 @@
 		6490748F198A5CD500D728CC /* CoreImage.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = CoreImage.framework; path = System/Library/Frameworks/CoreImage.framework; sourceTree = SDKROOT; };
 		64DDA75A188FD9140025590D /* tessdata */ = {isa = PBXFileReference; lastKnownFileType = folder; path = tessdata; sourceTree = "<group>"; };
 		64E40AB0180C6D4D00C36DDE /* libstdc++.6.0.9.dylib */ = {isa = PBXFileReference; lastKnownFileType = "compiled.mach-o.dylib"; name = "libstdc++.6.0.9.dylib"; path = "usr/lib/libstdc++.6.0.9.dylib"; sourceTree = SDKROOT; };
+		73BE4C411A5D83AB002C15F1 /* TesseractOCR.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = TesseractOCR.framework; path = "../build/Debug-iphoneos/TesseractOCR.framework"; sourceTree = "<group>"; };
 		73C0A7BE1A59565100D823D4 /* TesseractOCR.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; path = TesseractOCR.framework; sourceTree = BUILT_PRODUCTS_DIR; };
 /* End PBXFileReference section */
 
@@ -65,11 +51,11 @@
 			isa = PBXFrameworksBuildPhase;
 			buildActionMask = 2147483647;
 			files = (
+				73BE4C421A5D83AB002C15F1 /* TesseractOCR.framework in Frameworks */,
 				64907490198A5CD500D728CC /* CoreImage.framework in Frameworks */,
 				64E40AB1180C6D4D00C36DDE /* libstdc++.6.0.9.dylib in Frameworks */,
 				6400DDDB180C5DE000443362 /* CoreGraphics.framework in Frameworks */,
 				6400DDDD180C5DE000443362 /* UIKit.framework in Frameworks */,
-				73BE4C121A598F86002C15F1 /* TesseractOCR.framework in Frameworks */,
 				6400DDD9180C5DE000443362 /* Foundation.framework in Frameworks */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
@@ -97,6 +83,7 @@
 		6400DDD7180C5DE000443362 /* Frameworks */ = {
 			isa = PBXGroup;
 			children = (
+				73BE4C411A5D83AB002C15F1 /* TesseractOCR.framework */,
 				73C0A7BE1A59565100D823D4 /* TesseractOCR.framework */,
 				6490748F198A5CD500D728CC /* CoreImage.framework */,
 				64E40AB0180C6D4D00C36DDE /* libstdc++.6.0.9.dylib */,
@@ -144,7 +131,6 @@
 				6400DDD1180C5DE000443362 /* Sources */,
 				6400DDD2180C5DE000443362 /* Frameworks */,
 				6400DDD3180C5DE000443362 /* Resources */,
-				73BE4C141A598F86002C15F1 /* Embed Frameworks */,
 			);
 			buildRules = (
 			);
@@ -316,12 +302,17 @@
 					"$(inherited)",
 					"$(PROJECT_TEMP_DIR)/$(CONFIGURATION)$(EFFECTIVE_PLATFORM_NAME)",
 				);
+				GCC_INSTRUMENT_PROGRAM_FLOW_ARCS = NO;
 				GCC_PRECOMPILE_PREFIX_HEADER = YES;
 				GCC_PREFIX_HEADER = "Template Framework Project/Template Framework Project-Prefix.pch";
 				INFOPLIST_FILE = "Template Framework Project/Template Framework Project-Info.plist";
 				IPHONEOS_DEPLOYMENT_TARGET = 7.0;
 				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
 				ONLY_ACTIVE_ARCH = YES;
+				OTHER_LDFLAGS = (
+					"-lstdc++",
+					"-ObjC",
+				);
 				PRODUCT_NAME = "$(TARGET_NAME)";
 				WRAPPER_EXTENSION = app;
 			};
@@ -336,12 +327,17 @@
 					"$(inherited)",
 					"$(PROJECT_TEMP_DIR)/$(CONFIGURATION)$(EFFECTIVE_PLATFORM_NAME)",
 				);
+				GCC_INSTRUMENT_PROGRAM_FLOW_ARCS = NO;
 				GCC_PRECOMPILE_PREFIX_HEADER = YES;
 				GCC_PREFIX_HEADER = "Template Framework Project/Template Framework Project-Prefix.pch";
 				INFOPLIST_FILE = "Template Framework Project/Template Framework Project-Info.plist";
 				IPHONEOS_DEPLOYMENT_TARGET = 7.0;
 				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
 				ONLY_ACTIVE_ARCH = NO;
+				OTHER_LDFLAGS = (
+					"-lstdc++",
+					"-ObjC",
+				);
 				PRODUCT_NAME = "$(TARGET_NAME)";
 				WRAPPER_EXTENSION = app;
 			};

Tesseract OCR iOS.xcodeproj/project.pbxproj

@@ -581,8 +581,8 @@
 				COPY_PHASE_STRIP = NO;
 				GCC_C_LANGUAGE_STANDARD = gnu99;
 				GCC_DYNAMIC_NO_PIC = NO;
-				GCC_GENERATE_TEST_COVERAGE_FILES = YES;
-				GCC_INSTRUMENT_PROGRAM_FLOW_ARCS = YES;
+				GCC_GENERATE_TEST_COVERAGE_FILES = NO;
+				GCC_INSTRUMENT_PROGRAM_FLOW_ARCS = NO;
 				GCC_OPTIMIZATION_LEVEL = 0;
 				GCC_PREPROCESSOR_DEFINITIONS = (
 					"DEBUG=1",
@@ -664,6 +664,7 @@
 					"$(inherited)",
 					"$(PROJECT_DIR)/TesseractOCR/lib",
 				);
+				MACH_O_TYPE = staticlib;
 				MTL_ENABLE_DEBUG_INFO = YES;
 				PRODUCT_NAME = "$(TARGET_NAME)";
 				SKIP_INSTALL = YES;
@@ -704,6 +705,7 @@
 					"$(inherited)",
 					"$(PROJECT_DIR)/TesseractOCR/lib",
 				);
+				MACH_O_TYPE = staticlib;
 				MTL_ENABLE_DEBUG_INFO = NO;
 				PRODUCT_NAME = "$(TARGET_NAME)";
 				SKIP_INSTALL = YES;
@@ -714,13 +716,95 @@
 			};
 			name = Release;
 		};
+		F4EE1C791A612ADC007BB075 /* Coverage */ = {
+			isa = XCBuildConfiguration;
+			buildSettings = {
+				ALWAYS_SEARCH_USER_PATHS = NO;
+				CLANG_CXX_LANGUAGE_STANDARD = "gnu++0x";
+				CLANG_CXX_LIBRARY = "compiler-default";
+				CLANG_ENABLE_OBJC_ARC = YES;
+				CLANG_WARN_CONSTANT_CONVERSION = YES;
+				CLANG_WARN_EMPTY_BODY = YES;
+				CLANG_WARN_ENUM_CONVERSION = YES;
+				CLANG_WARN_INT_CONVERSION = YES;
+				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
+				COPY_PHASE_STRIP = NO;
+				GCC_C_LANGUAGE_STANDARD = gnu99;
+				GCC_DYNAMIC_NO_PIC = NO;
+				GCC_GENERATE_TEST_COVERAGE_FILES = YES;
+				GCC_INSTRUMENT_PROGRAM_FLOW_ARCS = YES;
+				GCC_OPTIMIZATION_LEVEL = 0;
+				GCC_PREPROCESSOR_DEFINITIONS = (
+					"DEBUG=1",
+					"$(inherited)",
+				);
+				GCC_SYMBOLS_PRIVATE_EXTERN = NO;
+				GCC_WARN_ABOUT_RETURN_TYPE = YES;
+				GCC_WARN_UNINITIALIZED_AUTOS = YES;
+				GCC_WARN_UNUSED_VARIABLE = YES;
+				INFOPLIST_FILE = "";
+				IPHONEOS_DEPLOYMENT_TARGET = 5.0;
+				ONLY_ACTIVE_ARCH = YES;
+				OTHER_LDFLAGS = (
+					"-ObjC",
+					"-lstdc++",
+				);
+				SDKROOT = iphoneos;
+			};
+			name = Coverage;
+		};
+		F4EE1C7A1A612ADC007BB075 /* Coverage */ = {
+			isa = XCBuildConfiguration;
+			buildSettings = {
+				CLANG_CXX_LIBRARY = "libc++";
+				CLANG_ENABLE_MODULES = YES;
+				CLANG_WARN_BOOL_CONVERSION = YES;
+				CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
+				CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
+				CLANG_WARN_UNREACHABLE_CODE = YES;
+				"CODE_SIGN_IDENTITY[sdk=iphoneos*]" = "iPhone Developer";
+				CURRENT_PROJECT_VERSION = 1;
+				DEFINES_MODULE = YES;
+				DYLIB_COMPATIBILITY_VERSION = 1;
+				DYLIB_CURRENT_VERSION = 1;
+				DYLIB_INSTALL_NAME_BASE = "@rpath";
+				ENABLE_STRICT_OBJC_MSGSEND = YES;
+				GCC_PREPROCESSOR_DEFINITIONS = (
+					"DEBUG=1",
+					"$(inherited)",
+				);
+				GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
+				GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
+				GCC_WARN_UNDECLARED_SELECTOR = YES;
+				GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
+				GCC_WARN_UNUSED_FUNCTION = YES;
+				INFOPLIST_FILE = "TesseractOCR/TesseractOCR-Info.plist";
+				INSTALL_PATH = "$(LOCAL_LIBRARY_DIR)/Frameworks";
+				IPHONEOS_DEPLOYMENT_TARGET = 7.0;
+				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks @loader_path/Frameworks";
+				LIBRARY_SEARCH_PATHS = (
+					"$(inherited)",
+					"$(PROJECT_DIR)/TesseractOCR/lib",
+				);
+				MACH_O_TYPE = staticlib;
+				MTL_ENABLE_DEBUG_INFO = YES;
+				PRODUCT_NAME = "$(TARGET_NAME)";
+				SKIP_INSTALL = YES;
+				TARGETED_DEVICE_FAMILY = "1,2";
+				VALID_ARCHS = "arm64 armv7 armv7s i386 x86_64";
+				VERSIONING_SYSTEM = "apple-generic";
+				VERSION_INFO_PREFIX = "";
+			};
+			name = Coverage;
+		};
 /* End XCBuildConfiguration section */
 
 /* Begin XCConfigurationList section */
 		64F74CC4172FD75F0068E657 /* Build configuration list for PBXProject "Tesseract OCR iOS" */ = {
 			isa = XCConfigurationList;
 			buildConfigurations = (
 				64F74CD5172FD75F0068E657 /* Debug */,
+				F4EE1C791A612ADC007BB075 /* Coverage */,
 				64F74CD6172FD75F0068E657 /* Release */,
 			);
 			defaultConfigurationIsVisible = 0;
@@ -730,6 +814,7 @@
 			isa = XCConfigurationList;
 			buildConfigurations = (
 				73C0A78F1A592C2B00D823D4 /* Debug */,
+				F4EE1C7A1A612ADC007BB075 /* Coverage */,
 				73C0A7901A592C2B00D823D4 /* Release */,
 			);
 			defaultConfigurationIsVisible = 0;

Tesseract OCR iOS.xcodeproj/xcshareddata/xcschemes/TesseractOCR.xcscheme

@@ -40,7 +40,7 @@
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
       shouldUseLaunchSchemeArgsEnv = "YES"
-      buildConfiguration = "Debug">
+      buildConfiguration = "Coverage">
       <Testables>
          <TestableReference
             skipped = "NO">

Tesseract-OCR-iOS.xcworkspace/xcshareddata/xcschemes/TestsProject.xcscheme

@@ -20,20 +20,6 @@
                ReferencedContainer = "container:TestsProject/TestsProject.xcodeproj">
             </BuildableReference>
          </BuildActionEntry>
-         <BuildActionEntry
-            buildForTesting = "YES"
-            buildForRunning = "YES"
-            buildForProfiling = "YES"
-            buildForArchiving = "YES"
-            buildForAnalyzing = "YES">
-            <BuildableReference
-               BuildableIdentifier = "primary"
-               BlueprintIdentifier = "64A0293917307C7E002B12E7"
-               BuildableName = "TesseractOCRAggregate"
-               BlueprintName = "TesseractOCRAggregate"
-               ReferencedContainer = "container:Tesseract OCR iOS.xcodeproj">
-            </BuildableReference>
-         </BuildActionEntry>
          <BuildActionEntry
             buildForTesting = "YES"
             buildForRunning = "YES"
@@ -54,7 +40,7 @@
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
       shouldUseLaunchSchemeArgsEnv = "YES"
-      buildConfiguration = "Debug">
+      buildConfiguration = "Coverage">
       <Testables>
          <TestableReference
             skipped = "NO">
@@ -82,7 +68,7 @@
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
       launchStyle = "0"
       useCustomWorkingDirectory = "NO"
-      buildConfiguration = "Debug"
+      buildConfiguration = "Coverage"
       ignoresPersistentStateOnLaunch = "NO"
       debugDocumentVersioning = "YES"
       allowLocationSimulation = "YES">

TesseractOCR/G8Tesseract.h

@@ -41,8 +41,9 @@
 @property (nonatomic, copy) NSString* language;
 
 /**
- * The path to the tessdata file, if it was specified in a call to initWithLanguage:configDictionary:configFileNames:cachesRelatedDataPath:engineMode: as a cachesRelatedDataPath
- * Otherwise it's supposed that the tessdata folder is located in the application bundle
+ *  The absolute path to the tessdata folder, which may exist in either the
+ *  application bundle or in the Caches directory depending on the argument to
+ *  `cachesRelatedDataPath` in the designated initializer.
  */
 @property (nonatomic, readonly, copy) NSString *absoluteDataPath;
 
@@ -112,6 +113,19 @@
  */
 @property (nonatomic, readonly) NSString *recognizedText;
 
+/**
+ *  Make an HTML-formatted string with hOCR markup from the internal Tesseract
+ *  data structures.
+ *  page_number is 0-based but will appear in the output as 1-based.
+ *
+ *  @param pageNumber The page number within the image of interest. If you
+ *                    aren't using a multipage image or don't know what this
+ *                    means, use `0` for `pageNumber`.
+ *
+ *  @return The HTML-formatted string with hOCR markup.
+ */
+- (NSString *)recognizedHOCRForPageNumber:(int)pageNumber;
+
 /**
  *  The result of Tesseract's orientation analysis of the target image. See
  *  `G8Orientation` in G8Constants.h for the possible orientations.
@@ -168,7 +182,7 @@
 
 /**
  *  Retrieve Tesseract's recognition result based on a provided resolution.
- *  For, example for the pageIteratorLevel == G8PageIteratorLevelSymbol it returns 
+ *  For example, the pageIteratorLevel == G8PageIteratorLevelSymbol returns
  *  an array of `G8RecognizedBlock`'s representing the characters recognized
  *  in the target image, including the bounding boxes for each character.
  *
@@ -177,8 +191,9 @@
  *                           resolution options.
  *
  *  @return An array of `G8RecognizedBlock`'s, each containing a confidence
- *          value and a bounding box for the text it represents. See G8RecognizedBlock.h for more
- *          information about the available fields for this data structure.
+ *          value and a bounding box for the text it represents. See 
+ *          G8RecognizedBlock.h for more information about the available fields 
+ *          for this data structure.
  */
 - (NSArray *)recognizedBlocksByIteratorLevel:(G8PageIteratorLevel)pageIteratorLevel;
 
@@ -231,17 +246,25 @@
 /**
  *  Initialize Tesseract with the provided language and engine mode.
  *
- *  @param language             The language to use in recognition. See `language`.
- *  @param configDictionary     A dictionary of the config variables
- *  @param configFileNames      An array of file names containing key-value config pairs. All the config
- *                              variables can be init only and debug time both. Furthermore they could be
- *                              specified at the same time, in such case tesseract will get variables from
- *                              every file and dictionary all together. 
- *                              The files are searched into two folders, which are tessdata/tessconfigs and tessdata/configs
- *  @param cachesRelatedPath    If the cachesRelatedDataPath is specified, the whole content of the tessdata from the
- *                              application bundle is copied to the Library/Caches/cachesRelatedDataPath/tessdata
- *                              and tesseract is initialized with that path.
- *  @param engineMode           The engine mode to use in recognition. See `engineMode`.
+ *  @param language             The language to use in recognition. See 
+ *                              `language`.
+ *  @param configDictionary     A dictionary of config variables to set.
+ *  @param configFileNames      An array of file names containing key-value 
+ *                              config pairs. Config settings can be set at
+ *                              initialization or run-time.  Furthermore, they 
+ *                              could be specified at the same time, in which 
+ *                              case Tesseract will get variables from every
+ *                              config file as well as the dictionary.
+ *                              The config files must exist in one of two 
+ *                              possible folders:  tessdata/tessconfigs or 
+ *                              tessdata/configs.
+ *  @param cachesRelatedPath    If the cachesRelatedDataPath is specified, the 
+ *                              whole contents of the tessdata folder in the
+ *                              application bundle will be copied to 
+ *                              Library/Caches/cachesRelatedDataPath/tessdata
+ *                              and Tesseract will be set to use that path.
+ *  @param engineMode           The engine mode to use in recognition. See 
+ *                              `engineMode`.
  *
  *  @return The initialized Tesseract object, or `nil` if there was an error.
  */
@@ -262,12 +285,13 @@
 - (void)setVariableValue:(NSString *)value forKey:(NSString *)key;
 
 /**
- *  Returns a Tesseract variable for the given key. See G8TesseractParameters.h for the available
- *  options.
+ *  Returns a Tesseract variable for the given key. See G8TesseractParameters.h 
+ *  for the available options.
  *
  *  @param key  The option to get.
  *
- *  @return     returns the variable value for the given key, if it's beeb set. nil otherwise.
+ *  @return     Returns the variable value for the given key, if it's been set. 
+ *              nil otherwise.
  */
 - (NSString*)variableValueForKey:(NSString *)key;