tesseract-ocr 0.0.3 → 0.0.4

data/examples/{nerdz-captcha-breaker.rb → nerdz-captcha-breaker/break.rb}

@@ -51,6 +51,8 @@ Tesseract::Engine.new.tap {|engine|
 			image.pixel_color x, y, 'black'
 		}
 
+		File.open('/tmp/lol.png', ?w) { |f| f.write(image.resize(10).to_blob) }
+
 		puts engine.text_for(image.resize 10).strip
 	}
 }

data/examples/nerdz-captcha-breaker/lol.gd-giant.exp.box

@@ -0,0 +1,112 @@
+T 31 46 39 56 0
+h 40 46 48 56 0
+e 49 46 57 53 0
+( 69 45 74 57 0
+q 76 44 84 53 0
+u 85 46 93 53 0
+i 95 46 101 56 0
+c 103 46 111 53 0
+k 113 46 120 56 0
+) 122 45 127 57 0
+b 139 46 147 56 0
+r 148 46 156 53 0
+o 157 46 165 53 0
+w 166 46 174 53 0
+n 175 46 183 53 0
+{ 195 46 200 56 0
+f 202 46 210 56 0
+o 211 46 219 53 0
+x 220 46 228 53 0
+} 230 46 235 56 0
+j 247 44 254 56 0
+u 256 46 264 53 0
+m 265 46 273 53 0
+p 274 44 282 53 0
+s 283 46 291 53 0
+! 295 45 297 56 0
+o 310 46 318 53 0
+v 319 46 327 53 0
+e 328 46 336 53 0
+r 337 46 345 53 0
+t 355 46 363 55 0
+h 364 46 372 56 0
+e 373 46 381 53 0
+$ 392 45 399 56 0
+3 400 46 408 56 0
+, 410 45 415 48 0
+4 418 46 426 56 0
+5 427 46 435 56 0
+6 436 46 444 56 0
+. 447 45 451 48 0
+7 454 46 462 56 0
+8 463 46 471 56 0
+< 482 46 488 56 0
+l 492 46 496 56 0
+a 499 46 507 53 0
+z 509 46 515 53 0
+y 517 44 525 53 0
+> 526 46 532 56 0
+# 544 47 551 55 0
+9 553 46 561 56 0
+0 562 46 570 56 0
+d 580 46 588 56 0
+o 589 46 597 53 0
+g 598 44 606 53 0
+& 616 46 624 56 0
+d 634 46 642 56 0
+u 643 46 651 53 0
+c 652 46 660 53 0
+k 662 46 669 56 0
+/ 670 46 678 56 0
+g 679 44 687 53 0
+o 688 46 696 53 0
+o 697 46 705 53 0
+s 706 46 714 53 0
+e 715 46 723 53 0
+, 725 45 730 48 0
+a 742 46 750 53 0
+s 751 46 759 53 0
+1 770 46 776 56 0
+2 778 46 786 56 0
+. 789 45 793 48 0
+5 796 46 804 56 0
+% 805 46 813 56 0
+o 823 46 831 53 0
+f 832 46 840 56 0
+E 850 46 857 56 0
+- 859 50 867 51 0
+m 868 46 876 53 0
+a 877 46 885 53 0
+i 887 46 893 56 0
+l 897 46 901 56 0
+f 913 46 921 56 0
+r 922 46 930 53 0
+o 931 46 939 53 0
+m 940 46 948 53 0
+a 958 46 966 53 0
+s 967 46 975 53 0
+p 976 44 984 53 0
+a 985 46 993 53 0
+m 994 46 1002 53 0
+m 1003 46 1011 53 0
+e 1012 46 1020 53 0
+r 1021 46 1029 53 0
+@ 1030 46 1038 56 0
+w 1039 46 1047 53 0
+e 1048 46 1056 53 0
+b 1057 46 1065 56 0
+s 1066 46 1074 53 0
+i 1076 46 1082 56 0
+t 1084 46 1092 55 0
+e 1093 46 1101 53 0
+. 1104 45 1108 48 0
+c 1111 46 1119 53 0
+o 1120 46 1128 53 0
+m 1129 46 1137 53 0
+i 1148 46 1154 56 0
+s 1156 46 1164 53 0
+s 1174 46 1182 53 0
+p 1183 44 1191 53 0
+a 1192 46 1200 53 0
+m 1201 46 1209 53 0
+? 1211 46 1218 56 0

data/lib/tesseract/api/iterator.rb

@@ -77,9 +77,10 @@ class Iterator
 		pointer = C::Iterator.get_utf8_text(to_ffi, C.for_enum(level))
 		result  = pointer.read_string
 		result.force_encoding 'UTF-8'
-		C.free_string(pointer)
 
 		result
+	ensure
+		C.free_array_of_char(pointer)
 	end
 
 	def confidence (level = :word)

data/lib/tesseract/api.rb

@@ -65,11 +65,11 @@ class API
 		C::BaseAPI.version(to_ffi)
 	end
 
-	def input_name= (name)
+	def set_input_name (name)
 		C::BaseAPI.set_input_name(to_ffi, name)
 	end
 
-	def output_name= (name)
+	def set_output_name (name)
 		C::BaseAPI.set_output_name(to_ffi, name)
 	end
 
@@ -125,6 +125,30 @@ class API
 		C::BaseAPI.set_rectangle(to_ffi, left, top, width, height)
 	end
 
+	def process_pages (name)
+		result = C.create_string
+
+		unless C::BaseAPI.process_pages(to_ffi, name, result)
+			raise 'process_pages failed'
+		end
+
+		C.string_content(result).read_string(C.string_length(result))
+	ensure
+		C.destroy_string(result)
+	end
+
+	def process_page (pix, page = 0, name = "")
+		result = C.create_string
+
+		unless C::BaseAPI.process_page(to_ffi, pix.is_a?(Image) ? pix.to_ffi : pix, page, name, result)
+			raise 'process_page failed'
+		end
+
+		C.string_content(result).read_string(C.string_length(result))
+	ensure
+		C.destroy_string(result)
+	end
+
 	def get_iterator
 		Iterator.new(C::BaseAPI.get_iterator(to_ffi))
 	end
@@ -133,27 +157,30 @@ class API
 		pointer = C::BaseAPI.get_utf8_text(to_ffi)
 		result  = pointer.read_string
 		result.force_encoding 'UTF-8'
-		C.free_string(pointer)
 
 		result
+	ensure
+		C.free_array_of_char(pointer)
 	end
 
 	def get_box (page = 0)
 		pointer = C::BaseAPI.get_box_text(to_ffi, page)
 		result  = pointer.read_string
 		result.force_encoding 'UTF-8'
-		C.free_string(pointer)
 
 		result
+	ensure
+		C.free_array_of_char(pointer)
 	end
 
 	def get_unlv
 		pointer = C::BaseAPI.get_unlv_text(to_ffi)
 		result  = pointer.read_string
 		result.force_encoding 'ISO8859-1'
-		C.free_string(pointer)
 
 		result
+	ensure
+		C.free_array_of_char(pointer)
 	end
 
 	def mean_text_confidence

data/lib/tesseract/c/baseapi.rb

@@ -184,6 +184,18 @@ module BaseAPI
 			}
 		}
 
+		cpp.function %{
+			bool process_pages (TessBaseAPI* api, const char* filename, STRING* output) {
+				return api->ProcessPages(filename, NULL, 0, output);
+			}
+		}
+
+		cpp.function %{
+			bool process_page (TessBaseAPI* api, Pix* pix, int page_index, const char* filename, STRING* output) {
+				return api->ProcessPage(pix, page_index, filename, NULL, 0, output);
+			}
+		}
+
 		cpp.function %{
 			ResultIterator* get_iterator (TessBaseAPI* api) {
 				return api->GetIterator();

data/lib/tesseract/c.rb

@@ -32,8 +32,11 @@ module C
 	extend FFI::Inliner
 
 	inline 'C++' do |cpp|
+		cpp.include   'tesseract/strngs.h'
+		cpp.libraries 'tesseract'
+
 		cpp.function %{
-			void free_string (char* pointer) {
+			void free_array_of_char (char* pointer) {
 				delete [] pointer;
 			}
 		}
@@ -43,6 +46,30 @@ module C
 				delete [] pointer;
 			}
 		}
+
+		cpp.function %{
+			STRING* create_string (void) {
+				return new STRING();
+			}
+		}
+
+		cpp.function %{
+			void destroy_string (STRING* value) {
+				delete value;
+			}
+		}
+
+		cpp.function %{
+			int string_length (STRING* value) {
+				return value->length();
+			}
+		}
+
+		cpp.function %{
+			const char* string_content (STRING* value) {
+				return value->string();
+			}
+		}
 	end
 
 	def self.for_enum (what)

data/lib/tesseract/engine.rb

@@ -75,6 +75,14 @@ class Engine
 		}
 	end
 
+	def input= (name)
+		@api.set_input_name(name)
+	end
+
+	def output= (name)
+		@api.set_output_name(name)
+	end
+
 	def set (name, value)
 		@variables[name] = value
 
@@ -118,7 +126,9 @@ class Engine
 	end
 
 	def page_segmentation_mode= (value)
-		@api.set_page_seg_mode C.for_enum(value)
+		@psm = C.for_enum(value)
+
+		@api.set_page_seg_mode @psm
 	end
 
 	def image= (image)
@@ -190,6 +200,16 @@ class Engine
 		end
 	}
 
+	def process (image, page = nil)
+		if page
+			@api.process_page(API.image_for(image), page)
+		else
+			raise ArgumentError, 'the path does not exist' unless File.exists?(image)
+
+			@api.process_pages(image)
+		end
+	end
+
 protected
 	def _init
 		@api.end
@@ -203,6 +223,8 @@ protected
 		@config.each {|conf|
 			@api.read_config_file(conf)
 		}
+
+		@api.set_page_seg_mode @psm if @psm
 	end
 
 	def _setup (image = nil, x = nil, y = nil, width = nil, height = nil)

data/lib/tesseract/version.rb

@@ -24,6 +24,6 @@
 
 module Tesseract
 	def self.version
-		'0.0.3'
+		'0.0.4'
 	end
 end

data/test/tesseract_spec.rb

@@ -78,9 +78,33 @@ describe Tesseract::Engine do
 		end
 	end
 
-	describe '#each_block' do
+	describe '#blocks' do
 		it 'works properly with first image' do
+			engine.blocks_for('first.png').first.to_s.should == "ABC\n"
+		end
+	end
+
+	describe '#paragraphs' do
+		it 'works properly with first image' do
+			engine.paragraphs_for('first.png').first.to_s.should == "ABC\n"
+		end
+	end
 
+	describe '#lines' do
+		it 'works properly with first image' do
+			engine.lines_for('first.png').first.to_s.should == "ABC\n"
+		end
+	end
+
+	describe '#words' do
+		it 'works properly with first image' do
+			engine.words_for('first.png').first.to_s.should == 'ABC'
+		end
+	end
+
+	describe '#symbols' do
+		it 'works properly with first image' do
+			engine.symbols_for('first.png').first.to_s.should == 'A'
 		end
 	end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tesseract-ocr
 version: !ruby/object:Gem::Version
-  version: 0.0.3
+  version: 0.0.4
   prerelease: 
 platform: ruby
 authors:
@@ -9,11 +9,11 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2011-11-29 00:00:00.000000000 Z
+date: 2011-11-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: namedic
-  requirement: &3194800 !ruby/object:Gem::Requirement
+  requirement: &12863500 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -21,10 +21,10 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *3194800
+  version_requirements: *12863500
 - !ruby/object:Gem::Dependency
   name: memoized
-  requirement: &3193660 !ruby/object:Gem::Requirement
+  requirement: &12861460 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -32,10 +32,10 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *3193660
+  version_requirements: *12861460
 - !ruby/object:Gem::Dependency
   name: iso-639
-  requirement: &3207460 !ruby/object:Gem::Requirement
+  requirement: &12888100 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -43,10 +43,10 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *3207460
+  version_requirements: *12888100
 - !ruby/object:Gem::Dependency
   name: ffi-extra
-  requirement: &3205940 !ruby/object:Gem::Requirement
+  requirement: &12886020 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -54,10 +54,10 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *3205940
+  version_requirements: *12886020
 - !ruby/object:Gem::Dependency
   name: ffi-inliner
-  requirement: &3205380 !ruby/object:Gem::Requirement
+  requirement: &12882760 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -65,7 +65,7 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *3205380
+  version_requirements: *12882760
 description: 
 email: meh@paranoici.org
 executables:
@@ -76,7 +76,9 @@ files:
 - README.md
 - Rakefile
 - bin/tesseract.rb
-- examples/nerdz-captcha-breaker.rb
+- examples/nerdz-captcha-breaker/break.rb
+- examples/nerdz-captcha-breaker/lol.gd-giant.exp.box
+- examples/nerdz-captcha-breaker/lol.gd-giant.exp.tif
 - lib/tesseract-ocr.rb
 - lib/tesseract.rb
 - lib/tesseract/api.rb