tesseract-ocr 0.0.4 → 0.1.0

data/examples/nerdz-captcha-breaker/test.rb

@@ -0,0 +1,11 @@
+#! /usr/bin/env ruby
+
+expected = File.read('captchas/captchas.txt').lines.to_a
+
+`./break.rb captchas/*.png`.lines.each_with_index {|line, index|
+	whole, path, output = line.match(/^(.*?): (.*?)$/).to_a
+
+	unless output == expected[index].chomp
+		puts "#{path}: expected #{expected[index].chomp} but got #{output}"
+	end
+}

data/lib/tesseract/engine.rb

@@ -178,6 +178,23 @@ class Engine
 			_iterator.__send__ "each_#{level}", &block
 		end
 
+		namedic :image, :x, :y, :width, :height,
+			:optional => 0 .. -1,
+			:alias    => { :w => :width, :h => :height }
+		define_method "each_#{level}_for" do |image = nil, x = nil, y = nil, width = nil, height = nil, &block|
+			self.image = image if image
+			select x, y, width, height
+
+			__send__ "each_#{level}", &block
+		end
+
+		namedic :x, :y, :width, :height,
+			:optional => 0 .. -1,
+			:alias    => { :w => :width, :h => :height }
+		define_method "each_#{level}_at" do |x = nil, y = nil, width = nil, height = nil, &block|
+			__send__ "each_#{level}_for", nil, x, y, width, height, &block
+		end
+
 		define_method "#{level}s" do
 			_iterator.__send__ "#{level}s"
 		end
@@ -189,14 +206,14 @@ class Engine
 			self.image = image if image
 			select x, y, width, height
 
-			__send__("#{level}s")
+			__send__ "#{level}s"
 		end
 
 		namedic :x, :y, :width, :height,
 			:optional => 0 .. -1,
 			:alias    => { :w => :width, :h => :height }
 		define_method "#{level}s_at" do |x = nil, y = nil, width = nil, height = nil|
-			__send__("#{level}s_for", nil, x, y, width, height)
+			__send__ "#{level}s_for", nil, x, y, width, height
 		end
 	}
 

data/lib/tesseract/extensions.rb

@@ -25,4 +25,3 @@
 require 'namedic'
 require 'memoized'
 require 'iso-639'
-require 'io/manage'

data/lib/tesseract/version.rb

@@ -24,6 +24,6 @@
 
 module Tesseract
 	def self.version
-		'0.0.4'
+		'0.1.0'
 	end
 end

data/test/tesseract_spec.rb

@@ -78,6 +78,15 @@ describe Tesseract::Engine do
 		end
 	end
 
+	describe '#page_segmentation_mode' do
+		it 'sets it correctly' do
+			engine.with {|e|
+				e.page_segmentation_mode = 8
+				e.whitelist              = [*'a'..'z', *'A'..'Z', *0..9, " ."].join
+			}.text_for('jsmj.png').strip.should == 'JSmj'
+		end
+	end
+
 	describe '#blocks' do
 		it 'works properly with first image' do
 			engine.blocks_for('first.png').first.to_s.should == "ABC\n"

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tesseract-ocr
 version: !ruby/object:Gem::Version
-  version: 0.0.4
+  version: 0.1.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,11 +9,11 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2011-11-30 00:00:00.000000000 Z
+date: 2011-12-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: namedic
-  requirement: &12863500 !ruby/object:Gem::Requirement
+  requirement: &20715740 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -21,10 +21,10 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *12863500
+  version_requirements: *20715740
 - !ruby/object:Gem::Dependency
   name: memoized
-  requirement: &12861460 !ruby/object:Gem::Requirement
+  requirement: &20730960 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -32,10 +32,10 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *12861460
+  version_requirements: *20730960
 - !ruby/object:Gem::Dependency
   name: iso-639
-  requirement: &12888100 !ruby/object:Gem::Requirement
+  requirement: &20729480 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -43,10 +43,10 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *12888100
+  version_requirements: *20729480
 - !ruby/object:Gem::Dependency
   name: ffi-extra
-  requirement: &12886020 !ruby/object:Gem::Requirement
+  requirement: &20727420 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -54,10 +54,10 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *12886020
+  version_requirements: *20727420
 - !ruby/object:Gem::Dependency
   name: ffi-inliner
-  requirement: &12882760 !ruby/object:Gem::Requirement
+  requirement: &20726120 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -65,20 +65,126 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *12882760
+  version_requirements: *20726120
 description: 
 email: meh@paranoici.org
 executables:
+- tesseract-train.rb
 - tesseract.rb
 extensions: []
 extra_rdoc_files: []
 files:
 - README.md
 - Rakefile
+- bin/tesseract-train.rb
 - bin/tesseract.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
+- examples/nerdz-captcha-breaker/captchas/001.png
+- examples/nerdz-captcha-breaker/captchas/002.png
+- examples/nerdz-captcha-breaker/captchas/003.png
+- examples/nerdz-captcha-breaker/captchas/004.png
+- examples/nerdz-captcha-breaker/captchas/005.png
+- examples/nerdz-captcha-breaker/captchas/006.png
+- examples/nerdz-captcha-breaker/captchas/007.png
+- examples/nerdz-captcha-breaker/captchas/008.png
+- examples/nerdz-captcha-breaker/captchas/009.png
+- examples/nerdz-captcha-breaker/captchas/010.png
+- examples/nerdz-captcha-breaker/captchas/011.png
+- examples/nerdz-captcha-breaker/captchas/012.png
+- examples/nerdz-captcha-breaker/captchas/013.png
+- examples/nerdz-captcha-breaker/captchas/014.png
+- examples/nerdz-captcha-breaker/captchas/015.png
+- examples/nerdz-captcha-breaker/captchas/016.png
+- examples/nerdz-captcha-breaker/captchas/017.png
+- examples/nerdz-captcha-breaker/captchas/018.png
+- examples/nerdz-captcha-breaker/captchas/019.png
+- examples/nerdz-captcha-breaker/captchas/020.png
+- examples/nerdz-captcha-breaker/captchas/021.png
+- examples/nerdz-captcha-breaker/captchas/022.png
+- examples/nerdz-captcha-breaker/captchas/023.png
+- examples/nerdz-captcha-breaker/captchas/024.png
+- examples/nerdz-captcha-breaker/captchas/025.png
+- examples/nerdz-captcha-breaker/captchas/026.png
+- examples/nerdz-captcha-breaker/captchas/027.png
+- examples/nerdz-captcha-breaker/captchas/028.png
+- examples/nerdz-captcha-breaker/captchas/029.png
+- examples/nerdz-captcha-breaker/captchas/030.png
+- examples/nerdz-captcha-breaker/captchas/031.png
+- examples/nerdz-captcha-breaker/captchas/032.png
+- examples/nerdz-captcha-breaker/captchas/033.png
+- examples/nerdz-captcha-breaker/captchas/034.png
+- examples/nerdz-captcha-breaker/captchas/035.png
+- examples/nerdz-captcha-breaker/captchas/036.png
+- examples/nerdz-captcha-breaker/captchas/037.png
+- examples/nerdz-captcha-breaker/captchas/038.png
+- examples/nerdz-captcha-breaker/captchas/039.png
+- examples/nerdz-captcha-breaker/captchas/040.png
+- examples/nerdz-captcha-breaker/captchas/041.png
+- examples/nerdz-captcha-breaker/captchas/042.png
+- examples/nerdz-captcha-breaker/captchas/043.png
+- examples/nerdz-captcha-breaker/captchas/044.png
+- examples/nerdz-captcha-breaker/captchas/045.png
+- examples/nerdz-captcha-breaker/captchas/046.png
+- examples/nerdz-captcha-breaker/captchas/047.png
+- examples/nerdz-captcha-breaker/captchas/048.png
+- examples/nerdz-captcha-breaker/captchas/049.png
+- examples/nerdz-captcha-breaker/captchas/050.png
+- examples/nerdz-captcha-breaker/captchas/051.png
+- examples/nerdz-captcha-breaker/captchas/052.png
+- examples/nerdz-captcha-breaker/captchas/053.png
+- examples/nerdz-captcha-breaker/captchas/054.png
+- examples/nerdz-captcha-breaker/captchas/055.png
+- examples/nerdz-captcha-breaker/captchas/056.png
+- examples/nerdz-captcha-breaker/captchas/057.png
+- examples/nerdz-captcha-breaker/captchas/058.png
+- examples/nerdz-captcha-breaker/captchas/059.png
+- examples/nerdz-captcha-breaker/captchas/060.png
+- examples/nerdz-captcha-breaker/captchas/061.png
+- examples/nerdz-captcha-breaker/captchas/062.png
+- examples/nerdz-captcha-breaker/captchas/063.png
+- examples/nerdz-captcha-breaker/captchas/064.png
+- examples/nerdz-captcha-breaker/captchas/065.png
+- examples/nerdz-captcha-breaker/captchas/066.png
+- examples/nerdz-captcha-breaker/captchas/067.png
+- examples/nerdz-captcha-breaker/captchas/068.png
+- examples/nerdz-captcha-breaker/captchas/069.png
+- examples/nerdz-captcha-breaker/captchas/070.png
+- examples/nerdz-captcha-breaker/captchas/071.png
+- examples/nerdz-captcha-breaker/captchas/072.png
+- examples/nerdz-captcha-breaker/captchas/073.png
+- examples/nerdz-captcha-breaker/captchas/074.png
+- examples/nerdz-captcha-breaker/captchas/075.png
+- examples/nerdz-captcha-breaker/captchas/076.png
+- examples/nerdz-captcha-breaker/captchas/077.png
+- examples/nerdz-captcha-breaker/captchas/078.png
+- examples/nerdz-captcha-breaker/captchas/079.png
+- examples/nerdz-captcha-breaker/captchas/080.png
+- examples/nerdz-captcha-breaker/captchas/081.png
+- examples/nerdz-captcha-breaker/captchas/082.png
+- examples/nerdz-captcha-breaker/captchas/083.png
+- examples/nerdz-captcha-breaker/captchas/084.png
+- examples/nerdz-captcha-breaker/captchas/085.png
+- examples/nerdz-captcha-breaker/captchas/086.png
+- examples/nerdz-captcha-breaker/captchas/087.png
+- examples/nerdz-captcha-breaker/captchas/088.png
+- examples/nerdz-captcha-breaker/captchas/089.png
+- examples/nerdz-captcha-breaker/captchas/090.png
+- examples/nerdz-captcha-breaker/captchas/091.png
+- examples/nerdz-captcha-breaker/captchas/092.png
+- examples/nerdz-captcha-breaker/captchas/093.png
+- examples/nerdz-captcha-breaker/captchas/094.png
+- examples/nerdz-captcha-breaker/captchas/095.png
+- examples/nerdz-captcha-breaker/captchas/096.png
+- examples/nerdz-captcha-breaker/captchas/097.png
+- examples/nerdz-captcha-breaker/captchas/098.png
+- examples/nerdz-captcha-breaker/captchas/099.png
+- examples/nerdz-captcha-breaker/captchas/100.png
+- examples/nerdz-captcha-breaker/captchas/captchas.txt
+- examples/nerdz-captcha-breaker/tessdata/generate.rb
+- examples/nerdz-captcha-breaker/tessdata/lol.box
+- examples/nerdz-captcha-breaker/tessdata/lol.tif
+- examples/nerdz-captcha-breaker/tessdata/lol.traineddata
+- examples/nerdz-captcha-breaker/test.rb
 - lib/tesseract-ocr.rb
 - lib/tesseract.rb
 - lib/tesseract/api.rb
@@ -99,6 +205,7 @@ files:
 - lib/tesseract/version.rb
 - tesseract-ocr.gemspec
 - test/first.png
+- test/jsmj.png
 - test/second.png
 - test/tesseract_bench.rb
 - test/tesseract_spec.rb
@@ -130,6 +237,7 @@ specification_version: 3
 summary: A wrapper library to the tesseract-ocr API.
 test_files:
 - test/first.png
+- test/jsmj.png
 - test/second.png
 - test/tesseract_bench.rb
 - test/tesseract_spec.rb

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

@@ -1,112 +0,0 @@
-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