rhocr 0.1.1 → 0.1.3

data/Manifest

@@ -17,7 +17,6 @@ lib/ocr_document.rb
 lib/ocr_element.rb
 lib/ocr_page.rb
 lib/rhocr.rb
-rhocr.gemspec
 spec/hocr_box_spec.rb
 spec/ocr_document_spec.rb
 spec/ocr_element_spec.rb

data/Rakefile

@@ -2,7 +2,7 @@ require 'rubygems'
 require 'rake'
 require 'echoe'
 
-Echoe.new('rhocr', '0.1.1') do |p|
+Echoe.new('rhocr', '0.1.3') do |p|
   p.description    = "Manipulate and use OCR data encoded in hOCR-Format see: http://code.google.com/p/hocr-tools/"
   p.url            = "http://github.com/daandi/rhocr"
   p.author         = "Andreas Neumann"

data/lib/hocr_box.rb

@@ -2,15 +2,13 @@
 
 class HOCRBox
     
-    attr_reader :left, :top, :right, :bottom, :upper_left, :lower_right, :coordinates
+    attr_reader :left, :top, :right, :width, :height, :bottom, :coordinates
     def initialize(* coordinates)
         
         @left, @top, @right, @bottom = coordinates.flatten.collect { |x| x.to_i}
         
         @height  = @bottom - @top
         @width   = @right - @left
-        @upper_left = [ @left, @top]
-        @lower_rigth = [ @right, @bottom ]
         @coordinates = [ @left, @top,@right, @bottom ]
         
         if left > right || top > bottom then

data/lib/ocr_document.rb

@@ -11,7 +11,8 @@ class OCRDocument
     end
     
     def add_pages( list_o_pages  )
-        for file in list_o_pages do
+        raise "no files given" if list_o_pages.empty?
+        list_o_pages.each do |file|
             add_page(file)
         end
     end

data/lib/ocr_element.rb

@@ -86,16 +86,28 @@ class OCRElement < HOCRBox
         "<span style='color: #{color}'>#{to_s}</span>"
     end
     
-    def to_image_html(dipslay_class = @ocr_class)
+    def css_class_string
+        if @features.empty?
+            "#{@ocr_class}"
+        else
+            "#{@ocr_class}-#{features_to_css_class}"
+        end
+    end
+    
+    def to_image_html(dipslay_class =css_class_string)
         children_html = @children.map {|c| c.to_image_html}.join("")
         "<span class='#{ dipslay_class }' style='#{ to_css_style }' ></span>#{ children_html }"
     end
     
-    def to_html( display_class = @ocr_class, style = nil )
+    def to_html( display_class = css_class_string, style = nil )
          children_html = @children.map {|c| c.to_html}.join("")
         "<span class='#{ display_class }'> #{ children_html } </span>"
     end
     
+    def features_to_css_class
+        @features.uniq.sort.join('_')
+    end
+    
 end
 
 class OCRWord < OCRElement
@@ -109,11 +121,11 @@ class OCRWord < OCRElement
     end
     
     def to_image_html
-        "<span class='#{ @ocr_class }' style='#{ to_css_style }'>#{ text }</span>"
+        "<span class='#{ css_class_string}' style='#{ to_css_style }'>#{ text }</span>"
     end
     
     def to_html
-        "<span class='#{ @ocr_class }'>#{ text }</span>"
+        "<span class='#{ css_class_string }'>#{ text }</span>"
     end
     
 end

data/lib/ocr_page.rb

@@ -52,18 +52,6 @@ class OCRPage < OCRElement
         end
     end
     
-    #deprecated
-    def lines
-        unless @lines then
-            @lines = []
-            
-            each_line do |line|
-                @lines << line
-            end
-            
-        end
-        @lines
-    end
     
     def extract_bbox_ppageno( ocr_html_text_fragment )
         bbox, ppageno = ocr_html_text_fragment.split(';')
@@ -77,7 +65,7 @@ class OCRPage < OCRElement
     end
         
     def to_text
-        lines.map {|line| line.to_text}.join("\n")
+        Enumerator.new(self,:each_line).map {|line| line.to_text}.join("\n")
     end
     
     def to_image_html(dipslay_class = @ocr_class)
@@ -86,13 +74,16 @@ class OCRPage < OCRElement
     end
     
     def enclosed_words(ocr_box)
-        a = []
+        enum = Enumerator.new(self,:each_enclosed_word,ocr_box)
+        enum.inject([]) { |acc,w| acc << w}
+    end
+    
+    def each_enclosed_word(ocr_box)
         each_word do |w|
             if w.enclosed_by? ocr_box then
-                a << w
+                yield w
             end
         end
-        a
     end
     
 end

data/lib/rhocr.rb

@@ -3,29 +3,16 @@
 require_relative "ocr_document"
 class RHOCR < OCRDocument
     
-    attr_reader :words, :lines
-    
     def add_folder(path)
         add_files Dir[path]
-        compute_lines
-        compute_words
-        self
     end
     
-    #should be called if new pages are added
-    def compute_words
-        @words = []
-        each_word do |w|
-            @words << w
-        end
+    def words
+        Enumerator.new(self,:each_word)
     end
     
-    #should be called if new pages are added
-    def compute_lines
-        @lines = []
-        each_line do |l|
-            @lines << l
-        end
+    def lines
+        Enumerator.new(self,:each_line)
     end
     
 end

data/rhocr.gemspec

@@ -2,15 +2,15 @@
 
 Gem::Specification.new do |s|
   s.name = %q{rhocr}
-  s.version = "0.1.1"
+  s.version = "0.1.3"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 1.2") if s.respond_to? :required_rubygems_version=
   s.authors = [%q{Andreas Neumann}]
-  s.date = %q{2011-09-30}
+  s.date = %q{2011-10-10}
   s.description = %q{Manipulate and use OCR data encoded in hOCR-Format see: http://code.google.com/p/hocr-tools/}
   s.email = %q{andreas@neumann.biz}
   s.extra_rdoc_files = [%q{README}, %q{lib/hocr_box.rb}, %q{lib/ocr_document.rb}, %q{lib/ocr_element.rb}, %q{lib/ocr_page.rb}, %q{lib/rhocr.rb}]
-  s.files = [%q{Manifest}, %q{README}, %q{Rakefile}, %q{data/Seite_Die_Gartenlaube_242.html}, %q{data/Seite_Tagebuch_H_C_Lang_08.html}, %q{data/Seite_Tagebuch_H_C_Lang_08.jpg}, %q{data/test.html}, %q{data/test.png}, %q{example/example_server.rb}, %q{example/public/OCRTest.css}, %q{example/public/OCRTest.html}, %q{example/public/OCRTest_marker.js}, %q{example/public/Seite_Tagebuch_H_C_Lang_08.jpg}, %q{example/public/img/Seite_Tagebuch_H_C_Lang_08.jpg}, %q{lib/hocr_box.rb}, %q{lib/ocr_document.rb}, %q{lib/ocr_element.rb}, %q{lib/ocr_page.rb}, %q{lib/rhocr.rb}, %q{rhocr.gemspec}, %q{spec/hocr_box_spec.rb}, %q{spec/ocr_document_spec.rb}, %q{spec/ocr_element_spec.rb}, %q{spec/ocr_page_spec.rb}, %q{spec/rhocr_spec.rb}, %q{test.html}]
+  s.files = [%q{Manifest}, %q{README}, %q{Rakefile}, %q{data/Seite_Die_Gartenlaube_242.html}, %q{data/Seite_Tagebuch_H_C_Lang_08.html}, %q{data/Seite_Tagebuch_H_C_Lang_08.jpg}, %q{data/test.html}, %q{data/test.png}, %q{example/example_server.rb}, %q{example/public/OCRTest.css}, %q{example/public/OCRTest.html}, %q{example/public/OCRTest_marker.js}, %q{example/public/Seite_Tagebuch_H_C_Lang_08.jpg}, %q{example/public/img/Seite_Tagebuch_H_C_Lang_08.jpg}, %q{lib/hocr_box.rb}, %q{lib/ocr_document.rb}, %q{lib/ocr_element.rb}, %q{lib/ocr_page.rb}, %q{lib/rhocr.rb}, %q{spec/hocr_box_spec.rb}, %q{spec/ocr_document_spec.rb}, %q{spec/ocr_element_spec.rb}, %q{spec/ocr_page_spec.rb}, %q{spec/rhocr_spec.rb}, %q{test.html}, %q{rhocr.gemspec}]
   s.homepage = %q{http://github.com/daandi/rhocr}
   s.rdoc_options = [%q{--line-numbers}, %q{--inline-source}, %q{--title}, %q{Rhocr}, %q{--main}, %q{README}]
   s.require_paths = [%q{lib}]

data/spec/hocr_box_spec.rb

@@ -12,6 +12,24 @@ describe HOCRBox do
         it 'should have coordinates' do
             @box.coordinates.should == [1,2,20,8]
         end
+        it 'should have #left' do
+            @box.left.should == 1
+        end
+        it 'should have #right' do
+            @box.right.should == 20
+        end
+        it 'should have #top' do
+            @box.top.should == 2
+        end
+        it 'should have #bottom' do
+            @box.bottom.should == 8
+        end
+        it 'should have height' do
+            @box.height == 7
+        end
+        it 'should have width' do
+            @box.width.should == 19
+        end
     end
     
     describe "#to_s" do
@@ -91,7 +109,7 @@ describe HOCRBox do
     end
     
     describe '#top_distance_to(element)' do
-        it 'box should be 2px below of element' do
+        it 'box should be 9px below of element' do
              HOCRBox.new(109,241,206,274).top_distance_to(HOCRBox.new(160,196,1117,232)).should == 9
         end
     end

data/spec/ocr_element_spec.rb

@@ -16,7 +16,7 @@ describe OCRElement do
   describe '#initialize and Object' do
       it 'should create an element given ocr_class, children and coordiantes' do
           test_element = OCRElement.new('test', [], %w{10, 11, 20, 21})
-          test_element.children.should == []
+          test_element.children.should be_empty
           test_element.coordinates.should == [10,11,20,21]
           test_element.ocr_class.should == 'test'
       end
@@ -63,6 +63,29 @@ describe OCRElement do
       it 'should have a #to_html method' do
         @ocr_element.to_html.should == "<span class='ocr_par'> <span class='ocr_line'> <span class='ocrx_word'>Athenobius,</span><span class='ocrx_word'>Der</span><span class='ocrx_word'>von</span><span class='ocrx_word'>der</span><span class='ocrx_word'>Göttin</span><span class='ocrx_word'>Minerva</span><span class='ocrx_word'>lebt,</span><span class='ocrx_word'>oder:</span><span class='ocrx_word'>Mi»</span> </span><span class='ocr_line'> <span class='ocrx_word'>nerva</span><span class='ocrx_word'>Bogen.</span> </span> </span>"
       end
+  
+      it 'should add features to css class' do
+         @ocr_element.lines[0].words[5].features << :test
+         @ocr_element.lines[0].words[5].to_html.should == "<span class='ocrx_word-test'>Minerva</span>"
+      end
+      
+      it 'should transform features to an css class #features_to_css_class' do
+        test_elem = @ocr_element.lines[0]
+        test_elem.features_to_css_class.should == ''
+        test_elem.features << :test 
+        test_elem.features << :stuff
+        test_elem.features_to_css_class.should == 'stuff_test'
+        test_elem.features << :test 
+        test_elem.features_to_css_class.should == 'stuff_test'
+      end
+      
+      it 'should use plain css class if elemetn has no features #css_class_string' do
+        test = @ocr_element.lines[1]
+        test.css_class_string.should == 'ocr_line'
+        test.features << :abc
+        test.css_class_string.should == 'ocr_line-abc'
+      end
+
   end
   
   describe 'ocr_line' do

data/spec/ocr_page_spec.rb

@@ -46,41 +46,24 @@ describe OCRPage do
   
   describe '#Iterators' do
       it 'should have a block iterator #each_block' do
-          a = []
-          @test_page.each_block do |block|
-             a << block
-          end
-          a.length.should == 1
+          Enumerator.new(@test_page,:each_block).
+          inject([]) {|acc,elem| acc << elem}.length.should == 1
       end
       it 'should have a paragraph iterator #each_paragraph' do
-          a = []
-          @test_page.each_paragraph do |paragraph|
-              a << paragraph
-          end
-          a.length.should == 12
+          Enumerator.new(@test_page,:each_paragraph).
+          inject([]) {|acc,elem| acc << elem}.length.should == 12
       end
       it 'should have a line iterator #each_line' do
-          a = []
-          @test_page.each_line do |line|
-                a << line
-            end
-          a.length.should == 45
+          Enumerator.new(@test_page,:each_line).
+          inject([]) {|acc,elem| acc << elem}.length.should == 45
       end
       it 'should have a word iterator #each_word' do
-        a = []
-        @test_page.each_word do |word|
-            a << word
-        end
-        a.length.should == 415
+            Enumerator.new(@test_page,:each_word).
+            inject([]) {|acc,elem| acc << elem}.length.should == 415
       end
     
   end
   
-  describe 'convinience methods' do
-      it 'should have a method #lines' do
-        @test_page.lines[5].children.length.should == 11
-      end
-  end
   
   describe ' display methods' do
     
@@ -111,6 +94,8 @@ describe OCRPage do
         words = @output_page.enclosed_words( HOCRBox.new 0,0, 300,300 )
         words.length.should == 6
     end
+    
+    it 'should have an #enclosed_word method to iterate over ewords enclosed by given box'
   end
 
 end

data/spec/rhocr_spec.rb

@@ -13,16 +13,17 @@ describe RHOCR do
     end
     
     before(:each) do
-        @rhocr_doc ||= RHOCR.new.add_folder "data/*.html"
+        @rhocr_doc ||= RHOCR.new
+        @rhocr_doc.add_folder "data/*.html"
     end
     
     describe 'methods to iterate' do
         it 'should have lines' do
-            @rhocr_doc.lines.length.should == 237
+            @rhocr_doc.lines.inject([]){|acc,l| acc <<l}.length.should == 237
         end
         
         it 'should have words' do
-            @rhocr_doc.words.length.should == 2071
+            @rhocr_doc.words.inject([]){|acc,w| acc <<w}.length.should == 2071
         end
         
         it 'should support common iterator methods throug enumerable for word an line arrays' do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: rhocr
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.1.3
   prerelease: 
 platform: ruby
 authors:
@@ -9,11 +9,11 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2011-09-30 00:00:00.000000000Z
+date: 2011-10-10 00:00:00.000000000Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogiri
-  requirement: &70233340332980 !ruby/object:Gem::Requirement
+  requirement: &70118176813440 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -21,7 +21,7 @@ dependencies:
         version: '0'
   type: :runtime
   prerelease: false
-  version_requirements: *70233340332980
+  version_requirements: *70118176813440
 description: ! 'Manipulate and use OCR data encoded in hOCR-Format see: http://code.google.com/p/hocr-tools/'
 email: andreas@neumann.biz
 executables: []
@@ -53,13 +53,13 @@ files:
 - lib/ocr_element.rb
 - lib/ocr_page.rb
 - lib/rhocr.rb
-- rhocr.gemspec
 - spec/hocr_box_spec.rb
 - spec/ocr_document_spec.rb
 - spec/ocr_element_spec.rb
 - spec/ocr_page_spec.rb
 - spec/rhocr_spec.rb
 - test.html
+- rhocr.gemspec
 homepage: http://github.com/daandi/rhocr
 licenses: []
 post_install_message: