html_gen 0.0.1 → 0.0.3

data/VERSION

@@ -1 +1 @@
-0.0.1
+0.0.3

data/html_gen.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{html_gen}
-  s.version = "0.0.1"
+  s.version = "0.0.3"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Kasper Johansen"]
-  s.date = %q{2012-08-10}
+  s.date = %q{2012-08-11}
   s.description = %q{A small framework for generating HTML.}
   s.email = %q{k@spernj.org}
   s.extra_rdoc_files = [
@@ -28,6 +28,9 @@ Gem::Specification.new do |s|
     "html_gen.gemspec",
     "lib/html_gen.rb",
     "lib/html_gen_element.rb",
+    "lib/html_gen_parser.rb",
+    "lib/html_gen_text_ele.rb",
+    "spec/html_gen_parser_spec.rb",
     "spec/html_gen_spec.rb",
     "spec/spec_helper.rb"
   ]

data/lib/html_gen_element.rb

@@ -40,6 +40,12 @@ class Html_gen::Element
   #  element.html #=> "<b>Te<i>s</i>t</b>"
   attr_accessor :str_html
   
+  #An array holding all the sub-elements of this element.
+  attr_accessor :eles
+  
+  #The name of the element. "a" for <a> and such.
+  attr_accessor :name
+  
   #You can give various arguments as shortcuts to calling the methods. You can also decide what should be used for newline and indentation.
   #  Html_gen::Element.new(:b, {
   #    :css => {"font-weight" => "bold"},
@@ -177,4 +183,39 @@ class Html_gen::Element
     #Returns the string.
     return str
   end
+  
+  #Returns the names of all sub-elements in an array.
+  def eles_names
+    names = []
+    @eles.each do |ele|
+      names << ele.name
+    end
+    
+    return names
+  end
+  
+  #Converts the content of the 'style'-attribute to css-hash-content.
+  def convert_style_to_css
+    if !@attr[:style].to_s.strip.empty?
+      style = @attr[:style]
+    elsif !@attr["style"].to_s.strip.empty?
+      style = @attr["style"]
+    else
+      raise "No style set in element."
+    end
+    
+    loop do
+      if match = style.match(/\A\s*(\S+?):\s*(.+?)\s*(;|\Z)/)
+        style.gsub!(match[0], "")
+        key = match[1]
+        val = match[2]
+        raise "Such a key already exists in CSS-hash: '#{key}'." if @css.key?(key)
+        @css[key] = val
+      elsif match = style.slice!(/\A\s*\Z/)
+        break
+      else
+        raise "Dont know what to do with style-variable: '#{style}'."
+      end
+    end
+  end
 end

data/lib/html_gen_parser.rb

@@ -0,0 +1,185 @@
+#A simple, lightweight and pure-Ruby class for parsing HTML-strings into elements.
+#===Examples
+#  doc = Html_gen::Parser.new(:str => a_html_variable)
+#  html_ele = doc.eles.first
+#  html_ele.name #=> "html"
+class Html_gen::Parser
+  #An array that holds all the parsed root-elements.
+  attr_reader :eles
+  
+  #The constructor. See class documentation for usage of this.
+  def initialize(args)
+    if args[:io]
+      @io = args[:io]
+    elsif args[:str]
+      @io = StringIO.new(args[:str])
+    else
+      raise "Dont know how to handle given arguments."
+    end
+    
+    raise "No ':io' was given." if !@io
+    @eof = false
+    @buffer = ""
+    @eles = []
+    @eles_t = []
+    @debug = args[:debug]
+    
+    while !@eof or !@buffer.empty?
+      parse_tag
+    end
+  end
+  
+  private
+  
+  #Ensures at least 16kb of data is loaded into the buffer.
+  def ensure_buffer
+    while @buffer.length < 16384 and !@eof
+      str = @io.gets(16384)
+      if !str
+        @eof = true
+      else
+        @buffer << str
+      end
+    end
+  end
+  
+  #Searches for a given regex. If found the contents is removed from the buffer.
+  def search(regex)
+    ensure_buffer
+    
+    if match = @buffer.match(regex)
+      @buffer.gsub!(regex, "")
+      ensure_buffer
+      return match
+    end
+    
+    return false
+  end
+  
+  #Asumes a tag is the next to be parsed and adds it to document-data.
+  def parse_tag(args = {})
+    if match = search(/\A\s*<\s*(\/|)\s*(\S+?)(\s+|\/\s*>|>)/)
+      tag_name = match[2].to_s.strip.downcase
+      start_sign = match[1].to_s.strip.downcase
+      end_sign = match[3].to_s.strip.downcase
+      
+      raise "Dont know how to handle start-sign: '#{start_sign}' for tag: '#{tag_name}'." if !start_sign.empty?
+      
+      ele = Html_gen::Element.new(tag_name)
+      
+      if @eles_t.empty?
+        puts "Adding element '#{tag_name}' to root elements." if @debug
+        @eles << ele
+      else
+        puts "Adding element '#{tag_name}' to last t-element: '#{@eles_t.last.name}'." if @debug
+        @eles_t.last.eles << ele
+      end
+      
+      @eles_t << ele
+      puts "New element-match: #{match.to_a}" if @debug
+      
+      if end_sign.match(/^\/\s*>$/)
+        puts "End of element '#{tag_name}' for '#{@eles_t.last.name}'." if @debug
+        ele = @eles_t.pop
+        raise "Expected ele-name to be: '#{tag_name}' but it wasnt: '#{ele.name}'." if ele.name.to_s != tag_name
+        return ele
+      elsif end_sign.to_s.strip.empty?
+        parse_attr_of_tag(ele, tag_name)
+        ele.convert_style_to_css if ele.attr.key?("style") or ele.attr.key?(:style)
+        return ele
+      else
+        parse_content_of_tag(ele, tag_name)
+        return ele
+      end
+    else
+      if args[:false]
+        return false
+      else
+        raise "Dont know what to do with buffer: '#{@buffer}'."
+      end
+    end
+  end
+  
+  def parse_attr_of_tag(ele, tag_name)
+    loop do
+      if match = search(/\A\s*(\S+)=(\"|'|)/)
+        attr_name = match[1]
+        raise "Attribute already exists on element: '#{attr_name}'." if ele.attr.key?(attr_name)
+        
+        if match[2].to_s.empty?
+          quote_char = /\s+/
+          quote_val = :whitespace
+        else
+          quote_char = /#{Regexp.escape(match[2])}/
+          quote_val = :normal
+        end
+        
+        attr_val = parse_attr_until_quote(quote_char, quote_val)
+        
+        puts "Parsed attribute '#{attr_name}' with value '#{attr_val}'." if @debug
+        ele.attr[attr_name] = attr_val
+      elsif search(/\A\s*>/)
+        parse_content_of_tag(ele, tag_name)
+        break
+      else
+        raise "Dont know what to do with buffer when parsing attributes: '#{@buffer}'."
+      end
+    end
+  end
+  
+  def parse_attr_until_quote(quote_char, quote_val)
+    val = ""
+    
+    loop do
+      ensure_buffer
+      char = @buffer.slice!(0)
+      break if !char
+      
+      if char == "\\"
+        val << char
+        val << @buffer.slice!(0)
+      elsif char =~ quote_char
+        break
+      elsif char == ">" and quote_val == :whitespace
+        @buffer = char + @buffer
+        break
+      else
+        val << char
+      end
+    end
+    
+    return val
+  end
+  
+  #Assumes some content of a tag is next to be parsed and parses it.
+  def parse_content_of_tag(ele, tag_name)
+    raise "Empty tag-name given: '#{tag_name}'." if tag_name.to_s.strip.empty?
+    raise "No 'ele' was given." if !ele
+    
+    loop do
+      if search(/\A\s*\Z/)
+        raise "Could not find end of tag: '#{tag_name}'."
+      elsif match = search(/\A\s*<\s*\/\s*#{Regexp.escape(tag_name)}\s*>\s*/i)
+        puts "Found end: '#{match.to_a}' for '#{@eles_t.last.name}'." if @debug
+        ele = @eles_t.pop
+        raise "Expected ele-name to be: '#{tag_name}' but it wasnt: '#{ele.name}'." if ele.name.to_s != tag_name
+        
+        break
+      elsif new_ele = parse_tag(:false => true)
+        puts "Found new element '#{new_ele.name}' and adding it to '#{ele.name}'." if @debug
+        #ele.eles << new_ele
+      elsif match = search(/\A(.+?)(<|\Z)/)
+        puts "Text-content-match: '#{match.to_a}'." if @debug
+        
+        #Put end back into buffer.
+        @buffer = match[2] + @buffer
+        puts "Buffer after text-match: #{@buffer}" if @debug
+        
+        #Add text element to list as finished.
+        ele.eles << Html_gen::Text_ele.new(:str => match[1])
+      else
+        raise "Dont know what to do with buffer: '#{@buffer}'."
+      end
+    end
+  end
+end

data/lib/html_gen_text_ele.rb

@@ -0,0 +1,12 @@
+class Html_gen::Text_ele
+  attr_reader :args
+  
+  def initialize(args)
+    @str = args[:str]
+  end
+  
+  #Returns the text that this element holds.
+  def str
+    return @args[:str]
+  end
+end

data/spec/html_gen_parser_spec.rb

@@ -0,0 +1,25 @@
+require File.expand_path(File.dirname(__FILE__) + '/spec_helper')
+
+describe "Parser" do
+  it "should be able generate elements from HTML" do
+    parser = Html_gen::Parser.new(:str => "<html><head><title>Test</title></head><body>This is the body</body></html>")
+    raise "Expected 1 root element but got: '#{parser.eles.length}'." if parser.eles.length != 1
+    
+    html = parser.eles.first
+    raise "Expected 2 elements of HTML element but got: '#{html.eles.length}'. #{html.eles_names}" if html.eles.length != 2
+    
+    head = html.eles.first
+    title = head.eles.first
+    raise "Expected name to be 'title' but it wasnt: '#{title.name}'." if title.name != "title"
+    
+    doc = Html_gen::Parser.new(:str => "<td colspan=\"2\" style=\"font-weight: bold;\" width='100px' height=50px>test</td>")
+    td = doc.eles.first
+    
+    raise "Expected name of element to be 'td' but it wasnt: '#{td.name}'." if td.name != "td"
+    raise "Expected colspan to be '2' but it wasnt: '#{td.attr["colspan"]}'." if td.attr["colspan"] != "2"
+    raise "Expected width to be '100px' but it wasnt: '#{td.attr["width"]}'." if td.attr["width"] != "100px"
+    raise "Expected height to be '50px' but it wasnt: '#{td.attr["height"]}'." if td.attr["height"] != "50px"
+    raise "Expected CSS-font-weight to be 'bold' but it wasnt: '#{td.css["font-weight"]}'." if td.css["font-weight"] != "bold"
+    raise "Expected style to be empty but it wasnt: '#{td.attr["style"]}'." if !td.attr["style"].to_s.empty?
+  end
+end

metadata

@@ -2,7 +2,7 @@
 name: html_gen
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 0.0.1
+  version: 0.0.3
 platform: ruby
 authors: 
 - Kasper Johansen
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2012-08-10 00:00:00 +02:00
+date: 2012-08-11 00:00:00 +02:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -78,6 +78,9 @@ files:
 - html_gen.gemspec
 - lib/html_gen.rb
 - lib/html_gen_element.rb
+- lib/html_gen_parser.rb
+- lib/html_gen_text_ele.rb
+- spec/html_gen_parser_spec.rb
 - spec/html_gen_spec.rb
 - spec/spec_helper.rb
 has_rdoc: true
@@ -94,7 +97,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
-      hash: 3676011512365822361
+      hash: 4461972116095674458
       segments: 
       - 0
       version: "0"