Added paragraph alignment and font size recognition

Author: higginsdragon

lib/docx/containers/paragraph.rb

@@ -15,9 +15,11 @@ def self.tag
 
         # Child elements: pPr, r, fldSimple, hlink, subDoc
         # http://msdn.microsoft.com/en-us/library/office/ee364458(v=office.11).aspx
-        def initialize(node)
+        def initialize(node, document_properties = {})
           @node = node
           @properties_tag = 'pPr'
+          @document_properties = document_properties
+          @font_size = @document_properties[:font_size]
         end
 
         # Set text of paragraph
@@ -41,15 +43,41 @@ def to_s
 
         # Array of text runs contained within paragraph
         def text_runs
-          @node.xpath('w:r|w:hyperlink/w:r').map {|r_node| Containers::TextRun.new(r_node) }
+          @node.xpath('w:r|w:hyperlink/w:r').map { |r_node| Containers::TextRun.new(r_node, @document_properties) }
         end
 
         # Iterate over each text run within a paragraph
         def each_text_run
           text_runs.each { |tr| yield(tr) }
         end
+
+        def aligned_left?
+          ['left', nil].include?(alignment)
+        end
+
+        def aligned_right?
+          alignment == 'right'
+        end
+
+        def aligned_center?
+          alignment == 'center'
+        end
+
+        def font_size
+          size_tag = @node.xpath('w:pPr//w:sz').first
+          size_tag ? size_tag.attributes['val'].value.to_i / 2 : @font_size
+        end
 
         alias_method :text, :to_s
+
+        private
+
+        # Returns the alignment if any, or nil if left
+        def alignment
+          alignment_tag = @node.xpath('.//w:jc').first
+          alignment_tag ? alignment_tag.attributes['val'].value : nil
+        end
+
       end
     end
   end

lib/docx/containers/text_run.rb

@@ -20,12 +20,14 @@ def self.tag
         attr_reader :text
         attr_reader :formatting
 
-        def initialize(node)
+        def initialize(node, document_properties = {})
           @node = node
           @text_nodes = @node.xpath('w:t').map {|t_node| Elements::Text.new(t_node) }
           @properties_tag = 'rPr'
           @text       = parse_text || ''
           @formatting = parse_formatting || DEFAULT_FORMATTING
+          @document_properties = document_properties
+          @font_size = @document_properties[:font_size]
         end
 
         # Set text of text run
@@ -66,6 +68,11 @@ def bolded?
         def underlined?
           @formatting[:underline]
         end
+
+        def font_size
+          size_tag = @node.xpath('w:rPr//w:sz').first
+          size_tag ? size_tag.attributes['val'].value.to_i / 2 : @font_size
+        end
       end
     end
   end

lib/docx/document.rb

@@ -18,19 +18,28 @@ module Docx
   #     puts d.text
   #   end
   class Document
-    attr_reader :xml, :doc, :zip
+    attr_reader :xml, :doc, :zip, :styles
 
     def initialize(path, &block)
       @replace = {}
       @zip = Zip::ZipFile.open(path)
-      @xml = @zip.read('word/document.xml')
-      @doc = Nokogiri::XML(@xml)
+      @document_xml = @zip.read('word/document.xml')
+      @doc = Nokogiri::XML(@document_xml)
+      @styles_xml = @zip.read('word/styles.xml')
+      @styles = Nokogiri::XML(@styles_xml)
       if block_given?
         yield self
         @zip.close
       end
     end
-    
+
+    # This stores the current global document properties, for now
+    def document_properties
+      {
+        font_size: font_size
+      }
+    end
+
     # With no associated block, Docx::Document.open is a synonym for Docx::Document.new. If the optional code block is given, it will be passed the opened +docx+ file as an argument and the Docx::Document oject will automatically be closed when the block terminates. The values of the block will be returned from Docx::Document.open.
     # call-seq:
     #   open(filepath) => file
@@ -56,6 +65,13 @@ def tables
       @doc.xpath('//w:document//w:body//w:tbl').map { |t_node| parse_table_from t_node }
     end
 
+    # Some documents have this set, others don't.
+    # Values are returned as half-points, so to get points, that's why it's divided by 2.
+    def font_size
+      size_tag = @styles.xpath('//w:docDefaults//w:rPrDefault//w:rPr//w:sz').first
+      size_tag ? size_tag.attributes['val'].value.to_i / 2 : nil
+    end
+
     ##
     # *Deprecated*
     # 
@@ -106,7 +122,7 @@ def update
 
     # generate Elements::Containers::Paragraph from paragraph XML node
     def parse_paragraph_from(p_node)
-      Elements::Containers::Paragraph.new(p_node)
+      Elements::Containers::Paragraph.new(p_node, document_properties)
     end
 
     # generate Elements::Bookmark from bookmark XML node

spec/docx/document_spec.rb

@@ -166,13 +166,18 @@
     end
 
     it 'should have the correct text' do
-      @doc.paragraphs.size.should eq 6
+      @doc.paragraphs.size.should eq 11
       @doc.paragraphs[0].text.should eq 'Normal'
       @doc.paragraphs[1].text.should eq 'Italic'
       @doc.paragraphs[2].text.should eq 'Bold'
       @doc.paragraphs[3].text.should eq 'Underline'
       @doc.paragraphs[4].text.should eq 'Normal'
       @doc.paragraphs[5].text.should eq 'This is a sentence with all formatting options in the middle of the sentence.'
+      @doc.paragraphs[6].text.should eq 'This is a centered paragraph.'
+      @doc.paragraphs[7].text.should eq 'This paragraph is aligned left.'
+      @doc.paragraphs[8].text.should eq 'This paragraph is aligned right.'
+      @doc.paragraphs[9].text.should eq 'This paragraph is 14 points.'
+      @doc.paragraphs[10].text.should eq 'This paragraph has a word at 16 points.'
     end
 
     it 'should contain a paragraph with multiple text runs' do
@@ -225,6 +230,47 @@
       @doc.paragraphs[5].text_runs[2].bolded?.should be_false
       @doc.paragraphs[5].text_runs[2].underlined?.should be_false
     end
+
+    it 'should detect centered paragraphs' do
+      @doc.paragraphs[5].aligned_center?.should be_false
+      @doc.paragraphs[6].aligned_center?.should be_true
+      @doc.paragraphs[7].aligned_center?.should be_false
+    end
+
+    it 'should detect left justified paragraphs' do
+      @doc.paragraphs[6].aligned_left?.should be_false
+      @doc.paragraphs[7].aligned_left?.should be_true
+      @doc.paragraphs[8].aligned_left?.should be_false
+    end
+
+    it 'should detect right justified paragraphs' do
+      @doc.paragraphs[7].aligned_right?.should be_false
+      @doc.paragraphs[8].aligned_right?.should be_true
+      @doc.paragraphs[9].aligned_right?.should be_false
+    end
+
+    # ECMA-376 Office Open XML spec (4th edition), 17.3.2.38, size is
+    # defined in half-points, meaning 14pt text returns a value of 28.
+    # http://www.ecma-international.org/publications/standards/Ecma-376.htm
+    it 'should return proper font size for paragraphs' do
+      @doc.font_size.should eq 11
+      @doc.paragraphs[5].font_size.should eq 11
+      paragraph = @doc.paragraphs[9]
+      paragraph.font_size.should eq 14
+      paragraph.text_runs[0].font_size.should eq 14
+    end
+
+    it 'should return proper font size for runs' do
+      @doc.font_size.should eq 11
+      paragraph = @doc.paragraphs[10]
+      paragraph.font_size.should eq 11
+      text_runs = paragraph.text_runs
+      text_runs[0].font_size.should eq 11
+      text_runs[1].font_size.should eq 16
+      text_runs[2].font_size.should eq 11
+      text_runs[3].font_size.should eq 11
+      text_runs[4].font_size.should eq 11
+    end
   end
 
   describe 'saving' do