ooxml_parser 0.2.0 → 0.3.0

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_paragraph_run/docx_paragraph_run_helpers.rb

@@ -61,21 +61,11 @@ module OoxmlParser
           unless node_child.attribute('fill').value == 'auto' || node_child.attribute('fill').value == '' || node_child.attribute('fill').value == 'null'
             self.background_color = Color.new(parent: self).parse_hex_string(node_child.attribute('fill').value)
           end
-        when 'uCs'
+        when 'u', 'uCs'
           font_style.underlined = Underline.new
           if option_enabled?(node_child)
             font_style.underlined.style = node_child.attribute('val').value
-            unless node_child.attribute('color').nil?
-              font_style.underlined.color = Color.new(parent: font_style.underlined).parse_hex_string(node_child.attribute('color').value)
-            end
-          end
-        when 'u'
-          font_style.underlined = Underline.new
-          if option_enabled?(node_child)
-            font_style.underlined.style = node_child.attribute('val').value
-            unless node_child.attribute('color').nil?
-              font_style.underlined.color = Color.new(parent: font_style.underlined).parse_hex_string(node_child.attribute('color').value)
-            end
+            font_style.underlined.color = Color.new(parent: font_style.underlined).parse_hex_string(node_child.attribute('color').value) unless node_child.attribute('color').nil?
           end
         when 'strike'
           font_style.strike = :single if option_enabled?(node_child)

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/inserted.rb

@@ -1,3 +1,5 @@
+require 'date'
+
 module OoxmlParser
   # Class for parsing `w:ins` tag - Inserted Run Content
   class Inserted < OOXMLDocumentObject
@@ -23,7 +25,7 @@ module OoxmlParser
         when 'author'
           @author = value.value.to_s
         when 'date'
-          @date = DateTime.parse(value.value.to_s)
+          @date = parse_date(value.value.to_s)
         when 'oouserid'
           @user_id = value.value.to_s
         end
@@ -37,5 +39,17 @@ module OoxmlParser
       end
       self
     end
+
+    private
+
+    # Parse date and handle incorrect dates
+    # @param value [Sting] value of date
+    # @return [DateTime, String] if date correct or incorrect
+    def parse_date(value)
+      DateTime.parse(value)
+    rescue ArgumentError
+      warn "Date #{value} is incorrect format"
+      value
+    end
   end
 end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/sdt/sdt_content.rb

@@ -0,0 +1,30 @@
+module OoxmlParser
+  # Class for parsing `w:w:sdtContent` tags
+  class SDTContent < OOXMLDocumentObject
+    # @return [Array, ParagraphRun] runs of sdt
+    attr_reader :runs
+    # @return [Array, ParagraphRun] paragraphs of sdt
+    attr_reader :paragraphs
+
+    def initialize(parent: nil)
+      @runs = []
+      @paragraphs = []
+      @parent = parent
+    end
+
+    # Parse SDTContent object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [SDTContent] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'p'
+          @paragraphs << DocxParagraph.new(parent: self).parse(node_child)
+        when 'r'
+          @runs << ParagraphRun.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/sdt/sdt_properties.rb

@@ -0,0 +1,28 @@
+module OoxmlParser
+  # Class for parsing `w:sdtPr` tags
+  class SDTProperties < OOXMLDocumentObject
+    # @return [ValuedChild] alias value
+    attr_reader :alias
+    # @return [ValuedChild] tag value
+    attr_reader :tag
+    # @return [ValuedChild] Locking Setting
+    attr_reader :lock
+
+    # Parse SDTProperties object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [SDTProperties] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'alias'
+          @alias = ValuedChild.new(:string, parent: self).parse(node_child)
+        when 'tag'
+          @tag = ValuedChild.new(:string, parent: self).parse(node_child)
+        when 'lock'
+          @lock = ValuedChild.new(:symbol, parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/structured_document_tag.rb

@@ -0,0 +1,26 @@
+require_relative 'sdt/sdt_content'
+require_relative 'sdt/sdt_properties'
+module OoxmlParser
+  # Class for parsing `w:std`
+  class StructuredDocumentTag < OOXMLDocumentObject
+    # @return [SDTContent] content of sdt
+    attr_reader :sdt_content
+    # @return [SDTProperties] properties of sdt
+    attr_reader :properties
+
+    # Parse StructuredDocumentTag object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [StructuredDocumentTag] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'sdtContent'
+          @sdt_content = SDTContent.new(parent: self).parse(node_child)
+        when 'sdtPr'
+          @properties = SDTProperties.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/header_footer.rb

@@ -46,7 +46,7 @@ module OoxmlParser
         next unless footnote.attribute('id').value.to_i == @id
         paragraph_number = 0
         footnote.xpath('w:p').each do |paragraph|
-          @elements << DocumentStructure.default_paragraph_style.copy.parse(paragraph, paragraph_number, DocumentStructure.default_run_style, parent: self)
+          @elements << DocumentStructure.default_paragraph_style.dup.parse(paragraph, paragraph_number, DocumentStructure.default_run_style, parent: self)
           paragraph_number += 1
         end
       end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/page_properties/note.rb

@@ -6,15 +6,15 @@ module OoxmlParser
       @elements = []
     end
 
-    def self.parse(default_paragraph, default_character, target, assigned_to, type, parent: nil)
+    def self.parse(params)
       note = Note.new
-      note.type = type
-      note.assigned_to = assigned_to
-      note.parent = parent
-      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + "word/#{target}"))
-      if type.include?('footer')
+      note.type = params[:type]
+      note.assigned_to = params[:assigned_to]
+      note.parent = params[:parent]
+      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + "word/#{params[:target]}"))
+      if note.type.include?('footer')
         xpath_note = '//w:ftr'
-      elsif type.include?('header')
+      elsif note.type.include?('header')
         xpath_note = '//w:hdr'
       else
         raise "Cannot parse unknown Note type: #{type}"
@@ -23,19 +23,19 @@ module OoxmlParser
         number = 0
         ftr.xpath('*').each do |sub_element|
           if sub_element.name == 'p'
-            note.elements << default_paragraph.copy.parse(sub_element, number, default_character, parent: note)
+            note.elements << params[:default_paragraph].dup.parse(sub_element, number, params[:default_character], parent: note)
             number += 1
           elsif sub_element.name == 'tbl'
             note.elements << Table.new(parent: note).parse(sub_element, number)
             number += 1
           elsif sub_element.name == 'std'
             sub_element.xpath('w:p').each do |p|
-              note.elements << default_paragraph.copy.parse(p, number, default_character)
+              note.elements << params[:default_paragraph].copy.parse(p, number, params[:default_character])
               number += 1
             end
             sub_element.xpath('w:sdtContent').each do |sdt_content|
               sdt_content.xpath('w:p').each do |p|
-                note.elements << default_paragraph.copy.parse(p, number, default_character)
+                note.elements << params[:default_paragraph].copy.parse(p, number, params[:default_character])
                 number += 1
               end
             end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/page_properties/page_margins.rb

@@ -3,22 +3,15 @@ module OoxmlParser
   class PageMargins < OOXMLDocumentObject
     attr_accessor :top, :bottom, :left, :right, :footer, :gutter, :header
 
-    def initialize(top: nil,
-                   bottom: nil,
-                   left: nil,
-                   right: nil,
-                   header: nil,
-                   footer: nil,
-                   gutter: nil,
-                   parent: nil)
-      @top = top
-      @bottom = bottom
-      @left = left
-      @right = right
-      @header = header
-      @footer = footer
-      @gutter = gutter
-      @parent = parent
+    def initialize(params)
+      @top = params[:top]
+      @bottom = params[:bottom]
+      @left = params[:left]
+      @right = params[:right]
+      @header = params[:header]
+      @footer = params[:footer]
+      @gutter = params[:gutter]
+      @parent = params[:parent]
     end
 
     # Parse BordersProperties

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/page_properties/page_properties.rb

@@ -10,6 +10,8 @@ module OoxmlParser
                   :notes
     # @return [FootnoteProperties] properties of footnote
     attr_accessor :footnote_properties
+    # @return [True, False] Specifies whether the section should have a different header and footer for its first page
+    attr_reader :title_page
 
     def initialize(parent: nil)
       @notes = []
@@ -26,12 +28,8 @@ module OoxmlParser
           @size = PageSize.new.parse(pg_size_subnode)
         when 'pgBorders'
           page_borders = Borders.new
-          unless pg_size_subnode.attribute('display').nil?
-            page_borders.display = pg_size_subnode.attribute('display').value.to_sym
-          end
-          unless pg_size_subnode.attribute('offsetFrom').nil?
-            page_borders.offset_from = pg_size_subnode.attribute('offsetFrom').value.to_sym
-          end
+          page_borders.display = pg_size_subnode.attribute('display').value.to_sym unless pg_size_subnode.attribute('display').nil?
+          page_borders.offset_from = pg_size_subnode.attribute('offsetFrom').value.to_sym unless pg_size_subnode.attribute('offsetFrom').nil?
           pg_size_subnode.xpath('w:bottom').each do |bottom|
             page_borders.bottom = BordersProperties.new(parent: page_borders).parse(bottom)
           end
@@ -57,14 +55,18 @@ module OoxmlParser
           @text_direction = pg_size_subnode.attribute('val').value
         when 'docGrid'
           @document_grid = DocumentGrid.new(parent: self).parse(pg_size_subnode)
+        when 'titlePg'
+          @title_page = option_enabled?(pg_size_subnode)
         when 'cols'
           @columns = Columns.new.parse(pg_size_subnode)
         when 'headerReference', 'footerReference'
           target = OOXMLDocumentObject.get_link_from_rels(pg_size_subnode.attribute('id').value)
           OOXMLDocumentObject.add_to_xmls_stack("word/#{target}")
-          note = Note.parse(default_paragraph, default_character, target,
-                            pg_size_subnode.attribute('type').value,
-                            File.basename(target).sub('.xml', ''),
+          note = Note.parse(default_paragraph: default_paragraph,
+                            default_character: default_character,
+                            target: target,
+                            assigned_to: pg_size_subnode.attribute('type').value,
+                            type: File.basename(target).sub('.xml', ''),
                             parent: self)
           @notes << note
           OOXMLDocumentObject.xmls_stack.pop

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation.rb

@@ -7,6 +7,8 @@ module OoxmlParser
   class Presentation < CommonDocumentStructure
     include PresentationHelpers
     attr_accessor :slides, :theme, :slide_size, :comments
+    # @return [Relationships] relationships of presentation
+    attr_accessor :relationships
 
     def initialize(slides = [], theme = nil, comments = [])
       @slides = slides
@@ -32,18 +34,19 @@ module OoxmlParser
             silde_id_node.attribute_nodes.select { |node| id = node.to_s if node.namespace && node.namespace.prefix == 'r' }
             presentation.slides << Slide.new(parent: presentation,
                                              xml_path: "#{OOXMLDocumentObject.root_subfolder}/#{OOXMLDocumentObject.get_link_from_rels(id)}")
-                                   .parse
+                                        .parse
           end
         end
       end
       presentation.comments = PresentationComment.parse_list
       OOXMLDocumentObject.xmls_stack.pop
+      presentation.relationships = Relationships.parse_rels("#{OOXMLDocumentObject.path_to_folder}/ppt/_rels/presentation.xml.rels")
       presentation
     end
 
     class << self
       # @return [FontStyle] current font_style
-      attr_accessor :current_font_style
+      attr_writer :current_font_style
 
       def current_font_style
         @current_font_style = FontStyle.new if @current_font_style.nil?

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/presentation_theme/theme_color.rb

@@ -14,9 +14,7 @@ module OoxmlParser
       else
         all_instance_variables = instance_variables
         all_instance_variables.each do |current_attributes|
-          unless instance_variable_get(current_attributes) == other.instance_variable_get(current_attributes)
-            return false
-          end
+          return false unless instance_variable_get(current_attributes) == other.instance_variable_get(current_attributes)
         end
         true
       end

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide.rb

@@ -1,5 +1,8 @@
 require_relative 'slide/presentation_alternate_content'
 require_relative 'slide/background'
+require_relative 'slide/common_slide_data'
+require_relative 'slide/connection_shape.rb'
+require_relative 'slide/presentation_notes.rb'
 require_relative 'slide/graphic_frame/graphic_frame'
 require_relative 'slide/slide/shapes_grouping'
 require_relative 'slide/slide/timing'
@@ -8,50 +11,47 @@ module OoxmlParser
   # Class for parsing `slide.xml`
   class Slide < OOXMLDocumentObject
     include SlideHelper
-    attr_accessor :elements, :background, :transition, :timing, :alternate_content
+    attr_accessor :transition, :timing, :alternate_content
+    # @return [CommonSlideData] common slide data
+    attr_reader :common_slide_data
+    # @return [Relationships] relationships of slide
+    attr_reader :relationships
+    # @return [String] name of slide
+    attr_reader :name
+    # @return [Notes] note of slide
+    attr_reader :note
 
     def initialize(parent: nil, xml_path: nil)
-      @elements = []
-      @background = nil
       @parent = parent
       @xml_path = xml_path
     end
 
     def with_data?
-      return true unless @background.nil?
-      @elements.each do |current_element|
+      return true unless background.nil?
+      elements.each do |current_element|
         return true if current_element.with_data?
       end
       false
     end
 
+    def elements
+      @common_slide_data.shape_tree.elements
+    end
+
+    def background
+      @common_slide_data.background
+    end
+
     # Parse Slide object
     # @return [Slide] result of parsing
     def parse
       OOXMLDocumentObject.add_to_xmls_stack(@xml_path)
+      @name = File.basename(@xml_path, '.*')
       node = Nokogiri::XML(File.open(OOXMLDocumentObject.current_xml))
       node.xpath('//p:sld/*').each do |node_child|
         case node_child.name
         when 'cSld'
-          node_child.xpath('*').each do |common_slide_node_child|
-            case common_slide_node_child.name
-            when 'spTree'
-              common_slide_node_child.xpath('*').each do |slide_element_node|
-                case slide_element_node.name
-                when 'sp'
-                  @elements << DocxShape.new(parent: self).parse(slide_element_node).dup
-                when 'pic'
-                  @elements << DocxPicture.new(parent: self).parse(slide_element_node)
-                when 'graphicFrame'
-                  @elements << GraphicFrame.new(parent: self).parse(slide_element_node)
-                when 'grpSp'
-                  @elements << ShapesGrouping.new(parent: self).parse(slide_element_node)
-                end
-              end
-            when 'bg'
-              @background = Background.new(parent: self).parse(common_slide_node_child)
-            end
-          end
+          @common_slide_data = CommonSlideData.new(parent: self).parse(node_child)
         when 'timing'
           @timing = Timing.new(parent: self).parse(node_child)
         when 'transition'
@@ -61,7 +61,18 @@ module OoxmlParser
         end
       end
       OOXMLDocumentObject.xmls_stack.pop
+      @relationships = Relationships.parse_rels("#{OOXMLDocumentObject.path_to_folder}#{File.dirname(@xml_path)}/_rels/#{@name}.xml.rels")
+      parse_note
       self
     end
+
+    private
+
+    # Parse slide notes if present
+    def parse_note
+      notes_target = @relationships.target_by_type('notes')
+      return nil unless notes_target
+      @note = PresentationNotes.new(parent: self).parse("#{OOXMLDocumentObject.path_to_folder}#{File.dirname(@xml_path)}/#{notes_target}")
+    end
   end
 end

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide/common_slide_data.rb

@@ -0,0 +1,26 @@
+require_relative 'common_slide_data/shape_tree'
+
+module OoxmlParser
+  # Class for parsing `cSld` tag
+  class CommonSlideData < OOXMLDocumentObject
+    # @return [ShapeTree] tree of shape
+    attr_reader :shape_tree
+    # @return [Background] background
+    attr_reader :background
+
+    # Parse CommonSlideData object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [CommonSlideData] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'spTree'
+          @shape_tree = ShapeTree.new(parent: self).parse(node_child)
+        when 'bg'
+          @background = Background.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end