ooxml_parser 0.2.0 → 0.3.0

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/comments_extended/comment_extended.rb

@@ -0,0 +1,24 @@
+module OoxmlParser
+  # Class for parsing `w15:commentEx` tag
+  class CommentExtended < OOXMLDocumentObject
+    # @return [Integer] id of paragraph
+    attr_accessor :paragraph_id
+    # @return [True, False] is done?
+    attr_accessor :done
+
+    # Parse CommentExtended object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [CommentExtended] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'paraId'
+          @paragraph_id = value.value.to_i
+        when 'done'
+          @done = attribute_enabled?(value.value)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -1,9 +1,11 @@
 module OoxmlParser
   # Class for describing Document Background `w:background`
   class DocumentBackground < OOXMLDocumentObject
-    attr_accessor :color1, :size, :color2, :type
+    attr_reader :color1, :size, :color2, :type
     # @return [FileReference] image structure
-    attr_accessor :file_reference
+    attr_reader :file_reference
+    # @return [Fill] fill data
+    attr_reader :fill
 
     def initialize(parent: nil)
       @color1 = nil
@@ -17,16 +19,11 @@ module OoxmlParser
     def parse(node)
       @color1 = Color.new(parent: self).parse_hex_string(node.attribute('color').value)
       node.xpath('v:background').each do |second_color|
-        unless second_color.attribute('targetscreensize').nil?
-          @size = second_color.attribute('targetscreensize').value.sub(',', 'x')
-        end
-        second_color.xpath('v:fill').each do |fill|
-          if !fill.attribute('color2').nil?
-            @color2 = Color.new(parent: self).parse_hex_string(fill.attribute('color2').value.split(' ').first.delete('#'))
-            @type = fill.attribute('type').value
-          elsif !fill.attribute('id').nil?
-            @file_reference = FileReference.new(parent: self).parse(fill)
-            @type = 'image'
+        @size = second_color.attribute('targetscreensize').value.sub(',', 'x') unless second_color.attribute('targetscreensize').nil?
+        second_color.xpath('*').each do |node_child|
+          case node_child.name
+          when 'fill'
+            @fill = Fill.new(parent: self).parse(node_child)
           end
         end
       end

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

@@ -68,7 +68,7 @@ module OoxmlParser
         when 'rPr'
           @run_properties = DocxParagraphRun.new.parse_properties(subnode)
         when 'pPr'
-          @paragraph_properties = DocxParagraph.new(parent: self).parse_paragraph_style(subnode)
+          @paragraph_properties = ParagraphProperties.new(parent: self).parse(subnode)
         when 'tblPr'
           @table_properties = TableProperties.new(parent: self).parse(subnode)
         when 'trPr'

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

@@ -1,21 +1,32 @@
 # noinspection RubyTooManyInstanceVariablesInspection
 require_relative 'docx_paragraph/bookmark'
+require_relative 'docx_paragraph/docx_paragraph_helper'
 require_relative 'docx_paragraph/docx_paragraph_run'
 require_relative 'docx_paragraph/indents'
 require_relative 'docx_paragraph/inserted'
+require_relative 'docx_paragraph/structured_document_tag'
 require_relative 'docx_paragraph/frame_properties'
 require_relative 'docx_paragraph/docx_formula'
 require_relative 'docx_paragraph/style_parametres'
 module OoxmlParser
   class DocxParagraph < OOXMLDocumentObject
+    include DocxParagraphHelper
     attr_accessor :number, :bookmark_start, :bookmark_end, :align, :spacing, :background_color, :ind, :numbering,
                   :character_style_array, :horizontal_line, :page_break, :kinoku, :borders, :keep_lines,
                   :contextual_spacing, :sector_properties, :page_numbering, :section_break, :style, :keep_next,
                   :orphan_control, :frame_properties
+    # @return [Hyperlink] hyperlink in paragraph
+    attr_reader :hyperlink
     # @return [ParagraphProperties] Properties of current paragraph
     attr_accessor :paragraph_properties
     # @return [Inserted] data inserted by review
     attr_accessor :inserted
+    # @return [Integer] id of paragraph (for comment)
+    attr_accessor :paragraph_id
+    # @return [Integer] id of text (for comment)
+    attr_accessor :text_id
+    # @return [StructuredDocumentTag] structured document tag data
+    attr_accessor :sdt
 
     def initialize(parent: nil)
       @number = 0
@@ -23,53 +34,26 @@ module OoxmlParser
       @bookmark_end = []
       @align = 'left'
       @spacing = Spacing.new
-      @background_color = nil
       @ind = Indents.new
       @kinoku = false
-      @numbering = nil
       @character_style_array = []
       @horizontal_line = false
       @page_break = false
       @borders = Borders.new
       @keep_lines = false
       @contextual_spacing = false
-      @sector_properties = nil
       @page_numbering = false
-      @section_break = nil
-      @style = nil
       @keep_next = false
       @orphan_control = true
-      @frame_properties = nil
       @parent = parent
     end
 
-    def copy
-      paragraph = DocxParagraph.new
-      paragraph.number = number
-      paragraph.bookmark_start = @bookmark_start.dup
-      paragraph.bookmark_end = @bookmark_end.dup
-      paragraph.align = @align
-      paragraph.spacing = @spacing.copy
-      paragraph.background_color = @background_color
-      paragraph.ind = @ind.dup
-      paragraph.numbering = @numbering
-      paragraph.character_style_array = @character_style_array
-      paragraph.horizontal_line = @horizontal_line
-      paragraph.page_break = @page_break
-      paragraph.kinoku = @kinoku
-      paragraph.borders = @borders
-      paragraph.keep_lines = @keep_lines
-      paragraph.contextual_spacing = @contextual_spacing
-      paragraph.sector_properties = @sector_properties
-      paragraph.page_numbering = @page_numbering
-      paragraph.section_break = @section_break
-      paragraph.style = @style
-      paragraph.keep_next = @keep_next
-      paragraph.orphan_control = @orphan_control
-      paragraph.frame_properties = @frame_properties
-      paragraph.paragraph_properties = @paragraph_properties
-      paragraph.inserted = @inserted
-      paragraph
+    def initialize_copy(source)
+      super
+      @bookmark_start = source.bookmark_start.clone
+      @bookmark_end = source.bookmark_end.clone
+      @character_style_array = source.character_style_array.clone
+      @spacing = source.spacing.clone
     end
 
     def nonempty_runs
@@ -79,6 +63,7 @@ module OoxmlParser
               !cur_run.alternate_content.nil? ||
               !cur_run.drawing.nil? ||
               !cur_run.object.nil? ||
+              !cur_run.shape.nil? ||
               !cur_run.footnote.nil? ||
               !cur_run.endnote.nil?
           )
@@ -88,6 +73,8 @@ module OoxmlParser
       end
     end
 
+    alias runs character_style_array
+
     # @return [True, false] if structure contain any user data
     def with_data?
       !nonempty_runs.empty?
@@ -101,24 +88,30 @@ module OoxmlParser
     end
 
     def ==(other)
-      ignored_attributes = [:@number, :@parent]
+      ignored_attributes = %i[@number @parent]
       all_instance_variables = instance_variables
       significan_attribues = all_instance_variables - ignored_attributes
       significan_attribues.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
 
     def parse(node, par_number = 0, default_character = DocxParagraphRun.new, parent: nil)
       @parent = parent
-      default_character_style = default_character.copy
+      default_character_style = default_character.dup
       character_styles_array = []
-      custom_character_style = default_character_style.copy
+      custom_character_style = default_character_style.dup
       char_number = 0
       comments = []
+      node.attributes.each do |key, value|
+        case key
+        when 'paraId'
+          @paragraph_id = value.value.to_i
+        when 'textId'
+          @text_id = value.value.to_i
+        end
+      end
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'bookmarkStart'
@@ -139,37 +132,34 @@ module OoxmlParser
           instruction = node_child.attribute('instr').to_s
           @page_numbering = true if instruction.include?('PAGE')
           node_child.xpath('w:r').each do |r_tag|
-            character_style = default_character_style.copy
+            character_style = default_character_style.dup
             character_style.parse(r_tag, char_number, parent: parent)
             character_style.page_number = @page_numbering
             character_style.instruction = instruction
-            character_styles_array << character_style.copy
+            character_styles_array << character_style.dup
             char_number += 1
           end
         when 'r'
-          character_style = custom_character_style.copy
+          character_style = custom_character_style.dup
           node_child.xpath('w:instrText').each do |insrt_text|
             @page_numbering = true if insrt_text.text.include?('PAGE')
           end
           character_style.parse(node_child, char_number, parent: self)
           character_style.comments = comments.dup
-          character_styles_array << character_style.copy
-          unless character_style.shape.nil?
-            character_styles_array.last.shape = character_style.shape
-          end
+          character_styles_array << character_style.dup
+          character_styles_array.last.shape = character_style.shape unless character_style.shape.nil?
           char_number += 1
         when 'hyperlink'
-          character_style = default_character_style.copy
+          @hyperlink = Hyperlink.new(parent: self).parse(node_child)
+          character_style = default_character_style.dup
           if !node_child.attribute('id').nil?
             character_style.link = Hyperlink.new(parent: character_style).parse(node_child)
           else
-            unless node_child.attribute('anchor').nil?
-              character_style.link = node_child.attribute('anchor').value
-            end
+            character_style.link = node_child.attribute('anchor').value unless node_child.attribute('anchor').nil?
           end
           node_child.xpath('w:r').each do |r_tag|
             character_style.parse(r_tag, char_number, parent: parent)
-            character_styles_array << character_style.copy
+            character_styles_array << character_style.dup
             char_number += 1
           end
           node_child.xpath('w:fldSimple').each do |simple_field|
@@ -179,7 +169,7 @@ module OoxmlParser
               character_style.parse(r_tag, char_number, parent: self)
               character_style.page_number = @page_numbering
               character_style.instruction = instruction
-              character_styles_array << character_style.copy
+              character_styles_array << character_style.dup
               char_number += 1
             end
           end
@@ -196,12 +186,12 @@ module OoxmlParser
           end
         when 'ins'
           @inserted = Inserted.new(parent: self).parse(node_child)
+        when 'sdt'
+          @sdt = StructuredDocumentTag.new(parent: self).parse(node_child)
         end
       end
       @number = par_number
-      if character_styles_array.last.class == DocxParagraphRun
-        character_styles_array.last.text = character_styles_array.last.text.rstrip
-      end
+      character_styles_array.last.text = character_styles_array.last.text.rstrip if character_styles_array.last.class == DocxParagraphRun
       @character_style_array = character_styles_array
       @parent = parent
       self
@@ -211,18 +201,14 @@ module OoxmlParser
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'pageBreakBefore'
-          if node_child.attribute('val').nil? || node_child.attribute('val').value != 'false'
-            @page_break = true
-          end
+          @page_break = true if node_child.attribute('val').nil? || node_child.attribute('val').value != 'false'
         when 'pBdr'
           @borders = ParagraphBorders.new(parent: self).parse(node_child)
         when 'keepLines'
           if node_child.attribute('val').nil?
             @keep_lines = true
           else
-            unless node_child.attribute('val').value == 'false'
-              @keep_lines = true
-            end
+            @keep_lines = true unless node_child.attribute('val').value == 'false'
           end
         when 'widowControl'
           @orphan_control = option_enabled?(node_child)
@@ -233,9 +219,7 @@ module OoxmlParser
         when 'shd'
           background_color_string = node_child.attribute('fill').value
           @background_color = Color.new(parent: self).parse_hex_string(background_color_string)
-          unless node_child.attribute('val').nil?
-            @background_color.set_style(node_child.attribute('val').value)
-          end
+          @background_color.set_style(node_child.attribute('val').value.to_sym) unless node_child.attribute('val').nil?
         when 'pStyle'
           DocxParagraph.parse_paragraph_style_xml(node_child.attribute('val').value, self, default_char_style)
         when 'ind'
@@ -250,15 +234,9 @@ module OoxmlParser
           @align = node_child.attribute('val').value.to_sym unless node_child.attribute('val').nil?
           @align = :justify if node_child.attribute('val').value == 'both'
         when 'spacing'
-          unless node_child.attribute('before').nil?
-            @spacing.before = (node_child.attribute('before').value.to_f / 566.9).round(2)
-          end
-          unless node_child.attribute('after').nil?
-            @spacing.after = (node_child.attribute('after').value.to_f / 566.9).round(2)
-          end
-          unless node_child.attribute('lineRule').nil?
-            @spacing.line_rule = node_child.attribute('lineRule').value.sub('atLeast', 'at_least').to_sym
-          end
+          @spacing.before = (node_child.attribute('before').value.to_f / 566.9).round(2) unless node_child.attribute('before').nil?
+          @spacing.after = (node_child.attribute('after').value.to_f / 566.9).round(2) unless node_child.attribute('after').nil?
+          @spacing.line_rule = node_child.attribute('lineRule').value.sub('atLeast', 'at_least').to_sym unless node_child.attribute('lineRule').nil?
           unless node_child.attribute('line').nil?
             @spacing.line = (@spacing.line_rule == :auto ? (node_child.attribute('line').value.to_f / 240.0).round(2) : (node_child.attribute('line').value.to_f / 566.9).round(2))
           end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/delimeter.rb

@@ -19,9 +19,9 @@ module OoxmlParser
           node_child.xpath('*').each do |node_child_child|
             case node_child_child.name
             when 'begChr'
-              @begin_character = node_child_child.attribute('val').value
+              @begin_character = ValuedChild.new(:string, parent: self).parse(node_child_child)
             when 'endChr'
-              @end_character = node_child_child.attribute('val').value
+              @end_character = ValuedChild.new(:string, parent: self).parse(node_child_child)
             end
           end
         end

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

@@ -0,0 +1,8 @@
+# Stuff for help working with paragraph
+module DocxParagraphHelper
+  # @return [Nil, CommentExtended] extended data for this comment
+  def comment_extend_data
+    return if @paragraph_id.nil?
+    root_object.comments_extended.by_id(@paragraph_id)
+  end
+end

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

@@ -26,31 +26,21 @@ module OoxmlParser
       @vertical_align = :baseline
       @size = 11
       @font_color = Color.new
-      @background_color = nil
       @font_style = FontStyle.new
       @text = ''
       @drawings = []
-      @link = nil
-      @highlight = nil
-      @effect = nil
-      @caps = nil
       @w = false
       @position = 0.0
-      @em = nil
       @spacing = 0.0
       @break = false
-      @touch = nil
-      @footnote = nil
-      @endnote = nil
-      @fld_char = nil
-      @style = nil
       @comments = []
-      @alternate_content = nil
       @page_number = false
-      @text_outline = nil
-      @text_fill = nil
-      @instruction = nil
-      @object = nil
+    end
+
+    def initialize_copy(source)
+      super
+      @drawings = source.drawings.clone
+      @comments = source.comments.clone
     end
 
     def drawing
@@ -58,50 +48,12 @@ module OoxmlParser
       @drawings.empty? ? nil : drawings.first
     end
 
-    def copy
-      character = DocxParagraphRun.new
-      character.number = number
-      character.font = font
-      character.vertical_align = vertical_align
-      character.size = size
-      character.font_color = font_color
-      character.background_color = @background_color
-      character.font_style = @font_style
-      character.text = @text
-      character.drawings = @drawings.clone
-      character.link = @link
-      character.highlight = @highlight
-      character.effect = @effect
-      character.caps = @caps
-      character.w = @w
-      character.position = @position
-      character.em = @em
-      character.spacing = @spacing
-      character.break = @break
-      character.touch = @touch
-      character.footnote = @footnote
-      character.endnote = @endnote
-      character.fld_char = @fld_char
-      character.style = @style
-      character.comments = @comments.clone
-      character.alternate_content = @alternate_content
-      character.page_number = @page_number
-      character.text_outline = @text_outline
-      character.text_fill = @text_fill
-      character.instruction = @instruction
-      character.run_properties = @run_properties
-      character.object = @object
-      character
-    end
-
     def ==(other)
-      ignored_attributes = [:@number]
+      ignored_attributes = %i[@number @parent]
       all_instance_variables = instance_variables
       significan_attribues = all_instance_variables - ignored_attributes
       significan_attribues.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