ooxml_parser 0.2.0 → 0.3.0

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragraph_properties.rb

@@ -2,11 +2,12 @@ require_relative 'paragrpah_properties/numbering_properties'
 require_relative 'paragrpah_properties/paragraph_borders'
 require_relative 'paragrpah_properties/paragraph_spacing'
 require_relative 'paragrpah_properties/spacing'
-require_relative 'paragrpah_properties/tab_list'
 require_relative 'paragrpah_properties/tabs'
 module OoxmlParser
   class ParagraphProperties < OOXMLDocumentObject
     attr_accessor :align, :numbering, :level, :spacing, :spacing_before, :spacing_after, :indent, :margin_left, :margin_right
+    # @return [FrameProperties] frame properties
+    attr_reader :frame_properties
     # @return [Tabs] list of tabs
     attr_accessor :tabs
     # @return [RunProperties] properties of run
@@ -68,10 +69,12 @@ module OoxmlParser
         when 'buAutoNum'
           @numbering.type = node_child.attribute('type').value.to_sym
           @numbering.start_at = node_child.attribute('startAt').value if node_child.attribute('startAt')
+        when 'framePr'
+          @frame_properties = FrameProperties.new(parent: self).parse(node_child)
         when 'tabs'
           @tabs = Tabs.new(parent: self).parse(node_child)
         when 'tabLst'
-          @tabs = TabList.new(parent: self).parse(node_child)
+          @tabs = Tabs.new(parent: self).parse(node_child)
         when 'ind'
           @indent = Indents.new(parent: self).parse(node_child)
         when 'rPr'

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragraph_run.rb

@@ -1,8 +1,13 @@
 require_relative 'paragraph_run/run_properties'
+require_relative 'paragraph_run/text'
 module OoxmlParser
   # Class for parsing `r` tags
   class ParagraphRun < OOXMLDocumentObject
     attr_accessor :properties, :text
+    # @return [Text] text of run
+    attr_reader :t
+    # @return [Tab] tab of paragraph
+    attr_reader :tab
 
     def initialize(properties = RunProperties.new, text = '', parent: nil)
       @properties = properties
@@ -19,7 +24,10 @@ module OoxmlParser
         when 'rPr'
           @properties = RunProperties.new(parent: self).parse(node_child)
         when 't'
-          @text = node_child.text
+          @t = Text.new(parent: self).parse(node_child)
+          @text = t.text
+        when 'tab'
+          @tab = Tab.new(parent: self).parse(node_child)
         end
       end
       self

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragraph_run/run_properties.rb

@@ -89,7 +89,7 @@ module OoxmlParser
         when 'spacing'
           @spacing = RunSpacing.new(parent: self).parse(node_child)
         when 'color'
-          @color = Color.parse_color_tag(node_child)
+          @color = OoxmlColor.new(parent: self).parse(node_child)
         when 'solidFill'
           @font_color = Color.parse_color(node_child.xpath('*').first)
         when 'latin'

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragraph_run/text.rb

@@ -0,0 +1,23 @@
+module OoxmlParser
+  # Class for parsing `t` tags
+  class Text < OOXMLDocumentObject
+    # @return [Symbol] space values
+    attr_reader :space
+    # @return [String] text value
+    attr_reader :text
+
+    # Parse Text object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Text] result of parsing
+    def parse(node)
+      @text = node.text
+      node.attributes.each do |key, value|
+        case key
+        when 'space'
+          @space = value_to_symbol(value)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragrpah_properties/paragraph_borders.rb

@@ -22,8 +22,8 @@ module OoxmlParser
       result << :bottom if @bottom.val == :single
       result << :inner if @between.val == :single
       return :none if result == []
-      return :all if result == [:left, :right, :top, :bottom, :inner]
-      return :outer if result == [:left, :right, :top, :bottom]
+      return :all if result == %i[left right top bottom inner]
+      return :outer if result == %i[left right top bottom]
       result.first if result.size == 1
     end
 

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragrpah_properties/spacing.rb

@@ -1,4 +1,3 @@
-# endcoding: utf-8
 # @author Pavel.Lobashov
 
 # Class to describe spacing

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragrpah_properties/tabs.rb

@@ -1,4 +1,4 @@
-require_relative 'tabs/paragraph_tab'
+require_relative 'tabs/tab'
 module OoxmlParser
   # Class for working with `w:tabs`
   class Tabs < OOXMLDocumentObject
@@ -27,7 +27,7 @@ module OoxmlParser
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'tab'
-          @tabs_array << ParagraphTab.new(parent: self).parse(node_child)
+          @tabs_array << Tab.new(parent: self).parse(node_child)
         end
       end
       self

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragrpah_properties/{tab_list/presentation_tab.rb → tabs/tab.rb}

@@ -1,8 +1,10 @@
 module OoxmlParser
-  # Class for storing `a:tab` data
-  class PresentationTab < OOXMLDocumentObject
+  # Class for storing `w:tab`, `a:tab` data
+  class Tab < OOXMLDocumentObject
     # @return [Symbol] Specifies the style of the tab.
     attr_accessor :value
+    # @return [Symbol] Specifies the leader symbol of tab
+    attr_reader :leader
     # @return [OOxmlSize] Specifies the position of the tab stop.
     attr_accessor :position
 
@@ -14,13 +16,24 @@ module OoxmlParser
     def parse(node)
       node.attributes.each do |key, value|
         case key
-        when 'algn'
+        when 'algn', 'val'
           @value = value_to_symbol(value)
+        when 'leader'
+          @leader = value_to_symbol(value)
         when 'pos'
-          @position = OoxmlSize.new(value.value.to_f, :emu)
+          @position = OoxmlSize.new(value.value.to_f, position_unit(node))
         end
       end
       self
     end
+
+    private
+
+    # @param node [Nokogiri::XML:Element] node to determine size
+    # @return [Symbol] type of size unit
+    def position_unit(node)
+      return :emu if node.namespace.prefix == 'a'
+      :dxa
+    end
   end
 end

data/lib/ooxml_parser/common_parser/common_data/table.rb

@@ -28,7 +28,7 @@ module OoxmlParser
     def parse(node,
               number = 0,
               default_table_properties = TableProperties.new)
-      table_properties = default_table_properties.copy
+      table_properties = default_table_properties.dup
       table_properties.jc = :left
       node.xpath('*').each do |node_child|
         case node_child.name

data/lib/ooxml_parser/common_parser/common_data/table/margins/table_margins.rb

@@ -16,9 +16,7 @@ module OoxmlParser
       instance_variables.each do |current_attribute|
         next if current_attribute == :@parent
         next if current_attribute == :@is_default
-        unless instance_variable_get(current_attribute) == other.instance_variable_get(current_attribute)
-          return false
-        end
+        return false unless instance_variable_get(current_attribute) == other.instance_variable_get(current_attribute)
       end
       true
     end

data/lib/ooxml_parser/common_parser/common_data/table/properties/table_style.rb

@@ -10,45 +10,39 @@ module OoxmlParser
     # @param style_id [String] id to parse
     # @return [TableStyle] result of parsing
     def parse(style_id: nil)
-      unless style_id.nil?
-        nodes = get_style_node(style_id)
-        return nil if nodes.nil?
-        xpath = '*'
-        attributes = 'style_node_child.name'
-      end
-      begin
-        nodes.xpath(xpath).each do |style_node_child|
-          case instance_eval(attributes)
-          when 'wholeTbl'
-            @whole_table = TableElement.new(parent: self).parse(style_node_child)
-          when 'band1H'
-            @banding_1_horizontal = TableElement.new(parent: self).parse(style_node_child)
-          when 'band2H', 'band2Horz'
-            @banding_2_horizontal = TableElement.new(parent: self).parse(style_node_child)
-          when 'band1V'
-            @banding_1_vertical = TableElement.new(parent: self).parse(style_node_child)
-          when 'band2V'
-            @banding_2_vertical = TableElement.new(parent: self).parse(style_node_child)
-          when 'lastCol'
-            @last_column = TableElement.new(parent: self).parse(style_node_child)
-          when 'firstCol'
-            @first_column = TableElement.new(parent: self).parse(style_node_child)
-          when 'lastRow'
-            @last_row = TableElement.new(parent: self).parse(style_node_child)
-          when 'firstRow'
-            @first_row = TableElement.new(parent: self).parse(style_node_child)
-          when 'seCell'
-            @southeast_cell = TableElement.new(parent: self).parse(style_node_child)
-          when 'swCell'
-            @southwest_cell = TableElement.new(parent: self).parse(style_node_child)
-          when 'neCell'
-            @northeast_cell = TableElement.new(parent: self).parse(style_node_child)
-          when 'nwCell'
-            @northwest_cell = TableElement.new(parent: self).parse(style_node_child)
-          end
+      nodes = get_style_node(style_id)
+      return nil unless nodes
+      nodes.xpath('*').each do |style_node_child|
+        case style_node_child.name
+        when 'wholeTbl'
+          @whole_table = TableElement.new(parent: self).parse(style_node_child)
+        when 'band1H'
+          @banding_1_horizontal = TableElement.new(parent: self).parse(style_node_child)
+        when 'band2H', 'band2Horz'
+          @banding_2_horizontal = TableElement.new(parent: self).parse(style_node_child)
+        when 'band1V'
+          @banding_1_vertical = TableElement.new(parent: self).parse(style_node_child)
+        when 'band2V'
+          @banding_2_vertical = TableElement.new(parent: self).parse(style_node_child)
+        when 'lastCol'
+          @last_column = TableElement.new(parent: self).parse(style_node_child)
+        when 'firstCol'
+          @first_column = TableElement.new(parent: self).parse(style_node_child)
+        when 'lastRow'
+          @last_row = TableElement.new(parent: self).parse(style_node_child)
+        when 'firstRow'
+          @first_row = TableElement.new(parent: self).parse(style_node_child)
+        when 'seCell'
+          @southeast_cell = TableElement.new(parent: self).parse(style_node_child)
+        when 'swCell'
+          @southwest_cell = TableElement.new(parent: self).parse(style_node_child)
+        when 'neCell'
+          @northeast_cell = TableElement.new(parent: self).parse(style_node_child)
+        when 'nwCell'
+          @northwest_cell = TableElement.new(parent: self).parse(style_node_child)
         end
       end
-      @id = style_id unless style_id.nil?
+      @id = style_id
       self
     end
 

data/lib/ooxml_parser/common_parser/common_data/table/properties/table_style_properties.rb

@@ -10,6 +10,8 @@ module OoxmlParser
     attr_accessor :table_cell_properties
     # @return [TableProperties] properties of table
     attr_accessor :table_properties
+    # @return [TableRowProperties] properties of table row
+    attr_reader :table_row_properties
     # @return [ParagraphProperties] properties of paragraph
     attr_accessor :paragraph_properties
     alias cell_properties table_cell_properties
@@ -40,6 +42,8 @@ module OoxmlParser
           @table_cell_properties = CellProperties.new(parent: self).parse(node_child)
         when 'tblPr'
           @table_properties = TableProperties.new(parent: self).parse(node_child)
+        when 'trPr'
+          @table_row_properties = TableRowProperties.new(parent: self).parse(node_child)
         when 'pPr'
           @paragraph_properties = ParagraphProperties.new(parent: self).parse(node_child)
         end

data/lib/ooxml_parser/common_parser/common_data/table/properties/table_style_properties/table_style_properties_helper.rb

@@ -15,7 +15,7 @@ module OoxmlParser
                                 first_row: :firstRow,
                                 last_row: :lastRow }.freeze
 
-    TABLE_STYLES_NAMES_HASH.keys.each do |table_style_name|
+    TABLE_STYLES_NAMES_HASH.each_key do |table_style_name|
       define_method(table_style_name) do
         @table_style_properties_list.each do |table_style|
           return table_style if table_style.type == TABLE_STYLES_NAMES_HASH[table_style_name]
@@ -29,7 +29,7 @@ module OoxmlParser
     # with `ooxml_parser` 0.1.2 and earlier
     # @return [Nothing]
     def fill_empty_table_styles
-      TABLE_STYLES_NAMES_HASH.values.each do |current_table_style|
+      TABLE_STYLES_NAMES_HASH.each_value do |current_table_style|
         style_exist = false
         @table_style_properties_list.each do |existing_style|
           style_exist = true if existing_style.type == current_table_style

data/lib/ooxml_parser/common_parser/common_data/table/row/cell/cell.rb

@@ -34,10 +34,10 @@ module OoxmlParser
         when 'tcPr'
           @properties = CellProperties.new(parent: self).parse(node_child)
         when 'p'
-          @elements << DocumentStructure.default_table_paragraph_style.copy.parse(node_child,
-                                                                                  0,
-                                                                                  DocumentStructure.default_table_run_style,
-                                                                                  parent: self)
+          @elements << DocumentStructure.default_table_paragraph_style.dup.parse(node_child,
+                                                                                 0,
+                                                                                 DocumentStructure.default_table_run_style,
+                                                                                 parent: self)
         when 'tbl'
           @elements << Table.new(parent: self).parse(node_child)
         end

data/lib/ooxml_parser/common_parser/common_data/table/row/cell/properties/border.rb

@@ -18,7 +18,7 @@ module OoxmlParser
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'color'
-          @color = Color.parse_color_tag(node_child)
+          @color = OoxmlColor.new(parent: self).parse(node_child)
         end
       end
       self

data/lib/ooxml_parser/common_parser/common_data/table/table_properties.rb

@@ -10,7 +10,7 @@ require_relative 'table_properties/table_borders'
 module OoxmlParser
   # Class for parsing `w:tblPr` tags
   class TableProperties < OOXMLDocumentObject
-    attr_accessor :jc, :table_width, :table_borders, :table_properties, :table_positon, :table_cell_margin, :table_indent, :stretching, :table_style, :row_banding_size,
+    attr_accessor :jc, :table_width, :table_borders, :table_positon, :table_cell_margin, :table_indent, :stretching, :table_style, :row_banding_size,
                   :column_banding_size, :table_look, :grid_column, :style
     # @return [TableStyleColumnBandSize] table style column band size
     attr_accessor :table_style_column_band_size
@@ -35,21 +35,6 @@ module OoxmlParser
       @parent = parent
     end
 
-    def copy
-      table = TableProperties.new
-      table.jc = @jc
-      table.table_width = @table_width
-      table.shade = @shade
-      table.stretching = @stretching
-      table.table_borders = @table_borders
-      table.table_properties = @table_properties
-      table.table_cell_margin = @table_cell_margin
-      table.table_indent = @table_indent
-      table.grid_column = @grid_column
-      table.style = style
-      table
-    end
-
     # Parse TableProperties object
     # @param node [Nokogiri::XML:Element] node to parse
     # @return [TableProperties] result of parsing

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

@@ -1,5 +1,6 @@
 # noinspection RubyInstanceMethodNamingConvention
 require_relative 'document_structure/comment'
+require_relative 'document_structure/comments_extended'
 require_relative 'document_structure/docx_paragraph'
 require_relative 'document_structure/document_background'
 require_relative 'document_structure/document_properties'
@@ -25,6 +26,8 @@ module OoxmlParser
     attr_accessor :theme_colors
     # @return [DocumentSettings] settings
     attr_accessor :settings
+    # @return [CommentsExtended] extended comments
+    attr_accessor :comments_extended
 
     def initialize
       @elements = []
@@ -147,19 +150,22 @@ module OoxmlParser
         end
         document.xpath('w:body').each do |body|
           body.xpath('*').each do |element|
-            if element.name == 'p'
+            case element.name
+            when 'p'
               child = element.child
               unless child.nil? && doc_structure.elements.last.class == Table
-                paragraph_style = DocumentStructure.default_paragraph_style.copy.parse(element, number, DocumentStructure.default_run_style, parent: doc_structure)
+                paragraph_style = DocumentStructure.default_paragraph_style.dup.parse(element, number, DocumentStructure.default_run_style, parent: doc_structure)
                 number += 1
-                doc_structure.elements << paragraph_style.copy
+                doc_structure.elements << paragraph_style.dup
               end
-            elsif element.name == 'tbl'
+            when 'tbl'
               table = Table.new(parent: doc_structure).parse(element,
                                                              number,
                                                              TableProperties.new)
               number += 1
               doc_structure.elements << table
+            when 'sdt'
+              doc_structure.elements << StructuredDocumentTag.new(parent: self).parse(element)
             end
           end
           body.xpath('w:sectPr').each do |sect_pr|
@@ -172,7 +178,8 @@ module OoxmlParser
       end
       OOXMLDocumentObject.xmls_stack.pop
       doc_structure.document_properties = DocumentProperties.new(parent: doc_structure).parse
-      doc_structure.comments = Comment.parse_list
+      doc_structure.comments = Comment.parse_list(parent: doc_structure)
+      doc_structure.comments_extended = CommentsExtended.new(parent: doc_structure).parse
       doc_structure.settings = DocumentSettings.new(parent: doc_structure).parse
       doc_structure
     end
@@ -194,9 +201,9 @@ module OoxmlParser
           end
         end
       end
-      DocumentStructure.default_table_paragraph_style = DocumentStructure.default_paragraph_style.copy
+      DocumentStructure.default_table_paragraph_style = DocumentStructure.default_paragraph_style.dup
       DocumentStructure.default_table_paragraph_style.spacing = Spacing.new(0, 0, 1, :auto)
-      DocumentStructure.default_table_run_style = DocumentStructure.default_run_style.copy
+      DocumentStructure.default_table_run_style = DocumentStructure.default_run_style.dup
       doc.search('//w:style').each do |style|
         next if style.attribute('default').nil?
         next unless (style.attribute('default').value == '1' || style.attribute('default').value == 'on' || style.attribute('default').value == 'true') && style.attribute('type').value == 'table'

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

@@ -8,7 +8,7 @@ module OoxmlParser
       @paragraphs = paragraphs
     end
 
-    def self.parse_list
+    def self.parse_list(parent: nil)
       comments = []
       comments_filename = "#{OOXMLDocumentObject.path_to_folder}/#{OOXMLDocumentObject.root_subfolder}/comments.xml"
       return [] unless File.exist?(comments_filename)
@@ -17,7 +17,7 @@ module OoxmlParser
         document.xpath('w:comment').each do |comment_tag|
           comment = Comment.new(comment_tag.attribute('id').value)
           comment_tag.xpath('w:p').each_with_index do |p, index|
-            comment.paragraphs << DocxParagraph.new.parse(p, index)
+            comment.paragraphs << DocxParagraph.new.parse(p, index, parent: parent)
           end
           comments << comment.dup
         end

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

@@ -0,0 +1,39 @@
+require_relative 'comments_extended/comment_extended'
+module OoxmlParser
+  # Class for parsing `commentsExtended.xml` file
+  class CommentsExtended < OOXMLDocumentObject
+    def initialize(parent: nil)
+      @comments_extended_array = []
+      @parent = parent
+    end
+
+    # @return [Array, CommentsExtended] accessor
+    def [](key)
+      @comments_extended_array[key]
+    end
+
+    # Parse CommentsExtended object
+    # @return [CommentsExtended] result of parsing
+    def parse
+      file_to_parse = OOXMLDocumentObject.path_to_folder + 'word/commentsExtended.xml'
+      return nil unless File.exist?(file_to_parse)
+      doc = Nokogiri::XML(File.open(file_to_parse))
+      doc.xpath('w15:commentsEx/*').each do |node_child|
+        case node_child.name
+        when 'commentEx'
+          @comments_extended_array << CommentExtended.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+
+    # @param id [Integer] id of comment
+    # @return [CommentExtended] comment by id
+    def by_id(id)
+      @comments_extended_array.each do |cur_comment|
+        return cur_comment if cur_comment.paragraph_id == id
+      end
+      nil
+    end
+  end
+end