ooxml_parser 0.32.0 → 0.34.0

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

@@ -4,7 +4,7 @@ require_relative 'grid_span'
 module OoxmlParser
   # Class for parsing 'w:tcPr' element
   class CellProperties < OOXMLDocumentObject
-    attr_accessor :fill, :color, :borders, :text_direction, :anchor, :table_cell_width, :borders_properties, :vertical_align
+    attr_accessor :fill, :color, :borders, :text_direction, :anchor, :table_cell_width, :borders_properties
     # @return [GridSpan] data about grid span
     attr_accessor :grid_span
     # @return [TableMargins] margins
@@ -22,6 +22,8 @@ module OoxmlParser
     # width is set to auto or pct, then the content of the cell will not wrap.
     # > ECMA-376, 3rd Edition (June, 2011), Fundamentals and Markup Language Reference 17.4.30.
     attr_accessor :no_wrap
+    # @return [ValuedChild] vertical align type
+    attr_reader :vertical_align_object
 
     alias table_cell_borders borders_properties
 
@@ -38,15 +40,16 @@ module OoxmlParser
         when 'vMerge'
           @vertical_merge = ValuedChild.new(:symbol, parent: self).parse(node_child)
         when 'vAlign'
-          @vertical_align = node_child.attribute('val').value.to_sym
+          @vertical_align_object = ValuedChild.new(:symbol, parent: self).parse(node_child)
         when 'gridSpan'
           @grid_span = GridSpan.new(parent: self).parse(node_child)
         when 'tcW'
-          @table_cell_width = OoxmlSize.new(node_child.attribute('w').value.to_f)
+          @table_cell_width = OoxmlSize.new.parse(node_child)
         when 'tcMar'
           @table_cell_margin = TableMargins.new(parent: self).parse(node_child)
         when 'textDirection'
-          @text_direction = value_to_symbol(node_child.attribute('val'))
+          @text_direction_object = ValuedChild.new(:string, parent: self).parse(node_child)
+          @text_direction = value_to_symbol(@text_direction_object)
         when 'noWrap'
           @no_wrap = option_enabled?(node_child)
         when 'shd'
@@ -85,5 +88,12 @@ module OoxmlParser
       end
       self
     end
+
+    # @return [nil, Symbol] vertical align of cell
+    def vertical_align
+      return nil unless @vertical_align_object
+
+      @vertical_align_object.value
+    end
   end
 end

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

@@ -59,7 +59,8 @@ module OoxmlParser
         when 'tblW'
           @table_width = OoxmlSize.new.parse(node_child)
         when 'jc'
-          @jc = node_child.attribute('val').text.to_sym
+          @justification_object = ValuedChild.new(:symbol, parent: self).parse(node_child)
+          @jc = @justification_object.value
         when 'shd'
           @shade = Shade.new(parent: self).parse(node_child)
         when 'solidFill'
@@ -67,7 +68,7 @@ module OoxmlParser
         when 'tblLook'
           @table_look = TableLook.new(parent: self).parse(node_child)
         when 'tblInd'
-          @table_indent = OoxmlSize.new(node_child.attribute('w').text.to_f)
+          @table_indent = OoxmlSize.new(parent: self).parse(node_child)
         when 'tblpPr'
           @table_positon = TablePosition.new(parent: self).parse(node_child)
         when 'tblCellMar'

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

@@ -3,7 +3,11 @@
 module OoxmlParser
   # Class for parsing `u` tags
   class Underline < OOXMLDocumentObject
-    attr_accessor :style, :color
+    attr_accessor :style
+    # @return [Color] color of underline
+    attr_accessor :color
+    # @return [Symbol] value of underline
+    attr_reader :value
 
     def initialize(style = :none, color = nil, parent: nil)
       @style = style == 'single' ? :single : style
@@ -38,6 +42,8 @@ module OoxmlParser
     # @param node [Nokogiri::XML:Element] node to parse
     # @return [Underline] result of parsing
     def parse(node)
+      parse_attributes(node) if node.is_a?(Nokogiri::XML::Element)
+
       case node
       when 'sng'
         @style = :single
@@ -46,5 +52,20 @@ module OoxmlParser
       end
       self
     end
+
+    private
+
+    # Parse attributes
+    # @param node [Nokogiri::XML:Element] node to parse
+    def parse_attributes(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'color'
+          @color = Color.new(parent: self).parse_hex_string(value.value)
+        when 'val'
+          @value = value.value.to_sym
+        end
+      end
+    end
   end
 end

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

@@ -23,9 +23,23 @@ module OoxmlParser
           @value = value.value.to_s if type == :string
           @value = value_to_symbol(value) if type == :symbol
           @value = value.value.to_i if type == :integer
+          @value = value.value.to_f if type == :float
+          @value = parse_boolean(value.value.to_s) if type == :boolean
         end
       end
       self
     end
+
+    private
+
+    # Handle boolean value
+    # @param [String] value to parse
+    # @return [Boolean] result
+    def parse_boolean(value)
+      return true if value == '1'
+      return true if value == 'true'
+      return false if value == '0'
+      return false if value == 'false'
+    end
   end
 end

data/lib/ooxml_parser/common_parser/parser/ooxml_file.rb

@@ -56,6 +56,7 @@ module OoxmlParser
     # @param password [String] password to file
     # @return [OoxmlFile] path to decrypted file
     def decrypt(password)
+      check_decryption_support
       file_name = File.basename(@path)
       tmp_folder = Dir.mktmpdir('ruby-ooxml-parser')
       decrypted_path = "#{tmp_folder}/#{file_name}"
@@ -64,5 +65,14 @@ module OoxmlParser
 
       OoxmlFile.new(decrypted_path)
     end
+
+    private
+
+    # Check if file can be decrypted
+    # @return [Boolean] is file encrypted
+    def check_decryption_support
+      # Because of https://github.com/woodbusy/ooxml_decrypt/issues/6
+      raise NotImplementedError, 'Cannot decrypt file on `jruby` platform' if RUBY_PLATFORM == 'java'
+    end
   end
 end

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

@@ -0,0 +1,79 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Helper methods for parsing default style
+  module DefaultStyleHelper
+    # Parse default style
+    # @return [void]
+    def parse_default_style
+      parse_default_paragraph_style
+      parse_default_character_style
+      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.dup
+      parse_default_table_style
+    end
+
+    # Perform parsing styles.xml
+    def parse_styles
+      file = "#{root_object.unpacked_folder}/word/styles.xml"
+      DocumentStructure.default_paragraph_style = DocxParagraph.new(parent: self)
+      DocumentStructure.default_table_paragraph_style = DocxParagraph.new(parent: self)
+      DocumentStructure.default_run_style = DocxParagraphRun.new(parent: self)
+      DocumentStructure.default_table_run_style = DocxParagraphRun.new(parent: self)
+
+      return unless File.exist?(file)
+
+      @styles = Styles.new(parent: self).parse
+      @numbering = Numbering.new(parent: self).parse
+
+      if @styles&.document_defaults&.paragraph_properties_default
+        DocumentStructure.default_paragraph_style = DocxParagraph.new(parent: self)
+                                                                 .parse(@styles.document_defaults
+                                                                               .paragraph_properties_default
+                                                                               .raw_node, 0)
+      end
+      if @styles&.document_defaults&.run_properties_default
+        DocumentStructure.default_run_style = DocxParagraphRun.new(parent: self)
+                                                              .parse_properties(@styles
+                                                                                  .document_defaults
+                                                                                  .run_properties_default
+                                                                                  .run_properties
+                                                                                  .raw_node)
+      end
+      parse_default_style
+    end
+
+    private
+
+    # Parse default paragraph style
+    def parse_default_paragraph_style
+      return unless @styles&.default_style(:paragraph)&.paragraph_properties_node
+
+      DocxParagraph.new.parse_paragraph_style(@styles.default_style(:paragraph)
+                                                     .paragraph_properties_node,
+                                              DocumentStructure.default_run_style)
+      DocumentStructure.default_run_style
+                       .parse_properties(@styles.default_style(:paragraph)
+                                                .run_properties_node)
+    end
+
+    # Parse default character style
+    def parse_default_character_style
+      return unless @styles&.default_style(:character)&.run_properties_node
+
+      DocumentStructure.default_run_style
+                       .parse_properties(@styles.default_style(:character)
+                                                .run_properties_node)
+    end
+
+    # Parse default table style
+    def parse_default_table_style
+      return unless @styles&.default_style(:table)&.run_properties_node
+
+      DocumentStructure.default_table_run_style
+                       .parse_properties(@styles.default_style(:table)
+                                                .run_properties_node)
+    end
+  end
+end

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

@@ -19,7 +19,12 @@ module OoxmlParser
     # @param node [Nokogiri::XML:Element] node to parse
     # @return [DocumentBackground] result of parsing
     def parse(node)
-      @color1 = Color.new(parent: self).parse_hex_string(node.attribute('color').value)
+      node.attributes.each do |key, value|
+        case key
+        when 'color'
+          @color1 = Color.new(parent: self).parse_hex_string(value.value)
+        end
+      end
       node.xpath('v:background').each do |second_color|
         @size = second_color.attribute('targetscreensize').value.sub(',', 'x') unless second_color.attribute('targetscreensize').nil?
         second_color.xpath('*').each do |node_child|

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

@@ -39,6 +39,8 @@ module OoxmlParser
     # Used to determine if current style is visible in style list in editors
     # According to http://www.wordarticles.com/Articles/WordStyles/LatentStyles.php
     attr_accessor :q_format
+    # @return [Nokogiri::XML:Element] raw node value
+    attr_reader :raw_node
 
     alias visible? q_format
 
@@ -61,6 +63,7 @@ module OoxmlParser
     # Parse single document style
     # @return [DocumentStyle]
     def parse(node)
+      @raw_node = node
       node.attributes.each do |key, value|
         case key
         when 'type'
@@ -74,11 +77,14 @@ module OoxmlParser
       node.xpath('*').each do |subnode|
         case subnode.name
         when 'name'
-          @name = subnode.attribute('val').value
+          @name_object = ValuedChild.new(:string, parent: self).parse(subnode)
+          @name = @name_object.value
         when 'basedOn'
-          @based_on = subnode.attribute('val').value
+          @based_on_object = ValuedChild.new(:string, parent: self).parse(subnode)
+          @based_on = @based_on_object.value
         when 'next'
-          @next_style = subnode.attribute('val').value
+          @next_style_object = ValuedChild.new(:string, parent: self).parse(subnode)
+          @next_style = @next_style_object.value
         when 'rPr'
           @run_properties_node = subnode
           @run_properties = DocxParagraphRun.new(parent: self).parse_properties(@run_properties_node)

data/lib/ooxml_parser/docx_parser/document_structure/docx_paragraph/docx_formula/accent.rb

@@ -3,7 +3,9 @@
 module OoxmlParser
   # Class for `acc` data
   class Accent < OOXMLDocumentObject
-    attr_accessor :symbol, :element
+    # @return [ValuedChild] symbol object
+    attr_reader :symbol_object
+    attr_accessor :element
 
     # Parse Accent object
     # @param node [Nokogiri::XML:Element] node to parse
@@ -15,7 +17,7 @@ module OoxmlParser
           node_child.xpath('*').each do |node_child_child|
             case node_child_child.name
             when 'chr'
-              @symbol = node_child_child.attribute('val').value
+              @symbol_object = ValuedChild.new(:string, parent: self).parse(node_child_child)
             end
           end
         end
@@ -23,5 +25,10 @@ module OoxmlParser
       @element = DocxFormula.new(parent: self).parse(node)
       self
     end
+
+    # @return [String] symbol value
+    def symbol
+      @symbol_object.value
+    end
   end
 end

data/lib/ooxml_parser/docx_parser/document_structure/docx_paragraph/docx_formula/bar.rb

@@ -3,10 +3,11 @@
 module OoxmlParser
   # Class for `bar` data
   class Bar < OOXMLDocumentObject
-    attr_accessor :position, :element
+    # @return [ValuedChild] position object
+    attr_reader :position_object
+    attr_accessor :element
 
     def initialize(parent: nil)
-      @position = :bottom
       super
     end
 
@@ -20,7 +21,7 @@ module OoxmlParser
           node_child.xpath('*').each do |node_child_child|
             case node_child_child.name
             when 'pos'
-              @position = node_child_child.attribute('val').value.to_sym
+              @position_object = ValuedChild.new(:symbol, parent: self).parse(node_child_child)
             end
           end
         end
@@ -28,5 +29,10 @@ module OoxmlParser
       @element = DocxFormula.new(parent: self).parse(node)
       self
     end
+
+    # @return [Symbol] position of bar
+    def position
+      @position_object.value
+    end
   end
 end

data/lib/ooxml_parser/docx_parser/document_structure/docx_paragraph/docx_formula/group_char.rb

@@ -3,7 +3,13 @@
 module OoxmlParser
   # Class for `groupChr` data
   class GroupChar < OOXMLDocumentObject
-    attr_accessor :symbol, :position, :vertical_align, :element
+    # @return [ValuedChild] symbol object
+    attr_reader :symbol_object
+    # @return [ValuedChild] position object
+    attr_reader :position_object
+    # @return [ValuedChild] vertical align object
+    attr_reader :vertical_align_object
+    attr_accessor :element
 
     # Parse GroupChar object
     # @param node [Nokogiri::XML:Element] node to parse
@@ -15,11 +21,11 @@ module OoxmlParser
           node_child.xpath('*').each do |node_child_child|
             case node_child_child.name
             when 'chr'
-              @symbol = node_child_child.attribute('val').value
+              @symbol_object = ValuedChild.new(:string, parent: self).parse(node_child_child)
             when 'pos'
-              @position = node_child_child.attribute('val').value
+              @position_object = ValuedChild.new(:string, parent: self).parse(node_child_child)
             when 'vertJc'
-              @vertical_align = node_child_child.attribute('val').value
+              @vertical_align_object = ValuedChild.new(:string, parent: self).parse(node_child_child)
             end
           end
         end
@@ -27,5 +33,20 @@ module OoxmlParser
       @element = DocxFormula.new(parent: self).parse(node)
       self
     end
+
+    # @return [String] symbol value
+    def symbol
+      @symbol_object.value
+    end
+
+    # @return [String] position value
+    def position
+      @position_object.value
+    end
+
+    # @return [String] vertical align value
+    def vertical_align
+      @vertical_align_object.value
+    end
   end
 end

data/lib/ooxml_parser/docx_parser/document_structure/docx_paragraph/docx_formula/matrix.rb

@@ -22,7 +22,8 @@ module OoxmlParser
           mcs.xpath('m:mc').each do |mc|
             mc.xpath('m:mcPr').each do |mc_pr|
               mc_pr.xpath('m:count').each do |count|
-                columns_count = count.attribute('val').value.to_i
+                count_object = ValuedChild.new(:integer, parent: self).parse(count)
+                columns_count = count_object.value
               end
             end
           end

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

@@ -11,33 +11,40 @@ module OoxmlParser
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'rFonts'
+          run_fonts = RunFonts.new(parent: self).parse(node_child)
           node_child.attributes.each do |font_attribute, value|
             case font_attribute
             when 'asciiTheme'
-              theme = node_child.attribute('asciiTheme').value
               next unless root_object.theme
 
-              self.font = root_object.theme.font_scheme.major_font.latin.typeface if theme.include?('major')
-              self.font = root_object.theme.font_scheme.minor_font.latin.typeface if theme.include?('minor')
+              self.font = root_object.theme.font_scheme.major_font.latin.typeface if run_fonts.ascii_theme.include?('major')
+              self.font = root_object.theme.font_scheme.minor_font.latin.typeface if run_fonts.ascii_theme.include?('minor')
             when 'ascii'
               self.font = value.value
               break
             end
           end
         when 'sz'
-          self.size = node_child.attribute('val').value.to_i / 2.0
+          @size_object = ValuedChild.new(:integer, parent: self).parse(node_child)
+          self.size = @size_object.value.to_i / 2.0
         when 'highlight'
-          self.highlight = node_child.attribute('val').value
+          @highlight_object = ValuedChild.new(:string, parent: self).parse(node_child)
+          self.highlight = @highlight_object.value
         when 'vertAlign'
-          self.vertical_align = node_child.attribute('val').value.to_sym
+          @vertical_align_object = ValuedChild.new(:symbol, parent: self).parse(node_child)
+          self.vertical_align = @vertical_align_object.value
         when 'effect'
-          self.effect = node_child.attribute('val').value
+          @effect_object = ValuedChild.new(:string, parent: self).parse(node_child)
+          self.effect = @effect_object.value
         when 'position'
-          self.position = (node_child.attribute('val').value.to_f / (28.0 + (1.0 / 3.0)) / 2.0).round(1)
+          @position_object = ValuedChild.new(:integer, parent: self).parse(node_child)
+          self.position = (@position_object.value.to_f / (28.0 + (1.0 / 3.0)) / 2.0).round(1)
         when 'em'
-          self.em = node_child.attribute('val').value
+          @em_object = ValuedChild.new(:string, parent: self).parse(node_child)
+          self.em = @em_object.value
         when 'spacing'
-          self.spacing = (node_child.attribute('val').value.to_f / 566.9).round(1)
+          @spacing_object = RunSpacing.new(parent: self).parse(node_child)
+          self.spacing = (@spacing_object.value.value.to_f / 566.9).round(1)
         when 'textFill'
           self.text_fill = TextFill.new(parent: self).parse(node_child)
         when 'textOutline'
@@ -78,17 +85,15 @@ module OoxmlParser
     # @param node [Nokogiri::XML:Element] node to parse
     # @return [nil]
     def parse_color_tag(node)
-      node.attributes.each do |key, value|
+      color = OoxmlColor.new(parent: self).parse(node)
+      node.attributes.sort.to_h.each do |key, value|
         case key
         when 'val'
           self.font_color = Color.new(parent: self).parse_hex_string(value.value)
         when 'themeColor'
-          if root_object.theme && root_object.theme.color_scheme[value.value.to_sym]
-            break if value.value == 'text2' || value.value == 'background2' || value.value.include?('accent') # Don't know why. Just works
-
-            self.font_color = root_object.theme.color_scheme[value.value.to_sym].color.dup
-          end
+          self.font_color = root_object.theme.color_scheme[value.value.to_sym].color.dup if root_object.theme && root_object.theme.color_scheme[value.value.to_sym]
         when 'themeShade'
+          self.font_color = color.value if font_color.none?
           font_color.calculate_with_shade!(value.value.hex.to_f / 255.0)
         when 'themeTint'
           font_color.calculate_with_tint!(1.0 - (value.value.hex.to_f / 255.0))
@@ -100,12 +105,13 @@ module OoxmlParser
     # @param node [Nokogiri::XML:Element] node to parse
     # @return [nil]
     def parse_underline(node)
+      underline_object = Underline.new.parse(node)
       font_style.underlined = Underline.new
 
-      return unless option_enabled?(node)
+      return unless underline_object.value
 
-      font_style.underlined.style = node.attribute('val').value
-      font_style.underlined.color = Color.new(parent: font_style.underlined).parse_hex_string(node.attribute('color').value) unless node.attribute('color').nil?
+      font_style.underlined.style = underline_object.value
+      font_style.underlined.color = underline_object.color if underline_object.color
     end
   end
 end

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

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `w:fldChar` object
+  class FieldChar < OOXMLDocumentObject
+    # @return [Symbol] type of field char
+    attr_accessor :type
+
+    # Parse FieldChar
+    # @param [Nokogiri::XML:Node] node with FieldChar
+    # @return [FieldChar] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'fldCharType'
+          @type = value.value.to_sym
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -1,6 +1,7 @@
 # frozen_string_literal: true
 
 require_relative 'docx_paragraph_run/docx_paragraph_run_helpers'
+require_relative 'docx_paragraph_run/field_char'
 require_relative 'docx_paragraph_run/instruction_text'
 require_relative 'docx_paragraph_run/object'
 require_relative 'docx_paragraph_run/text_outline'
@@ -103,7 +104,8 @@ module OoxmlParser
             @text = '*PAGE NUMBER*'
           end
         when 'fldChar'
-          @fld_char = node_child.attribute('fldCharType').value.to_sym
+          @field_char_object = FieldChar.new(parent: self).parse(node_child)
+          @fld_char = @field_char_object.type
         when 't'
           @text += node_child.text
         when 'noBreakHyphen'

data/lib/ooxml_parser/docx_parser/document_structure/docx_paragraph/sdt/sdt_properties/checkbox/checkbox_state.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `checkedState` tag
+  class CheckBoxState < OOXMLDocumentObject
+    # @return [Integer] number of state symbol
+    attr_reader :value
+    # @return [String] font name
+    attr_reader :font
+
+    # Parse CheckBoxState object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [CheckBoxState] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'val'
+          @value = value.value.to_i
+        when 'font'
+          @font = value.value
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+require_relative 'checkbox/checkbox_state'
+module OoxmlParser
+  # Class for parsing `checkbox` tag
+  class CheckBox < OOXMLDocumentObject
+    # @return [True, False] specifies if checkbox is checked
+    attr_reader :checked
+    # @return [CheckBoxState] options of checked state
+    attr_reader :checked_state
+    # @return [CheckBoxState] options of unchecked state
+    attr_reader :unchecked_state
+    # @return [ValuedChild] group key
+    attr_reader :group_key
+
+    # Parse CheckBox object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [CheckBox] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'checked'
+          @checked = option_enabled?(node_child, 'hidden')
+        when 'checkedState'
+          @checked_state = CheckBoxState.new(parent: self).parse(node_child)
+        when 'uncheckedState'
+          @unchecked_state = CheckBoxState.new(parent: self).parse(node_child)
+        when 'groupKey'
+          @group_key = ValuedChild.new(:string, parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/document_structure/docx_paragraph/sdt/sdt_properties/combobox/list_item.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `listItem` tag
+  class ListItem < OOXMLDocumentObject
+    # @return [String] item text
+    attr_reader :text
+    # @return [String] item value
+    attr_reader :value
+
+    # Parse ListItem object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [ListItem] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'displayText'
+          @text = value.value
+        when 'value'
+          @value = value.value
+        end
+      end
+      self
+    end
+  end
+end