ooxml_parser 0.1.1

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

@@ -0,0 +1,16 @@
+# Class for describing strikeout
+module OoxmlParser
+  class Strikeout
+    # Parse Strikeout
+    # @param [Nokogiri::XML:Attr] node with Strikeout
+    # @return [Symbol] value of Strikeout
+    def self.parse(node)
+      case node.value
+      when 'sngStrike'
+        :single
+      when 'dblStrike'
+        :double
+      end
+    end
+  end
+end

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

@@ -0,0 +1,79 @@
+require_relative 'run_properties/outline'
+require_relative 'run_properties/strikeout'
+module OoxmlParser
+  class RunProperties < OOXMLDocumentObject
+    attr_accessor :font_style, :font_color, :space, :dirty, :font_name, :font_size, :baseline, :hyperlink, :caps,
+                  :vertical_align, :outline
+
+    def initialize(font_name = '', font_style = FontStyle.new, font_color = nil, space = nil, baseline = :baseline)
+      @font_name = font_name
+      @font_style = font_style
+      @font_color = font_color
+      @space = space
+      @baseline = baseline
+    end
+
+    def self.parse(character_props_node)
+      character_properties = RunProperties.new
+      character_properties.font_style = Presentation.current_font_style.dup
+      character_props_node.attributes.each do |key, value|
+        case key
+        when 'sz'
+          character_properties.font_size = value.value.to_f / 100.0
+        when 'spc'
+          character_properties.space = (value.value.to_f / 2_834.0).round(2)
+        when 'b'
+          character_properties.font_style.bold = OOXMLDocumentObject.option_enabled?(character_props_node, 'b')
+        when 'i'
+          character_properties.font_style.italic = OOXMLDocumentObject.option_enabled?(character_props_node, 'i')
+        when 'u'
+          character_properties.font_style.underlined = Underline.parse(value.value)
+        when 'strike'
+          character_properties.font_style.strike = Strikeout.parse(value)
+        when 'baseline'
+          case value.value.to_i
+          when -25_000, -30_000
+            character_properties.baseline = :subscript
+          when 30_000
+            character_properties.baseline = :superscript
+          when 0
+            character_properties.baseline = :baseline
+          end
+        when 'cap'
+          character_properties.caps = value.value.to_sym
+        end
+      end
+      character_properties.font_color = DocxColorScheme.parse(character_props_node)
+      character_props_node.xpath('*').each do |properties_element|
+        case properties_element.name
+        when 'solidFill'
+          character_properties.font_color = Color.parse_color(properties_element.xpath('*').first)
+        when 'latin'
+          character_properties.font_name = properties_element.attribute('typeface').value
+        when 'b'
+          character_properties.font_style.bold = OOXMLDocumentObject.option_enabled?(properties_element)
+        when 'i'
+          character_properties.font_style.italic = OOXMLDocumentObject.option_enabled?(properties_element, 'i')
+        when 'u'
+          character_properties.font_style.underlined = Underline.parse(properties_element.attribute('val').value)
+        when 'vertAlign'
+          character_properties.vertical_align = properties_element.attribute('val').value.to_sym
+        when 'rFont'
+          character_properties.font_name = properties_element.attribute('val').value
+        when 'color'
+          character_properties.font_color = Color.parse_color_tag(properties_element)
+        when 'strike'
+          character_properties.font_style.strike = OOXMLDocumentObject.option_enabled?(properties_element)
+        when 'hlinkClick'
+          character_properties.hyperlink = Hyperlink.parse(properties_element)
+        when 'ln'
+          character_properties.outline = Outline.parse(properties_element)
+        end
+      end
+      character_properties.font_color = DocxColorScheme.parse(character_props_node)
+      character_properties.font_name = Presentation.default_font_typeface if character_properties.font_name.empty?
+      character_properties.font_size = Presentation.default_font_size if character_properties.font_size.nil?
+      character_properties
+    end
+  end
+end

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

@@ -0,0 +1,24 @@
+require_relative 'paragraph_run/run_properties'
+module OoxmlParser
+  class ParagraphRun < OOXMLDocumentObject
+    attr_accessor :properties, :text
+
+    def initialize(properties = RunProperties.new, text = '')
+      @properties = properties
+      @text = text
+    end
+
+    def self.parse(character_node)
+      character = ParagraphRun.new
+      character_node.xpath('*').each do |character_node_child|
+        case character_node_child.name
+        when 'rPr'
+          character.properties = RunProperties.parse(character_node_child)
+        when 't'
+          character.text = character_node_child.text
+        end
+      end
+      character
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragrpah_properties/numbering/numbering_properties/numbering_level.rb

@@ -0,0 +1,25 @@
+module OoxmlParser
+  class NumberingLevel
+    attr_accessor :ilvl, :start, :num_format, :level_text, :level_jc, :ind, :font
+
+    def initialize(ilvl = nil, start = nil, num_fmt = nil, lvl_text = nil, lvl_jc = nil, ind = nil, font = nil)
+      @ilvl = ilvl
+      @start = start
+      @num_format = num_fmt
+      @level_text = lvl_text
+      @level_jc = lvl_jc
+      @ind = ind
+      @font = font
+    end
+
+    def ==(other)
+      if @font.nil? || other.font.nil? # Ilya Kirillov change logic of numbering font of bullets, so check should be correct
+        @start.to_s == other.start.to_s && @num_format == other.num_format && @level_text == other.level_text &&
+          @ind.equal_with_round(other.ind)
+      else
+        @start.to_s == other.start.to_s && @num_format == other.num_format && @level_text == other.level_text &&
+          @ind.equal_with_round(other.ind) && @font == other.font
+      end
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragrpah_properties/numbering/numbering_properties.rb

@@ -0,0 +1,69 @@
+require_relative 'numbering_properties/numbering_level'
+module OoxmlParser
+  class NumberingProperties < OOXMLDocumentObject
+    attr_accessor :id, :nsid, :multilevel_type, :tmpl, :ilvls
+
+    def initialize(id = nil, multilevel_type = nil, ilvls = [])
+      @id = id
+      @multilevel_type = multilevel_type
+      @ilvls = ilvls
+    end
+
+    def self.parse(id)
+      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + 'word/numbering.xml'), 'r:UTF-8')
+      numbering = NumberingProperties.new
+      numbering.id = id
+      doc.search('//w:num').each do |num|
+        next unless id == num.attribute('numId').value
+        num.xpath('w:abstractNumId').each do |abstract_num_id|
+          abstract_num_id = abstract_num_id.attribute('val').value
+          doc.search('//w:abstractNum').each do |abstract_num|
+            next unless abstract_num_id == abstract_num.attribute('abstractNumId').value
+            abstract_num.xpath('w:multiLevelType').each do |multilevel_type|
+              numbering.multilevel_type = multilevel_type.attribute('val').value
+            end
+            lvls = []
+            abstract_num.xpath('w:lvl').each do |lvl|
+              numbering_lvl = NumberingLevel.new
+              numbering_lvl.ilvl = lvl.attribute('ilvl').value
+              lvl.xpath('w:start').each do |start|
+                numbering_lvl.start = start.attribute('val').value
+              end
+              lvl.xpath('w:numFmt').each do |num_format|
+                numbering_lvl.num_format = num_format.attribute('val').value
+              end
+              lvl.xpath('w:lvlText').each do |level_text|
+                numbering_lvl.level_text = level_text.attribute('val').value
+              end
+              lvl.xpath('w:lvlJc').each do |level_jc|
+                numbering_lvl.level_jc = level_jc.attribute('val').value
+              end
+              lvl.xpath('w:pPr').each do |p_pr|
+                p_pr.xpath('w:ind').each do |ind|
+                  numbering_lvl.ind = Indents.parse(ind)
+                end
+                lvl.xpath('w:rPr').each do |r_pr|
+                  r_pr.xpath('w:rFonts').each do |r_fonts|
+                    if !r_fonts.attribute('ascii').nil?
+                      numbering_lvl.font = r_fonts.attribute('ascii').value
+                    elsif !r_fonts.attribute('hAnsi').nil?
+                      numbering_lvl.font = r_fonts.attribute('hAnsi').value
+                    elsif !r_fonts.attribute('eastAsia').nil?
+                      numbering_lvl.font = r_fonts.attribute('eastAsia').value
+                    else
+                      numbering_lvl.font = r_fonts.attribute('cs').value unless r_fonts.attribute('cs').nil?
+                    end
+                  end
+                end
+              end
+              lvls << numbering_lvl
+            end
+            numbering.ilvls = lvls
+          end
+        end
+      end
+
+      numbering
+    end
+  end
+end

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

@@ -0,0 +1,18 @@
+require_relative 'numbering/numbering_properties'
+module OoxmlParser
+  class Numbering
+    attr_accessor :size, :font, :symbol, :start_at, :type, :ilvl, :numbering_properties
+
+    def initialize(ilvl = 0, numbering_properties = nil)
+      @ilvl = ilvl
+      @numbering_properties = numbering_properties
+    end
+
+    def numbering_level_current
+      @numbering_properties.ilvls.each do |current_ilvl|
+        return current_ilvl if current_ilvl.ilvl == @ilvl
+      end
+      nil
+    end
+  end
+end

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

@@ -0,0 +1,115 @@
+# endcoding: utf-8
+# @author Pavel.Lobashov
+
+# Class to describe spacing
+module OoxmlParser
+  class Spacing
+    # @return [Float] Spacing before paragraph
+    attr_accessor :before
+    # @return [Float] Spacing after paragraph
+    attr_accessor :after
+    # @return [Float] Spacing between lines
+    attr_accessor :line
+    # @return [String] Spacing line rule
+    attr_accessor :line_rule
+
+    def initialize(before = nil, after = 0.35, line = nil, line_rule = nil)
+      @before = before
+      @after = after
+      @line = line
+      @line_rule = line_rule
+    end
+
+    def ==(other)
+      self.line_rule = :at_least if line_rule == 'atLeast'
+      self.line_rule = :multiple if line_rule == :auto
+      other.line_rule = :multiple if other.line_rule == :auto
+      if self.class == NilClass || other.class == NilClass
+        return true if self.class == NilClass && other.class == NilClass
+        return false
+      else
+        self.line_rule = line_rule.to_sym if line_rule.instance_of?(String)
+
+        if @before.to_f.round(2) == other.before.to_f.round(2) && @after.to_f.round(2) == other.after.to_f.round(2) &&
+           @line.to_f.round(2) == other.line.to_f.round(2) && @line_rule.to_s == other.line_rule.to_s
+          return true
+        else
+          return false
+        end
+      end
+    end
+
+    def to_s
+      result_string = ''
+      variables = instance_variables
+      variables.each do |current_variable|
+        result_string += current_variable.to_s.sub('@', '') + ': ' + instance_variable_get(current_variable).to_s + "\n"
+      end
+      result_string
+    end
+
+    def copy
+      Spacing.new(@before, @after, @line, @line_rule)
+    end
+
+    def self.default_spacing_canvas(line_spacing = 1.15)
+      Spacing.new(0.0, 0.35277777777777775, line_spacing)
+    end
+
+    def round(count_of_digits = 1)
+      before = @before.to_f.round(count_of_digits)
+      after = @after.to_f.round(count_of_digits)
+      line = @line.to_f.round(count_of_digits)
+      Spacing.new(before, after, line, @line_rule)
+    end
+
+    # Parse data for Spacing
+    # @param [Nokogiri::XML:Element] node with Spacing
+    # @return [Nothing]
+    def parse(node)
+      node.xpath('*').each do |spacing_node_child|
+        case spacing_node_child.name
+        when 'lnSpc'
+          self.line = Spacing.parse_spacing_value(spacing_node_child)
+          self.line_rule = Spacing.parse_spacing_rule(spacing_node_child)
+        when 'spcBef'
+          self.before = Spacing.parse_spacing_value(spacing_node_child)
+        when 'spcAft'
+          self.after = Spacing.parse_spacing_value(spacing_node_child)
+        end
+      end
+    end
+
+    # Parse values of spacing number
+    # @param [Nokogiri::XML:Element] node with Spacing
+    # @return [Float] value of spacing number
+    def self.parse_spacing_value(node)
+      node.xpath('*').each do |spacing_node_child|
+        case spacing_node_child.name
+        when 'spcPct'
+          return (spacing_node_child.attribute('val').value.to_f / 100_000.0).round(2)
+        when 'spcPts'
+          return (spacing_node_child.attribute('val').value.to_f / 2_834.0).round(2)
+        else
+          return (spacing_node_child.attribute('val').value.to_f / 1_000.0).round(2)
+        end
+      end
+    end
+
+    # Parse spacing rule
+    # @param [Nokogiri::XML:Element] node with Spacing
+    # @return [Symbol] type of spacing rule
+    def self.parse_spacing_rule(node)
+      node.xpath('*').each do |spacing_node_child|
+        case spacing_node_child.name
+        when 'spcPct'
+          return :multiple
+        when 'spcPts'
+          return :exact
+        else
+          return :at_least
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,35 @@
+require_relative 'paragraph/paragraph_properties'
+require_relative 'paragraph/paragraph_run'
+module OoxmlParser
+  class Paragraph
+    attr_accessor :properties, :runs, :text_field, :formulas
+
+    def initialize(runs = [], formulas = [])
+      @runs = runs
+      @formulas = formulas
+      @runs = []
+    end
+
+    alias characters runs
+    alias character_style_array runs
+    alias characters= runs=
+    alias character_style_array= runs=
+
+    def self.parse(paragraph_node)
+      paragraph = Paragraph.new
+      paragraph_node.xpath('*').each do |paragraph_node_child|
+        case paragraph_node_child.name
+        when 'pPr'
+          paragraph.properties = ParagraphProperties.parse(paragraph_node_child)
+        when 'fld'
+          paragraph.text_field = TextField.parse(paragraph_node_child)
+        when 'r'
+          paragraph.characters << ParagraphRun.parse(paragraph_node_child)
+        when 'm'
+          # TODO: add parsing formulas in paragraph
+        end
+      end
+      paragraph
+    end
+  end
+end

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

@@ -0,0 +1,35 @@
+module OoxmlParser
+  class ParagraphMargins < TableMargins
+    def initialize(top = 0.0, bottom = 0.0, left = 0.0, right = 0.0)
+      @top = top
+      @bottom = bottom
+      @left = left
+      @right = right
+    end
+
+    def round(count_of_digits = 1)
+      top = @top.round(count_of_digits)
+      bottom = @bottom.round(count_of_digits)
+      left = @left.round(count_of_digits)
+      right = @right.round(count_of_digits)
+      ParagraphMargins.new(top, bottom, left, right)
+    end
+
+    def self.parse(text_body_props_node)
+      margins = ParagraphMargins.new(0.127, 0.127, 0.254, 0.254)
+      text_body_props_node.attributes.each do |key, value|
+        case key
+        when 'bIns', 'marB'
+          margins.bottom = (value.value.to_f / 360_000.0).round(3)
+        when 'tIns', 'marT'
+          margins.top = (value.value.to_f / 360_000.0).round(3)
+        when 'lIns', 'marL'
+          margins.left = (value.value.to_f / 360_000.0).round(3)
+        when 'rIns', 'marR'
+          margins.right = (value.value.to_f / 360_000.0).round(3)
+        end
+      end
+      margins
+    end
+  end
+end

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

@@ -0,0 +1,47 @@
+module OoxmlParser
+  class TableMargins
+    attr_accessor :is_default, :top, :bottom, :left, :right
+
+    def initialize(is_default = true, top = nil, bottom = nil, left = nil, right = nil)
+      @is_default = is_default
+      @top = top
+      @bottom = bottom
+      @left = left
+      @right = right
+    end
+
+    def ==(other)
+      (@top - other.top).round(2) == 0 && (@bottom - other.bottom).round(2) == 0 &&
+        (@left - other.left).round(2) == 0 && (@right - other.right).round(2) == 0
+    end
+
+    def to_s
+      'Default: ' + is_default.to_s + ' top: ' + @top.to_s + ', bottom: ' + @bottom.to_s + ', left: ' + @left.to_s + ', right: ' + @right.to_s
+    end
+
+    def round(count_of_digits = 1)
+      top = @top.round(count_of_digits)
+      bottom = @bottom.round(count_of_digits)
+      left = @left.round(count_of_digits)
+      right = @right.round(count_of_digits)
+      TableMargins.new(@is_default, top, bottom, left, right)
+    end
+
+    def self.parse(margin_node)
+      cell_margins = TableMargins.new
+      margin_node.xpath('*').each do |cell_margin_node|
+        case cell_margin_node.name
+        when 'left'
+          cell_margins.left = (cell_margin_node.attribute('w').value.to_f / 566.9).round(2)
+        when 'top'
+          cell_margins.top = (cell_margin_node.attribute('w').value.to_f / 566.9).round(2)
+        when 'right'
+          cell_margins.right = (cell_margin_node.attribute('w').value.to_f / 566.9).round(2)
+        when 'bottom'
+          cell_margins.bottom = (cell_margin_node.attribute('w').value.to_f / 566.9).round(2)
+        end
+      end
+      cell_margins
+    end
+  end
+end

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

@@ -0,0 +1,20 @@
+# Table Cell Margin data
+module OoxmlParser
+  class TableCellMargin
+    attr_accessor :left, :right, :top, :bottom
+
+    def initialize(left = nil, right = nil, top = nil, bottom = nil)
+      @left = left
+      @top = top
+      @right = right
+      @bottom = bottom
+    end
+
+    def ==(other)
+      @left == other.left &&
+        @top == other.top &&
+        @right == other.right &&
+        @bottom == other.bottom
+    end
+  end
+end

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

@@ -0,0 +1,20 @@
+require_relative 'whole_table/table_cell_style'
+# Describe single table element
+module OoxmlParser
+  class TableElement
+    attr_accessor :cell_style
+
+    alias cell_properties cell_style
+
+    def self.parse(whole_table_node)
+      table_element = TableElement.new
+      whole_table_node.xpath('*').each do |whole_table_node_child|
+        case whole_table_node_child.name
+        when 'tcStyle', 'tcPr'
+          table_element.cell_style = CellProperties.parse(whole_table_node_child)
+        end
+      end
+      table_element
+    end
+  end
+end

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

@@ -0,0 +1,43 @@
+# Describe look of table
+module OoxmlParser
+  class TableLook < OOXMLDocumentObject
+    attr_accessor :first_row, :first_column, :last_row, :last_column, :banding_row, :banding_column, :no_horizontal_banding,
+                  :no_vertical_banding
+
+    def initialize
+      @first_row = false
+      @first_column = false
+      @last_row = false
+      @last_column = false
+      @banding_row = false
+      @banding_column = false
+      @no_horizontal_banding = false
+      @no_horizontal_banding = false
+    end
+
+    def self.parse(table_look_node)
+      table_look = TableLook.new
+      table_look_node.attributes.each do |key, value|
+        case key
+        when 'firstRow'
+          table_look.first_row = OOXMLDocumentObject.option_enabled?(value)
+        when 'lastRow'
+          table_look.last_row = OOXMLDocumentObject.option_enabled?(value)
+        when 'firstCol', 'firstColumn'
+          table_look.first_column = OOXMLDocumentObject.option_enabled?(value)
+        when 'lastCol', 'lastColumn'
+          table_look.last_column = OOXMLDocumentObject.option_enabled?(value)
+        when 'noHBand'
+          table_look.no_horizontal_banding = OOXMLDocumentObject.option_enabled?(value)
+        when 'noVBand'
+          table_look.no_vertical_banding = OOXMLDocumentObject.option_enabled?(value)
+        when 'bandRow'
+          table_look.banding_row = OOXMLDocumentObject.option_enabled?(value)
+        when 'bandCol'
+          table_look.banding_column = OOXMLDocumentObject.option_enabled?(value)
+        end
+      end
+      table_look
+    end
+  end
+end

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

@@ -0,0 +1,49 @@
+# Table Position Data
+module OoxmlParser
+  class TablePosition
+    attr_accessor :left, :right, :top, :bottom, :position_x, :position_y, :horizontal_anchor, :vertical_anchor, :vertical_align_from_anchor,
+                  :horizontal_align_from_anchor
+
+    def initialize
+      @left = nil
+      @right = nil
+      @top = nil
+      @bottom = nil
+      @position_x = nil
+      @position_y = nil
+    end
+
+    def to_s
+      'Table position left: ' + @left.to_s + ', right: ' + @right.to_s + ', top: ' + @top.to_s + ', bottom: ' + @bottom.to_s + 'position x: ' + position_x.to_s + 'position y: ' + position_x.to_s
+    end
+
+    def self.parse(node)
+      table_p_properties = TablePosition.new
+      node.attributes.each do |key, value|
+        case key
+        when 'leftFromText'
+          table_p_properties.left = value.value.to_f / 567.5
+        when 'rightFromText'
+          table_p_properties.right = value.value.to_f / 567.5
+        when 'topFromText'
+          table_p_properties.top = value.value.to_f / 567.5
+        when 'bottomFromText'
+          table_p_properties.bottom = value.value.to_f / 567.5
+        when 'tblpX'
+          table_p_properties.position_x = (value.value.to_f / 566.9).round(3)
+        when 'tblpY'
+          table_p_properties.position_y = (value.value.to_f / 566.9).round(3)
+        when 'vertAnchor'
+          table_p_properties.vertical_anchor = value.value.to_sym
+        when 'horzAnchor'
+          table_p_properties.horizontal_anchor = value.value.to_sym
+        when 'tblpXSpec'
+          table_p_properties.horizontal_align_from_anchor = value.value.to_sym
+        when 'tblpYSpec'
+          table_p_properties.vertical_align_from_anchor = value.value.to_sym
+        end
+      end
+      table_p_properties
+    end
+  end
+end