ooxml_parser 0.1.1

data/README.md

@@ -0,0 +1,26 @@
+# ooxml_parser
+
+_ooxml_parser_ is a Ooxml files (docx, xlsx, pptx) parser written in Ruby.
+
+## Installation
+
+    $ gem install parser
+    
+## Usage
+
+Parse a docx file
+
+    require 'ooxml_parser'
+    docx = OoxmlParser::DocxParser.parse_docx('spec/docx_examples/document_properties/page_count.docx')
+    p docx.document_properties.pages # 2
+    
+    
+## Configuration
+
+Set units of mesaurements:
+Possible units are `:points`, `:centimeters`, `:dxa`, `:inches`, `:emu`
+Default is `:centimeters`
+
+    OoxmlParser.configure do |config|
+      config.units = :points
+    end

data/circle.yml

@@ -0,0 +1,3 @@
+machine:
+  ruby:
+    version: 2.3.0

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

@@ -0,0 +1,56 @@
+# Class for describing Alignment type
+module OoxmlParser
+  class Alignment
+    # Parse TransitionDirection
+    # @param [Nokogiri::XML:Element] node with Alignment
+    # @return [Symbol] value of Alignment
+    def self.parse(node)
+      case node.value
+      when 'l'
+        :left
+      when 'ctr'
+        :center
+      when 'r'
+        :right
+      when 'just'
+        :justify
+      when 'b'
+        :bottom
+      when 't'
+        :top
+      when 'tr'
+        :top_right
+      when 'tl'
+        :top_left
+      when 'br'
+        :bottom_right
+      when 'bl'
+        :bottom_left
+      when 'dist'
+        :distributed
+      when 'tb'
+        :horizontal
+      when 'rl'
+        :rotate_on_90
+      when 'lr'
+        :rotate_on_270
+      when 'inset'
+        :in
+      when 'lu'
+        :left_up
+      when 'ru'
+        :right_up
+      when 'ld'
+        :left_down
+      when 'rd'
+        :right_down
+      when 'd'
+        :down
+      when 'u'
+        :up
+      else
+        node.value.to_sym
+      end
+    end
+  end
+end

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

@@ -0,0 +1,34 @@
+require_relative 'chart/chart'
+require_relative 'drawing/docx_drawing'
+require_relative 'shape/shape'
+require_relative 'picture/old_docx_picture'
+# Class for storing fallback graphic elements
+module OoxmlParser
+  class AlternateContent
+    attr_accessor :office2010_content, :office2007_content
+
+    def self.parse(alt_content_node)
+      alternate_content = AlternateContent.new
+      alt_content_node.xpath('*').each do |alternate_content_node_child|
+        begin
+          alternate_content_node_child.xpath('w:drawing')
+        rescue Nokogiri::XML::XPath::SyntaxError # This mean it is Chart
+          case alternate_content_node_child.name
+          when 'Choice'
+            alternate_content.office2010_content = Office2010ChartStyle.parse(alternate_content_node_child)
+          when 'Fallback'
+            alternate_content.office2007_content = Office2007ChartStyle.parse(alternate_content_node_child)
+          end
+          next
+        end
+        case alternate_content_node_child.name
+        when 'Choice'
+          alternate_content.office2010_content = DocxDrawing.parse(alternate_content_node_child.xpath('w:drawing').first) unless alternate_content_node_child.xpath('w:drawing').first.nil?
+        when 'Fallback'
+          alternate_content.office2007_content = OldDocxPicture.parse(alternate_content_node_child.xpath('w:pict').first) unless alternate_content_node_child.xpath('w:pict').first.nil?
+        end
+      end
+      alternate_content
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/chart/chart.rb

@@ -0,0 +1,123 @@
+require_relative 'chart_axis'
+require_relative 'chart_cells_range'
+require_relative 'chart_legend'
+require_relative 'chart_point'
+require_relative 'display_labels_properties'
+require_relative 'chart_style/office_2007_chart_style'
+require_relative 'chart_style/office_2010_chart_style'
+
+module OoxmlParser
+  class Chart < OOXMLDocumentObject
+    attr_accessor :type, :data, :grouping, :title, :legend, :display_labels, :axises, :alternate_content, :shape_properties
+
+    def initialize
+      @type = ''
+      @data = []
+      @grouping = nil
+      @title = nil
+      @legend = nil
+      @axises = []
+    end
+
+    def parse_properties(chart_prop_node)
+      chart_prop_node.xpath('*').each do |chart_props_node_child|
+        case chart_props_node_child.name
+        when 'grouping'
+          @grouping = chart_props_node_child.attribute('val').value.to_sym
+        when 'ser'
+          case @type
+          when :bar, :line, :area, :pie, :doughnut, :radar, :stock, :surface_3d, :column
+            val = chart_props_node_child.xpath('c:val')[0]
+          when :point, :bubble
+            val = chart_props_node_child.xpath('c:yVal')[0]
+          else
+            val = nil
+          end
+          @data << ChartCellsRange.parse(val.xpath('c:numRef').first).dup unless val.nil?
+        when 'dLbls'
+          @display_labels = DisplayLabelsProperties.parse(chart_props_node_child)
+        end
+      end
+    end
+
+    def self.parse
+      chart = Chart.new
+      chart_xml = Nokogiri::XML(File.open(OOXMLDocumentObject.current_xml))
+      chart_xml.xpath('c:chartSpace/*', 'xmlns:c' => 'http://schemas.openxmlformats.org/drawingml/2006/chart').each do |chart_space_node_child|
+        case chart_space_node_child.name
+        when 'AlternateContent'
+          chart.alternate_content = AlternateContent.parse(chart_space_node_child)
+        when 'spPr'
+          chart.shape_properties = DocxShapeProperties.parse chart_space_node_child
+        when 'chart'
+          chart_space_node_child.xpath('*').each do |chart_node_child|
+            case chart_node_child.name
+            when 'plotArea'
+              chart_node_child.xpath('*').each do |plot_area_node_child|
+                next unless chart.type.empty?
+                case plot_area_node_child.name
+                when 'barChart'
+                  chart.type = :bar
+                  bar_dir_node = plot_area_node_child.xpath('c:barDir')
+                  unless bar_dir_node.first.nil?
+                    chart.type = :column if bar_dir_node.first.attribute('val').value == 'col'
+                  end
+                  chart.parse_properties(plot_area_node_child)
+                when 'lineChart'
+                  chart.type = :line
+                  chart.parse_properties(plot_area_node_child)
+                when 'areaChart'
+                  chart.type = :area
+                  chart.parse_properties(plot_area_node_child)
+                when 'bubbleChart'
+                  chart.type = :bubble
+                  chart.parse_properties(plot_area_node_child)
+                when 'doughnutChart'
+                  chart.type = :doughnut
+                  chart.parse_properties(plot_area_node_child)
+                when 'pieChart'
+                  chart.type = :pie
+                  chart.parse_properties(plot_area_node_child)
+                when 'scatterChart'
+                  chart.type = :point
+                  chart.parse_properties(plot_area_node_child)
+                when 'radarChart'
+                  chart.type = :radar
+                  chart.parse_properties(plot_area_node_child)
+                when 'stockChart'
+                  chart.type = :stock
+                  chart.parse_properties(plot_area_node_child)
+                when 'surface3DChart'
+                  chart.type = :surface_3d
+                  chart.parse_properties(plot_area_node_child)
+                when 'line3DChart'
+                  chart.type = :line_3d
+                  chart.parse_properties(plot_area_node_child)
+                when 'bar3DChart'
+                  chart.type = :bar_3d
+                  chart.parse_properties(plot_area_node_child)
+                when 'pie3DChart'
+                  chart.type = :pie_3d
+                  chart.parse_properties(plot_area_node_child)
+                end
+              end
+              chart_node_child.xpath('*').each do |plot_area_node_child|
+                case plot_area_node_child.name
+                when 'catAx'
+                  chart.axises << ChartAxis.parse(plot_area_node_child)
+                when 'valAx'
+                  chart.axises << ChartAxis.parse(plot_area_node_child)
+                end
+              end
+            when 'title'
+              chart.title = ChartAxisTitle.parse(chart_node_child)
+            when 'legend'
+              chart.legend = ChartLegend.parse(chart_node_child)
+            end
+          end
+        end
+      end
+      chart
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/chart/chart_axis.rb

@@ -0,0 +1,34 @@
+require_relative 'chart_axis_title'
+# Chart Axis
+module OoxmlParser
+  class ChartAxis < OOXMLDocumentObject
+    attr_accessor :title, :display, :position, :major_grid_lines, :minor_grid_lines
+
+    def initialize(title = ChartAxisTitle.new, display = true, major_grid_lines = false, minor_grid_lines = false)
+      @title = title
+      @display = display
+      @minor_grid_lines = minor_grid_lines
+      @major_grid_lines = major_grid_lines
+    end
+
+    def self.parse(axis_node)
+      axis = ChartAxis.new
+      axis_node.xpath('*').each do |axis_node_child|
+        case axis_node_child.name
+        when 'delete'
+          axis.display = false if axis_node_child.attribute('val').value == '1'
+        when 'title'
+          axis.title = ChartAxisTitle.parse(axis_node_child)
+        when 'majorGridlines'
+          axis.major_grid_lines = true
+        when 'minorGridlines'
+          axis.minor_grid_lines = true
+        when 'axPos'
+          axis.position = Alignment.parse(axis_node_child.attribute('val'))
+        end
+      end
+      axis.display = false if axis.title.nil?
+      axis
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/chart/chart_axis_title.rb

@@ -0,0 +1,38 @@
+# Chart Axis Title
+module OoxmlParser
+  class ChartAxisTitle < OOXMLDocumentObject
+    attr_accessor :layout, :overlay, :elements
+
+    def initialize(layout = nil, overlay = nil, elements = [])
+      @layout = layout
+      @overlay = overlay
+      @elements = elements
+    end
+
+    def nil?
+      @layout.nil? && @overlay.nil? && @elements.empty?
+    end
+
+    def self.parse(axis_title_node)
+      title = ChartAxisTitle.new
+      axis_title_node.xpath('*').each do |title_node_child|
+        case title_node_child.name
+        when 'tx'
+          title_node_child.xpath('c:rich/*').each do |rich_node_child|
+            case rich_node_child.name
+            when 'p'
+              Presentation.current_font_style = FontStyle.new(true) # Default font style for chart title always bold
+              title.elements << Paragraph.parse(rich_node_child)
+              Presentation.current_font_style = FontStyle.new
+            end
+          end
+        when 'layout'
+          title.layout = OOXMLDocumentObject.option_enabled?(title_node_child)
+        when 'overlay'
+          title.overlay = OOXMLDocumentObject.option_enabled?(title_node_child)
+        end
+      end
+      title
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/chart/chart_cells_range.rb

@@ -0,0 +1,23 @@
+# Cell Range of Chart
+module OoxmlParser
+  class ChartCellsRange
+    attr_accessor :list, :points
+
+    def initialize(list = '', points = [])
+      @list = list
+      @points = points
+    end
+
+    def self.parse(num_ref_node)
+      chart_range = ChartCellsRange.new
+      chart_range.list = num_ref_node.xpath('c:f')[0].text.split('!').first
+      coordinates = Coordinates.parser_coordinates_range(num_ref_node.xpath('c:f')[0].text) # .split('!')[1].gsub('$', ''))
+      num_ref_node.xpath('c:numCache/c:pt').each_with_index do |point_node, index|
+        point = ChartPoint.new(coordinates[index])
+        point.value = point_node.xpath('c:v').first.text.to_f unless point_node.xpath('c:v').first.nil?
+        chart_range.points << point
+      end
+      chart_range
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/chart/chart_legend.rb

@@ -0,0 +1,33 @@
+# Legend of Chart
+module OoxmlParser
+  class ChartLegend
+    attr_accessor :position, :overlay
+
+    def initialize(position = :right, overlay = false)
+      @position = position
+      @overlay = overlay
+    end
+
+    # Return combined data from @position and @overlay
+    # If there is no overlay - return :right f.e.
+    # If there is overlay - return :right_overlay
+    # @return [Symbol] overlay and position type
+    def position_with_overlay
+      return "#{@position}_overlay".to_sym if overlay
+      @position
+    end
+
+    def self.parse(legend_node)
+      legend = ChartLegend.new
+      legend_node.xpath('*').each do |legend_node_child|
+        case legend_node_child.name
+        when 'legendPos'
+          legend.position = Alignment.parse(legend_node_child.attribute('val'))
+        when 'overlay'
+          legend.overlay = true if legend_node_child.attribute('val').value.to_s == '1'
+        end
+      end
+      legend
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/chart/chart_point.rb

@@ -0,0 +1,11 @@
+# Single Chart Point
+module OoxmlParser
+  class ChartPoint
+    attr_accessor :coordinates, :value
+
+    def initialize(coordinates = nil, value = nil)
+      @coordinates = coordinates
+      @value = value
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/chart/chart_style/office_2007_chart_style.rb

@@ -0,0 +1,17 @@
+# XLSX 2007 Chart Style
+module OoxmlParser
+  class Office2007ChartStyle
+    attr_accessor :style_number
+
+    def self.parse(chart_style_node)
+      chart_style = Office2007ChartStyle.new
+      chart_style_node.xpath('*').each do |chart_style_node_child|
+        case chart_style_node_child.name
+        when 'style'
+          chart_style.style_number = chart_style_node_child.attribute('val').value.to_i
+        end
+      end
+      chart_style
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/chart/chart_style/office_2010_chart_style.rb

@@ -0,0 +1,17 @@
+# XLSX 2010 Chart Style
+module OoxmlParser
+  class Office2010ChartStyle
+    attr_accessor :style_number
+
+    def self.parse(chart_style_node)
+      chart_style = Office2010ChartStyle.new
+      chart_style_node.xpath('*').each do |chart_style_node_child|
+        case chart_style_node_child.name
+        when 'style'
+          chart_style.style_number = chart_style_node_child.attribute('val').value.to_i
+        end
+      end
+      chart_style
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/chart/display_labels_properties.rb

@@ -0,0 +1,26 @@
+# Chart Label Properties
+module OoxmlParser
+  class DisplayLabelsProperties
+    attr_accessor :position, :show_legend_key, :show_values, :show_x_axis_name, :show_y_axis_name
+
+    def initialize(show_legend_key = false, show_values = false)
+      @show_legend_key = show_legend_key
+      @show_values = show_values
+    end
+
+    def self.parse(display_labels_node)
+      display_labels_properties = DisplayLabelsProperties.new
+      display_labels_node.xpath('*').each do |display_label_property_node|
+        case display_label_property_node.name
+        when 'dLblPos'
+          display_labels_properties.position = Alignment.parse(display_label_property_node.attribute('val'))
+        when 'showLegendKey'
+          display_labels_properties.show_legend_key = true if display_label_property_node.attribute('val').value == '1'
+        when 'showVal'
+          display_labels_properties.show_values = true if display_label_property_node.attribute('val').value == '1'
+        end
+      end
+      display_labels_properties
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/docx_drawing.rb

@@ -0,0 +1,49 @@
+require_relative 'drawing_properties/docx_drawing_properties'
+require_relative 'graphic/docx_graphic'
+# Docx Drawing Data
+module OoxmlParser
+  class DocxDrawing
+    attr_accessor :type, :properties, :graphic
+
+    alias picture graphic
+
+    def initialize(properties = DocxDrawingProperties.new)
+      @properties = properties
+    end
+
+    def self.parse(drawing_node)
+      drawing = DocxDrawing.new
+      drawing_node.xpath('*').each do |drawing_node_child|
+        case drawing_node_child.name
+        when 'anchor'
+          drawing.type = :flow
+        when 'inline'
+          drawing.type = :inline
+        end
+        drawing.properties.distance_from_text = DocxDrawingDistanceFromText.parse(drawing_node_child)
+        drawing.properties.wrap = DocxWrapDrawing.parse(drawing_node_child)
+        drawing_node_child.attributes.each do |key, value|
+          case key
+          when 'relativeHeight'
+            drawing.properties.relative_height = value.value.to_i
+          end
+        end
+        drawing_node_child.xpath('*').each do |content_node_child|
+          case content_node_child.name
+          when 'simplePos'
+            drawing.properties.simple_position = OOXMLCoordinates.parse(content_node_child)
+          when 'extent'
+            drawing.properties.object_size = OOXMLCoordinates.parse(content_node_child, x_attr: 'cx', y_attr: 'cy')
+          when 'graphic'
+            drawing.graphic = DocxGraphic.parse(content_node_child)
+          when 'positionV'
+            drawing.properties.vertical_position = DocxDrawingPosition.parse(content_node_child)
+          when 'positionH'
+            drawing.properties.horizontal_position = DocxDrawingPosition.parse(content_node_child)
+          end
+        end
+      end
+      drawing
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/drawing_properties/docx_drawing_distance_from_text.rb

@@ -0,0 +1,23 @@
+# Docx Drawing Distance From Text
+module OoxmlParser
+  class DocxDrawingDistanceFromText
+    attr_accessor :top, :bottom, :left, :right
+
+    def self.parse(distance_node)
+      distance_from_text = DocxDrawingDistanceFromText.new
+      distance_node.attributes.each do |key, value|
+        case key
+        when 'distT'
+          distance_from_text.top = (value.value.to_f / 360_000.0).round(2)
+        when 'distB'
+          distance_from_text.bottom = (value.value.to_f / 360_000.0).round(2)
+        when 'distL'
+          distance_from_text.left = (value.value.to_f / 360_000.0).round(2)
+        when 'distR'
+          distance_from_text.right = (value.value.to_f / 360_000.0).round(2)
+        end
+      end
+      distance_from_text
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/drawing_properties/docx_drawing_position.rb

@@ -0,0 +1,35 @@
+# Docx Drawing Position
+module OoxmlParser
+  class DocxDrawingPosition
+    attr_accessor :relative_from, :offset, :align
+
+    def self.parse(drawing_position_node)
+      position = DocxDrawingPosition.new
+      position.relative_from = case drawing_position_node.attribute('relativeFrom').value
+                               when 'leftMargin'
+                                 :left_margin
+                               when 'rightMargin'
+                                 :right_margin
+                               when 'bottomMargin'
+                                 :bottom_margin
+                               when 'topMargin'
+                                 :top_margin
+                               when 'insideMargin'
+                                 :inside_margin
+                               when 'outsideMargin'
+                                 :outside_margin
+                               else
+                                 drawing_position_node.attribute('relativeFrom').value.to_sym
+                               end
+      drawing_position_node.xpath('*').each do |position_node_child|
+        case position_node_child.name
+        when 'posOffset', 'pctPosHOffset', 'pctPosVOffset'
+          position.offset = (position_node_child.text.to_f / 360_000.0).round(3)
+        when 'align'
+          position.align = position_node_child.text.to_sym
+        end
+      end
+      position
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/drawing_properties/docx_drawing_properties.rb

@@ -0,0 +1,11 @@
+require_relative 'ooxml_coordinates'
+require_relative 'docx_drawing_distance_from_text'
+require_relative 'docx_drawing_position'
+require_relative 'docx_wrap_drawing'
+# Docx Drawing Properties
+module OoxmlParser
+  class DocxDrawingProperties
+    attr_accessor :distance_from_text, :simple_position, :wrap, :object_size, :vertical_position, :horizontal_position,
+                  :relative_height
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/drawing_properties/docx_wrap_drawing.rb

@@ -0,0 +1,39 @@
+# Docx Wrap Drawing
+module OoxmlParser
+  class DocxWrapDrawing
+    attr_accessor :wrap_text, :distance_from_text
+
+    def initialize(wrap_text = :none)
+      @wrap_text = wrap_text
+    end
+
+    def self.parse(drawing_node)
+      wrap = DocxWrapDrawing.new
+      unless drawing_node.attribute('behindDoc').nil?
+        wrap.wrap_text = :behind if drawing_node.attribute('behindDoc').value == '1'
+        wrap.wrap_text = :infront if drawing_node.attribute('behindDoc').value == '0'
+      end
+      drawing_node.xpath('*').each do |wrap_node|
+        case wrap_node.name
+        when 'wrapSquare'
+          wrap.wrap_text = :square
+          wrap.distance_from_text = DocxDrawingDistanceFromText.parse(wrap_node)
+          break
+        when 'wrapTight'
+          wrap.wrap_text = :tight
+          wrap.distance_from_text = DocxDrawingDistanceFromText.parse(wrap_node)
+          break
+        when 'wrapThrough'
+          wrap.wrap_text = :through
+          wrap.distance_from_text = DocxDrawingDistanceFromText.parse(wrap_node)
+          break
+        when 'wrapTopAndBottom'
+          wrap.wrap_text = :topbottom
+          wrap.distance_from_text = DocxDrawingDistanceFromText.parse(wrap_node)
+          break
+        end
+      end
+      wrap
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/drawing_properties/ooxml_coordinates.rb

@@ -0,0 +1,32 @@
+# Docx Coordinates
+module OoxmlParser
+  class OOXMLCoordinates
+    attr_accessor :x, :y
+
+    def initialize(x, y)
+      @x = x
+      @y = y
+    end
+
+    def to_s
+      '(' + @x.to_s + '; ' + @y.to_s + ')'
+    end
+
+    # Compare two OOXMLCoordinates objects
+    # @param other [OOXMLCoordinates] other object
+    # @return [True, False] result of comparasion
+    def ==(other)
+      x == other.x && y == other.y
+    end
+
+    # Parse OOXMLCoordinates object
+    # @param position_node [Nokogiri::XML:Element] node to parse
+    # @param x_attr [String] name of x attribute
+    # @param y_attr [String] name of y attribute
+    # @param delimiter [Float] delimiter to devise values
+    # @return [OOXMLCoordinates] result of parsing
+    def self.parse(position_node, x_attr: 'x', y_attr: 'y', delimiter: OoxmlParser.configuration.units_delimiter)
+      OOXMLCoordinates.new((position_node.attribute(x_attr).value.to_f / delimiter).round(3), (position_node.attribute(y_attr).value.to_f / delimiter).round(3))
+    end
+  end
+end