ooxml_parser 0.2.0 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
-SHA1:
-  metadata.gz: 17ff610450bb961ad457f38aa206097940893f58
-  data.tar.gz: bcfe11a7a961f9a103eb8c047fb503e324d05bb3
+SHA256:
+  metadata.gz: 071bdcbc2fb31371918e97a11e168bcdd849a3a606d880a6f8f29f03d3fc54c9
+  data.tar.gz: 48631f0dcb24d64688065bbb226d8e0c3dd13f18473a680aa33fcf284751b195
 SHA512:
-  metadata.gz: 03f3590091b03d708883a052300a754e9e94b64c55fbc85564514d3194d8bf3e9e1927069eaa30ee2aea0a095e77a089c0a899029eceb5f60394064869e6c59c
-  data.tar.gz: 4c7eb4fa0cf09ced780de674a48acd4939b29b01bb09b0e240627cc1654a1f90a945c738dba76d0e59db3c88ee859028f3dda38b4990c8165bc9adf9b200ef67
+  metadata.gz: 175a3bb85a4c2b85cf577c498c50c021c180c9054440451a55b81d93752eb6211908e2309738dc8486ab0adfdd0cdcd3f5afbf75910226141f179ae878c62e6a
+  data.tar.gz: 031f7cf2aa5d53223df37d425fa0050a5ff875d3646ea5e82c56064262927ac728d5ed6563c557a9fc74d8da6f3a1b34ce230ed41b536914450ee7efe2a58831

data/README.md

@@ -4,7 +4,23 @@ _ooxml_parser_ is a Ooxml files (docx, xlsx, pptx) parser written in Ruby.
 
 ## Installation
 
-    $ gem install ooxml_parser
+1. Install `magic` lib, required by `ruby-filemagic`  
+   Mac OS: 
+   ```
+   brew install libmagic
+   ```  
+   Debian-Based Linux: 
+   ```
+   sudo apt-get install libmagic-dev
+   ```
+   Centos-Based Linux: 
+   ```
+   sudo yum install file-devel
+   ```  
+2. Install gem by command: 
+    ```
+    gem install ooxml_parser
+    ```   
     
 ## Usage
 

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

@@ -2,6 +2,8 @@ module OoxmlParser
   # Chart Style
   class ChartStyle < OOXMLDocumentObject
     attr_accessor :style_number
+    # @return [OleObject] ole object
+    attr_accessor :ole_object
 
     # Parse ChartStyle
     # @param [Nokogiri::XML:Node] node with Relationships
@@ -11,6 +13,8 @@ module OoxmlParser
         case node_child.name
         when 'style'
           @style_number = node_child.attribute('val').value.to_i
+        when 'oleObject'
+          @ole_object = OleObject.new(parent: self).parse(node_child)
         end
       end
       self

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

@@ -37,6 +37,7 @@ module OoxmlParser
           else
             val = chart_props_node_child.xpath('c:val')[0]
           end
+          next unless val
           next if val.xpath('c:numRef').empty?
           @data << ChartCellsRange.new(parent: self).parse(val.xpath('c:numRef').first).dup
         when 'dLbls'

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

@@ -1,8 +1,13 @@
+require_relative 'chart_axis/scaling'
 require_relative 'chart_axis_title'
 module OoxmlParser
   # Parsing Chart axis tags 'catAx', 'valAx'
   class ChartAxis < OOXMLDocumentObject
     attr_accessor :title, :display, :position, :major_grid_lines, :minor_grid_lines
+    # @return [Scaling] scaling attribute
+    attr_reader :scaling
+    # @return [ValuedChild] the position of the tick labels
+    attr_reader :tick_label_position
 
     def initialize(title = ChartAxisTitle.new,
                    display = true,
@@ -30,6 +35,10 @@ module OoxmlParser
           @major_grid_lines = true
         when 'minorGridlines'
           @minor_grid_lines = true
+        when 'scaling'
+          @scaling = Scaling.new(parent: self).parse(node_child)
+        when 'tickLblPos'
+          @tick_label_position = ValuedChild.new(:symbol, parent: self).parse(node_child)
         when 'axPos'
           @position = value_to_symbol(node_child.attribute('val'))
         end

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

@@ -0,0 +1,20 @@
+module OoxmlParser
+  # Parsing Scaling tag 'scaling'
+  class Scaling < OOXMLDocumentObject
+    # @return [ValuedChild] orientation value
+    attr_reader :orientation
+
+    # Parse Scaling object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Scaling] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'orientation'
+          @orientation = ValuedChild.new(:symbol, parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -63,7 +63,7 @@ module OoxmlParser
       when :percent
         OoxmlSize.new(@value * 100_000, :one_100000th_percent)
       when :pct
-        OoxmlSize.new(@value * 1_000, :one_100000th_percent)
+        OoxmlSize.new(@value * 100_000 / 50, :one_100000th_percent)
       when :one_1000th_percent
         OoxmlSize.new(@value * 100, :one_100000th_percent)
       else

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/graphic/picture/docx_blip/file_reference.rb

@@ -15,9 +15,10 @@ module OoxmlParser
           @resource_id = value.value
         end
       end
-      return self if @resource_id.nil?
+      return self unless @resource_id
+      return self if @resource_id.empty?
       @path = OOXMLDocumentObject.get_link_from_rels(@resource_id)
-      if @path.empty?
+      if !@path || @path.empty?
         warn "Cant find path to media file by id: #{@resource_id}"
         return self
       end

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/graphic/shape/shape_body_properties/ooxml_shape_body_properties.rb

@@ -6,6 +6,10 @@ module OoxmlParser
     attr_accessor :margins, :anchor, :wrap, :preset_text_warp
     # @return [Symbol] Vertical Text
     attr_accessor :vertical
+    # @return [Symbol] Number of Columns
+    attr_reader :number_columns
+    # @return [Symbol] Spacing between columns
+    attr_reader :space_columns
 
     alias vertical_align anchor
 
@@ -25,6 +29,10 @@ module OoxmlParser
           @anchor = value_to_symbol(value)
         when 'vert'
           @vertical = value_to_symbol(value)
+        when 'numCol'
+          @number_columns = value.value.to_i
+        when 'spcCol'
+          @space_columns = OoxmlSize.new(value.value.to_f, :emu)
         end
       end
       node.xpath('*').each do |node_child|

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/graphic/shape/shape_properties/docx_shape_properties/text_box.rb

@@ -13,7 +13,7 @@ module OoxmlParser
           when 'p'
             DocumentStructure.default_paragraph_style = DocxParagraph.new
             DocumentStructure.default_paragraph_style.spacing = Spacing.new(0, 0.35, 1.15, :multiple)
-            elements << DocumentStructure.default_paragraph_style.copy.parse(textbox_element, i, parent: parent)
+            elements << DocumentStructure.default_paragraph_style.dup.parse(textbox_element, i, parent: parent)
           when 'tbl'
             elements << Table.new(parent: parent).parse(textbox_element, i)
           end

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

@@ -1,7 +1,7 @@
 # Border Properties Data
 module OoxmlParser
   class BordersProperties < OOXMLDocumentObject
-    attr_accessor :color, :space, :sz, :val, :space, :shadow, :frame, :side
+    attr_accessor :color, :space, :sz, :val, :shadow, :frame, :side
 
     def initialize(color = :auto, sz = 0, val = :none, space = 0, parent: nil)
       @color = color
@@ -28,7 +28,7 @@ module OoxmlParser
 
     def visible?
       return false if nil?
-      !(val == 'none')
+      val != 'none'
     end
 
     # Parse BordersProperties

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

@@ -1,4 +1,5 @@
 require_relative 'color/color_helper'
+require_relative 'color/ooxml_color'
 require_relative 'colors/color_alpha_channel'
 require_relative 'colors/hsl_color'
 require_relative 'colors/scheme_color'
@@ -13,7 +14,7 @@ module OoxmlParser
     # List of color duplicated from `OpenXML Sdk IndexedColors` class
     # See https://msdn.microsoft.com/en-us/library/documentformat.openxml.spreadsheet.indexedcolors.aspx
     COLOR_INDEXED =
-      %w(
+      %w[
         000000
         FFFFFF
         FF0000
@@ -80,7 +81,7 @@ module OoxmlParser
         333333
         n/a
         n/a
-      ).freeze
+      ].freeze
 
     # @return [Integer] Value of Red Part
     attr_accessor :red
@@ -133,7 +134,12 @@ module OoxmlParser
     alias to_int16 to_hex
 
     def none?
-      (@red == VALUE_FOR_NONE_COLOR) && (@green == VALUE_FOR_NONE_COLOR) && (@blue == VALUE_FOR_NONE_COLOR)
+      (@red == VALUE_FOR_NONE_COLOR) && (@green == VALUE_FOR_NONE_COLOR) && (@blue == VALUE_FOR_NONE_COLOR) ||
+        (style == :nil)
+    end
+
+    def any?
+      !none?
     end
 
     def white?
@@ -171,7 +177,8 @@ module OoxmlParser
       color_to_check = Color.parse(color_to_check.to_s) if color_to_check.is_a?(Symbol)
       color_to_check = Color.parse(color_to_check.value) if color_to_check.is_a?(DocxColor)
       return true if none? && color_to_check.nil?
-      return false if none? && !color_to_check.none?
+      return true if none? && color_to_check.none?
+      return false if none? && color_to_check.any?
       return false if !none? && color_to_check.none?
       return true if self == color_to_check
       red = color_to_check.red
@@ -253,22 +260,6 @@ module OoxmlParser
         end
       end
 
-      def parse_color_tag(node)
-        return if node.nil?
-        node.attributes.each do |key, value|
-          case key
-          when 'val'
-            return Color.new.parse_hex_string(value.value.to_s)
-          when 'theme'
-            return ThemeColors.parse_color_theme(node)
-          when 'rgb'
-            return Color.new.parse_hex_string(value.value.to_s)
-          when 'indexed'
-            return Color.get_rgb_by_color_index(value.value.to_i)
-          end
-        end
-      end
-
       def parse_scheme_color(scheme_color_node)
         color = ThemeColors.list[scheme_color_node.attribute('val').value.to_sym]
         hls = HSLColor.rgb_to_hsl(color)

data/lib/ooxml_parser/common_parser/common_data/color/color_helper.rb

@@ -3,7 +3,7 @@ module ColorHelper
   # Parse string in hex
   # @param [String] hex_string with or without alpha-channel
   def parse_hex_string(hex_string)
-    return self if hex_string == 'auto' || hex_string == 'null'
+    return self if %w[auto null].include?(hex_string)
 
     char_array = hex_string.split(//)
     if char_array.length == 3

data/lib/ooxml_parser/common_parser/common_data/color/ooxml_color.rb

@@ -0,0 +1,57 @@
+module OoxmlParser
+  # Class for parsing `color` tag
+  class OoxmlColor < OOXMLDocumentObject
+    # @return [Color] value of color
+    attr_reader :value
+    # @return [Symbol] theme color name
+    attr_reader :theme_color
+    # @return [Symbol] theme shade
+    attr_reader :theme_shade
+    # @return [Integer] theme index
+    attr_reader :theme
+    # @return [Float] tint
+    attr_reader :tint
+    # @return [Float] Indexed id
+    attr_reader :indexed
+    # @return [Color] rgb color
+    attr_reader :rgb
+
+    def to_color
+      return Color.get_rgb_by_color_index(indexed) if indexed
+      return ThemeColors.parse_color_theme(theme, tint) if theme
+      return rgb if rgb
+      value
+    end
+
+    def ==(other)
+      return to_color == other if other.is_a?(Color)
+      return to_color == other if other.is_a?(Symbol)
+      super
+    end
+
+    # Parse Hyperlink object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Hyperlink] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'val'
+          @value = Color.new.parse_hex_string(value.value.to_s)
+        when 'themeColor'
+          @theme_color = value.value.to_sym
+        when 'themeShade'
+          @theme_shade = Integer("0x#{value.value}")
+        when 'theme'
+          @theme = value.value.to_i
+        when 'tint'
+          @tint = value.value.to_f
+        when 'indexed'
+          @indexed = value.value.to_i
+        when 'rgb'
+          @rgb = Color.new.parse_hex_string(value.value)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/colors/theme_colors.rb

@@ -5,16 +5,13 @@ module OoxmlParser
       # @return [Hash] list of colors
       attr_accessor :list
 
-      def parse_color_theme(color_theme_node)
+      def parse_color_theme(theme, tint)
         themes_array = ThemeColors.list.to_a
         # TODO: if no swap performed - incorrect color parsing. But don't know why it needed
         themes_array[0], themes_array[1] = themes_array[1], themes_array[0]
         themes_array[2], themes_array[3] = themes_array[3], themes_array[2]
-        hls = HSLColor.rgb_to_hsl(themes_array[color_theme_node.attribute('theme').value.to_i][1])
-        tint = 0
-        unless color_theme_node.attribute('tint').nil?
-          tint = color_theme_node.attribute('tint').value.to_f
-        end
+        hls = HSLColor.rgb_to_hsl(themes_array[theme][1])
+        tint = 0 if tint.nil?
         hls.calculate_rgb_with_tint(tint)
       end
     end

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

@@ -27,10 +27,14 @@ module OoxmlParser
       end
 
       def parser_coordinates_range(arguments_string)
+        return parse_coordinates_array(arguments_string) if arguments_string.include?(',')
+        return warn "Formulas with # is unsupported: #{arguments_string}" if arguments_string.include?('#')
+        return warn 'Formulas consists from `!` only' if arguments_string == '!'
         sheet_name = 'Sheet1'
 
         if arguments_string.include?('!')
-          sheet_name, arguments_string = arguments_string.split('!')
+          sheet_name = arguments_string.match(/.*!/).to_s
+          arguments_string = arguments_string.sub(sheet_name, '')
         end
 
         range = arguments_string.split(':')
@@ -70,6 +74,18 @@ module OoxmlParser
         arguments_array
       end
 
+      # Parse array of coordinates
+      # @param arguments_string [String] string
+      # @return [Array] result
+      def parse_coordinates_array(arguments_string)
+        result = []
+        coord_array = arguments_string.split(',')
+        coord_array.each do |current_coord|
+          result << parser_coordinates_range(current_coord)
+        end
+        result
+      end
+
       # This method check is argument contains coordinate
       # @param [String] string
       def coordinates?(string)

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

@@ -1,9 +1,11 @@
 module OoxmlParser
   # Class for parsing `hlinkClick`, `hyperlink` tags
   class Hyperlink < OOXMLDocumentObject
-    attr_accessor :url, :tooltip, :coordinates, :id, :highlight_click, :action
+    attr_accessor :url, :tooltip, :coordinates, :highlight_click, :action
     attr_accessor :action_link
     attr_accessor :id
+    # @return [Array<ParagraphRun>] run of paragraph
+    attr_reader :runs
 
     def initialize(link = nil,
                    tooltip = nil,
@@ -13,6 +15,7 @@ module OoxmlParser
       @tooltip = tooltip
       @coordinates = coordinates
       @parent = parent
+      @runs = []
     end
 
     alias link url
@@ -41,6 +44,14 @@ module OoxmlParser
           @highlight_click = attribute_enabled?(value)
         end
       end
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'r'
+          @runs << ParagraphRun.new(parent: self).parse(node_child)
+        end
+      end
+
       case @action_link
       when 'ppaction://hlinkshowjump?jump=previousslide'
         @action = :previous_slide

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

@@ -17,9 +17,7 @@ module OoxmlParser
     def ==(other)
       instance_variables.each do |current_attribute|
         next if current_attribute == :@parent
-        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
@@ -42,7 +40,9 @@ module OoxmlParser
         # Support of Encrtypted status in `file` util was introduced in file v5.20
         # but LTS version of ubuntu before 16.04 uses older `file` and it return `Composite Document`
         # https://github.com/file/file/blob/master/ChangeLog#L217
-        if file_result.include?('encrypted') || file_result.include?('Composite Document File V2 Document, No summary info')
+        if file_result.include?('encrypted') ||
+           file_result.include?('Composite Document File V2 Document, No summary info') ||
+           file_result.include?('application/CDFV2-corrupt')
           warn("File #{path_to_file} is encrypted. Can't parse it")
           return true
         end

data/lib/ooxml_parser/common_parser/common_data/ooxml_document_object/ooxml_document_object_helper.rb

@@ -1,6 +1,28 @@
-# frozen_string_literal: true
 # Module for helper methods for OOXMLDocumentObject
 module OoxmlDocumentObjectHelper
+  # Convert object to hash
+  # @return [Hash]
+  def to_hash
+    result_hash = {}
+    instance_variables.each do |current_attribute|
+      next if current_attribute == :@parent
+      attribute_value = instance_variable_get(current_attribute)
+      next unless attribute_value
+      if attribute_value.is_a?(Array)
+        attribute_value.each_with_index do |object_element, index|
+          result_hash["#{current_attribute}_#{index}".to_sym] = object_element.to_hash
+        end
+      else
+        result_hash[current_attribute.to_sym] = if attribute_value.respond_to?(:to_hash)
+                                                  attribute_value.to_hash
+                                                else
+                                                  attribute_value.to_s
+                                                end
+      end
+    end
+    result_hash
+  end
+
   private
 
   VALUE_TO_SYMBOL_HASH = { l: :left,
@@ -13,6 +35,7 @@ module OoxmlDocumentObjectHelper
                            tl: :top_left,
                            br: :bottom_right,
                            bl: :bottom_left,
+                           contentLocked: :content_locked,
                            dist: :distributed,
                            tb: :horizontal,
                            rl: :rotate_on_90,
@@ -44,6 +67,8 @@ module OoxmlDocumentObjectHelper
                            mediumDashDotDot: :medium_dash_dot_dot,
                            lgDashDot: :large_dash_dot,
                            lgDashDotDot: :large_dash_dot_dot,
+                           sdtLocked: :sdt_locked,
+                           sdtContentLocked: :sdt_content_locked,
                            sysDash: :system_dash,
                            sysDot: :system_dot,
                            sysDashDot: :system_dash_dot,
@@ -58,7 +83,11 @@ module OoxmlDocumentObjectHelper
                            tbRl: :top_to_bottom_right_to_left,
                            btLr: :bottom_to_top_left_to_right,
                            singlelevel: :single_level,
+                           maxMin: :max_min,
+                           middleDot: :middle_dot,
+                           minMax: :min_max,
                            multilevel: :multi_level,
+                           nextTo: :next_to,
                            hybridMultilevel: :hybrid_multi_level,
                            rnd: :round,
                            sq: :square }.freeze