ooxml_parser 0.4.0 → 0.7.0

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide/slide/timing/time_node/common_timing.rb

@@ -1,5 +1,8 @@
-require_relative 'common_timing/condition'
+# frozen_string_literal: true
+
+require_relative 'common_timing/condition_list'
 module OoxmlParser
+  # Class for data of CommonTiming
   class CommonTiming
     attr_accessor :id, :duration, :restart, :children, :start_conditions, :end_conditions
 
@@ -28,9 +31,9 @@ module OoxmlParser
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'stCondLst'
-          @start_conditions = Condition.parse_list(node_child)
+          @start_conditions = ConditionList.new(parent: self).parse(node_child)
         when 'endCondLst'
-          @end_conditions = Condition.parse_list(node_child)
+          @end_conditions = ConditionList.new(parent: self).parse(node_child)
         when 'childTnLst'
           @children = TimeNode.parse_list(node_child)
         end

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide/slide/timing/time_node/common_timing/{condition.rb → codition_list/condition.rb}

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `cond` tags
   class Condition < OOXMLDocumentObject
@@ -12,18 +14,10 @@ module OoxmlParser
         when 'evt'
           @event = value.value
         when 'delay'
-          @delay = value.value
+          @delay = value_to_symbol(value)
         end
       end
       self
     end
-
-    def self.parse_list(conditions_list_node)
-      conditions = []
-      conditions_list_node.xpath('p:cond').each do |condition_node|
-        conditions << Condition.new(parent: conditions).parse(condition_node)
-      end
-      conditions
-    end
   end
 end

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide/slide/timing/time_node/common_timing/condition_list.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+require_relative 'codition_list/condition'
+module OoxmlParser
+  # Class for parsing `stCondLst` tags
+  class ConditionList < OOXMLDocumentObject
+    attr_reader :list
+
+    def initialize(parent: nil)
+      @list = []
+      @parent = parent
+    end
+
+    # Parse ConditionList object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [ConditionList] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'cond'
+          @list << Condition.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide/slide_helper.rb

@@ -1,10 +1,14 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Methods to help working with slide data
   module SlideHelper
+    # @return [Array] list of not empty element on slide
     def nonempty_elements
       elements.reject { |cur_shape| cur_shape.text_body.paragraphs.first.characters.empty? }
     end
 
+    # @return [Array<GraphicFrame>] list GraphicFrame elements on slide
     def graphic_frames
       elements.select { |cur_element| cur_element.is_a?(GraphicFrame) }
     end
@@ -15,7 +19,7 @@ module OoxmlParser
     def transform_of_object(object)
       case object
       when :image
-        elements.find { |e| e.is_a? Picture }.properties.transform
+        elements.find { |e| e.is_a? DocxPicture }.properties.transform
       when :chart
         elements.find { |e| e.is_a? GraphicFrame }.transform
       when :table
@@ -27,22 +31,36 @@ module OoxmlParser
       end
     end
 
+    # Get horizontal align of object on slide
+    # @param object [Symbol] object to get
+    # @param slide_size [SlideSize] size of slide
+    # @return [Symbol] type of align
     def content_horizontal_align(object, slide_size)
       transform = transform_of_object(object)
       return :left if transform.offset.x.zero?
       return :center if OoxmlSize.new((slide_size.width.value / 2) - (transform.extents.x.value / 2)) == OoxmlSize.new(transform.offset.x.value)
       return :right if OoxmlSize.new((slide_size.width.value - transform.extents.x.value)) == OoxmlSize.new(transform.offset.x.value)
+
       :unknown
     end
 
+    # Get vertical align of object on slide
+    # @param object [Symbol] object to get
+    # @param slide_size [SlideSize] size of slide
+    # @return [Symbol] type of align
     def content_vertical_align(object, slide_size)
       transform = transform_of_object(object)
       return :top if transform.offset.y.zero?
       return :middle if OoxmlSize.new((slide_size.height.value / 2) - (transform.extents.y.value / 2)) == OoxmlSize.new(transform.offset.y.value)
       return :bottom if OoxmlSize.new(slide_size.height.value - transform.extents.y.value) == OoxmlSize.new(transform.offset.y.value)
+
       :unknown
     end
 
+    # Get content distribution of object
+    # @param object [Symbol] object to get
+    # @param slide_size [SlideSize] size of slide
+    # @return [Array<Symbol>] type of align
     def content_distribute(object, slide_size)
       return %i[horizontally vertically] if content_horizontal_align(object, slide_size) == :center && content_vertical_align(object, slide_size) == :middle
       return [:horizontally] if content_horizontal_align(object, slide_size) == :center

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide/transition/transition.rb

@@ -1,9 +1,15 @@
+# frozen_string_literal: true
+
 require_relative 'transition/sound_action'
 require_relative 'transition_properties/transition_properties'
 module OoxmlParser
+  # Class for data of Transition
   class Transition < OOXMLDocumentObject
     attr_accessor :speed, :properties, :sound_action, :advance_on_click, :delay, :duration
 
+    # Parse Transition object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Transition] result of parsing
     def parse(node)
       node.xpath('*').each do |node_child|
         @properties = TransitionProperties.new(parent: self).parse(node_child)

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide/transition/transition/sound_action.rb

@@ -1,19 +1,19 @@
+# frozen_string_literal: true
+
 require_relative 'sound_action/sound'
 module OoxmlParser
   # Class for parsing SoundAction
   class SoundAction < OOXMLDocumentObject
     attr_accessor :start_sound, :end_sound
 
-    # Parse HyperlinkForHover
+    # Parse SoundAction
     # @param [Nokogiri::XML:Node] node with NumberingProperties
-    # @return [HyperlinkForHover] result of parsing
+    # @return [SoundAction] result of parsing
     def parse(node)
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'stSnd'
-          @start_sound = Sound.new(parent: self).parse(node_child.xpath('p:snd').first) unless node_child.xpath('p:snd').first.nil?
-        when 'endSnd'
-          @end_sound = Sound.new(parent: self).parse(node_child.xpath('p:snd').first) unless node_child.xpath('p:snd').first.nil?
+          @start_sound = Sound.new(parent: self).parse(node_child.xpath('p:snd').first) if node_child.xpath('p:snd').first
         end
       end
       self

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide/transition/transition/sound_action/sound.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `snd` tags
   class Sound < OOXMLDocumentObject

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide/transition/transition_properties/transition_properties.rb

@@ -1,6 +1,12 @@
+# frozen_string_literal: true
+
+require_relative 'transition_properties/wheel'
 module OoxmlParser
+  # Class for data for TransitionProperties
   class TransitionProperties < OOXMLDocumentObject
     attr_accessor :type, :through_black, :direction, :orientation, :spokes
+    # @return [Wheel] Wheel Slide Transition
+    attr_reader :wheel
 
     # Parse TransitionProperties object
     # @param node [Nokogiri::XML:Element] node to parse
@@ -15,8 +21,8 @@ module OoxmlParser
       when 'split'
         @direction = value_to_symbol(node.attribute('dir')) if node.attribute('dir')
         @orientation = node.attribute('orient').value.to_sym if node.attribute('orient')
-      when 'wheel', 'wheelReverse'
-        @spokes = option_enabled?(node, 'spokes')
+      when 'wheel'
+        @wheel = Wheel.new(parent: self).parse(node)
       end
       self
     end

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide/transition/transition_properties/transition_properties/wheel.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for Wheel Slide Transition `p:wheel` tag
+  class Wheel < OOXMLDocumentObject
+    # @return [Integer] count of spokes
+    attr_reader :spokes
+
+    # Parse Wheel object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Wheel] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'spokes'
+          @spokes = value.value.to_i
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide_size.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `sldSz` tag
   class SlideSize < OOXMLDocumentObject

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/table_styles.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `tableStyles.xml` file
+  class TableStyles < OOXMLDocumentObject
+    # @return [Array<TableStyle>] list of table styles
+    attr_reader :table_style_list
+
+    def initialize(parent: nil)
+      @table_style_list = []
+      @parent = parent
+    end
+
+    # Parse TableStyles object
+    # @param file [Nokogiri::XML:Element] node to parse
+    # @return [TableStyles] result of parsing
+    def parse(file = "#{OOXMLDocumentObject.path_to_folder}/#{OOXMLDocumentObject.root_subfolder}/tableStyles.xml")
+      return nil unless File.exist?(file)
+
+      document = parse_xml(file)
+      node = document.xpath('*').first
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'tblStyle'
+          @table_style_list << TableStyle.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+
+    # @param id [String] style to find
+    # @return [TableStyle] style by this id
+    def style_by_id(id)
+      table_style_list.detect { |style| style.id == id }
+    end
+  end
+end

data/lib/ooxml_parser/pptx_parser/pptx_parser.rb

@@ -1,10 +1,16 @@
+# frozen_string_literal: true
+
 require_relative 'pptx_data/presentation.rb'
 
 module OoxmlParser
+  # Basic class for parsing pptx
   class PptxParser
+    # Parse pptx file
+    # @param path_to_file [String] file path
+    # @return [Presentation] result of parse
     def self.parse_pptx(path_to_file)
       Parser.parse_format(path_to_file) do
-        Presentation.parse
+        Presentation.new.parse
       end
     end
   end

data/lib/ooxml_parser/version.rb

@@ -1,6 +1,9 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # This module holds the RuboCop version information.
   module Version
-    STRING = '0.4.0'.freeze
+    # [String] Version of Gem
+    STRING = '0.7.0'
   end
 end

data/lib/ooxml_parser/xlsx_parser/xlsx_data/view_model/workbook.rb

@@ -1,9 +1,13 @@
+# frozen_string_literal: true
+
 require_relative 'workbook/chartsheet'
+require_relative 'workbook/pivot_cache'
+require_relative 'workbook/shared_string_table'
 require_relative 'workbook/style_sheet'
 require_relative 'workbook/worksheet'
 require_relative 'workbook/workbook_helpers'
-# Class for storing XLSX Workbook
 module OoxmlParser
+  # Class for storing XLSX Workbook
   class XLSXWorkbook < CommonDocumentStructure
     include WorkbookHelpers
     attr_accessor :worksheets
@@ -13,17 +17,29 @@ module OoxmlParser
     attr_accessor :relationships
     # @return [StyleSheet] styles of book
     attr_accessor :style_sheet
+    # @return [SharedStringTable] styles of book
+    attr_accessor :shared_strings_table
+    # @return [Array<PivotCache>] list of pivot caches
+    attr_accessor :pivot_caches
 
-    def initialize(worksheets = [])
-      @worksheets = worksheets
+    def initialize(params = {})
+      @worksheets = []
+      @pivot_caches = []
+      super
     end
 
+    # Return cell by coordinates
+    # @param column [String, Integer] number or numeric digit of column
+    # @param row [Integer] row to find
+    # @param sheet [Integer] number of sheet
+    # @return [XlsxCell] result
     def cell(column, row, sheet = 0)
-      column = Coordinates.new(row, column).get_column_number unless StringHelper.numeric?(column.to_s)
+      column = Coordinates.new(row, column).column_number unless StringHelper.numeric?(column.to_s)
 
       if StringHelper.numeric?(sheet.to_s)
         row = @worksheets[sheet].rows[row.to_i - 1]
         return nil if row.nil?
+
         return row.cells[column.to_i - 1]
       elsif sheet.is_a?(String)
         @worksheets.each do |worksheet|
@@ -36,10 +52,6 @@ module OoxmlParser
       raise "Error. Wrong sheet value: #{sheet}"
     end
 
-    def difference(other)
-      Hash.object_to_hash(self).diff(Hash.object_to_hash(other))
-    end
-
     # Get all values of formulas.
     # @param [Fixnum] precision of formulas counting
     # @return [Array, String] all formulas
@@ -47,11 +59,15 @@ module OoxmlParser
       formulas = []
       worksheets.each do |c_sheet|
         next unless c_sheet
+
         c_sheet.rows.each do |c_row|
           next unless c_row
+
           c_row.cells.each do |c_cell|
             next unless c_cell
             next unless c_cell.formula
+            next if c_cell.formula.empty?
+
             text = c_cell.raw_text
             if StringHelper.numeric?(text)
               text = text.to_f.round(10).to_s[0..precision]
@@ -72,48 +88,51 @@ module OoxmlParser
       formulas
     end
 
-    def self.parse
-      workbook = XLSXWorkbook.new
-      workbook.relationships = Relationships.parse_rels("#{OOXMLDocumentObject.path_to_folder}xl/_rels/workbook.xml.rels")
+    # Do work for parsing shared strings file
+    def parse_shared_strings
+      shared_strings_target = relationships.target_by_type('sharedString')
+      return if shared_strings_target.empty?
+
+      shared_string_file = "#{OOXMLDocumentObject.path_to_folder}/xl/#{shared_strings_target.first}"
+      @shared_strings_table = SharedStringTable.new(parent: self).parse(shared_string_file)
+    end
+
+    # Parse content of Workbook
+    # @return [XLSXWorkbook]
+    def parse
+      @content_types = ContentTypes.new(parent: self).parse
+      @relationships = Relationships.new(parent: self).parse_file("#{OOXMLDocumentObject.path_to_folder}xl/_rels/workbook.xml.rels")
+      parse_shared_strings
       OOXMLDocumentObject.xmls_stack = []
       OOXMLDocumentObject.root_subfolder = 'xl/'
-      self.shared_strings = nil
       OOXMLDocumentObject.add_to_xmls_stack('xl/workbook.xml')
-      doc = Nokogiri::XML.parse(File.open(OOXMLDocumentObject.current_xml))
-      XLSXWorkbook.styles_node = Nokogiri::XML(File.open("#{OOXMLDocumentObject.path_to_folder}/#{OOXMLDocumentObject.root_subfolder}/styles.xml"))
-      workbook.theme = PresentationTheme.parse("xl/#{link_to_theme_xml}") if link_to_theme_xml
-      workbook.style_sheet = StyleSheet.new(parent: self).parse
-      doc.xpath('xmlns:workbook/xmlns:sheets/xmlns:sheet').each do |sheet|
-        file = workbook.relationships.target_by_id(sheet.attribute('id').value)
+      @doc = Nokogiri::XML.parse(File.open(OOXMLDocumentObject.current_xml))
+      @theme = PresentationTheme.parse("xl/#{link_to_theme_xml}") if link_to_theme_xml
+      @style_sheet = StyleSheet.new(parent: self).parse
+      @doc.xpath('xmlns:workbook/xmlns:sheets/xmlns:sheet').each do |sheet|
+        file = @relationships.target_by_id(sheet.attribute('id').value)
         if file.start_with?('worksheets')
-          workbook.worksheets << Worksheet.parse(file, parent: workbook)
-          workbook.worksheets.last.name = sheet.attribute('name').value
+          @worksheets << Worksheet.new(parent: self).parse(file)
+          @worksheets.last.name = sheet.attribute('name').value
         elsif file.start_with?('chartsheets')
-          workbook.worksheets << Chartsheet.new(parent: workbook).parse(file)
+          @worksheets << Chartsheet.new(parent: self).parse(file)
         end
       end
+      parse_pivot_cache
       OOXMLDocumentObject.xmls_stack.pop
-      workbook
+      self
     end
 
-    class << self
-      # @return [Array, Nokogiri::XML::Eelement] list of shared strings
-      attr_writer :shared_strings
-      attr_accessor :styles_node
+    private
 
-      # Accessor for shared string. Initialization for this array
-      def shared_strings
-        if @shared_strings.nil?
-          shared_strings_file = "#{OOXMLDocumentObject.path_to_folder}/#{OOXMLDocumentObject.root_subfolder}/sharedStrings.xml"
-          @shared_strings = Nokogiri::XML(File.open(shared_strings_file)).xpath('//xmlns:si') if File.exist?(shared_strings_file)
-        end
-        @shared_strings
-      end
+    def link_to_theme_xml
+      relationships.target_by_type('theme').first
+    end
 
-      def link_to_theme_xml
-        file = File.open(OOXMLDocumentObject.path_to_folder + 'xl/_rels/workbook.xml.rels')
-        relationships = Relationships.parse_rels(file)
-        relationships.target_by_type('theme')
+    # Perform parsing of pivot cache
+    def parse_pivot_cache
+      @doc.xpath('xmlns:workbook/xmlns:pivotCaches/xmlns:pivotCache').each do |pivot_cache|
+        @pivot_caches << PivotCache.new(parent: self).parse(pivot_cache)
       end
     end
   end