ooxml_parser 0.4.1 → 0.5.0

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 +1,13 @@
+# frozen_string_literal: true
+
 require_relative 'pptx_data/presentation.rb'
 
 module OoxmlParser
+  # Basic class for parsing pptx
   class PptxParser
     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 +1,8 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # This module holds the RuboCop version information.
   module Version
-    STRING = '0.4.1'.freeze
+    STRING = '0.5.0'
   end
 end

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

@@ -1,9 +1,12 @@
+# frozen_string_literal: true
+
 require_relative 'workbook/chartsheet'
+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 +16,21 @@ module OoxmlParser
     attr_accessor :relationships
     # @return [StyleSheet] styles of book
     attr_accessor :style_sheet
+    # @return [SharedStringTable] styles of book
+    attr_accessor :shared_strings_table
 
-    def initialize(worksheets = [])
-      @worksheets = worksheets
+    def initialize(params = {})
+      @worksheets = []
+      super
     end
 
     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 +43,6 @@ 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 +50,15 @@ 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]
@@ -72,49 +79,42 @@ 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
+
+    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
+      @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 = workbook.relationships.target_by_id(sheet.attribute('id').value)
+        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
       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
-        file = File.open(OOXMLDocumentObject.path_to_folder + 'xl/_rels/workbook.xml.rels')
-        relationships = Relationships.parse_rels(file)
-        relationships.target_by_type('theme')
-      end
+    def link_to_theme_xml
+      relationships.target_by_type('theme').first
     end
   end
 end

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for storing data of chartsheets files
   class Chartsheet < OOXMLDocumentObject
@@ -11,7 +13,7 @@ module OoxmlParser
 
     def parse(file)
       OOXMLDocumentObject.add_to_xmls_stack(OOXMLDocumentObject.root_subfolder + file)
-      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.current_xml))
+      doc = parse_xml(OOXMLDocumentObject.current_xml)
       node = doc.xpath('//xmlns:chartsheet').first
       node.xpath('*').each do |node_child|
         case node_child.name

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

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+require_relative 'shared_string_table/string_index'
+module OoxmlParser
+  # Class for parsing shared string table
+  class SharedStringTable < OOXMLDocumentObject
+    # @return [Integer] count of shared strings
+    attr_reader :count
+    # @return [Integer] unique count of shared strings
+    attr_reader :unique_count
+    # @return [Array<StringIndex>] String Index
+    attr_reader :string_indexes
+
+    def initialize(parent: nil)
+      @string_indexes = []
+      @parent = parent
+    end
+
+    # Parse Shared string table file
+    # @param file [String] path to file
+    # @return [SharedStringTable]
+    def parse(file)
+      return nil unless File.exist?(file)
+
+      document = parse_xml(file)
+      node = document.xpath('*').first
+
+      node.attributes.each do |key, value|
+        case key
+        when 'count'
+          @count = value.value.to_i
+        when 'uniqueCount'
+          @unique_count = value.value.to_i
+        end
+      end
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'si'
+          @string_indexes << StringIndex.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing string index `si` tag
+  class StringIndex < OOXMLDocumentObject
+    # @return [ParagraphRun] run of text
+    attr_reader :run
+    # @return [String] text
+    attr_reader :text
+
+    # Parse StringIndex data
+    # @param [Nokogiri::XML:Element] node with StringIndex data
+    # @return [StringIndex] value of StringIndex data
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 't'
+          @text = node_child.text
+        when 'r'
+          @run = ParagraphRun.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -1,6 +1,10 @@
+# frozen_string_literal: true
+
+require_relative 'style_sheet/cell_xfs'
 require_relative 'style_sheet/fills'
 require_relative 'style_sheet/fonts'
 require_relative 'style_sheet/number_formats'
+require_relative 'style_sheet/xlsx_borders'
 module OoxmlParser
   # Parsing file styles.xml
   class StyleSheet < OOXMLDocumentObject
@@ -10,6 +14,10 @@ module OoxmlParser
     attr_accessor :fonts
     # @return [Fills] fills
     attr_accessor :fills
+    # @return [CellXfs] Cell XFs
+    attr_reader :cell_xfs
+    # @return [XlsxBorders] Cell XFs
+    attr_reader :borders
 
     def initialize(parent: nil)
       @number_formats = NumberFormats.new(parent: self)
@@ -19,7 +27,7 @@ module OoxmlParser
     end
 
     def parse
-      doc = Nokogiri::XML(File.open("#{OOXMLDocumentObject.path_to_folder}/#{OOXMLDocumentObject.root_subfolder}/styles.xml"))
+      doc = parse_xml("#{OOXMLDocumentObject.path_to_folder}/#{OOXMLDocumentObject.root_subfolder}/styles.xml")
       doc.root.xpath('*').each do |node_child|
         case node_child.name
         when 'numFmts'
@@ -28,6 +36,10 @@ module OoxmlParser
           @fonts.parse(node_child)
         when 'fills'
           @fills.parse(node_child)
+        when 'cellXfs'
+          @cell_xfs = CellXfs.new(parent: self).parse(node_child)
+        when 'borders'
+          @borders = XlsxBorders.new(parent: self).parse(node_child)
         end
       end
       self

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

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+require_relative 'cell_xfs/xf'
+module OoxmlParser
+  # Class for parsing `cellXfs`
+  class CellXfs < OOXMLDocumentObject
+    # @return [Integer] count of cell xfs
+    attr_reader :count
+    # @return [Array<Xf>] list of xf's
+    attr_reader :xf_array
+
+    def initialize(parent: nil)
+      @xf_array = []
+      @parent = parent
+    end
+
+    # Parse CellXfs data
+    # @param [Nokogiri::XML:Element] node with CellXfs data
+    # @return [CellXfs] value of CellXfs data
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'count'
+          @count = value.value.to_i
+        end
+      end
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'xf'
+          @xf_array << Xf.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/xlsx_parser/xlsx_data/view_model/workbook/{worksheet/xlsx_row/xlsx_cell → style_sheet/cell_xfs}/cell_style/alignment.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Character Alignment in XLSX
   class XlsxAlignment < OOXMLDocumentObject

data/lib/ooxml_parser/xlsx_parser/xlsx_data/view_model/workbook/style_sheet/cell_xfs/xf.rb

@@ -0,0 +1,145 @@
+# frozen_string_literal: true
+
+require_relative 'cell_style/alignment'
+module OoxmlParser
+  # Class for parsing `xf` object
+  class Xf < OOXMLDocumentObject
+    ALL_FORMAT_VALUE = ['General',
+                        '0',
+                        '0.00',
+                        '#,##0',
+                        '#,##0.00',
+                        '$#,##0_);($#,##0)',
+                        '$#,##0_);[Red]($#,##0)',
+                        '$#,##0.00_);($#,##0.00)',
+                        '$#,##0.00_);[Red]($#,##0.00)',
+                        '0%', '0.00%',
+                        '0.00E+00',
+                        '# ?/?',
+                        '# ??/??',
+                        'm/d/yyyy',
+                        'd-mmm-yy',
+                        'd-mmm',
+                        'mmm-yy',
+                        'h:mm AM/PM',
+                        'h:mm:ss AM/PM',
+                        'h:mm',
+                        'h:mm:ss',
+                        'm/d/yyyy h:mm',
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        '#,##0_);(#,##0)',
+                        '#,##0_);[Red](#,##0)',
+                        '#,##0.00_);(#,##0.00)',
+                        '#,##0.00_);[Red](#,##0.00)',
+                        nil,
+                        nil,
+                        nil,
+                        nil,
+                        'mm:ss',
+                        '[h]:mm:ss',
+                        'mm:ss.0',
+                        '##0.0E+0',
+                        '@'].freeze
+
+    attr_reader :alignment
+    # @return [True, False] check if style should add QuotePrefix (' symbol) to start of the string
+    attr_reader :quote_prefix
+    # @return [True, False] is font applied
+    attr_reader :apply_font
+    # @return [True, False] is border applied
+    attr_reader :apply_border
+    # @return [True, False] is fill applied
+    attr_reader :apply_fill
+    # @return [True, False] is number format applied
+    attr_reader :apply_number_format
+    # @return [True, False] is alignment applied
+    attr_reader :apply_alignment
+    # @return [Integer] id of font
+    attr_reader :font_id
+    # @return [Integer] id of border
+    attr_reader :border_id
+    # @return [Integer] id of fill
+    attr_reader :fill_id
+    # @return [Integer] id of number format
+    attr_reader :number_format_id
+
+    def initialize(parent: nil)
+      @numerical_format = 'General'
+      @alignment = XlsxAlignment.new
+      @parent = parent
+    end
+
+    # Parse Xf object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Xf] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'applyFont'
+          @apply_font = attribute_enabled?(value)
+        when 'applyBorder'
+          @apply_border = attribute_enabled?(value)
+        when 'applyFill'
+          @apply_fill = attribute_enabled?(value)
+        when 'applyNumberFormat'
+          @apply_number_format = attribute_enabled?(value)
+        when 'applyAlignment'
+          @apply_alignment = attribute_enabled?(value)
+        when 'fontId'
+          @font_id = value.value.to_i
+        when 'borderId'
+          @border_id = value.value.to_i
+        when 'fillId'
+          @fill_id = value.value.to_i
+        when 'numFmtId'
+          @number_format_id = value.value.to_i
+        when 'quotePrefix'
+          @quote_prefix = attribute_enabled?(value)
+        end
+      end
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'alignment'
+          @alignment.parse(node_child) if @apply_alignment
+        end
+      end
+      self
+    end
+
+    def font
+      root_object.style_sheet.fonts[@font_id]
+    end
+
+    def borders
+      root_object.style_sheet.borders.borders_array[@border_id] if @apply_border
+    end
+
+    def fill_color
+      root_object.style_sheet.fills[@fill_id] if @apply_fill
+    end
+
+    def numerical_format
+      return @numerical_format unless @apply_number_format
+
+      format = root_object.style_sheet.number_formats.format_by_id(@number_format_id)
+      if format
+        format.format_code
+      else
+        ALL_FORMAT_VALUE[@number_format_id]
+      end
+    end
+  end
+end