ooxml_parser 0.3.0 → 0.6.0

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

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'fills/fill'
 module OoxmlParser
   # Parsing `fonts` tag

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'fill/pattern_fill'
 module OoxmlParser
   # Parsing `fill` tag

data/lib/ooxml_parser/xlsx_parser/xlsx_data/view_model/workbook/style_sheet/fills/fill/pattern_fill.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Parsing `patternFill` tag
   class PatternFill < OOXMLDocumentObject

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'fonts/font'
 module OoxmlParser
   # Parsing `fonts` tag
@@ -10,14 +12,14 @@ module OoxmlParser
       @parent = parent
     end
 
-    # @return [Array, Font] accessor
+    # @return [Array<Font>] accessor
     def [](key)
       @fonts_array[key]
     end
 
-    # Parse NumberFormats data
+    # Parse Fonts data
     # @param [Nokogiri::XML:Element] node with NumberFormats data
-    # @return [NumberFormats] value of NumberFormats data
+    # @return [Fonts] value of NumberFormats data
     def parse(node)
       node.attributes.each do |key, value|
         case key

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Parsing `fonts` tag
   class Font < OOXMLDocumentObject

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'number_formats/number_format'
 module OoxmlParser
   # Parsing `numFmts` tag
@@ -10,11 +12,6 @@ module OoxmlParser
       @parent = parent
     end
 
-    # @return [Array, NumberFormats] accessor
-    def [](key)
-      @number_formats_array[key]
-    end
-
     # Parse NumberFormats data
     # @param [Nokogiri::XML:Element] node with NumberFormats data
     # @return [NumberFormats] value of NumberFormats data

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Parsing `numFmt` tag
   class NumberFormat < OOXMLDocumentObject

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

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+require_relative 'xlsx_borders/xlsx_border'
+module OoxmlParser
+  # Class for parsing `borders`
+  class XlsxBorders < OOXMLDocumentObject
+    # @return [Integer] count of xlsx borders
+    attr_reader :count
+    # @return [Array<XlsxBorder>] list of xlsx borders
+    attr_reader :borders_array
+
+    def initialize(parent: nil)
+      @borders_array = []
+      @parent = parent
+    end
+
+    # Parse XlsxBorders data
+    # @param [Nokogiri::XML:Element] node with XlsxBorders data
+    # @return [XlsxBorders] value of XlsxBorders 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 'border'
+          @borders_array << XlsxBorder.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/xlsx_borders/xlsx_border.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `border`
+  class XlsxBorder < OOXMLDocumentObject
+    # @return [Border] value of border
+    attr_reader :left
+    # @return [Border] value of border
+    attr_reader :right
+    # @return [Border] value of border
+    attr_reader :top
+    # @return [Border] value of border
+    attr_reader :bottom
+
+    # Parse XlsxBorder data
+    # @param [Nokogiri::XML:Element] node with XlsxBorder data
+    # @return [XlsxBorder] value of XlsxBorder data
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'bottom'
+          @bottom = Border.new(parent: self).parse(node_child)
+        when 'top'
+          @top = Border.new(parent: self).parse(node_child)
+        when 'right'
+          @right = Border.new(parent: self).parse(node_child)
+        when 'left'
+          @left = Border.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # method to help to work with DocumentStructure
   module WorkbookHelpers
@@ -5,6 +7,7 @@ module OoxmlParser
     def with_data?
       return true if @worksheets.length > 1
       return true if @worksheets.first.with_data?
+
       false
     end
   end

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

@@ -1,14 +1,19 @@
+# frozen_string_literal: true
+
 require_relative 'worksheet/excel_comments'
 require_relative 'worksheet/ole_objects'
+require_relative 'worksheet/page_setup'
 require_relative 'worksheet/sheet_format_properties'
 require_relative 'worksheet/sheet_view'
 require_relative 'worksheet/table_part'
+require_relative 'worksheet/worksheet_helper'
 require_relative 'worksheet/xlsx_column_properties'
 require_relative 'worksheet/xlsx_drawing'
 require_relative 'worksheet/xlsx_row'
-# Properties of worksheet
 module OoxmlParser
+  # Properties of worksheet
   class Worksheet < OOXMLDocumentObject
+    include WorksheetHelper
     attr_accessor :name, :rows, :merge, :charts, :hyperlinks, :drawings, :comments, :columns, :sheet_format_properties,
                   :autofilter, :table_parts, :sheet_views
     # @return [String] xml name of sheet
@@ -17,10 +22,14 @@ module OoxmlParser
     attr_accessor :relationships
     # @return [Relationships] array of ole objects
     attr_accessor :ole_objects
+    # @return [PageMargins] page margins settings
+    attr_reader :page_margins
+    # @return [PageSetup] page setup settings
+    attr_reader :page_setup
     # @return [ExtensionList] list of extensions
     attr_accessor :extension_list
 
-    def initialize
+    def initialize(parent: nil)
       @columns = []
       @name = ''
       @rows = []
@@ -30,87 +39,104 @@ module OoxmlParser
       @drawings = []
       @sheet_views = []
       @table_parts = []
+      @parent = parent
     end
 
     def parse_relationships
       OOXMLDocumentObject.add_to_xmls_stack("#{OOXMLDocumentObject.root_subfolder}/worksheets/_rels/#{@xml_name}.rels")
-      @relationships = Relationships.parse_rels(OOXMLDocumentObject.current_xml) if File.exist?(OOXMLDocumentObject.current_xml)
+      @relationships = Relationships.new(parent: self).parse_file(OOXMLDocumentObject.current_xml) if File.exist?(OOXMLDocumentObject.current_xml)
       OOXMLDocumentObject.xmls_stack.pop
     end
 
     # @return [True, false] if structure contain any user data
     def with_data?
       return true unless @rows.empty?
-      return true unless @columns.empty?
+      return true unless default_columns?
       return true unless @drawings.empty?
       return true unless @charts.empty?
       return true unless @hyperlinks.empty?
+
       false
     end
 
     # Parse list of drawings in file
     def parse_drawing
-      drawing_node = Nokogiri::XML(File.open(OOXMLDocumentObject.current_xml))
+      drawing_node = parse_xml(OOXMLDocumentObject.current_xml)
       drawing_node.xpath('xdr:wsDr/*').each do |drawing_node_child|
         @drawings << XlsxDrawing.new(parent: self).parse(drawing_node_child)
       end
     end
 
-    def self.parse(path_to_xml_file, parent: nil)
-      worksheet = Worksheet.new
-      worksheet.xml_name = File.basename path_to_xml_file
-      worksheet.parse_relationships
-      worksheet.parent = parent
+    def parse(path_to_xml_file)
+      @xml_name = File.basename path_to_xml_file
+      parse_relationships
       OOXMLDocumentObject.add_to_xmls_stack("#{OOXMLDocumentObject.root_subfolder}/worksheets/#{File.basename(path_to_xml_file)}")
-      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.current_xml))
+      doc = parse_xml(OOXMLDocumentObject.current_xml)
       sheet = doc.search('//xmlns:worksheet').first
       sheet.xpath('*').each do |worksheet_node_child|
         case worksheet_node_child.name
         when 'sheetData'
           worksheet_node_child.xpath('xmlns:row').each do |row_node|
-            worksheet.rows[row_node.attribute('r').value.to_i - 1] = XlsxRow.new(parent: worksheet).parse(row_node)
-            worksheet.rows[row_node.attribute('r').value.to_i - 1].style = CellStyle.new(parent: worksheet).parse(row_node.attribute('s').value) unless row_node.attribute('s').nil?
+            @rows[row_node.attribute('r').value.to_i - 1] = XlsxRow.new(parent: self).parse(row_node)
+            @rows[row_node.attribute('r').value.to_i - 1].style = root_object.style_sheet.cell_xfs.xf_array[row_node.attribute('s').value.to_i] if row_node.attribute('s')
           end
         when 'sheetFormatPr'
           if !worksheet_node_child.attribute('defaultColWidth').nil? && !worksheet_node_child.attribute('defaultRowHeight').nil?
-            worksheet.sheet_format_properties = SheetFormatProperties.new(parent: worksheet).parse(worksheet_node_child)
+            @sheet_format_properties = SheetFormatProperties.new(parent: self).parse(worksheet_node_child)
           end
         when 'mergeCells'
           worksheet_node_child.xpath('xmlns:mergeCell').each do |merge_node|
-            worksheet.merge << merge_node.attribute('ref').value.to_s
+            @merge << merge_node.attribute('ref').value.to_s
           end
         when 'drawing'
           path_to_drawing = OOXMLDocumentObject.get_link_from_rels(worksheet_node_child.attribute('id').value)
           unless path_to_drawing.nil?
             OOXMLDocumentObject.add_to_xmls_stack(path_to_drawing)
-            worksheet.parse_drawing
+            parse_drawing
             OOXMLDocumentObject.xmls_stack.pop
           end
         when 'hyperlinks'
           worksheet_node_child.xpath('xmlns:hyperlink').each do |hyperlink_node|
-            worksheet.hyperlinks << Hyperlink.new(parent: worksheet).parse(hyperlink_node).dup
+            @hyperlinks << Hyperlink.new(parent: self).parse(hyperlink_node).dup
           end
         when 'cols'
-          worksheet.columns = XlsxColumnProperties.parse_list(worksheet_node_child, parent: worksheet)
+          @columns = XlsxColumnProperties.parse_list(worksheet_node_child, parent: self)
         when 'autoFilter'
-          worksheet.autofilter = Autofilter.new(parent: self).parse(worksheet_node_child)
+          @autofilter = Autofilter.new(parent: self).parse(worksheet_node_child)
         when 'tableParts'
           worksheet_node_child.xpath('*').each do |part_node|
-            worksheet.table_parts << TablePart.new(parent: worksheet).parse(part_node)
+            @table_parts << TablePart.new(parent: self).parse(part_node)
           end
         when 'sheetViews'
           worksheet_node_child.xpath('*').each do |view_child|
-            worksheet.sheet_views << SheetView.new(parent: worksheet).parse(view_child)
+            @sheet_views << SheetView.new(parent: self).parse(view_child)
           end
         when 'oleObjects'
-          worksheet.ole_objects = OleObjects.new(parent: worksheet).parse(worksheet_node_child)
+          @ole_objects = OleObjects.new(parent: self).parse(worksheet_node_child)
+        when 'pageMargins'
+          @page_margins = PageMargins.new(parent: self).parse(worksheet_node_child, :inch)
+        when 'pageSetup'
+          @page_setup = PageSetup.new(parent: self).parse(worksheet_node_child)
         when 'extLst'
-          worksheet.extension_list = ExtensionList.new(parent: self).parse(worksheet_node_child)
+          @extension_list = ExtensionList.new(parent: self).parse(worksheet_node_child)
         end
       end
-      worksheet.comments = ExcelComments.parse_file(File.basename(path_to_xml_file), OOXMLDocumentObject.path_to_folder)
+      parse_comments
       OOXMLDocumentObject.xmls_stack.pop
-      worksheet
+      self
+    end
+
+    private
+
+    # Do work for parsing shared comments file
+    def parse_comments
+      return unless relationships
+
+      comments_target = relationships.target_by_type('comment')
+      return if comments_target.empty?
+
+      comments_file = "#{OOXMLDocumentObject.path_to_folder}/#{OOXMLDocumentObject.root_subfolder}/#{comments_target.first.gsub('..', '')}"
+      @comments = ExcelComments.new(parent: self).parse(comments_file)
     end
   end
 end