ooxml_parser 0.1.2 → 0.2.0

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

@@ -1,19 +1,41 @@
 require_relative 'xlsx_row/xlsx_cell'
-# Single Row of XLSX
 module OoxmlParser
+  # Single Row of XLSX
   class XlsxRow < OOXMLDocumentObject
     attr_accessor :cells, :height, :style, :hidden
+    # @return [True, False] true if the row height has been manually set.
+    attr_accessor :custom_height
+    # @return [Integer] Indicates to which row in the sheet this <row> definition corresponds.
+    attr_accessor :index
 
-    def initialize(cells = [])
-      @cells = cells
+    def initialize(parent: nil)
+      @cells = []
+      @parent = parent
     end
 
-    def self.parse(row_node)
-      row = XlsxRow.new
-      row.height = row_node.attribute('ht').value if OOXMLDocumentObject.option_enabled?(row_node, 'customHeight') && row_node.attribute('ht')
-      row.hidden = OOXMLDocumentObject.option_enabled?(row_node, 'hidden')
-      row_node.xpath('xmlns:c').each { |cell_node| row.cells[Coordinates.parse_coordinates_from_string(cell_node.attribute('r').value.to_s).get_column_number.to_i - 1] = XlsxCell.parse(cell_node) }
-      row
+    # Parse XlsxRow object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [XlsxRow] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'customHeight'
+          @custom_height = option_enabled?(node, 'customHeight')
+        when 'ht'
+          @height = OoxmlSize.new(value.value.to_f, :point)
+        when 'hidden'
+          @hidden = option_enabled?(node, 'hidden')
+        when 'r'
+          @index = value.value.to_i
+        end
+      end
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'c'
+          @cells[Coordinates.parse_coordinates_from_string(node_child.attribute('r').value.to_s).get_column_number.to_i - 1] = XlsxCell.new(parent: self).parse(node_child)
+        end
+      end
+      self
     end
   end
 end

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

@@ -1,10 +1,10 @@
 require_relative 'worksheet/excel_comments'
 require_relative 'worksheet/sheet_format_properties'
 require_relative 'worksheet/sheet_view'
+require_relative 'worksheet/table_part'
 require_relative 'worksheet/xlsx_column_properties'
 require_relative 'worksheet/xlsx_drawing'
 require_relative 'worksheet/xlsx_row'
-require_relative 'worksheet/xlsx_table'
 # Properties of worksheet
 module OoxmlParser
   class Worksheet < OOXMLDocumentObject
@@ -29,14 +29,33 @@ module OoxmlParser
 
     def parse_relationships
       OOXMLDocumentObject.add_to_xmls_stack("#{OOXMLDocumentObject.root_subfolder}/worksheets/_rels/#{@xml_name}.rels")
-      @relationships = Relationships.parse(Nokogiri::XML(File.open(OOXMLDocumentObject.current_xml)).xpath('*').first) if File.exist?(OOXMLDocumentObject.current_xml)
+      @relationships = Relationships.parse_rels(OOXMLDocumentObject.current_xml) if File.exist?(OOXMLDocumentObject.current_xml)
       OOXMLDocumentObject.xmls_stack.pop
     end
 
-    def self.parse(path_to_xml_file)
+    # @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 @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.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
       OOXMLDocumentObject.add_to_xmls_stack("#{OOXMLDocumentObject.root_subfolder}/worksheets/#{File.basename(path_to_xml_file)}")
       doc = Nokogiri::XML(File.open(OOXMLDocumentObject.current_xml))
       sheet = doc.search('//xmlns:worksheet').first
@@ -44,12 +63,12 @@ module OoxmlParser
         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.parse(row_node)
-            worksheet.rows[row_node.attribute('r').value.to_i - 1].style = CellStyle.parse(row_node.attribute('s').value) unless row_node.attribute('s').nil?
+            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?
           end
         when 'sheetFormatPr'
           if !worksheet_node_child.attribute('defaultColWidth').nil? && !worksheet_node_child.attribute('defaultRowHeight').nil?
-            worksheet.sheet_format_properties = SheetFormatProperties.parse(worksheet_node_child)
+            worksheet.sheet_format_properties = SheetFormatProperties.new(parent: worksheet).parse(worksheet_node_child)
           end
         when 'mergeCells'
           worksheet_node_child.xpath('xmlns:mergeCell').each do |merge_node|
@@ -59,22 +78,25 @@ module OoxmlParser
           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)
-            XlsxDrawing.parse_list(worksheet)
+            worksheet.parse_drawing
             OOXMLDocumentObject.xmls_stack.pop
           end
         when 'hyperlinks'
           worksheet_node_child.xpath('xmlns:hyperlink').each do |hyperlink_node|
-            worksheet.hyperlinks << Hyperlink.parse(hyperlink_node).dup
+            worksheet.hyperlinks << Hyperlink.new(parent: worksheet).parse(hyperlink_node).dup
           end
         when 'cols'
-          worksheet.columns = XlsxColumnProperties.parse_list(worksheet_node_child)
+          worksheet.columns = XlsxColumnProperties.parse_list(worksheet_node_child, parent: worksheet)
         when 'autoFilter'
           worksheet.autofilter = Coordinates.parser_coordinates_range(worksheet_node_child.attribute('ref').value.to_s)
         when 'tableParts'
-          worksheet.table_parts = []
-          worksheet_node_child.xpath('xmlns:tablePart').each { |table_part_node| worksheet.table_parts << XlsxTable.parse(table_part_node) }
+          worksheet_node_child.xpath('*').each do |part_node|
+            worksheet.table_parts << TablePart.new(parent: worksheet).parse(part_node)
+          end
         when 'sheetViews'
-          worksheet_node_child.xpath('xmlns:sheetView').each { |sheet_view_node| worksheet.sheet_views << SheetView.parse(sheet_view_node) }
+          worksheet_node_child.xpath('*').each do |view_child|
+            worksheet.sheet_views << SheetView.new(parent: worksheet).parse(view_child)
+          end
         end
       end
       worksheet.comments = ExcelComments.parse_file(File.basename(path_to_xml_file), OOXMLDocumentObject.path_to_folder)

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

@@ -1,10 +1,18 @@
+require_relative 'workbook/chartsheet'
+require_relative 'workbook/style_sheet'
 require_relative 'workbook/worksheet'
+require_relative 'workbook/workbook_helpers'
 # Class for storing XLSX Workbook
 module OoxmlParser
   class XLSXWorkbook < CommonDocumentStructure
+    include WorkbookHelpers
     attr_accessor :worksheets
     # @return [PresentationTheme] theme of Workbook
     attr_accessor :theme
+    # @return [Relationships] rels of book
+    attr_accessor :relationships
+    # @return [StyleSheet] styles of book
+    attr_accessor :style_sheet
 
     def initialize(worksheets = [])
       @worksheets = worksheets
@@ -66,6 +74,7 @@ module OoxmlParser
 
     def self.parse
       workbook = XLSXWorkbook.new
+      workbook.relationships = Relationships.parse_rels("#{OOXMLDocumentObject.path_to_folder}xl/_rels/workbook.xml.rels")
       OOXMLDocumentObject.xmls_stack = []
       OOXMLDocumentObject.root_subfolder = 'xl/'
       self.shared_strings = nil
@@ -73,9 +82,15 @@ module OoxmlParser
       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|
-        workbook.worksheets << Worksheet.parse('xl/' + OOXMLDocumentObject.get_link_from_rels(sheet.attribute('id').value))
-        workbook.worksheets.last.name = sheet.attribute('name').value
+        file = workbook.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
+        elsif file.start_with?('chartsheets')
+          workbook.worksheets << Chartsheet.new(parent: workbook).parse(file)
+        end
       end
       OOXMLDocumentObject.xmls_stack.pop
       workbook
@@ -96,13 +111,9 @@ module OoxmlParser
       end
 
       def link_to_theme_xml
-        doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + 'xl/_rels/workbook.xml.rels'))
-        doc.xpath('xmlns:Relationships/xmlns:Relationship').each do |relationship_node|
-          if relationship_node.attribute('Target').value.include?('theme')
-            return relationship_node.attribute('Target').value
-          end
-        end
-        nil
+        file = File.open(OOXMLDocumentObject.path_to_folder + 'xl/_rels/workbook.xml.rels')
+        relationships = Relationships.parse_rels(file)
+        relationships.target_by_type('theme')
       end
     end
   end

data/lib/ooxml_parser/xlsx_parser/xlsx_parser.rb

@@ -3,7 +3,7 @@ require_relative 'xlsx_data/view_model/workbook'
 module OoxmlParser
   class XlsxParser
     def self.parse_xlsx(path_to_file)
-      Parser.parse(path_to_file) do
+      Parser.parse_format(path_to_file) do
         XLSXWorkbook.parse
       end
     end

data/lib/ooxml_parser.rb

@@ -1,5 +1,4 @@
 require_relative 'ooxml_parser/common_parser/common_data/ooxml_document_object'
-require_relative 'ooxml_parser/common_parser/common_data/alingment'
 require_relative 'ooxml_parser/common_parser/common_data/color'
 require_relative 'ooxml_parser/common_parser/common_data/coordinates'
 require_relative 'ooxml_parser/common_parser/common_data/font_style'