ooxml_parser 0.4.1 → 0.7.1

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
@@ -9,9 +11,12 @@ module OoxmlParser
       @parent = parent
     end
 
+    # Parse Chartsheet object
+    # @param file [String] file to parse
+    # @return [Chartsheet] result of parsing
     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/pivot_cache.rb

@@ -0,0 +1,41 @@
+# frozen_string_literal: true
+
+require_relative 'pivot_cache/pivot_cache_definition'
+
+module OoxmlParser
+  # Class for parsing <pivotCache> tag
+  class PivotCache < OOXMLDocumentObject
+    # @return [Integer] cacheId of pivot cache
+    attr_reader :cache_id
+    # @return [String] id of pivot cache
+    attr_reader :id
+    # @return [PivotCacheDefinition] parsed pivot cache definition
+    attr_reader :pivot_cache_definition
+
+    # Parse Pivot Cache data
+    # @param [Nokogiri::XML:Element] node with Pivot Cache data
+    # @return [PivotCache] value of PivotCache
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'cacheId'
+          @cache_id = value.value.to_i
+        when 'id'
+          @id = value.value.to_s
+        end
+      end
+      parse_pivot_cache_definition
+      self
+    end
+
+    private
+
+    # @return [PivotCacheDefinition] pivot cache definition for current pivot cache
+    def parse_pivot_cache_definition
+      definition_file = root_object.relationships.target_by_id(id)
+      full_file_path = "#{OOXMLDocumentObject.path_to_folder}/xl/#{definition_file}"
+      @pivot_cache_definition = PivotCacheDefinition.new(parent: root_object)
+                                                    .parse(full_file_path)
+    end
+  end
+end

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

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+require_relative 'pivot_cache_definition/cache_source'
+module OoxmlParser
+  # Class for parsing <pivotCacheDefinition> file
+  class PivotCacheDefinition < OOXMLDocumentObject
+    # @return [String] id of pivot cache definition
+    attr_reader :id
+    # @return [CacheSource] source of pivot cache
+    attr_reader :cache_source
+
+    # Parse PivotCacheDefinition file
+    # @param file [String] path to file
+    # @return [PivotCacheDefinition]
+    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 'id'
+          @id = value.value.to_s
+        end
+      end
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'cacheSource'
+          @cache_source = CacheSource.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/xlsx_parser/xlsx_data/view_model/workbook/pivot_cache/pivot_cache_definition/cache_source.rb

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+require_relative 'cache_source/worksheet_source'
+
+module OoxmlParser
+  # Class for parsing <cacheSource> file
+  class CacheSource < OOXMLDocumentObject
+    # @return [String] type
+    attr_reader :type
+    # @return [WorksheetSource] source of worksheet data
+    attr_reader :worksheet_source
+
+    # Parse `<cacheSource>` tag
+    # # @param [Nokogiri::XML:Element] node with CacheSource data
+    # @return [CacheSource]
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'type'
+          @type = value.value.to_s
+        end
+      end
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'worksheetSource'
+          @worksheet_source = WorksheetSource.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/xlsx_parser/xlsx_data/view_model/workbook/pivot_cache/pivot_cache_definition/cache_source/worksheet_source.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing <worksheetSource> file
+  class WorksheetSource < OOXMLDocumentObject
+    # @return [String] ref of cells
+    attr_reader :ref
+    # @return [String] sheet name
+    attr_reader :sheet
+
+    # Parse `<worksheetSource>` tag
+    # # @param [Nokogiri::XML:Element] node with WorksheetSource data
+    # @return [WorksheetSource]
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'ref'
+          @ref = value.value.to_s
+        when 'sheet'
+          @sheet = value.value.to_s
+        end
+      end
+      self
+    end
+  end
+end

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)
@@ -18,8 +26,10 @@ module OoxmlParser
       @parent = parent
     end
 
+    # Parse StyleSheet object
+    # @return [StyleSheet] result of parsing
     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 +38,10 @@ 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 +1,150 @@
+# frozen_string_literal: true
+
+require_relative 'cell_style/alignment'
+module OoxmlParser
+  # Class for parsing `xf` object
+  class Xf < OOXMLDocumentObject
+    # [Array<String,nil>] list of predefined format values
+    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
+
+    # @return [Font] font of object
+    def font
+      root_object.style_sheet.fonts[@font_id]
+    end
+
+    # @return [XlsxBorder] border of object
+    def borders
+      root_object.style_sheet.borders.borders_array[@border_id] if @apply_border
+    end
+
+    # @return [Fill] fill color of object
+    def fill_color
+      root_object.style_sheet.fills[@fill_id] if @apply_fill
+    end
+
+    # @return [String] numerical format of object
+    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