Added some minor performance improvements to date format parsing and …

…to column index generation as well as adding enumeration to workbook and worksheet.

lib/rubyXL/cell.rb

@@ -30,7 +30,7 @@ def is_date?
         if @workbook.num_fmts
           num_fmt_id = xf_id()[:numFmtId]
           num_fmt = @workbook.num_fmts[:numFmt].select { |f| f[:attributes][:numFmtId] == num_fmt_id }[0].andand[:attributes].andand[:formatCode]
-          if num_fmt && is_date_format?(num_fmt)
+          if num_fmt && workbook.date_num_fmt?(num_fmt)
             return true
           end
         end

lib/rubyXL/parser.rb

@@ -7,22 +7,29 @@ module RubyXL
 
   class Parser
     attr_reader :data_only, :num_sheets
-
+    @@parsed_column_hash ={}
     # converts cell string (such as "AA1") to matrix indices
     def Parser.convert_to_index(cell_string)
       index = Array.new(2)
       index[0]=-1
       index[1]=-1
       if(cell_string =~ /^([A-Z]+)(\d+)$/)
-        one = $1.to_s
-        row = Integer($2) - 1 #-1 for 0 indexing
+
+        one = $1
+        row = $2.to_i - 1 #-1 for 0 indexing
         col = 0
         i = 0
-        one = one.reverse #because of 26^i calculation
-        one.each_byte do |c|
-          int_val = c - 64 #converts A to 1
-          col += int_val * 26**(i)
-          i=i+1
+        if @@parsed_column_hash[one].nil?
+          puts "||#{one}||"
+          two = one.reverse #because of 26^i calculation
+          two.each_byte do |c|
+            int_val = c - 64 #converts A to 1
+            col += int_val * 26**(i)
+            i=i+1
+          end
+          @@parsed_column_hash[one] = col
+        else
+          col = @@parsed_column_hash[one]
         end
         col -= 1 #zer0 index
         index[0] = row

lib/rubyXL/workbook.rb

@@ -13,11 +13,13 @@
 
 module RubyXL
   class Workbook
+    include Enumerable
     attr_accessor :worksheets, :filepath, :creator, :modifier, :created_at,
       :modified_at, :company, :application, :appversion, :num_fmts, :fonts, :fills,
       :borders, :cell_xfs, :cell_style_xfs, :cell_styles, :shared_strings, :calc_chain,
       :num_strings, :size, :date1904, :external_links, :style_corrector, :drawings,
-      :worksheet_rels, :printer_settings, :macros, :colors, :shared_strings_XML, :defined_names
+      :worksheet_rels, :printer_settings, :macros, :colors, :shared_strings_XML, :defined_names, :column_lookup_hash
+
 
     APPLICATION = 'Microsoft Macintosh Excel'
     APPVERSION  = '12.0000'
@@ -57,6 +59,7 @@ def initialize(worksheets=[], filepath=nil, creator=nil, modifier=nil, created_a
       @colors             = nil
       @shared_strings_XML = nil
       @defined_names      = nil
+      @column_lookup_hash = {}
 
       begin
         @created_at       = DateTime.parse(created_at).strftime('%Y-%m-%dT%TZ')
@@ -75,6 +78,10 @@ def [](worksheet)
       return worksheets[worksheet]
     end
 
+    def each
+      worksheets.each{|i| yield i}
+    end
+
     #filepath of xlsx file (including file itself)
     def write(filepath=@filepath)
       validate_before_write
@@ -214,6 +221,70 @@ def num_to_date(num)
       compare_date - 1 + num
     end
 
+    def date_num_fmt?(num_fmt)
+      @num_fmt_date_hash ||= {}
+      if @num_fmt_date_hash[num_fmt].nil?
+        @num_fmt_date_hash[num_fmt] = is_date_format?(num_fmt)
+      end
+      return @num_fmt_date_hash[num_fmt]
+    end
+
+     def is_date_format?(num_fmt)
+      skip_chars = ['$', '-', '+', '/', '(', ')', ':', ' ']
+      num_chars = ['0', '#', '?']
+      non_date_formats = ['0.00E+00', '##0.0E+0', 'General', 'GENERAL', 'general', '@']
+      date_chars = ['y','m','d','h','s']
+
+      state = 0
+      s = ''
+      num_fmt.split(//).each do |c|
+        if state == 0
+          if c == '"'
+            state = 1
+          elsif ['\\', '_', '*'].include?(c)
+            state = 2
+          elsif skip_chars.include?(c)
+            next
+          else
+            s << c
+          end
+        elsif state == 1
+          if c == '"'
+            state = 0
+          end
+        elsif state == 2
+          state = 0
+        end
+      end
+      s.gsub!(/\[[^\]]*\]/, '')
+      if non_date_formats.include?(s)
+        return false
+      end
+      separator = ';'
+      got_sep = 0
+      date_count = 0
+      num_count = 0
+      s.split(//).each do |c|
+        if date_chars.include?(c)
+          date_count += 1
+        elsif num_chars.include?(c)
+          num_count += 1
+        elsif c == separator
+          got_sep = 1
+        end
+      end
+      if date_count > 0 && num_count == 0
+        return true
+      elsif num_count > 0 && date_count == 0
+        return false
+      elsif date_count
+        # ambiguous result
+      elsif got_sep == 0
+        # constant result
+      end
+      return date_count > num_count
+    end
+
     #gets style object from style array given index
     def get_style(style_index)
       if !@cell_xfs[:xf].is_a?Array

lib/rubyXL/worksheet.rb

@@ -1,5 +1,6 @@
 module RubyXL
 class Worksheet < PrivateClass
+  include Enumerable
 
   attr_accessor :sheet_name, :sheet_data, :cols, :merged_cells, :pane,
     :validations, :sheet_view, :legacy_drawing, :extLst, :workbook, :row_styles
@@ -26,6 +27,10 @@ def [](row=0)
     return @sheet_data[row]
   end
 
+  def each
+    @sheet_data.each {|i| yield i}
+  end
+
   #returns 2d array of just the cell values (without style or formula information)
   def extract_data
     return @sheet_data.map {|row| row.map {|c| if c.is_a?(Cell) then c.value else nil end}}
@@ -51,6 +56,8 @@ def get_table(headers=[])
   	header_row.each_with_index do |header_cell, index|
       next if header_cell.nil? || header_cell.value.nil?
       header = header_cell.value.to_s
+      table_hash[:sorted_headers]||=[]
+      table_hash[:sorted_headers] << header
   	  table_hash[header] = []
 
   	  original_row = row_num + 1