workbook 0.1.1

data/lib/workbook/readers/xls_reader.rb

@@ -0,0 +1,161 @@
+require 'spreadsheet'
+
+module Workbook
+  module Readers
+    module XlsReader
+      XLS_COLORS = {:xls_color_1=>'#000000',
+                    :xls_color_2=>'#FFFFFF',
+                    :xls_color_3=>'#FF0000',
+                    :xls_color_4=>'#00FF00',
+                    :xls_color_5=>'#0000FF',
+                    :xls_color_6=>'#FFFF00',
+                    :xls_color_7=>'#FF00FF',
+                    :xls_color_8=>'#00FFFF',
+                    :xls_color_9=>'#800000',
+                    :xls_color_10=>'#008000',
+                    :xls_color_11=>'#000080',
+                    :xls_color_12=>'#808000',
+                    :xls_color_13=>'#800080',
+                    :xls_color_14=>'#008080',
+                    :xls_color_15=>'#C0C0C0',
+                    :xls_color_16=>'#808080',
+                    :xls_color_17=>'#9999FF',
+                    :xls_color_18=>'#993366',
+                    :xls_color_19=>'#FFFFCC',
+                    :xls_color_20=>'#CCFFFF',
+                    :xls_color_21=>'#660066',
+                    :xls_color_22=>'#FF8080',
+                    :xls_color_23=>'#0066CC',
+                    :xls_color_24=>'#CCCCFF',
+                    :xls_color_25=>'#000080',
+                    :xls_color_26=>'#FF00FF',
+                    :xls_color_27=>'#FFFF00',
+                    :xls_color_28=>'#00FFFF',
+                    :xls_color_29=>'#800080',
+                    :xls_color_30=>'#800000',
+                    :xls_color_31=>'#008080',
+                    :xls_color_32=>'#0000FF',
+                    :xls_color_33=>'#00CCFF',
+                    :xls_color_34=>'#CCFFFF',
+                    :xls_color_35=>'#CCFFCC',
+                    :xls_color_36=>'#FFFF99',
+                    :xls_color_37=>'#99CCFF',
+                    :xls_color_38=>'#FF99CC',
+                    :xls_color_39=>'#CC99FF',
+                    :xls_color_40=>'#FFCC99',
+                    :xls_color_41=>'#3366FF',
+                    :xls_color_42=>'#33CCCC',
+                    :xls_color_43=>'#99CC00',
+                    :xls_color_44=>'#FFCC00',
+                    :xls_color_45=>'#FF9900',
+                    :xls_color_46=>'#FF6600',
+                    :xls_color_47=>'#666699',
+                    :xls_color_48=>'#969696',
+                    :xls_color_49=>'#003366',
+                    :xls_color_50=>'#339966',
+                    :xls_color_51=>'#003300',
+                    :xls_color_52=>'#333300',
+                    :xls_color_53=>'#993300',
+                    :xls_color_54=>'#993366',
+                    :xls_color_55=>'#333399',
+                    :xls_color_56=>'#333333',
+                    :black=>'#000000',
+                    :white=>'#FFFFFF',
+                    :red=>'#FF0000',
+                    :green=>'#00FF00',
+                    :blue=>'#0000FF',
+                    :yellow=>'#FFFF00',
+                    :magenta=>'#FF00FF',
+                    :cyan=>'#00FFFF',
+                    :border=>'#FFFFFF',
+                    :text=>'#000000',
+                    :lime=>'#00f94c'}
+
+        
+
+      def load_xls file_obj
+        sp = Spreadsheet.open(file_obj, mode='rb')
+        template.add_raw sp
+        parse_xls sp
+      end
+      
+      
+      def parse_xls xls_spreadsheet=template.raws[Spreadsheet::Excel::Workbook], options={}
+        options = {:additional_type_parsing=>true}.merge options
+        number_of_worksheets = xls_spreadsheet.worksheets.count
+        (0..number_of_worksheets-1).each do |si|
+          xls_sheet = xls_spreadsheet.worksheets[si]
+          begin
+            number_of_rows = xls_spreadsheet.worksheets[si].count
+            s = create_or_open_sheet_at(si)    
+            (0..number_of_rows-1).each do |ri|
+              xls_row = xls_sheet.row(ri)
+              r = s.table.create_or_open_row_at(ri)
+              col_widths = xls_sheet.columns.collect{|c| c.width if c}
+              xls_row.each_with_index do |xls_cell,ci|
+              
+                begin
+                  r[ci] = Workbook::Cell.new xls_cell    
+                  r[ci].parse!    
+                rescue ArgumentError => e
+                  if e.message.match('not a Spreadsheet::Formula')
+                    v = xls_cell.value
+                    if v.class == Float and xls_row.format(ci).date?
+                      xls_row[ci] = v
+                      v = xls_row.datetime(ci)
+                    end
+                    if v.is_a? Spreadsheet::Excel::Error
+                      v = "----!"
+                    end 
+                    r[ci] = Workbook::Cell.new v
+                  elsif e.message.match('not a Spreadsheet::Link')
+                    r[ci] = Workbook::Cell.new xls_cell.to_s
+                  elsif e.message.match('not a Spreadsheet::Link')
+                    r[ci] = Workbook::Cell.new xls_cell.to_s
+                  elsif e.message.match('not a Spreadsheet::Excel::Error')
+                    r[ci] = "._."        
+                  else 
+                    r[ci] = "._."  # raise e (we're going to be silent for now)
+                  end
+                end
+                xls_format = xls_row.format(ci)
+                col_width = nil
+              
+                if ri == 0
+                  col_width = col_widths[ci]
+                end
+                f = template.create_or_find_format_by "object_id_#{xls_format.object_id}",col_width
+                f[:width]= col_width
+                f[:rotation] = xls_format.rotation if xls_format.rotation 
+                f[:background_color] = xls_color_to_html_hex(xls_format.pattern_fg_color)
+                f[:number_format] = ms_formatting_to_strftime(xls_format.number_format)
+                f[:text_direction] = xls_format.text_direction
+                f[:font_family] = "#{xls_format.font.name}, #{xls_format.font.family}"
+                f[:font_weight] = xls_format.font.weight
+                f[:color] = xls_color_to_html_hex(xls_format.font.color)
+
+                f.add_raw xls_format
+     
+                r[ci].format = f
+              end
+            end
+          rescue TypeError
+            puts "WARNING: Failed at worksheet (#{si})... ignored"
+            #ignore SpreadsheetGem can be buggy...
+          end
+        end
+      end
+      
+      private 
+      def xls_color_to_html_hex color_sym
+        XLS_COLORS[color_sym] ? XLS_COLORS[color_sym] : "#000000"
+      end
+      
+      def ms_formatting_to_strftime ms_nr_format       
+        ms_nr_format = ms_nr_format.downcase
+        return nil if ms_nr_format == 'general'
+        ms_nr_format.gsub('yyyy','%Y').gsub('dddd','%A').gsub('mmmm','%B').gsub('ddd','%a').gsub('mmm','%b').gsub('yy','%y').gsub('dd','%d').gsub('mm','%m').gsub('y','%y').gsub('%%y','%y').gsub('d','%e').gsub('%%e','%d').gsub('m','%m').gsub('%%m','%m').gsub(';@','').gsub('\\','')
+      end
+    end
+  end
+end

data/lib/workbook/row.rb

@@ -0,0 +1,101 @@
+module Workbook
+  class Row < Array
+    alias_method :compare_without_header, :<=>
+    
+    #used in compares
+    attr_accessor :placeholder
+    attr_accessor :format
+    
+    
+    def initialize cells=[], table=nil, options={}
+      options=options ? {:parse_cells_on_batch_creation=>false,:cell_parse_options=>{}}.merge(options) : {} 
+      cells = [] if cells==nil
+      self.table= table
+      cells.each do |c| 
+        if c.is_a? Workbook::Cell
+          c = c
+        else
+          c = Workbook::Cell.new(c)
+          c.parse!(options[:cell_parse_options]) if options[:parse_cells_on_batch_creation]
+        end
+        push c
+      end
+    end
+    
+    def placeholder?
+      placeholder ? true : false
+    end
+    
+    def table
+      @table
+    end
+    
+    # @param [Workbook::Table] Reference to the table this row belongs to (and adds the row to this table)
+    def table= t
+      raise ArgumentError, "table should be a Workbook::Table (you passed a #{t.class})" unless t.is_a?(Workbook::Table) or t == nil
+      if t
+        @table = t
+        table << self
+      end
+    end
+    
+    def [](index_or_hash)
+      if index_or_hash.is_a? Fixnum
+        return to_a[index_or_hash]
+      elsif index_or_hash.is_a? Symbol
+        rv = nil
+        begin
+          rv = to_hash[index_or_hash]
+        rescue NoMethodError
+        end
+        return rv
+      end
+    end
+    
+    # find_cells_by_color returns an array of cells allows you to find cells by a given color, normally a string containing a hex
+    def find_cells_by_background_color color=:any, options={}
+      options = {:hash_keys=>true}.merge(options)
+      cells = self.collect {|c| c if c.format.has_background_color?(color) }.compact
+      r = Row.new cells
+      options[:hash_keys] ? r.to_symbols : r
+    end
+    
+    def header?
+      table != nil and self.object_id == table.header.object_id
+    end
+    
+    def to_symbols
+      collect{|c| c.to_sym}
+    end
+    
+    def to_a
+      self.collect{|c| c}
+    end
+    
+    def to_hash
+      return @hash if @hash
+      keys = table.header.to_symbols
+      values = self
+      @hash = {}
+      keys.each_with_index {|k,i| @hash[k]=values[i]}
+      return @hash
+    end
+    
+    def <=> other
+      a = self.header? ? 0 : 1
+      b = other.header? ? 0 : 1
+      return (a <=> b) if (a==0 or b==0)
+      compare_without_header other
+    end
+    
+    def key
+      first
+    end
+    
+    # Compact detaches the row from the table
+    def compact
+      r = self.clone
+      r = r.collect{|c| c unless c.nil?}.compact
+    end
+  end
+end

data/lib/workbook/sheet.rb

@@ -0,0 +1,22 @@
+module Workbook
+  class Sheet < Array
+    attr_accessor :book
+    
+    def initialize table=Workbook::Table.new([], self), book=nil, options={}
+      if table.is_a? Workbook::Table
+        push table
+      else
+        push Workbook::Table.new(table, self, options)
+      end
+      self.book = book
+    end
+    
+    def has_contents?
+      table.has_contents?
+    end
+        
+    def table
+      first
+    end
+  end
+end

data/lib/workbook/table.rb

@@ -0,0 +1,67 @@
+require 'workbook/modules/table_diff_sort'
+require 'workbook/writers/csv_table_writer'
+
+
+module Workbook  
+  class Table < Array
+    include Workbook::Modules::TableDiffSort
+    include Workbook::Writers::CsvTableWriter
+    attr_accessor :sheet
+    attr_accessor :header
+    
+    def initialize row_cel_values=[], sheet=nil, options={}
+      #@rows = []
+      row_cel_values = [] if row_cel_values == nil
+      row_cel_values.each do |r|
+        if r.is_a? Workbook::Row
+          r.table = self
+        else
+          r = Workbook::Row.new(r,self, options)
+        end
+      end
+      self.sheet = sheet
+      # Column data is considered as a 'row' with 'cells' that contain 'formatting'
+
+    end
+    
+    def header
+      if @header == false
+        false
+      elsif @header
+        @header
+      else
+        first
+      end
+    end
+    
+    # factory pattern...?
+    def new_row cel_values=[]
+      r = Workbook::Row.new(cel_values,self)
+      return r
+    end
+    
+    def create_or_open_row_at index
+      r = self[index]
+      if r == nil
+        r = Workbook::Row.new
+        r.table=(self)
+      end
+      r 
+    end  
+    
+    def remove_empty_lines!
+      self.delete_if{|r| r.nil? or r.compact.empty?}
+      self
+    end
+    
+    def has_contents?
+      self.clone.remove_empty_lines!.count != 0
+    end
+    
+    def contains_row? row
+      raise ArgumentError, "table should be a Workbook::Row (you passed a #{t.class})" unless row.is_a?(Workbook::Row)
+      self.collect{|r| r.object_id}.include? row.object_id
+    end
+    
+  end
+end

data/lib/workbook/template.rb

@@ -0,0 +1,52 @@
+require 'workbook/modules/raw_objects_storage'
+
+module Workbook
+	class Template
+    include Workbook::Modules::RawObjectsStorage
+    
+    def initialize 
+      @formats = {}
+      @has_header = true
+    end
+    
+    def has_header?
+      @has_header
+    end
+    
+    def has_header= boolean
+      if format.is_a? TrueClass or format.is_a? FalseClass
+        @has_header = boolean
+      else
+        raise ArgumentError, "format should be a boolean, true of false"
+      end  
+    end
+      
+    def add_format format
+      if format.is_a? Workbook::Format
+        @formats[format.name]=format
+      else
+        raise ArgumentError, "format should be a Workboot::Format"
+      end      
+    end
+    
+    def formats
+      @formats
+    end
+    
+    def create_or_find_format_by name, variant=:default
+      fs = @formats[name]
+      fs = @formats[name] = {} if fs.nil?
+      f = fs[variant]
+      if f.nil? 
+        f = Workbook::Format.new 
+        if variant != :default and fs[:default]
+          f = fs[:default].clone
+        end
+        @formats[name][variant] = f
+      end
+      return @formats[name][variant]
+    end
+    
+    
+	end
+end

data/lib/workbook/writers/csv_table_writer.rb

@@ -0,0 +1,23 @@
+require 'CSV'
+
+module Workbook
+  module Writers
+    module CsvTableWriter
+      def to_csv options={}
+        csv = ""
+        options = {}.merge options
+        self.each_with_index do |r, ri|
+          line=nil
+          begin
+            line = CSV::generate_line(r.collect{|c| c.value if c},{:row_sep=>""})
+          rescue TypeError
+            line = CSV::generate_line(r.collect{|c| c.value if c})
+          end
+          csv += "#{line}\n"
+        end
+        csv
+      end
+      
+    end
+  end
+end

data/lib/workbook/writers/xls_writer.rb

@@ -0,0 +1,172 @@
+require 'spreadsheet'
+
+module Workbook
+  module Writers
+    module XlsWriter
+      # TODO: find better way to dupe
+      XLS_COLORS = {:xls_color_1=>'#000000',
+                    :xls_color_2=>'#FFFFFF',
+                    :xls_color_3=>'#FF0000',
+                    :xls_color_4=>'#00FF00',
+                    :xls_color_5=>'#0000FF',
+                    :xls_color_6=>'#FFFF00',
+                    :xls_color_7=>'#FF00FF',
+                    :xls_color_8=>'#00FFFF',
+                    :xls_color_9=>'#800000',
+                    :xls_color_10=>'#008000',
+                    :xls_color_11=>'#000080',
+                    :xls_color_12=>'#808000',
+                    :xls_color_13=>'#800080',
+                    :xls_color_14=>'#008080',
+                    :xls_color_15=>'#C0C0C0',
+                    :xls_color_16=>'#808080',
+                    :xls_color_17=>'#9999FF',
+                    :xls_color_18=>'#993366',
+                    :xls_color_19=>'#FFFFCC',
+                    :xls_color_20=>'#CCFFFF',
+                    :xls_color_21=>'#660066',
+                    :xls_color_22=>'#FF8080',
+                    :xls_color_23=>'#0066CC',
+                    :xls_color_24=>'#CCCCFF',
+                    :xls_color_25=>'#000080',
+                    :xls_color_26=>'#FF00FF',
+                    :xls_color_27=>'#FFFF00',
+                    :xls_color_28=>'#00FFFF',
+                    :xls_color_29=>'#800080',
+                    :xls_color_30=>'#800000',
+                    :xls_color_31=>'#008080',
+                    :xls_color_32=>'#0000FF',
+                    :xls_color_33=>'#00CCFF',
+                    :xls_color_34=>'#CCFFFF',
+                    :xls_color_35=>'#CCFFCC',
+                    :xls_color_36=>'#FFFF99',
+                    :xls_color_37=>'#99CCFF',
+                    :xls_color_38=>'#FF99CC',
+                    :xls_color_39=>'#CC99FF',
+                    :xls_color_40=>'#FFCC99',
+                    :xls_color_41=>'#3366FF',
+                    :xls_color_42=>'#33CCCC',
+                    :xls_color_43=>'#99CC00',
+                    :xls_color_44=>'#FFCC00',
+                    :xls_color_45=>'#FF9900',
+                    :xls_color_46=>'#FF6600',
+                    :xls_color_47=>'#666699',
+                    :xls_color_48=>'#969696',
+                    :xls_color_49=>'#003366',
+                    :xls_color_50=>'#339966',
+                    :xls_color_51=>'#003300',
+                    :xls_color_52=>'#333300',
+                    :xls_color_53=>'#993300',
+                    :xls_color_54=>'#993366',
+                    :xls_color_55=>'#333399',
+                    :xls_color_56=>'#333333',
+                    :black=>'#000000',
+                    :white=>'#FFFFFF',
+                    :red=>'#FF0000',
+                    :green=>'#00FF00',
+                    :blue=>'#0000FF',
+                    :yellow=>'#FFFF00',
+                    :magenta=>'#FF00FF',
+                    :cyan=>'#00FFFF',
+                    :border=>'#FFFFFF',
+                    :text=>'#000000',
+                    :lime=>'#00f94c'}
+      
+      def to_xls options={}
+        options = {:rewrite_header=>default_rewrite_header?}.merge options
+        book = init_spreadsheet_template
+        self.each_with_index do |s,si|
+          xls_sheet = book.worksheet si
+          xls_sheet = book.create_worksheet if xls_sheet == nil
+          s.table.each_with_index do |r, ri|
+            write_row = false
+            if r.header?
+              if options[:rewrite_header] == true
+                write_row = true
+              end
+            else
+              write_row = true
+            end
+            if write_row
+              xls_sheet.row(ri).height= r.format[:height] if r.format
+              r.each_with_index do |c, ci|
+                if c
+                  if r.header?
+                    xls_sheet.columns[ci] ||= Spreadsheet::Column.new(ci,nil)
+                    xls_sheet.columns[ci].width= c.format[:width]
+                  end
+                  xls_sheet.row(ri)[ci] = c.value
+                  xls_sheet.row(ri).set_format(ci, format_to_xls_format(c.format))
+                end
+              end
+            end
+          end
+        end
+        book
+      end
+      
+      def format_to_xls_format f
+        xlsfmt = nil
+        unless f.is_a? Workbook::Format
+          f = Workbook::Format.new f
+        end
+        xlsfmt = f.return_raw_for Spreadsheet::Format
+        unless xlsfmt
+          xlsfmt=Spreadsheet::Format.new :weight=>f[:font_weight]
+          xlsfmt.rotation = f[:rotation] if f[:rotation] 
+          xlsfmt.pattern_fg_color = html_color_to_xls_color(f[:background_color]) if html_color_to_xls_color(f[:background_color])
+          xlsfmt.pattern = 1 if html_color_to_xls_color(f[:background_color])
+          xlsfmt.number_format = strftime_to_ms_format(f[:number_format]) if f[:number_format]
+          xlsfmt.text_direction = f[:text_direction] if f[:text_direction]
+          xlsfmt.font.name = f[:font_family].split.first if f[:font_family]
+          xlsfmt.font.family = f[:font_family].split.last if f[:font_family]
+          xlsfmt.font.color = html_color_to_xls_color(f[:color]) if f[:color]
+          f.add_raw xlsfmt
+        end
+        return xlsfmt
+      end
+      
+      def html_color_to_xls_color hex
+        XLS_COLORS.each do |k,v|
+          return k if (v == hex or (hex and hex != "" and k == hex.to_sym))
+        end
+        return nil
+      end
+      
+      def strftime_to_ms_format numberformat
+        return nil if numberformat.nil?
+        numberformat.gsub('%Y','yyyy').gsub('%A','dddd').gsub('%B','mmmm').gsub('%a','ddd').gsub('%b','mmm').gsub('%y','yy').gsub('%d','dd').gsub('%m','mm').gsub('%y','y').gsub('%y','%%y').gsub('%e','d')
+      end
+      
+      def write_to_xls options={}
+        filename = options[:filename] ? options[:filename] : "#{title}.xls"
+        if to_xls(options).write(filename)
+          return filename
+        end
+      end
+    
+      def xls_sheet a
+        if xls_template.worksheet(a)
+          return xls_template.worksheet(a)
+        else
+          xls_template.create_worksheet
+          self.xls_sheet a
+        end
+      end
+      
+      def xls_template
+        return template.raws[Spreadsheet::Excel::Workbook] ? template.raws[Spreadsheet::Excel::Workbook] : template.raws[Spreadsheet::Workbook]
+      end
+            
+      def init_spreadsheet_template
+        if self.xls_template.is_a? Spreadsheet::Workbook
+          return self.xls_template
+        else
+          t = Spreadsheet::Workbook.new
+          template.add_raw t
+          return t
+        end
+      end
+    end
+  end
+end

data/lib/workbook.rb

@@ -0,0 +1,14 @@
+$KCODE="u"
+require 'workbook/book'
+require 'workbook/sheet'
+require 'workbook/table'
+require 'workbook/row'
+require 'workbook/cell'
+require 'workbook/format'
+require 'workbook/template'
+
+module Workbook
+  class << self
+
+  end
+end