datashift 0.16.0 → 0.40.0

data/lib/applications/jexcel_file.rb

@@ -1,31 +1,29 @@
 # Copyright:: Autotelik Media Ltd
 # Author ::   Tom Statter
 # Date ::     July 2010
-# License::   
+# License::
 #
-# A wrapper around creating and directly manipulating Excel files. 
+# A wrapper around creating and directly manipulating Excel files.
 #
 # i.e Create and populate XSL files
-# 
+#
 # jar added to class path in manifest - 'poi-3.5-beta4-20081128.jar'
 #
-if(DataShift::Guards::jruby?)
-  
+if DataShift::Guards.jruby?
+
   require 'java'
-  
-  require 'excel_base'
-  require 'ruby_poi_translations'
-  
+  require 'poi-3.7-20101029.jar'
+
+  require_relative 'ruby_poi_translations'
+
   class JExcelFile
-    
-    include ExcelBase
-    
+
     include RubyPoiTranslations
     extend RubyPoiTranslations
-        
+
     include Enumerable
-    
-    include_class 'org.apache.poi.hssf.util.HSSFColor'
+
+    java_import 'org.apache.poi.hssf.util.HSSFColor'
     java_import 'org.apache.poi.poifs.filesystem.POIFSFileSystem'
 
     java_import 'org.apache.poi.hssf.usermodel.HSSFCell'
@@ -36,73 +34,73 @@ if(DataShift::Guards::jruby?)
     java_import 'org.apache.poi.hssf.usermodel.HSSFRichTextString'
 
     attr_accessor :workbook, :row, :date_style
-    attr_reader   :sheet, :current_sheet_index
+    attr_reader :sheet, :current_sheet_index
 
     # NOTE: this is the POI 3.7 HSSF maximum rows
     @@maxrows = 65535
 
     def self.maxrows
-      return @@maxrows
+      @@maxrows
     end
-  
+
     def self.date_format
-      HSSFDataFormat.getBuiltinFormat("m/d/yy h:mm")
+      HSSFDataFormat.getBuiltinFormat('m/d/yy h:mm')
     end
-    
-    def self.open(filename)
-      HSSFWorkbook.new(FileInputStream.new(filename))
+
+    def self.open(file_name)
+      HSSFWorkbook.new(FileInputStream.new(file_name))
     end
-    
+
     # NOTES :
     #   The HSSFWorkbook uses 0 based indexes
-  
-    def initialize()
+
+    def initialize
       @workbook = HSSFWorkbook.new
-      
+
       @sheet = nil
       @current_sheet_index = 0
-      
+
       # The @patriarchs hash is a workaround because HSSFSheet.getDrawingPatriarch()
       # causes a lot of issues (if it doesn't throw an exception!)
-      @patriarchs = Hash.new
-      
+      @patriarchs = {}
+
       @date_style = nil
     end
-   
-    def open(filename)
-      @workbook = JExcelFile.open(filename)
-      
+
+    def open(file_name)
+      @workbook = JExcelFile.open(file_name)
+
       @date_style = @workbook.createCellStyle
-      @date_style.setDataFormat( JExcelFile::date_format )
-      
+      @date_style.setDataFormat( JExcelFile.date_format )
+
       activate_sheet(0)
       @workbook
     end
-    
-    # Create and return a new worksheet. 
+
+    # Create and return a new worksheet.
     # Not set to the active worksheet
-    
+
     def create_worksheet( options = {} )
       sheet_name = options[:name]
-      
-      @workbook = HSSFWorkbook.new() if @workbook.nil?
-            
-      unless(sheet_name)
+
+      @workbook = HSSFWorkbook.new if @workbook.nil?
+
+      if sheet_name
+
+        name = sanitize_sheet_name( sheet_name )
+
+        return create_sheet_and_set_styles( name ) if @workbook.getSheetIndex(name) < 0 # Check sheet doesn't already exist
+
+        activate_sheet(name)
+
+      else
         i = 0
+        # there is no hard limit to no of sheets in Excel but at some point you will run out of memory!
         begin
           sheet_name = "Worksheet#{i += 1}"
-        end while(@workbook.getSheetIndex(sheet_name) >= 0) # there is no hard limit to no of sheets in Excel but at some point you will run out of memory!
-        
-        return create_sheet_and_set_styles( sheet_name )
-      else 
-        
-        name = sanitize_sheet_name( sheet_name )
+        end while(@workbook.getSheetIndex(sheet_name) >= 0)
 
-        if (@workbook.getSheetIndex(name) < 0)  #Check sheet doesn't already exist
-          return create_sheet_and_set_styles( name )
-        else
-          activate_sheet(name)
-        end
+        return create_sheet_and_set_styles( sheet_name )
       end
     end
 
@@ -110,11 +108,11 @@ if(DataShift::Guards::jruby?)
     # If no such sheet return current sheet
     def activate_sheet(term)
 
-      if(@workbook)
+      if @workbook
         x = term.is_a?(String) ? @workbook.getSheetIndex(term.to_java(java.lang.String)) : term
         @sheet = worksheet(x)
-        
-        if( @sheet )
+
+        if @sheet
           @current_sheet_index = x
           @workbook.setActiveSheet(@current_sheet_index)
           @sheet = @workbook.getSheetAt(@current_sheet_index)
@@ -125,14 +123,14 @@ if(DataShift::Guards::jruby?)
     end
 
     # Return a sheet by index
-    def worksheet( index )  
-      if(@workbook)
+    def worksheet( index )
+      if @workbook
         x = index.is_a?(String) ? @workbook.getSheetIndex(index.to_java(java.lang.String)) : index
-        return @workbook.getSheetAt(x) 
+        return @workbook.getSheetAt(x)
       end
       nil
     end
-    
+
     def worksheets
       (0...@workbook.getNumberOfSheets).collect { |i| @workbook.getSheetAt(i) }
     end
@@ -140,89 +138,95 @@ if(DataShift::Guards::jruby?)
     # Create new row (indexing in line with POI usage, start 0)
     def create_row(index)
       return nil if @sheet.nil?
-      raise "BAD INDEX: Row indexing starts at 0" if(index < 0)
+      raise 'BAD INDEX: Row indexing starts at 0' if index < 0
       @row = @sheet.createRow(index)
       @row
     end
-    
+
     def num_rows
       @sheet.getPhysicalNumberOfRows
     end
 
     # Process each row. Row type is org.apache.poi.hssf.usermodel.HSSFRow
-        
+
     # Currently ignores skip argument - TODO - this is how spreadsheet gem works
-    # #each iterates over all used Rows (from the first used Row until but omitting the first unused Row, see also #dimensions)
-    # If the argument skip is given, 
-    # #each iterates from that row until but omitting the first unused Row, effectively skipping the first skip Rows from the top of the Worksheet. 
+    # #each iterates over all used Rows (from the first used Row until
+    # but omitting the first unused Row, see also #dimensions)
+    # If the argument skip is given,
+    # #each iterates from that row until but omitting the first unused Row,
+    # effectively skipping the first skip Rows from the top of the Worksheet.
 
-    def each(skip = nil, &block) 
+    def each(_skip = nil, &block)
       @sheet.rowIterator.each(&block)
     end
 
     def row( index )
       @sheet.getRow(index) || create_row(index)
     end
-         
+
     # Get the enriched value of the Cell at row, column.
-    def cell(row, column) 
+    def cell(row, column)
       row = row(row)
-      cell_value( row.get_or_create_cell( column )  )
+      cell_value( row.get_or_create_cell( column ) )
     end
 
-    # Get the enriched value of the Cell at row, column. 
-    def [](row, column) 
-      cell(row, column) 
+    # Get the enriched value of the Cell at row, column.
+    def [](row, column)
+      cell(row, column)
     end
 
     def []=(row, column, value)
       set_cell(row, column, value)
     end
-    
+
     def set_cell(row, column, value)
       @row = row(row)
-      
+
       @row[column] = value
     end
-    
-    def write( filename = nil )
-      filename.nil? ? file = @filepath : file = filename
+
+    def sanitize_sheet_name( name )
+      name.gsub(/[\[\]:\*\/\\\?]/, '')
+    end
+
+    def write( file_name = nil )
+      file = file_name.nil? ? @filepath : file_name
       out = FileOutputStream.new(file)
       @workbook.write(out) unless @workbook.nil?
       out.close
     end
 
-    alias_method :save, :write
-    
-    def save_to_text( filename )
-      File.open( filename, 'w') {|f| f.write(to_s) }
+    alias save write
+
+    def save_to_text( file_name )
+      File.open( file_name, 'w') { |f| f.write(to_s) }
     end
 
     def to_s
       outs = ByteArrayOutputStream.new
-      @workbook.write(outs);
-      outs.close();
+      @workbook.write(outs)
+      outs.close
       String.from_java_bytes(outs.toByteArray)
     end
-    
+
     private
-       
+
     def create_sheet_and_set_styles( sheet_name )
-      
+
       name = sanitize_sheet_name(sheet_name)
-      
+
       @sheet = @workbook.createSheet( name )
 
-      @patriarchs.store(name, @sheet.createDrawingPatriarch())
-     
+      @patriarchs.store(name, @sheet.createDrawingPatriarch)
+
       @date_style = @workbook.createCellStyle
-      @date_style.setDataFormat( JExcelFile::date_format )
+      @date_style.setDataFormat( JExcelFile.date_format )
       @sheet
     end
-    
+
   end
-  
+
   require 'jexcel_file_extensions'
   require 'apache_poi_extensions'
-  
-end
+
+end

data/lib/applications/jexcel_file_extensions.rb

@@ -1,143 +1,136 @@
 # Copyright:: Autotelik Media Ltd
 # Author ::   Tom Statter
 # Date ::     July 2010
-# License::   
+# License::
 #
-# Details::   A helper module providing shortcuts for manipulating Excel files. 
-# 
+# Details::   A helper module providing shortcuts for manipulating Excel files.
+#
+
+if DataShift::Guards.jruby?
 
-if(DataShift::Guards::jruby?)
-    
   JExcelFile.class_eval do
-    
-    include_class 'org.apache.poi.hssf.util.HSSFColor'
-    
+    java_import 'org.apache.poi.hssf.util.HSSFColor'
+
     # Return the current active sheet
     # If term supplied find sheet and set active
-    # 
+    #
     def active_worksheet(term = nil)
-      if( term.nil? )
+      if  term.nil?
         @sheet ||= @workbook.getSheetAt(@current_sheet_index)
-      else 
+      else
         activate_sheet(term)
-      end if(@workbook)
+      end if @workbook
       @sheet
     end
-    
-    #  Populate a row  of cells with data in an array 
+
+    #  Populate a row  of cells with data in an array
     #  where the co-ordinates relate to row/column start position
-    #    
+    #
     def set_row( row, col, data, sheet_num = nil)
-    
+
       sheet(sheet_num)
-      
+
       create_row(row)
-      
+
       column = col
       data.each do |datum|
         set_cell(row, column, datum)
         column += 1
       end
     end
-    
-    def headers=( headers )  
-      headers.each_with_index do |header, column|
-        self[0, column] = (header || "")
-      end
-    end
-    
+
     # Convert array into a header row
     def set_headers(headers, apply_style = nil)
       create_row(0)
       return if headers.empty?
-    
-      style = apply_style || header_style()
 
-      headers.each_with_index do |datum, i|
+      style = apply_style || header_style
+
+      headers.sources.each_with_index do |datum, i|
         c = @row.createCell(i, poi_cell_type(datum))
         c.setCellValue(datum)
         c.setCellStyle(style)
       end
     end
-    
+
     attr_writer :header_background_colour
-    
+
     def header_background_colour
-      @header_background_colour ||= org.apache.poi.hssf.util.HSSFColor::LIGHT_CORNFLOWER_BLUE.index        
+      @header_background_colour ||= org.apache.poi.hssf.util.HSSFColor::LIGHT_CORNFLOWER_BLUE.index
     end
-    
+
     def header_style
       return @header_style if @header_style
-      @header_style = @workbook.createCellStyle();
+      @header_style = @workbook.createCellStyle
       @header_style.setBorderTop(6) # double lines border
       @header_style.setBorderBottom(1) # single line border
       @header_style.setFillBackgroundColor(header_background_colour)
-      
+
       @header_style
     end
-    
+
     def add_comment( cell, text )
-      raise "Please supply valid HSSFCell" unless cell.respond_to?('setCellComment')
+      raise 'Please supply valid HSSFCell' unless cell.respond_to?('setCellComment')
       return if @sheet.nil?
 
-      patriarch = @patriarchs[@sheet.getSheetName()]
+      patriarch = @patriarchs[@sheet.getSheetName]
 
-      anchor = HSSFClientAnchor.new(100, 50, 100, 50, cell.getColumnIndex(), cell.getRowIndex(), cell.getColumnIndex()+3, cell.getRowIndex()+4)
+      anchor = HSSFClientAnchor.new(100, 50, 100, 50, cell.getColumnIndex, cell.getRowIndex, cell.getColumnIndex + 3, cell.getRowIndex + 4)
       comment = patriarch.createCellComment(anchor)
 
       comment_text = HSSFRichTextString.new(text)
       comment.setString(comment_text)
-      comment.setAuthor("Mapping")
+      comment.setAuthor('Mapping')
 
       cell.setCellComment(comment)
     end
-    
+
     # Get a percentage style
-    def getPercentStyle()
-      if (@percentCellStyle.nil? && @workbook)
-        @percentCellStyle = @workbook.createCellStyle();
-        @percentCellStyle.setDataFormat(HSSFDataFormat.getBuiltinFormat("0.00%"));
+    def getPercentStyle
+      if @percentCellStyle.nil? && @workbook
+        @percentCellStyle = @workbook.createCellStyle
+        @percentCellStyle.setDataFormat(HSSFDataFormat.getBuiltinFormat('0.00%'))
       end
-      return @percentCellStyle
+      @percentCellStyle
     end
 
     # Auto size either the given column index or all columns
     def autosize(column = nil)
       return if @sheet.nil?
-      if (column.kind_of? Integer)
+      if column.is_a? Integer
         @sheet.autoSizeColumn(column)
       else
-        @sheet.getRow(0).cellIterator.each{|c| @sheet.autoSizeColumn(c.getColumnIndex)}
+        @sheet.getRow(0).cellIterator.each { |c| @sheet.autoSizeColumn(c.getColumnIndex) }
       end
     end
-    
-    def create_freeze_pane(row=1, column=0)
+
+    def create_freeze_pane(row = 1, column = 0)
       return if @sheet.nil?
       @sheet.createFreezePane(row, column)
     end
 
-    
     # Use execute to run sql query provided
     # and write to a csv file (path required)
     # header row is optional but default is on
     # Auto mapping of specified columns is optional
     # @mappings is a hash{column => map} of columns to a map{old_value => new_value}
-    def results_to_sheet( results, sheet, mappings=nil, header=true)
+    def results_to_sheet( results, sheet, mappings = nil, header = true)
       numrows = results.length
       sheet_name = sheet
 
-      if (numrows == 0)
+      if numrows == 0
         log :info, "WARNING - results are empty nothing written to sheet: #{sheet}"
         return
       end
 
-      #Check if we need to split the results into seperate sheets
-      if (numrows > @@maxrows )
+      # Check if we need to split the results into seperate sheets
+      if numrows > @@maxrows
         startrow = 0
-        while (numrows > 0)
+        while numrows > 0
           # Split the results and write to a new sheet
-          next_results = results.slice(startrow, @@maxrows > numrows ? numrows : @@maxrows)
-          self.results_to_sheet(next_results, "#{sheet_name}", mappings, header) if next_results
+          slice_at = @@maxrows > numrows ? numrows : @@maxrows
+          next_results = results.slice(startrow, slice_at)
+          results_to_sheet(next_results, sheet_name.to_s, mappings, header) if next_results
 
           # Increase counters
           numrows -= next_results.length
@@ -147,43 +140,43 @@ if(DataShift::Guards::jruby?)
       else
         log :info, "Writting #{numrows} rows to : #{sheet_name}"
         # Create required sheet
-        self.create(sheet)
+        create(sheet)
 
-        row_index = self.num_rows
+        row_index = num_rows
         # write header line
-        if (header && row_index==0 )
+        if header && row_index == 0
           header_row = @sheet.createRow(row_index)
           cell_index = 0
-          results[0].keys.each{ |h|
-            header_row.createCell(cell_index).setCellValue("#{h}")
-            @sheet.setDefaultColumnStyle(cell_index, self.getPercentStyle) if "#{h}".include? '%'
+          results[0].keys.each do |h|
+            header_row.createCell(cell_index).setCellValue(h.to_s)
+            @sheet.setDefaultColumnStyle(cell_index, getPercentStyle) if h.to_s.include? '%'
             cell_index += 1
-          }
+          end
           # Freeze the header row
           @sheet.createFreezePane( 0, 1, 0, 1 )
           row_index += 1
         end
 
         # write_results
-        results.each{ |row|
+        results.each do |row|
           sheet_row = @sheet.createRow(row_index)
           cell_index = 0
-          row.each{|k,v|
-            celltype = v.kind_of?(Numeric) ? HSSFCell::CELL_TYPE_NUMERIC : HSSFCell::CELL_TYPE_STRING
+          row.each do |_k, v|
+            celltype = v.is_a?(Numeric) ? HSSFCell::CELL_TYPE_NUMERIC : HSSFCell::CELL_TYPE_STRING
             cell = sheet_row.createCell(cell_index, celltype)
 
-            v.nil? ? value = "<NIL>" : value = v
+            value = v.nil? ? '<NIL>' : v
 
             cell.setCellValue(value)
 
-            cell_index +=1
-          }
-          #puts "#{sheet}: written row #{row_index}"
-          row_index +=1
-        }
+            cell_index += 1
+          end
+          # puts "#{sheet}: written row #{row_index}"
+          row_index += 1
+        end
       end
     end
-    
+
     module ExcelHelper
       require 'java'
 
@@ -197,16 +190,16 @@ if(DataShift::Guards::jruby?)
 
       # ActiveRecord Helper - Export model data to XLS file format
       #
-      def to_xls(items=[])
+      def to_xls(items = [])
 
         @excel = ExcelFile.new(items[0].class.name)
-    
+
         @excel.create_row(0)
-  
+
         sheet = @excel.sheet
 
         # header row
-        if !items.empty?
+        unless items.empty?
           row = sheet.createRow(0)
           cell_index = 0
           items[0].class.columns.each do |column|
@@ -218,16 +211,16 @@ if(DataShift::Guards::jruby?)
         # value rows
         row_index = 1
         items.each do |item|
-          row = sheet.createRow(row_index);
+          row = sheet.createRow(row_index)
 
           cell_index = 0
           item.class.columns.each do |column|
             cell = row.createCell(cell_index)
-            if column.sql_type =~ /date/ then
+            if column.sql_type =~ /date/
               millis = item.send(column.name).to_f * 1000
               cell.setCellValue(Date.new(millis))
-              cell.setCellStyle(dateStyle);
-            elsif column.sql_type =~ /int/ then
+              cell.setCellStyle(dateStyle)
+            elsif column.sql_type =~ /int/
               cell.setCellValue(item.send(column.name).to_i)
             else
               value = item.send(column.name)
@@ -241,4 +234,4 @@ if(DataShift::Guards::jruby?)
       end
     end
   end
-end
+end