roo 1.2.3 → 1.3.5

data/lib/roo/google.rb

@@ -1,4 +1,5 @@
 require 'gdata/spreadsheet'
+require 'xml'
 
 # overwrite some methods from the gdata-gem:
 module GData
@@ -45,7 +46,7 @@ XML
     def add_to_cell_roo(row,col,value, sheet_no=1)
       save_entry_roo(entry_roo(value,row,col), sheet_no)
     end
-
+   
     #-- new
     def get_one_sheet
       path = "/feeds/cells/#{@spreadsheet_id}/1/private/full"
@@ -70,11 +71,13 @@ XML
       doc = Hpricot(request(path))
       return doc
     end
+
   end # class
 end # module
 
 class Google < GenericSpreadsheet
-
+  attr_accessor :date_format, :datetime_format
+  
   # Creates a new Google spreadsheet object.
   def initialize(spreadsheetkey,user=nil,password=nil)
     @filename = spreadsheetkey
@@ -88,8 +91,8 @@ class Google < GenericSpreadsheet
       password = ENV['GOOGLE_PASSWORD']
     end
     @default_sheet = nil
-    @cell = Hash.new
-    @cell_type = Hash.new
+    @cell = Hash.new {|h,k| h[k]=Hash.new}
+    @cell_type = Hash.new {|h,k| h[k]=Hash.new}
     @formula = Hash.new
     @first_row = Hash.new
     @last_row = Hash.new
@@ -97,10 +100,12 @@ class Google < GenericSpreadsheet
     @last_column = Hash.new
     @cells_read = Hash.new
     @header_line = 1
-
+    @date_format = '%d/%m/%Y'
+    @datetime_format = '%d/%m/%Y %H:%M:%S' 
+    @time_format = '%H:%M:%S'
     @gs = GData::Spreadsheet.new(spreadsheetkey)
     @gs.authenticate(user, password)
-
+    @sheetlist = @gs.sheetlist
     #-- ----------------------------------------------------------------------
     #-- TODO: Behandlung von Berechtigungen hier noch einbauen ???
     #-- ----------------------------------------------------------------------
@@ -112,32 +117,42 @@ class Google < GenericSpreadsheet
 
   # returns an array of sheet names in the spreadsheet
   def sheets
-    return @gs.sheetlist
+    @sheetlist
   end
 
-  # is String a date with format DD/MM/YYYY
-  def Google.date?(string)
-    return false if string.class == Float
-    return true if string.class == Date
-    return string.strip =~ /^([0-9]+)\/([0-9]+)\/([0-9]+)$/
+  def date?(string)
+    begin
+      Date.strptime(string, @date_format)
+      true
+    rescue
+      false
+    end
   end
 
   # is String a time with format HH:MM:SS?
-  def Google.time?(string)
-    return false if string.class == Float
-    return true if string.class == Date
-    return string.strip =~ /^([0-9]+):([0-9]+):([0-9]+)$/
+  def time?(string)
+    begin
+      DateTime.strptime(string, @time_format)
+      true
+    rescue
+      false
+    end
   end
 
-  # is String a date+time with format DD/MM/YYYY HH:MM:SS
-  def Google.datetime?(string)
-    return false if string.class == Float
-    return true if string.class == Date
-    return string.strip =~ /^([0-9]+)\/([0-9]+)\/([0-9]+)\ ([0-9]+):([0-9]+):([0-9]+)$/
+  def datetime?(string)
+    begin
+      DateTime.strptime(string, @datetime_format)
+      true
+    rescue
+      false
+    end
   end
 
+  def numeric?(string)
+    string =~ /^[0-9]+[\.]*[0-9]*$/
+  end
 
-  def Google.timestring_to_seconds(value)
+  def timestring_to_seconds(value)
     hms = value.split(':')
     hms[0].to_i*3600 + hms[1].to_i*60 + hms[2].to_i
   end
@@ -151,24 +166,21 @@ class Google < GenericSpreadsheet
     check_default_sheet #TODO: 2007-12-16
     read_cells(sheet) unless @cells_read[sheet]
     row,col = normalize(row,col)
+    value = @cell[sheet]["#{row},#{col}"]
     if celltype(row,col,sheet) == :date
-      yyyy,mm,dd = @cell[sheet]["#{row},#{col}"].split('-')
       begin
-        return Date.new(yyyy.to_i,mm.to_i,dd.to_i)
+        return  Date.strptime(value, @date_format)
       rescue ArgumentError
-        raise "Invalid date parameter: #{yyyy}, #{mm}, #{dd}"
+        raise "Invalid Date #{sheet}[#{row},#{col}] #{value} using format '{@date_format}'"
       end
     elsif celltype(row,col,sheet) == :datetime
       begin
-        date_part,time_part = @cell[sheet]["#{row},#{col}"].split(' ')
-        yyyy,mm,dd = date_part.split('-')
-        hh,mi,ss = time_part.split(':')
-        return DateTime.civil(yyyy.to_i,mm.to_i,dd.to_i,hh.to_i,mi.to_i,ss.to_i)
+        return  DateTime.strptime(value, @datetime_format)
       rescue ArgumentError
-        raise "Invalid date parameter: #{yyyy}, #{mm}, #{dd}, #{hh}, #{mi}, #{ss}"
+        raise "Invalid DateTime #{sheet}[#{row},#{col}] #{value} using format '{@datetime_format}'"
       end
     end 
-    return @cell[sheet]["#{row},#{col}"]
+    return value
   end
 
   # returns the type of a cell:
@@ -291,8 +303,15 @@ class Google < GenericSpreadsheet
     end
     row,col = normalize(row,col)
     @gs.add_to_cell_roo(row,col,value,sheet_no)
+    # re-read the portion of the document that has changed
+    if @cells_read[sheet]
+      key = "#{row},#{col}"
+      (value, value_type) = determine_datatype(value.to_s)
+      @cell[sheet][key] = value 
+      @cell_type[sheet][key] = value_type 
+    end
   end
-
+  
   # returns the first non-empty row in a sheet
   def first_row(sheet=nil)
     sheet = @default_sheet unless sheet
@@ -335,77 +354,52 @@ class Google < GenericSpreadsheet
 
   private
 
-  # read all cells in a sheet
+  # read all cells in a sheet. 
   def read_cells(sheet=nil)
     sheet = @default_sheet unless sheet
     raise RangeError, "illegal sheet <#{sheet}>" unless sheets.index(sheet)
     sheet_no = sheets.index(sheet)+1
-    doc = @gs.fulldoc(sheet_no)
-    (doc/"gs:cell").each {|item|
+    xml = @gs.fulldoc(sheet_no).to_s
+    doc = XML::Parser.string(xml).parse
+    doc.find("//*[local-name()='cell']").each do |item|
       row = item['row']
       col = item['col']
-      value = item['inputvalue']
-      numericvalue = item['numericvalue']
-      if value[0,1] == '='
-        formula = value
+      key = "#{row},#{col}"
+      string_value =  item['inputvalue'] ||  item['inputValue'] 
+      numeric_value = item['numericvalue']  ||  item['numericValue'] 
+      (value, value_type) = determine_datatype(string_value, numeric_value)
+      @cell[sheet][key] = value unless value == "" or value == nil
+      @cell_type[sheet][key] = value_type 
+      @formula[sheet] = {} unless @formula[sheet]
+      @formula[sheet][key] = string_value if value_type == :formula
+    end
+    @cells_read[sheet] = true
+  end
+  
+  def determine_datatype(val, numval=nil)
+    if val[0,1] == '='
+      ty = :formula
+      if numeric?(numval)
+        val = numval.to_f
       else
-        formula = nil
+        val = numval
       end
-      @cell_type[sheet] = {} unless @cell_type[sheet]
-      if formula
-        ty = :formula
-        if numeric?(numericvalue)
-          value = numericvalue.to_f
-        else
-          value = numericvalue
-        end
-      elsif Google.date?(value)
-        ty = :date
-      elsif Google.datetime?(value)
+    else
+      if datetime?(val)
         ty = :datetime
-      elsif numeric?(value) # or o.class ???
+      elsif date?(val)
+        ty = :date
+      elsif numeric?(val) 
         ty = :float
-        value = value.to_f
-      elsif Google.time?(value)
+        val = val.to_f
+      elsif time?(val)
         ty = :time
-        value = Google.timestring_to_seconds(value)
+        val = timestring_to_seconds(val)
       else
         ty = :string
       end
-      key = "#{row},#{col}"
-      @cell[sheet] = {} unless @cell[sheet]
-      if ty == :date
-        dd,mm,yyyy = value.split('/')
-        @cell[sheet][key] = sprintf("%04d-%02d-%02d",yyyy.to_i,mm.to_i,dd.to_i)
-      elsif ty == :datetime
-        date_part,time_part = value.split(' ')
-        dd,mm,yyyy = date_part.split('/')
-        hh,mi,ss = time_part.split(':')
-        @cell[sheet][key] = sprintf("%04d-%02d-%02d %02d:%02d:%02d",
-          yyyy.to_i,mm.to_i,dd.to_i,hh.to_i,mi.to_i,ss.to_i)
-      else
-        @cell[sheet][key] = value unless value == "" or value == nil
-      end
-      @cell_type[sheet][key] = ty # Openoffice.oo_type_2_roo_type(vt)
-      @formula[sheet] = {} unless @formula[sheet]
-      @formula[sheet][key] = formula  if formula
-    }
-    @cells_read[sheet] = true
+    end  
+    return val, ty 
   end
-
-  def numeric?(string)
-    string =~ /^[0-9]+[\.]*[0-9]*$/
-  end
-
-  # convert string DD/MM/YYYY into a Date-object
-  #TODO: was ist mit verschiedenen Typen der Datumseingabe bei Google?
-  def Google.to_date(string)
-    if string.strip =~ /^([0-9]+)\/([0-9]+)\/([0-9]+)$/
-      return Date.new($3.to_i,$2.to_i,$1.to_i)
-    else
-      return nil
-    end
-  end
-
-
+  
 end # class

data/lib/roo/openoffice.rb

@@ -1,10 +1,9 @@
-
-require 'rubygems'
-require 'rexml/document'
+require 'xml'
 require 'fileutils'
 require 'zip/zipfilesystem'
 require 'date'
 require 'base64'
+require 'cgi'
 
 class Openoffice < GenericSpreadsheet
 
@@ -37,7 +36,7 @@ class Openoffice < GenericSpreadsheet
       @file_nr = @@nr
       extract_content
       file = File.new(File.join(@tmpdir, @file_nr.to_s+"_roo_content.xml"))
-      @doc = REXML::Document.new file
+      @doc = XML::Parser.io(file).parse
       file.close
     ensure
       #if ENV["roo_local"] != "thomas-p"
@@ -56,6 +55,9 @@ class Openoffice < GenericSpreadsheet
     @last_row = Hash.new
     @first_column = Hash.new
     @last_column = Hash.new
+    @style = Hash.new
+    @style_defaults = Hash.new { |h,k| h[k] = [] }
+    @style_definitions = Hash.new 
     @header_line = 1
   end
 
@@ -107,7 +109,32 @@ class Openoffice < GenericSpreadsheet
     row,col = normalize(row,col)
     formula(row,col) != nil
   end
+  
+  class Font
+    attr_accessor :bold, :italic, :underline
+    
+    def bold? 
+      @bold == 'bold'
+    end
+
+    def italic? 
+      @italic == 'italic'
+    end
+    
+    def underline? 
+      @underline != nil
+    end
+  end
 
+  # Given a cell, return the cell's style 
+  def font(row, col, sheet=nil)
+    sheet = @default_sheet unless sheet
+    read_cells(sheet) unless @cells_read[sheet]
+    row,col = normalize(row,col)
+    style_name = @style[sheet][[row,col]] || @style_defaults[sheet][col - 1] || 'Default'
+    @style_definitions[style_name]
+  end 
+  
   # set a cell to a certain value
   # (this will not be saved back to the spreadsheet file!)
   def set(row,col,value,sheet=nil) #:nodoc:
@@ -145,31 +172,14 @@ class Openoffice < GenericSpreadsheet
     end
   end
 
-  # returns an array of sheet names in the spreadsheet
   def sheets
     return_sheets = []
-    oo_document_count = 0
-    @doc.each_element do |oo_document|
-      oo_document_count += 1
-      oo_element_count = 0
-      oo_document.each_element do |oo_element|
-        oo_element_count += 1
-        if oo_element.name == "body"
-          oo_element.each_element do |be|
-            if be.name == "spreadsheet"
-              be.each_element do |se|
-                if se.name == "table"
-                  return_sheets << se.attributes['name']
-                end
-              end
-            end
-          end
-        end
-      end
+    @doc.find("//*[local-name()='table']").each do |sheet|
+      return_sheets << sheet.attributes['name']
     end
     return_sheets
   end
-
+    
   # version of the openoffice document
   # at 2007 this is always "1.0"
   def officeversion
@@ -211,32 +221,34 @@ class Openoffice < GenericSpreadsheet
 
   # read the version of the OO-Version
   def oo_version
-    @doc.each_element do |oo_document|
-      @officeversion = oo_document.attributes['version']
+    @doc.find("//*[local-name()='document-content']").each do |office|
+      @officeversion = office.attributes['version']
     end
   end
 
   # helper function to set the internal representation of cells
-  def set_cell_values(sheet,x,y,i,v,vt,formula,tr,str_v)
-    key = [y,x+i]
+  def set_cell_values(sheet,x,y,i,v,vt,formula,table_cell,str_v,style_name)
+    key = [y,x+i]    
     @cell_type[sheet] = {} unless @cell_type[sheet]
     @cell_type[sheet][key] = Openoffice.oo_type_2_roo_type(vt)
     @formula[sheet] = {} unless @formula[sheet]
     @formula[sheet][key] = formula  if formula
     @cell[sheet]    = {} unless @cell[sheet]
+    @style[sheet] = {} unless @style[sheet]
+    @style[sheet][key] = style_name
     case @cell_type[sheet][key]
     when :float
       @cell[sheet][key] = v.to_f
     when :string
       @cell[sheet][key] = str_v
     when :date
-      if tr.attributes['date-value'].size != "XXXX-XX-XX".size
+      if table_cell.attributes['date-value'].size != "XXXX-XX-XX".size
         #-- dann ist noch eine Uhrzeit vorhanden
         #-- "1961-11-21T12:17:18"
-        @cell[sheet][key] = DateTime.parse(tr.attributes['date-value'])
+        @cell[sheet][key] = DateTime.parse(table_cell.attributes['date-value'])
         @cell_type[sheet][key] = :datetime
       else
-        @cell[sheet][key] = tr.attributes['date-value']
+        @cell[sheet][key] = table_cell.attributes['date-value']
       end
     when :percentage
       @cell[sheet][key] = v.to_f
@@ -258,130 +270,115 @@ class Openoffice < GenericSpreadsheet
     sheet_found = false
     raise ArgumentError, "Error: sheet '#{sheet||'nil'}' not valid" if @default_sheet == nil and sheet==nil
     raise RangeError unless self.sheets.include? sheet
-    oo_document_count = 0
-    @doc.each_element do |oo_document|
-      # @officeversion = oo_document.attributes['version']
-      oo_document_count += 1
-      oo_element_count = 0
-      oo_document.each_element do |oo_element|
-        oo_element_count += 1
-        if oo_element.name == "body"
-          oo_element.each_element do |be|
-            if be.name == "spreadsheet"
-              be.each_element do |se|
-                if se.name == "table"
-                  if se.attributes['name']==sheet
-                    sheet_found = true
-                    x=1
-                    y=1
-                    se.each_element do |te|
-                      if te.name == "table-column"
-                        rep = te.attributes["number-columns-repeated"]
-                      elsif te.name == "table-row"
-                        if te.attributes['number-rows-repeated']
-                          skip_y = te.attributes['number-rows-repeated'].to_i
-                          y = y + skip_y - 1 # minus 1 because this line will be counted as a line element
-                        end
-                        te.each_element do |tr|
-                          if tr.name == 'table-cell'
-                            skip = tr.attributes['number-columns-repeated']
-                            formula = tr.attributes['formula']
-                            vt = tr.attributes['value-type']
-                            v  = tr.attributes['value']
-                            if vt == 'string'
-                              str_v  = ''
-                              # insert \n if there is more than one paragraph
-                              para_count = 0
-                              tr.each_element do |str|
-                                if str.name == 'p'
-                                  v = str.text
-                                  str_v += "\n" if para_count > 0
-                                  para_count += 1
-                                  if str.children.size > 1
-                                    str_v = children_to_string(str.children)
-                                  else
-                                    str.children.each {|child|
-                                      str_v = str_v + child.to_s #.text
-                                    }
-                                  end
-                                  str_v.gsub!(/&quot;/,'"')
-                                  str_v.gsub!(/&amp;/,'&')
-                                  str_v.gsub!(/&apos;/,"'")
-                                end # == 'p'
-                              end
-                            elsif vt == 'time'
-                              tr.each_element do |str|
-                                if str.name == 'p'
-                                  v = str.text
-                                end
-                              end
-                            elsif vt == '' or vt == nil
-                              #
-                            elsif vt == 'date'
-                              #
-                            elsif vt == 'percentage'
-                              #
-                            elsif vt == 'float'
-                              #
-                            elsif vt == 'boolean'
-                              #
-                            else
-                              # raise "unknown type #{vt}"
-                            end
-                            if skip
-                              if v != nil or tr.attributes['date-value']
-                                0.upto(skip.to_i-1) do |i|
-                                  set_cell_values(sheet,x,y,i,v,vt,formula,tr,str_v)
-                                end
-                              end
-                              x += (skip.to_i - 1)
-                            end # if skip
-                            set_cell_values(sheet,x,y,0,v,vt,formula,tr,str_v)
-                            x += 1
-                          end
-                        end
-                        y += 1
-                        x = 1
+
+    @doc.find("//*[local-name()='table']").each do |ws|
+      if sheet == ws.attributes['name']
+        sheet_found = true
+        col = 1
+        row = 1
+        ws.each_element do |table_element|
+          case table_element.name
+          when 'table-column'
+            @style_defaults[sheet] << table_element.attributes['default-cell-style-name'] 
+          when 'table-row'
+            if table_element.attributes['number-rows-repeated']
+              skip_row = table_element.attributes['number-rows-repeated'].to_i
+              row = row + skip_row - 1
+            end
+            table_element.each_element do |cell|
+              skip_col = cell.attributes['number-columns-repeated']
+              formula = cell.attributes['formula']
+              vt = cell.attributes['value-type']
+              v =  cell.attributes['value']
+              style_name = cell.attributes['style-name']
+              if vt == 'string'
+                str_v  = ''
+                # insert \n if there is more than one paragraph
+                para_count = 0
+                cell.each_element do |str|
+                  if str.name == 'p'
+                    v = str.content
+                    str_v += "\n" if para_count > 0
+                    para_count += 1
+                    if str.children.size > 1
+                      str_v += children_to_string(str.children)
+                    else
+                      str.children.each do |child|
+                        str_v += child.content #.text
                       end
                     end
-                  end # sheet
+                    str_v.gsub!(/&apos;/,"'")  # special case not supported by unescapeHTML
+                    str_v = CGI.unescapeHTML(str_v)
+                  end # == 'p'
+                 end
+              elsif vt == 'time'
+                cell.each_element do |str|
+                  if str.name == 'p'
+                    v = str.content
+                  end
                 end
+              elsif vt == '' or vt == nil
+                #
+              elsif vt == 'date'
+                #
+              elsif vt == 'percentage'
+                #
+              elsif vt == 'float'
+                #
+              elsif vt == 'boolean'
+                v = cell.attributes['boolean-value']
+                #
+              else
+                # raise "unknown type #{vt}"
               end
-            end
+              if skip_col
+                if v != nil or cell.attributes['date-value']
+                  0.upto(skip_col.to_i-1) do |i|
+                    set_cell_values(sheet,col,row,i,v,vt,formula,cell,str_v,style_name)
+                  end
+                end
+                col += (skip_col.to_i - 1)
+              end # if skip
+              set_cell_values(sheet,col,row,0,v,vt,formula,cell,str_v,style_name)
+              col += 1
+            end    
+            row += 1
+            col = 1
           end
         end
       end
     end
+    
+    @doc.find("//*[local-name()='automatic-styles']").each do |style|
+      read_styles(style)
+    end
     if !sheet_found
       raise RangeError
     end
     @cells_read[sheet] = true
   end
 
+  def read_styles(style_elements)
+    @style_definitions['Default'] = Openoffice::Font.new
+    style_elements.each do |style|
+      next unless style.name == 'style'
+      style_name = style.attributes['name']
+      style.each do |properties|
+        font = Openoffice::Font.new
+        font.bold = properties.attributes['font-weight']
+        font.italic = properties.attributes['font-style']
+        font.underline = properties.attributes['text-underline-style']
+        @style_definitions[style_name] = font
+      end    
+    end
+  end
+  
   # Checks if the default_sheet exists. If not an RangeError exception is
   # raised
   def check_default_sheet
     sheet_found = false
     raise ArgumentError, "Error: default_sheet not set" if @default_sheet == nil
-    @doc.each_element do |oo_document|
-      oo_element_count = 0
-      oo_document.each_element do |oo_element|
-        oo_element_count += 1
-        if oo_element.name == "body"
-          oo_element.each_element do |be|
-            if be.name == "spreadsheet"
-              be.each_element do |se|
-                if se.name == "table"
-                  if se.attributes['name'] == @default_sheet
-                    sheet_found = true
-                  end # sheet
-                end
-              end
-            end
-          end
-        end
-      end
-    end
+    sheet_found = true if sheets.include?(@default_sheet)
     if ! sheet_found
       raise RangeError, "sheet '#{@default_sheet}' not found"
     end
@@ -437,8 +434,8 @@ class Openoffice < GenericSpreadsheet
   def children_to_string(children)
     result = ''
     children.each {|child|
-      if child.class == REXML::Text
-        result = result + child.to_s
+      if child.text?
+        result = result + child.content
       else
         if child.name == 's'
           compressed_spaces = child.attributes['c'].to_i
@@ -448,7 +445,7 @@ class Openoffice < GenericSpreadsheet
           end
           result = result + " "*compressed_spaces
         else
-          result = result + child.to_s
+          result = result + child.content
         end
       end
     }