要从2GB的excel数据中抽取用户的email。不会C，java写起来太麻烦，ruby一下吧，分享一下成果，如果谁也遇到这个就别撞墙啦~~~~~
代码语言：javascript复制$KCODE = 'u'
require 'find'
class Auto_fix_mail_pros
def runexcel
#把win32ole包含进来后,就可以通过和windows下的excel api进行交互来对excel文件进行读写了.
filepath=File.dirname('D:/excelrun/temp')
filename=[]
filenameid=0
require 'win32ole'
require 'find'
Find.find(filepath) do |f|
if File.basename(f).index('@$$.') == nil
if (File.basename(f) =~ /.*.xls$/) || (File.basename(f) =~ /.*.XLS$/)
filename << f
filenameid=filenameid 1
end
end
end
if filenameid>0
for id in 0..filenameid-1
excel = WIN32OLE::new('excel.Application')
excel.visible = false # in case you want to see what happens
excel.Application.DisplayAlerts = false
#打开excel文件,对其中的sheet进行访问:
workbook = excel.Workbooks.Open(filename[id])
sheetcount=workbook.Sheets.Count
loadImage(filename[id])
require 'pathname'
filepathname=Pathname.new(File.dirname(filename[id])).realpath
#开始时间
begtime=Time.now
for sheetnum in 1..sheetcount
worksheet = workbook.Worksheets(sheetnum) #定位到第一个sheet
worksheet.Select
rows=worksheet.UsedRange.Rows.count
cols=worksheet.UsedRange.columns.count
#puts rows
#puts cols
#符合条件存入数组
data = []
i=0
for row in 1..rows
for col in 1..cols
if worksheet.Cells(row, col).value.to_s =~/w ([- .]w )*@w ([-.]w )*.w ([-.]w )*/
data << worksheet.Cells(row, col).value.to_s
i=i 1
end
end
end
worksheet=nil
#结果写入文件
if i>0
#require 'stringio'
basename=File.basename(filename[id], ".*") "_#{sheetnum}.txt"
$Sfilename=filepathname basename
loadImage ($Sfilename)
file =File.new($Sfilename,"w")
for j in 0..i-1
file.puts data[j]
end
#内存回收
data.clear
file.close
end
end
workbook.saved=true
excel.Workbooks.Close
workbook=nil
#结束会话
excel.Quit
#excel=nil
#重命名处理过的文件
newname=filepathname (File.basename(filename[id], ".*") "@$$.xls")
#File.rename(filename[id],newname)
#结束时间
endtime=Time.now
#执行时间
loadImage((endtime-begtime).to_s)
end
#结束会话
excel.Quit
excel=nil
#内存回收
filename.clear
data.clear
end
GC.start
end
def loadImage(file)
end
end
mail_pros_tools = Auto_fix_mail_pros.new
mail_pros_tools.runexcel
