【Python】实现从AWR 报表上抓取指定数据改进版

相对上一个脚本，该脚本修改了如下内容：

1 url的传入方式，只需将url保存到一个文本文件里，由程序自己读入并解析

2 增加了oracle 指标数据统计和分析，比较两周(我自己的需要)的数据变化趋势

#!/usr/bin/python
# -*- coding: UTF-8 -*-
#created by yangql 

#date @2011-12-13

#function 获取AWR报表中的oracle数据库指标，分析并得出数据趋势

import sys
import urllib
import HTMLParser
import string

####转换unicode 类型的字符串到 float 型的数值
def utof(s1):
   s2=s1.strip()
   s3=s2.encode( 'utf-8')
   s4=s3.split( ',')
   length= len(s4)
   if length <= 1 :
      t1= string.atof(s4[ 0])
      return t1
   elif length == 2:
      t1=string.atof(s4[ 1])+string.atof(s4[ 0])* 1000
      return t1
   elif length == 3:
      t1=string.atof(s4[ 2])+string.atoi(s4[ 1])* 1000+string.atoi(s4[ 0])* 1000000
      return t1
   else:
      return 0
###构造解析html的类并且获取整个HTML中除去标签的文本数据
urltext = []
class CustomParser(HTMLParser.HTMLParser):
  selected=( 'table', 'h1', 'font', 'ul', 'li', 'tr', 'td', 'a')
  def reset( self):
      HTMLParser.HTMLParser.reset( self)
      self._level_stack = []

  def handle_starttag( self, tag, attrs):
       if tag in CustomParser.selected:
         self._level_stack.append(tag)

  def handle_endtag( self, tag):
      if self._level_stack \
         and tag in CustomParser.selected \
         and tag == self._level_stack[- 1]:
         self._level_stack.pop()
 
  def handle_data( self, data):
     if "/".join( self._level_stack) in ( 'table/tr/td', 'table/tr/td/h1/font', 'table/tr/td/ul/li') and data != ' \n ':
        #print data
        urltext.append(data)    

####调用解析html 的类，并且将获取的文本数据传递给urltext 数组
def gethtml(url):
    content = unicode(urllib.urlopen(url,params).read(), 'UTF-8')
    parser = CustomParser()
    parser.feed(content)
    parser.close()

####获取指定的指标的值,
def calculcate(urltext):
    print '-----------------------------------------'
    global Logical
    global Physical_reads
    global Physical_writes
    global Executes
    global Transactions
    k= 0
    for item in urltext:
       k=k+ 1
       if k< 50 :
         continue
       elif item == 'Logical reads:' :
         Logical.append(utof(urltext[k])) 
         print 'Logical reads:     ' ,urltext[k].strip()
       elif item == 'Physical reads:' :
         Physical_reads.append(utof(urltext[k]))
         print 'Physical reads:    ',urltext[k].strip()
       elif item == 'Physical writes:' :
         Physical_writes.append(utof(urltext[k]))
         print 'Physical writes:   ' ,urltext[k].strip()
       elif item == 'Executes:':
         Executes.append(utof(urltext[k]))
         print 'Executes:          ' ,urltext[k].strip()
       elif item == 'Transactions:' :
         Transactions.append(utof(urltext[k]))
         print 'Transactions:      ',urltext[k].strip()
       elif k> 86:
         break

def result(url):
  global urltext
  print ' '
  gethtml(url)
  calculcate(urltext)
  urltext = []

def get_avg(List):
    print ' '
    sum1= 0
    sum2= 0
    avg1= 0
    avg2= 0
    trend= 0
    count= len(List)
    for i in range( 0,count):
       if i<count/ 2:
          sum1 =sum1+List[i]
       elif i>=count/ 2:
          sum2 +=List[i]
    avg1=sum1/count* 2
    avg2= float(sum2)/(count/ 2)
    trend=(avg1-avg2)/avg2* 100
    print '第一周的均值 ', '    第二周的均值 ', '      趋 势(%)'
    print " %-20.2f " %avg1 , " %-15.2f " %avg2 , " %-4.2f " %trend, '%'
    print ' '

if __name__ == '__main__':

     if len(sys.argv)> 1:
        params=urllib.urlencode({ 'ip':sys.argv[ 1], 'action': 2})
     else:
        params= None

     Logical=[]
     Physical_reads=[]
     Physical_writes=[]
     Executes=[]
     Transactions=[]
     count= 0
     f = file( '/root/python/1.txt')
     while True:
       url = f.readline()
       if len(url) == 0:
          break
       elif len(url.strip())== 0:
          break
       else:
          result(url)
          count=count+ 1
     f.close()
     print '共有',count, '组数据'
     print '-------------------Logical------------------ '
     get_avg(Logical)      
     print '----------------Physical_reads-------------- '
     get_avg(Physical_reads)
     print '----------------Physical_writes------------- '
     get_avg(Physical_writes)
     print '-------------------Executes----------------- '
     get_avg(Executes)
     print '-----------------Transactions---------------'
     get_avg(Transactions)