awk的RS怎样使用正则？

网中淫

以下日志，想用日期时间所在的行作为RS分割，怎么弄呢？

1. [oracle@test ~]$ cat test.log
   
2. Sat Aug 31 08:26:44 2013
   
3. Shutting down instance (abort)
   
4. Sat Aug 31 13:11:09 2013
   
5. Shutting down instance (abort)
   
6. Sat Aug 31 13:23:10 2013
   
7. Shutting down instance (abort)
   
8. Sat Aug 31 13:26:12 2013
   
9. Shutting down instance (abort)
   
10. Tue Nov  5 13:23:43 CST 2013
    
11. ORA-00001
    
12. Tue Nov  6 13:23:43 CST 2013
    
13. this is test
    
14. ORA-00904
    
15. Tue Nov  6 13:23:43 CST 2013
    
16. ok,you win
    
17. ORA-04031
    
18. [oracle@source ~]$ awk 'BEGIN{RS=".*Nov.*\n"}{print NR,NF}' test.log
    
19. 1 0

复制代码

Herowinter

有时候我也想要类似的功能，但好像不支持？

网中淫

回复 2# Herowinter

可以支持呢，我没弄好：

1. [oracle@test ~]$ awk 'BEGIN{RS="[^\n]*Nov[^\n]*"}{print NR,NF}' test.log
   
2. 1 36
   
3. 2 1
   
4. 3 4
   
5. 4 3

复制代码

Herowinter

回复 3# 网中淫
恩，是可以，那好像你自己已经解决问题了？

   

jason680

回复 2# Herowinter

The RS is support with RE(Regular Express) in awk(GNU Awk 3.1.8)
and you can check it with RT variable

$ man awk
# search RS by /RS and get below:

   Records
       Normally, records are separated by newline characters.  You can control
       how records are separated by assigning values to the built-in  variable
       RS.   If  RS is any single character, that character separates records.
       Otherwise, RS is a regular expression.  Text in the input that  matches
       this  regular expression separates the record.  However, in compatibil‐
       ity mode, only the first character of its string value is used for sep‐
       arating  records.   If  RS  is set to the null string, then records are
       separated by blank lines.  When RS is set to the null string, the  new‐
       line  character  always acts as a field separator, in addition to what‐
       ever value FS may have.

       RT          The record terminator.  Gawk sets RT to the input text that
                   matched  the  character  or regular expression specified by
                   RS.

   

jason680

回复 3# 网中淫

you also can use RT variable to check your RS pattern

$ awk 'BEGIN{RS="[^\n]*Nov[^\n]*";OFS=","}{print NR,NF,RT}' test.log
1,36,Tue Nov  5 13:23:43 CST 2013
2,1,Tue Nov  6 13:23:43 CST 2013
3,4,Tue Nov  6 13:23:43 CST 2013
4,3,
   

javametro

manpage 确实需要细心的看看

RT这个功能不错~~

elu_ligao

1. awk -vRS="[^\n]+:[0-9]+:[^\n]+" '{print NR,NF,RT}' RS  
   
2. 1 0 Sat Aug 31 08:26:44 2013
   
3. 2 4 Sat Aug 31 13:11:09 2013
   
4. 3 4 Sat Aug 31 13:23:10 2013
   
5. 4 4 Sat Aug 31 13:26:12 2013
   
6. 5 4 Tue Nov  5 13:23:43 CST 2013
   
7. 6 1 Tue Nov  6 13:23:43 CST 2013
   
8. 7 4 Tue Nov  6 13:23:43 CST 2013
   
9. 8 3

复制代码