数字排序问题

可恶的

小弟准备开始学习python，故晚上在python版转转，看到个问题，过来请教下。
文本文档如下（每行开头无空格）：
2009-06-24 00:00:00,0802,0000002001,0
2009-06-24 00:00:00,1201,0000002001,0
2009-06-24 00:00:00,1452,0000002001,0
2009-06-24 00:00:00,1755,0000002001,0
2009-06-24 00:00:00,1800,0000002001,0
2009-06-25 00:00:00,0741,0000002001,0
2009-06-25 00:00:00,1202,0000002001,0
2009-06-25 00:00:00,1447,0000002001,0
2009-06-25 00:00:00,1800,0000002001,0
2009-06-26 00:00:00,0746,0000002001,0
2009-06-26 00:00:00,1200,0000002001,0
2009-06-26 00:00:00,1310,0000002001,0
2009-06-26 00:00:00,1431,0000002001,0
2009-06-26 00:00:00,1801,0000002001,0
2009-06-27 00:00:00,0803,0000002001,0
2009-06-27 00:00:00,1213,0000002001,0
2009-06-27 00:00:00,1427,0000002001,0
2009-06-27 00:00:00,1800,0000002001,0

要求经过变换后成为以下格式：
2009-06-24,0000002001,0802,1201,1452,1755,1800
2009-06-25,0000002001,0741,1202,1447,1800
2009-06-26,0000002001,0746,1200,1310,1431,1801
2009-06-27,0000002001,0803,1213,1427,1800

小弟愚昧，awk使用不熟练，一不小心成为管道之王了，求更好方法。
awk -F '[ ,]+' '{if(a[$2]==0)print $2,$(NF-1);a[$2]++;if(a[$2]>=1)print $(NF-2)}'|sed -r ':A;N;/\n\s*[0-9]{4}-[0-9]{2}-[0-9]{2}\s*/!s/\n/\t/g;tA;D'|awk 'OFS=","{$1=$1;print}'

原帖地址如下：http://bbs2.chinaunix.net/thread-1506321-1-1.html

welcome008

$2是哪个字段？为啥要区分是否==0？

我看糊涂了

[ 本帖最后由 welcome008 于 2009-7-16 23:05 编辑 ]

blackold

与python有什么关系？

welcome008

印象中好像这里之前有一个类似的排序，很像
包括格式

blackold

这哪是排序问题，源数据已经排序了。

welcome008

awk -F '[ ,]+' '{print $1,$(NF-1),$(NF-2)}' filename|sort -u|awk '{a[$1]++;if(a[$1]==1) {if(NR!=1) printf "\n";printf $1" "$2" "$3;} else printf ","$3}END{print}'

取巧了一下，不过仍然用了两个管道

wtuter

[2]cat 3
2009-06-24 00:00:00,0802,0000002001,0
2009-06-24 00:00:00,1201,0000002001,0
2009-06-24 00:00:00,1452,0000002001,0
2009-06-24 00:00:00,1755,0000002001,0
2009-06-24 00:00:00,1800,0000002001,0
2009-06-25 00:00:00,0741,0000002001,0
2009-06-25 00:00:00,1202,0000002001,0
2009-06-25 00:00:00,1447,0000002001,0
2009-06-25 00:00:00,1800,0000002001,0
2009-06-26 00:00:00,0746,0000002001,0
2009-06-26 00:00:00,1200,0000002001,0
2009-06-26 00:00:00,1310,0000002001,0
2009-06-26 00:00:00,1431,0000002001,0
2009-06-26 00:00:00,1801,0000002001,0
2009-06-27 00:00:00,0803,0000002001,0
2009-06-27 00:00:00,1213,0000002001,0
2009-06-27 00:00:00,1427,0000002001,0
2009-06-27 00:00:00,1800,0000002001,0
[2]
[2]


[2]awk -F'[ ,]'  '{a[$1]=$1","$4;;b[$1]=b[$1]","$3}END{for(i in a)print a,b}' 3
2009-06-24,0000002001 ,0802,1201,1452,1755,1800
2009-06-25,0000002001 ,0741,1202,1447,1800
2009-06-26,0000002001 ,0746,1200,1310,1431,1801
2009-06-27,0000002001 ,0803,1213,1427,1800


是这样吗？

Minsic

如果文件原先没有排序的话……

1. awk 'BEGIN{
   
2.   FS="[ ,]"
   
3. }
   
4. {
   
5.   cnt=++a[$1","$4]
   
6.   b[$1","$4","cnt]=$3;
   
7. }
   
8. END{
   
9.   for(p in a){
   
10.     for(i=2;i<=a[p];i++){
    
11.       for(j=i;b[p","j-1]>b[p","j];--j){
    
12.         t=b[p","j]
    
13.         b[p","j]=b[p","j-1]
    
14.         b[p","j-1]=t
    
15.       }
    
16.     }
    
17.     printf p",";for(k=1;k<=a[p];k++){if(k<a[p])printf b[p","k]",";else printf b[p","k]"\n"}
    
18.   }
    
19. }' urfile|sort

复制代码

[ 本帖最后由 Minsic 于 2009-7-17 10:07 编辑 ]

kwokcn

只看顶楼给的文本的话：

1. 
   
2. awk -F"[ ,]" '!a[$1]++{print s;s=""}{s=s?s","$3:$1","$4","$3}END{print s}' URFILE
   

复制代码

如果没排序的话，我的想法是全都往awk数组里丢，END时候遍历一下，最后加一个sort管道。只能对整个文本进行排序的时候，sort的效率是很高的。

非凡公子

行转列
oracle很容易搞定..