提取符合内容的词

luofeiyu_cu

test文件如下：
The following packages have unmet dependencies:
libqt4-core : Depends: libqt4-dbus (= 4:4.8.2+dfsg-10) but 4:4.8.2+dfsg-2 is to be installed
               Depends: libqt4-network (= 4:4.8.2+dfsg-10) but 4:4.8.2+dfsg-2 is to be installed
               Depends: libqt4-script (= 4:4.8.2+dfsg-10) but it is not going to be installed
               Depends: libqt4-test (= 4:4.8.2+dfsg-10) but it is not going to be installed
               Depends: libqt4-xml (= 4:4.8.2+dfsg-10) but 4:4.8.2+dfsg-2 is to be installed
               Depends: libqtcore4 (= 4:4.8.2+dfsg-10) but 4:4.8.2+dfsg-2 is to be installed
libqt4-gui : Depends: libqt4-designer (= 4:4.8.2+dfsg-10) but it is not going to be installed
              Depends: libqt4-opengl (= 4:4.8.2+dfsg-10) but 4:4.8.2+dfsg-2 is to be installed
              Depends: libqt4-svg (= 4:4.8.2+dfsg-10) but it is not going to be installed
              Depends: libqtgui4 (= 4:4.8.2+dfsg-10) but 4:4.8.2+dfsg-2 is to be installed
E: Unmet dependencies. Try 'apt-get -f install' with no packages (or specify a solution).

我要提取每行s:  到  ( 之间的所有词，比如：libqt4-dbus  等等。
请问如何写shell?

yinyuemi

回复 1# luofeiyu_cu

1. grep -Po '(?<=s: )[^\(]+(?=\()' file
   
2. libqt4-dbus
   
3. libqt4-network
   
4. libqt4-script
   
5. libqt4-test
   
6. libqt4-xml
   
7. libqtcore4
   
8. libqt4-designer
   
9. libqt4-opengl
   
10. libqt4-svg
    
11. libqtgui4
    

复制代码

tyw09

回复 2# yinyuemi


    顶楼上，喜欢的awk用不了零宽，Python必须指定长度。

yinyuemi

回复 3# tyw09


    awk也可以很cool
awk -F's: |\\(' 'NF==3{print $2}'

asdf2110

1. grep Depends file | sed -r 's/.*s: (.*) \(.*/\1/g

复制代码

tyw09

回复 4# yinyuemi

哈，谢谢，收藏了
   

luofeiyu_cu

问题1：
grep -Po '(?<=s: )[^\(]+(?=\()'  test
得到正确结果，-P 表示使用perl方式的正则表达式  o  only  match
(?<=s: )[^\(]+(?=\()  #请详细解释一下这个式子
问题2：
awk -F's: |\\(' 'NF==3{print $2}'   #没有写完全吧，请写完整
问题3：
grep test | sed -r 's/.*s: (.*) \(.*/\1/g  #无法运行
sed -r 's/.*s: (.*) \(.*/\1/g'  test
The following packages have unmet dependencies:
libqt4-dbus
libqt4-network
libqt4-script
libqt4-test
libqt4-xml
libqtcore4
libqt4-designer
libqt4-opengl
libqt4-svg
libqtgui4
E: Unmet dependencies. Try 'apt-get -f install' with no packages (or specify a solution).
sed的方法还需要修改。

cao627

回复 7# luofeiyu_cu

(?<=s: )[^\(]+(?=\()  #请详细解释一下这个式子

(?<=s: )和(?=\()相当于创建两个标记s: 和(,告诉正则引擎，放在他们中间的表达式必须前有s: 后有(才算匹配。
[^\(]+  这个表达式匹配的是非( 的连续字符。
   

luofeiyu_cu

sed 's/.*Depends: \([^ ]*\).*/\1/' test
这个表达式也可以，但是我不理解，
/.*Depends: \([^ ]*\).*/  这个正则的/.*Depends:  我理解，\([^ ]*\)  这个是分组吧， \([^ ]*\).*/ 联合在一起我就不理解了。

cao627

回复 9# luofeiyu_cu

\(    \)可理解为捕获

s/aaa\(bb\)/\1/ 就是匹配aaabb，将匹配到的aaabb替换为捕获到的bb。\1是对获取的引用。如果有多个获取按左到右，依次用\1 \2..
表示