Centos command -- sed

1. Merge rows
Zj @ zj :~ /Script/blog_script $ cat test1
1
2
3
4
Merge the upper and lower rows
Zj @ zj :~ /Script/blog_script $ sed '$! N; s/\ n/\ t/'test1
1 2
3 4
Merge matching mode and next line
Zj @ zj :~ /Script/blog_script $ sed '/2/{N; s/\ n/\ t/}' test1
1
2 3
4
Merge all rows
Zj @ zj :~ /Script/blog_script $ sed ': a; N; s/\ n/\ t/; ba; 'test1
1 2 3 4

2. Exchange rows
2.1 Exchange two rows when the row number is known
Zj @ zj :~ /Script/blog_script $ cat test
Baidu music so terrible so bad
Microsoft haha
Yahoo byebye
Google princess so good
Here is the exchange of 1, 4 rows. Of course you can modify it as needed.
Zj @ zj :~ /Script/blog_script $ for (I = 1; I <= 4; I ++); do case $ I in 1) sed-n 4 p test; 4) sed-n 1 p test; *) sed-n $ {I} p test; esac; done
Google princess so good
Microsoft haha
Yahoo byebye
Baidu music so terrible so bad
Continuous saying:
Zj @ zj :~ /Script/blog_script $ sed '1 {h; d}; 2 {G} 'test
Microsoft haha
Baidu music so terrible so bad
Yahoo byebye
Google princess so good

2.2 do not know the row number
If the two rows to be exchanged are consecutive rows:
Zj @ zj :~ /Script/blog_script $ sed '/baidu/{h; d};/microsoft/{G}' test
Microsoft haha
Baidu music so terrible so bad
Yahoo byebye
Google princess so good
Ps: Exchange rows that contain bai and microsoft
The two rows are discontinuous:

Zj @ zj :~ /Script/blog_script $ sed '/baidu/{: a; N;/google /! Ba; s/\ ([^ \ n] * \) \ n \(. * \) \ n \(. * \)/\ 3 \ n \ 2 \ n \ 1/} 'test
Google princess so good
Microsoft haha
Yahoo byebye
Baidu music so terrible so bad
Ps: Exchange rows containing baidu and google

We have studied the following:
Zj @ zj :~ /Script/blog_script $ sed '/baidu/{: a; N;/microsoft /! Ba; /[^ \ n] * baidu [^ \ n] * \ n [^ \ n] * microsoft [^ \ n] * $/{s/\ ([^ \ n] * baidu [^ \ n] * \) \ n \(. * \)/\ 2 \ n \ 1/}; s/\ ([^ \ n] * \) \ n \(. * \) \ n \(. * \)/\ 3 \ n \ 2 \ n \ 1/} 'test
Microsoft haha
Baidu music so terrible so bad
Yahoo byebye
Google princess so good

Zj @ zj :~ /Script/blog_script $ sed '/baidu/{: a; N;/google /! Ba; /[^ \ n] * baidu [^ \ n] * \ n [^ \ n] * google [^ \ n] * $/{s/\ ([^ \ n] * baidu [^ \ n] * \) \ n \(. * \)/\ 2 \ n \ 1/;}; s/\ ([^ \ n] * \) \ n \(. * \) \ n \(. * \)/\ 3 \ n \ 2 \ n \ 1/} 'test
Google princess so good
Microsoft haha
Yahoo byebye
Baidu music so terrible so bad

Ps: the above Code
/Baidu/{...} encounters a line containing baidu and starts to execute the command sequence in {}
: A; N;/google /! Ba reads information cyclically until google is read.
/[^ \ N] * baidu [^ \ n] * \ n [^ \ n] * google [^ \ n] * $/this means that if there is only one \ n, that is, the two rows are continuous: {s/\ ([^ \ n] * baidu [^ \ n] * \) \ n \(. * \)/\ 2 \ n \ 1/;} exchange the two rows

If it does not match the above pattern, it means it is not continuous.
S/\ ([^ \ n] * \) \ n \(. * \) \ n \(. * \)/\ 3 \ n \ 2 \ n \ 1 /}
In the case of matching continuous rows, it is impossible to match the. OK ~~~~~ of the above expression ~~~~

3. Interactive parity
Zj @ zj :~ /Script/blog_script $ sed '$! N; s/\ ([^ \ n] * \) \ n \ ([^ \ n] * \)/\ 2 \ n \ 1/'test
Microsoft haha
Baidu music so terrible so bad
Google princess so good
Yahoo byebye

4. No need to delete the row.
D hello ....

5. Delete empty rows
Sed '/^ $/d' test2
Delete multiple empty rows
Sed '/^ $/{N;/^ \ n * $/D}' test2