test:
#! /bin/bash
New York
Hello Hello
Hello
terminal:
-n could stop sed from printing out:
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed 's/Hello/world/' ./test
#! /bin/bash
New York
world Hello
world
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed -n 's/Hello/world/' ./test
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$
-p could force sed from printing out. "Printing out" could be separated from 2 parts: whenver sed read one line from the input, after applying all commands, it will output the result automatically, this is the "first printing out" which is sth '-n' could stop. After that, sed could choose whether to print out the result again, this is sth controlled by -p, -p forces sed to print out again. In following example, the last 2 lines were caught and replaced, so they got printed out twice.
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed 's/Hello/world/p' ./test
#! /bin/bash
New York
world Hello
world Hello
world
world
-p normally get used together with '-n'. -n stop all automatic printing out, and -p force sed to print out only lines get changed
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed -n 's/Hello/world/p' ./test
world Hello
world
2. Specify the range
Specify the single line:
test is same as above
terminal:
The first pattern /New York/ means that we only pick up lines matching this pattern and then apply the command.
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed '/New York/ s/New York/Chicago/' ./test
#! /bin/bash
Chicago
Hello Hello
Hello
combined with command file:
command.sed contains the command "s/New York/&, the US Financial Capital"
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed '/New York/' -f command.sed ./test
sed: can't read /New York/: No such file or directory
#! /bin/bash
New York, the US Financial Capital
Hello Hello
Hello
Specify the line range with number:
test is same as above
terminal:
1) First command is specifying line 1 to line 5, which is picking up all lines to apply commands, so both last two lines get changed
2) Second command is specifying the last line, line 5 to get picked, only the last line get changed.
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed '1,5 s/Hello/world/g' ./test
#! /bin/bash
New York
world world
world
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed '5,5 s/Hello/world/g' ./test
#! /bin/bash
New York
Hello Hello
world
Specify the line range with pattern:
test:
#! /bin/bash
New York
Hello Hello
Hello
Chicago
terminal:
This command starts from the line matching pattern /New York/, end to the line matching pattern /Chicago/, replace all "Hello" with "world"
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed '/New York/,/Chicago/ s/Hello/world/g' ./test
#! /bin/bash
New York
world world
world
Chicago
specify the line range by negating the pattern
test is same as above.
terminal:
The command is picking up lines not matching "Hello", and then apply all commands to those lines.
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed '/Hello/!' -e 's/New York/&, US Financial Capital/' -e 's/Chicago/&, Middle Financial Center/' ./test
sed: can't read /Hello/!: No such file or directory
#! /bin/bash
New York, US Financial Capital
Hello Hello
Hello
Chicago, Middle Financial Center
3. Use different delimiter:
test:
#! /bin/bash
New York
Boston
Washington D.C
terminal:
1) First command, we want to recognize "New York" firstly, then replace it with Chicago, ":New York:", means we want to use ":" to separate the pattern, but shell doesn't recognize it.
2) Second command, we firstly use "\" to escape ":", so shell recognize the colon as the separator. Then shell could successfully catch the pattern "New York" and replace it with Chicago
3) Third command is same as second command, the only difference is we used a different delimiter.
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed ':New York: s;;Chicago;' ./test
sed: -e expression #1, char 6: unknown command: `Y'
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed '\:New York: s;;Chicago;' ./test
#! /bin/bash
Chicago
Boston
Washington D.C
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed '\/New York/ s;;Chicago;' ./test
#! /bin/bash
Chicago
Boston
Washington D.C
4. Note the "pattern range"
Shell will try to match the patter "as far as it can"
test:
#! /bin/bash
Jersey City
abc
terminal:
we aim to replace "Jersey" with "Union", and get the result "Union City". But the shell match "J.*y" to the entire string "Jersey City". And we get the result "Union".
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed -n '3 s/J.*y/Union/p' ./test
Union
"b*" means 0 or more b, which means it could indicate null string. So for the 4th line "abc", when sed apply the command to this string:
1) In the beginning, it will match null to "b*"
2) At 'a', not match
3) At 'b', match
4) At 'c', not match
5) At the end, match another null string
so the final result is "1a1c1"
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed -n '4 s/b*/1/p' ./test
1abc
aubinxia@aubinxia-VirtualBox:~/Desktop/xxdev$ sed -n '4 s/b*/1/pg' ./test
1a1c1
No comments:
Post a Comment