and
Common options
-n: do not output the contents of the pattern space to the screen
-e: multi-point editing (can be multiple commands)
-f: editing commands, one per line
-r: support for extended regular expressions
-i: directly edit the original file
Address delimitation
1, an empty address: the full text processing
2, a single address:
#: The specified row
/ Pattern /: this mode are matched to each row
3, address range
#，#：
#，+#：
# / Pat1 /
/ Pat1 /, / pat2 /
4, step: -
1-2: all the odd rows
2 ~ 2: all even lines
Editing commands:
d: Delete
p: displays the contents of the pattern space
a \ text: the line appends text to support the use of \ n additional multi-line
i \ text: the text is inserted before the row, supports \ n multi-row insert
c \ text: replace specified text here is to match lines
w / PATH / TO / SOMEFILE: Space-saving mode to the matching line to the specified file
r / PATH / FROM / SOMEFILE: reading the contents of the specified file is matched to the rear row of the pattern to the current file; document merger
=: For the pattern matching to the line number of the line printing (sed '/ ^ UUID / =' / etc / fstab)
!: Condition inverted (sed '/ ^ # / d!' / Etc / fstab)
s ///: Find Alternatively, the separator may specify their own
Replacement tags:
g: global replace
w / PATH / TO / SOMEFILE: replaces the successful results saved to the specified file
p: replace successful line display
Exercise 1: All whitespace characters /boot/grub2/grub.cfg delete all files that start with a blank character of line
sed 's/^[[:space:]]\+//g' /boot/grub2/grub.cfg
Exercise 2: All whitespace characters of all lines beginning with the # # # back and delete the / etc / fstab file
sed 's/^#[[:space:]]*//g' /etc/fstab
Exercise 3: Output absolute path to a sed command, remove its catalog, its behavior is similar to dirname
echo "/var/log/messages" | sed 's/[^/]\+$//'