Replace slash in Bash

问题:

Let's suppose I have this variable:

DATE="04/Jun/2014:15:54:26".

Therein I need to replace / with \/ in order to get the string:

"04\/Jun\/2014:15:54:26".

I tried tr as follows:

echo "04\Jun\2014:15:54:26" | tr '\' '\\/'

But this results in: "04\Jun\2014:15:54:26".

It does not satisfy me. Can anyone help?

回答1:

No need to use an echo + a pipe + sed.

A simple substitution variable is enough and faster:

echo ${DATE//\//\\/}

#> 04\/Jun\/2014:15:54:26

回答2:

Use SED for substitutions:

 sed 's#/#\\/#g' < filename.txt > newfilename.txt

You usually use "/" instead of the "#" but as long as it is there it doesn't matter.

I am writing this on a windows PC so I hope it is right, you may have to escape the slashes with another slash.

Sed explained, the "-e",lets you edit the file in place. You can use -i to create backup automatically.

sed -e s/STRING_TO_REPLACE/STRING_TO_REPLACE_IT/g index.html

回答3:

here you go:

kent$  echo "04/Jun/2014:15:54:26"|sed 's#/#\\/#g'  
04\/Jun\/2014:15:54:26

your tr line was not correct, you may mis-understand what tr does, tr 'abc' 'xyz' will change a->x, b->y, c->z,not changing whole abc->xyz..

回答4:

You can also escape the slashes, with a slightly less readable solution than with hashes:

echo "04/Jun/2014:15:54:26" | sed 's/\//\\\//g'

回答5:

This has not been said in other answers so I thought I'd add some clarifications:

tr uses two sets of characters for replacement, and the characters from the first set are replaced with those from the second set in a one-to-one correspondance. The manpage states that

SET2 is extended to length of SET1 by repeating its last character as necessary. Excess characters of SET2 are ignored.

Example:

echo abca | tr ab de    # produces decd
echo abca | tr a de     # produces dbcd, 'e' is ignored
echo abca | tr ab d     # produces ddcd, 'd' is interpreted as a replacement for 'b' too

When using sed for substitutions, you can use another character than '/' for the delimiter, which will make your expression clearer (I like to use ':', @n34_panda proposed '#' in their answer). Don't forget to use the /g modifier to replace all occurences: sed 's:/:\\/:g' with quotes or sed s:/:\\\\/:g without (backslashes have to be escaped twice).

Finally your shortest solution will probably be @Luc-Olivier's answer, involving substitution, in the following form (don't forget to escape forward slashes too when part of the expected pattern):

echo ${variable/expected/replacement}     # will replace one occurrence
echo ${variable//expected/replacement}    # will replace all occurrences