I wrote a script (replace.sh) that replaces a ~ with ~\n. I was wondering how can I call this script with multiple arguments.
#!/bin/bash
for i in "$*"
do
sed 's/~/~\n/g' "$i"
done
For example I would like to call ./replace.sh text1 text2. It cannot read
the contents of text1 are: 1 ~ 1 ~ 1. After calling the script it should be
1 ~
1 ~
1 ~
Use "$@" in your for loop:
for i in "$@"
do
sed -i.bak 's/~/~\n/g' "$i"
done
Though I am not really sure if sed is doing what you've described here.
in "$@" at all ...
for i alone implies in "$@". It's an extra 7 characters to type, but the payoff in clarity is huge.in "$@" vs the usual ìn $* implies either. That's the reason why I prefer to suggest the empty form which is IMHO easier to remember.This should suffice:
for i do
sed -i.bak 's/~/~\n/g' "$i"
done
Here a portable variant should work with any posix compliant shell and sed:
for i do
sed 's/~/~\n/g' "$i" > "$i.bak" && cp "$i.bak" "$i"
done
Note: The in ... part of the shell for loop is optional. When not used, for defaults to pick all the arguments (i.e "$@") which is precisely what you are expecting.
-i without a suffix. It means no backups. No backups -> very likely ouch. Also, maybe explain what it means if you use a for loop in a shell script with no in <x>?.bak to new scripters. Pet issue of mine since I first learned Perl one-liners.The other answers cover your real problem. But one thing worth saying more about is how $* and $@ differ in expansion when double quoted. tl;dr You almost always want "$@".
"$*" expands to a single string containing all the parameters with the first character of IFS in between them. In practice this usually means all params as one string separated by spaces. So if your params are file1.txt, file2.txt and file3.txt, then "$*" hands that to the loop as one thing, namely file1.txt file2.txt file3.txt. That one thing, of course, does not exist as such.
"$@" expands to each of the parameters, one at a time, treated as a single "word". This is good because it will treat file1.txt, file2.txt and file3.txt from the earlier example correctly as three items. The loop will receive them correctly one at a time. In addition, "$@" is good since even if your filenames have spaces in them, you will get the right behavior since the loop will handle, e.g. filename 1, filename 2, filename 2 and a half as three items, one by one. (Normally the shell treats a space as marking the end of a word, so filename 1 would look like two items to the shell, filename and 1.
When they aren't quoted both $@ and $* expand the words of all the positional parameters. So in that case, they are effectively identical (and both a bad idea since they don't protect you from oddly named files).
There's a good visualization of all possibilities here: http://wiki.bash-hackers.org/scripting/posparams#mass_usage.
$* expands to a single string is incorrect; it should be "$*" expands to a single string ... Yes, you mentioned double quotes in the previous paragraph, but omitting them in the next is misleading.
$*and$@: mywiki.wooledge.org/BashGuide/…. It's subtle but very important for shell scripting.