This question already has answers here:
How can I store a command in a variable in a shell script?
(12 answers)
Closed 4 years ago.
These work as advertised:
grep -ir 'hello world' .
grep -ir hello\ world .
These don't:
argumentString1="-ir 'hello world'"
argumentString2="-ir hello\\ world"
grep $argumentString1 .
grep $argumentString2 .
Despite 'hello world' being enclosed by quotes in the second example, grep interprets 'hello (and hello\) as one argument and world' (and world) as another, which means that, in this case, 'hello will be the search pattern and world' will be the search path.
Again, this only happens when the arguments are expanded from the argumentString variables. grep properly interprets 'hello world' (and hello\ world) as a single argument in the first example.
Can anyone explain why this is? Is there a proper way to expand a string variable that will preserve the syntax of each character such that it is correctly interpreted by shell commands?
Why
When the string is expanded, it is split into words, but it is not re-evaluated to find special characters such as quotes or dollar signs or ... This is the way the shell has 'always' behaved, since the Bourne shell back in 1978 or thereabouts.
Fix
In bash, use an array to hold the arguments:
argumentArray=(-ir 'hello world')
grep "${argumentArray[#]}" .
Or, if brave/foolhardy, use eval:
argumentString="-ir 'hello world'"
eval "grep $argumentString ."
On the other hand, discretion is often the better part of valour, and working with eval is a place where discretion is better than bravery. If you are not completely in control of the string that is eval'd (if there's any user input in the command string that has not been rigorously validated), then you are opening yourself to potentially serious problems.
Note that the sequence of expansions for Bash is described in Shell Expansions in the GNU Bash manual. Note in particular sections 3.5.3 Shell Parameter Expansion, 3.5.7 Word Splitting, and 3.5.9 Quote Removal.
When you put quote characters into variables, they just become plain literals (see http://mywiki.wooledge.org/BashFAQ/050; thanks #tripleee for pointing out this link)
Instead, try using an array to pass your arguments:
argumentString=(-ir 'hello world')
grep "${argumentString[#]}" .
In looking at this and related questions, I'm surprised that no one brought up using an explicit subshell. For bash, and other modern shells, you can execute a command line explicitly. In bash, it requires the -c option.
argumentString="-ir 'hello world'"
bash -c "grep $argumentString ."
Works exactly as original questioner desired. There are two restrictions to this technique:
You can only use single quotes within the command or argument strings.
Only exported environment variables will be available to the command
Also, this technique handles redirection and piping, and other shellisms work as well. You also can use bash internal commands as well as any other command that works at the command line, because you are essentially asking a subshell bash to interpret it directly as a command line. Here's a more complex example, a somewhat gratuitously complex ls -l variant.
cmd="prefix=`pwd` && ls | xargs -n 1 echo \'In $prefix:\'"
bash -c "$cmd"
I have built command processors both this way and with parameter arrays. Generally, this way is much easier to write and debug, and it's trivial to echo the command you are executing. OTOH, param arrays work nicely when you really do have abstract arrays of parameters, as opposed to just wanting a simple command variant.
Related
I have a Bash script which generates, stores and modifies values in an array. These values are later used as arguments for a command.
For a MCVE I thought of an arbitrary command bash -c 'echo 0="$0" ; echo 1="$1"' which explains my problem. I will call my command with two arguments -option1=withoutspace and -option2="with space". So it would look like this
> bash -c 'echo 0="$0" ; echo 1="$1"' -option1=withoutspace -option2="with space"
if the call to the command would be typed directly into the shell. It prints
0=-option1=withoutspace
1=-option2=with space
In my Bash script, the arguments are part of an array. However
#!/bin/bash
ARGUMENTS=()
ARGUMENTS+=('-option1=withoutspace')
ARGUMENTS+=('-option2="with space"')
bash -c 'echo 0="$0" ; echo 1="$1"' "${ARGUMENTS[#]}"
prints
0=-option1=withoutspace
1=-option2="with space"
which still shows the double quotes (because they are interpreted literally?). What works is
#!/bin/bash
ARGUMENTS=()
ARGUMENTS+=('-option1=withoutspace')
ARGUMENTS+=('-option2=with space')
bash -c 'echo 0="$0" ; echo 1="$1"' "${ARGUMENTS[#]}"
which prints again
0=-option1=withoutspace
1=-option2=with space
What do I have to change to make ARGUMENTS+=('-option2="with space"') work as well as ARGUMENTS+=('-option2=with space')?
(Maybe it's even entirely wrong to store arguments for a command in an array? I'm open for suggestions.)
Get rid of the single quotes. Write the options exactly as you would on the command line.
ARGUMENTS+=(-option1=withoutspace)
ARGUMENTS+=(-option2="with space")
Note that this is exactly equivalent to your second option:
ARGUMENTS+=('-option1=withoutspace')
ARGUMENTS+=('-option2=with space')
-option2="with space" and '-option2=with space' both evaluate to the same string. They're two ways of writing the same thing.
(Maybe it's even entirely wrong to store arguments for a command in an array? I'm open for suggestions.)
It's the exact right thing to do. Arrays are perfect for this. Using a flat string would be a mistake.
I am trying to do string substitution in bash, want to understand it better.
I crafted a success case like this:
a=abc_de_f
var=$a
echo ${var//_/-}
outout is abc-de-f. This works.
However, the following script fails:
a=abc_de_f
echo ${$a//_/-}
The error message is ${$a//_/-}: bad substitution.
It seems like related to how we can use a variable in substitution. Why this fails? How bash handles variables in this case?
Also, what is the best practice to handle escape characters in bash string substitution?
In the second case, you don't need the second $ as a is the string.
a=abc_de_f
echo ${a//_/-}
If you wanted to add a level of indirection, you can use ! before the variable as in
a=abc_de_f
b=a
echo ${b//_/-}
will output a, while
echo ${!b//_/-}
will output abc-de-f.
See here for a discussion on the art of escaping in BASH
I have a BASH script that has a long set of arguments and two ways of calling it:
my_script --option1 value --option2 value ... etc
or
my_script val1 val2 val3 ..... valn
This script in turn compiles and runs a large FORTRAN code suite that eventually produces a netcdf file as output. I already have all the metadata in the netcdf output global attributes, but it would be really nice to also include the full run command one used to create that experiment. Thus another user who receives the netcdf file could simply reenter the run command to rerun the experiment, without having to piece together all the options.
So that is a long way of saying, in my BASH script, how do I get the last command entered from the parent shell and put it in a variable? i.e. the script is asking "how was I called?"
I could try to piece it together from the option list, but the very long option list and two interface methods would make this long and arduous, and I am sure there is a simple way.
I found this helpful page:
BASH: echoing the last command run
but this only seems to work to get the last command executed within the script itself. The asker also refers to use of history, but the answers seem to imply that the history will only contain the command after the programme has completed.
Many thanks if any of you have any idea.
You can try the following:
myInvocation="$(printf %q "$BASH_SOURCE")$((($#)) && printf ' %q' "$#")"
$BASH_SOURCE refers to the running script (as invoked), and $# is the array of arguments; (($#)) && ensures that the following printf command is only executed if at least 1 argument was passed; printf %q is explained below.
While this won't always be a verbatim copy of your command line, it'll be equivalent - the string you get is reusable as a shell command.
chepner points out in a comment that this approach will only capture what the original arguments were ultimately expanded to:
For instance, if the original command was my_script $USER "$(date +%s)", $myInvocation will not reflect these arguments as-is, but will rather contain what the shell expanded them to; e.g., my_script jdoe 1460644812
chepner also points that out that getting the actual raw command line as received by the parent process will be (next to) impossible. Do tell me if you know of a way.
However, if you're prepared to ask users to do extra work when invoking your script or you can get them to invoke your script through an alias you define - which is obviously tricky - there is a solution; see bottom.
Note that use of printf %q is crucial to preserving the boundaries between arguments - if your original arguments had embedded spaces, something like $0 $* would result in a different command.
printf %q also protects against other shell metacharacters (e.g., |) embedded in arguments.
printf %q quotes the given argument for reuse as a single argument in a shell command, applying the necessary quoting; e.g.:
$ printf %q 'a |b'
a\ \|b
a\ \|b is equivalent to single-quoted string 'a |b' from the shell's perspective, but this example shows how the resulting representation is not necessarily the same as the input representation.
Incidentally, ksh and zsh also support printf %q, and ksh actually outputs 'a |b' in this case.
If you're prepared to modify how your script is invoked, you can pass $BASH_COMMANDas an extra argument: $BASH_COMMAND contains the raw[1]
command line of the currently executing command.
For simplicity of processing inside the script, pass it as the first argument (note that the double quotes are required to preserve the value as a single argument):
my_script "$BASH_COMMAND" --option1 value --option2
Inside your script:
# The *first* argument is what "$BASH_COMMAND" expanded to,
# i.e., the entire (alias-expanded) command line.
myInvocation=$1 # Save the command line in a variable...
shift # ... and remove it from "$#".
# Now process "$#", as you normally would.
Unfortunately, there are only two options when it comes to ensuring that your script is invoked this way, and they're both suboptimal:
The end user has to invoke the script this way - which is obviously tricky and fragile (you could however, check in your script whether the first argument contains the script name and error out, if not).
Alternatively, provide an alias that wraps the passing of $BASH_COMMAND as follows:
alias my_script='/path/to/my_script "$BASH_COMMAND"'
The tricky part is that this alias must be defined in all end users' shell initialization files to ensure that it's available.
Also, inside your script, you'd have to do extra work to re-transform the alias-expanded version of the command line into its aliased form:
# The *first* argument is what "$BASH_COMMAND" expanded to,
# i.e., the entire (alias-expanded) command line.
# Here we also re-transform the alias-expanded command line to
# its original aliased form, by replacing everything up to and including
# "$BASH_COMMMAND" with the alias name.
myInvocation=$(sed 's/^.* "\$BASH_COMMAND"/my_script/' <<<"$1")
shift # Remove the first argument from "$#".
# Now process "$#", as you normally would.
Sadly, wrapping the invocation via a script or function is not an option, because the $BASH_COMMAND truly only ever reports the current command's command line, which in the case of a script or function wrapper would be the line inside that wrapper.
[1] The only thing that gets expanded are aliases, so if you invoked your script via an alias, you'll still see the underlying script in $BASH_COMMAND, but that's generally desirable, given that aliases are user-specific.
All other arguments and even input/output redirections, including process substitutiions <(...) are reflected as-is.
"$0" contains the script's name, "$#" contains the parameters.
Do you mean something like echo $0 $*?
command 'which' shows the link to a command.
command 'less' open the file.
How can I 'less' the file as the output of 'which'?
I don't want to use two commands like below to do it.
=>which script
/file/to/script/fiel
=>less /file/to/script/fiel
This is a use case for command substitution:
less -- "$(which commandname)"
That said, if your shell is bash, consider using type -P instead, which (unlike the external command which) is built into the shell:
less -- "$(type -P commandname)"
Note the quotes: These are important for reliable operation. Without them, the command may not work correctly if the filename contains characters inside IFS (by default, whitespace) or can be evaluated as a glob expression.
The double dashes are likewise there for correctness: Any argument after them is treated as positional (as per POSIX Utility Syntax Guidelines), so even if a filename starting with a dash were to be returned (however unlikely this may be), it ensures that less treats that as a filename rather than as the beginning of a sequence of options or flags.
You may also wish to consider honoring the user's pager selection via the environment variable $PAGER, and using type without -P to look for aliases, shell functions and builtins:
cmdsource() {
local sourcefile
if sourcefile="$(type -P -- "$1")"; then
"${PAGER:-less}" -- "$sourcefile"
else
echo "Unable to find source for $1" >&2
echo "...checking for a shell builtin:" >&2
type -- "$1"
fi
}
This defines a function you can run:
cmdsource commandname
You should be able to just pipe it over, try this:
which script | less
I want to get some data from a HTTP server. What it sends me depends on what I put in a POST request.
What I put in the INPUT_TEXT field is a sequence of words. When I run the following command, I get good looking output.
$ curl http://localhost:59125/process -d INPUT_TEXT="here are some words"
I want a bash script to take some string as a command line argument, and pass it appropriately to curl. The first thing I tried was to put the following in a script:
sentence=$1
command="curl http://localhost:59125/process -d INPUT_TEXT=\"${sentence}\""
$command
I then run the script like so:
$ ./script "here are some words"
But then I get a curl Couldn't resolve host error for each of "are", "some", and "words". It would seem that "here" got correctly treated as the INPUT_TEXT, but the rest of the words were then considered to be hosts, and not part of the option.
So I tried:
command=("curl" "http://localhost:59125/process" "-d" "INPUT_TEXT='$sentence'")
${command[#]}
I got the same output as the first script. I finally got what I wanted with:
result=$(curl http://localhost:59125/process -d INPUT_TEXT="${sentence}")
echo $result
I'm still unsure as to what the distinction is. In the first two cases, when I echoed out the contents of command, I get exactly what I input from the interactive Bash prompt, which had worked fine. What caused the difference?
The following will work:
command=("curl" "http://localhost:59125/process"
"-d" "INPUT_TEXT=$sentence")
"${command[#]}"
That has two changes from yours:
I removed the incorrect quotes around $sentence since you don't want to send quotes to the server (as far as I can see).
I put double-quotes around the use of "${command[#]}". Without the double quotes, the array's elements are concatenated with spaces between them and then the result is word-split. With double quotes, the individual array elements are used as individual words.
The second point is well-explained in the bash FAQ and a bunch of SO answers dealing with quotes.
The important thing to understand is that quotes only quote when a command is parsed. A quote which is a character in a variable is just a character; it is not reinterpreted when the value of the variable expanded. Whitespace in the variable is used for word-splitting if the variable expansion is unquoted; the fact that the whitespace was quoted in the the command which defined the variable is completely irrelevant. In this sense, bash is just the same as any other programming language.