[SOLVED] Bash / grep regex validation for path names

jomann · 02-01-2012, 06:55 AM

Hi all,

it's about Bash & regular expressions...

I try to validate the user input of a path.
A valid path looks for instance like this:

//server/sharename
//server-01/share_in/user/
//ser_01-ver/sha-re5

What I have so far:

Code:

if ! echo $share | grep -q -e '[^-0-9A-Z_a-z]' -e '^[^0-9A-Za-z]' \
			-e '[^0-9A-Za-z]$' -e '-_' -e '_-' ; then

echo OK

fi

This validates all constellations of A-Z and 0-9, etc., but I have problems with the "/" (can appear somewhere in the middle and at the end of the path) and "//" (at the beginning of the path, leading double-slashes).

Any help very much appreciated.
Thanks & regards,
jomann.

Dr_P_Ross · 02-01-2012, 08:11 AM

I think you want something like this:

Code:

if echo $1 | grep -x -E -e  '//[-_A-Za-z0-9]+(/[-_A-Za-z0-9]*)*';
then
 echo ' ok:' $1
else
 echo 'bad:' $1
fi

The -x tells grep to only accept if the whole line is matched. The -E tells grep to use extended regular expressions, thus permitting the use of '+' to mean 'one or more occurrences'. The regular expression expects a double slash, followed by some non-empty text, optionally followed by zero or more path components each starting with a slash.

Of course, this will accept something like '//server_09/////' because a path component may be a slash by itself; up to you whether that is OK.

jomann · 02-01-2012, 08:42 AM

Thanks Dr_P_Ross,

this is working nicely for my purpose.

Best regards,
jomann.

grail · 02-01-2012, 09:47 AM

You could also just do it all in bash:

Code:

#!/bin/bash

regex='^//[-_A-Za-z0-9]+(/[-_A-Za-z0-9]*)*'

if [[ $1 =~ $regex ]]
then
	echo yes
else
	echo no
fi

sailnfool · 09-14-2022, 10:08 AM

I found the regex listed above unsatisfactory since it was not sufficiently generic.
I propose a modification to the regex:

Code:

    bre_filedirname='^([\/-_A-Za-z0-9\.]+[-_A-Za-z0-9\.]*)*$'

I offer a long test program that can be invoked with "bash regex2.sh -v"
However, I am not 100% happy because this expression succeeds when it should fail when presented with the sample:
/abad/file^name

Code:

#!/bin/bash
scriptname=${0##*/}

bre_filedirname='^([\/-_A-Za-z1-9\.]+[-_A-Za-z0-9\.]*)*$'

TESTNAME="Test of function regex (bfunc.regex) from"
TESTNAMELOC="https://github.com/sailnfool/bfunc"
USAGE="\n${0##*/} [-[hv]]\n
\t\tVerifies that the regular expression(s for filedirnames\n
\t\twork correctly.\n
\t\tNormally emits only PASS|FAIL message\n
\t-h\t\tPrint this message\n
\t-v\t\tVerbose mode to show values\n
"

localNUMARGS=0
optionargs="hv"
verbosemode="FALSE"
verboseflag=""
fail=0

while getopts ${optionargs} name
do
	case ${name} in
		h)
			echo -e "${USAGE}" >&2
			if [[ "${verbosemode}" == "TRUE" ]] ; then
				echo -e "${DEBUG_USAGE}" >&2
			fi
			exit 0
			;;
		v)
			verbosemode="TRUE"
			verboseflag="-v"
			;;
		\?)
			echo -e "invalid option: -$OPTARG" >&2
			echo -e "${USAGE}" >&2
			exit 1
			;;
	esac
done

shift $(( ${OPTIND} - 1 ))

if [[ $# -lt ${localNUMARGS} ]] ; then
	echo -e "${0##*/} Insufficient arguments, expcted "\
		"${localNUMARGS}, got $@" >&2
	echo -e "${USAGE}" >&2
	exit 2
fi

declare -a tv
declare -a tt
########################################################################
# tv is short for testvalue
# tt is short for testtype
########################################################################
tv+=("/")
tt+=(${bre_filedirname})

tv+=("//")
tt+=(${bre_filedirname})

tv+=("///")
tt+=(${bre_filedirname})

tv+=("/.")
tt+=(${bre_filedirname})

tv+=("/./")
tt+=(${bre_filedirname})

tv+=("/../.")
tt+=(${bre_filedirname})

tv+=("/usr/bin/b3sum")
tt+=(${bre_filedirname})

tv+=("/etc/dhcp/dhclient.conf")
tt+=(${bre_filedirname})

tv+=("/etc/dhcp/dhclient-exit-hooks.d")
tt+=(${bre_filedirname})

tv+=("/etc/gdm3/config-error-dialog.sh")
tt+=(${bre_filedirname})


for ((ti=0; ti<${#tv[@]}; ti++))
do
	if [[ ! "${tv[${ti}]}" =~ ${tt[${ti}]} ]] ; then
		((fail++))
		if [[ "${verbosemode}" == "TRUE" ]] ; then
			echo -e "Failed Test of ${tv[${ti}]}, " \
				"expected ${tt[${ti}]}" >&2
		fi
	fi
done

declare -a fv
declare -a ft
########################################################################
# fv is short for failvalue
# ft is short for failtype
########################################################################
fv+=("/a bad/filename")
ft+=(${bre_filedirname})

fv+=("/abad/file^name")
ft+=(${bre_filedirname})

########################################################################
# In this case if the pattern match succeeds then it
# is broken
########################################################################
for ((ti=0; ti<${#fv[@]}; ti++))
do
	if [[ "${fv[${ti}]}" =~ ${ft[${ti}]} ]] ; then
		((fail++))
		if [[ "${verbosemode}" == "TRUE" ]] ; then
			echo -e "Failed Test succeeded with "\
				"${fv[${ti}]},\n" \
				"expected fail for ${ft[${ti}]}" >&2
		fi
	fi
done

exit ${fail}

dugan · 09-14-2022, 10:40 AM

People….

Don’t do this. If you want to know if a path exists, just “stat” it.

astrogeek · 09-14-2022, 10:44 AM

@sailinfool - You have replied to a thread that has been dormant for 10+ years, with an example that is not especially relevant to the question.

If you have a problem that you think to be similar, please open your own thread and post a minimal code example which illustrates your specific problem so that those currently active may provide relevant replies.

Please review the Site FAQ for guidance in asking well formed questions. Especially visit the link from that page, How to Ask Questions the Smart Way for discussion of things to consider when asking others for help.