162 lines
4.6 KiB
Bash
162 lines
4.6 KiB
Bash
# bash
|
|
|
|
#
|
|
# Copyright (c) 2022 Wind River Systems, Inc.
|
|
#
|
|
# SPDX-License-Identifier: Apache-2.0
|
|
#
|
|
|
|
: ${_GLOB_UTILS_TEST:=0}
|
|
: ${_GLOB_UTILS_LOW_NOISE:=1}
|
|
|
|
#
|
|
# Convert a glob pattern to a basic (grep/sed) regex.
|
|
#
|
|
# This function doesn't treat "/" and "." specially.
|
|
#
|
|
glob_to_basic_regex() {
|
|
|
|
# disable "set -x" to reduce noise in jenkins jobs
|
|
if (
|
|
[[ "$_GLOB_UTILS_LOW_NOISE" == 1 ]] && \
|
|
shopt -po xtrace | grep -q -- -o >/dev/null
|
|
) 2>/dev/null
|
|
then
|
|
: "$FUNCNAME: disabling debug trace"
|
|
set +x
|
|
local restore_xtrace=1
|
|
else
|
|
local restore_xtrace=0
|
|
fi
|
|
|
|
local len="${#1}"
|
|
local i c c2
|
|
local range_start range_len
|
|
local neg_char
|
|
local res
|
|
for ((i=0; i<len; ++i)) ; do
|
|
c="${1:$i:1}"
|
|
|
|
# character range [...]
|
|
if [[ "$c" == '[' ]] ; then
|
|
# find end bracket
|
|
range_len=""
|
|
# check if its a negative range
|
|
# negative ranges start with "!" in glob and "^" in regex
|
|
let ++i
|
|
neg_char=
|
|
if [[ "${1:$i:1}" == '!' ]] ; then
|
|
let ++i
|
|
neg_char='^'
|
|
fi
|
|
# at this point i refers to the 1st char in range
|
|
# range can't be empty, so we need to skip
|
|
# this first char, then search for ']'
|
|
range_start=$i
|
|
for ((++i; i<len; ++i)) ; do
|
|
if [[ "${1:$i:1}" == ']' ]] ; then
|
|
let range_len=i-range_start
|
|
break
|
|
fi
|
|
done
|
|
# end bracket found: append the (possibly negative) range
|
|
if [[ -n "$range_len" ]] ; then
|
|
res+='['
|
|
res+="$neg_char"
|
|
res+="${1:$range_start:$range_len}"
|
|
res+=']'
|
|
let i=range_start+range_len
|
|
# end bracket not found: append '\['
|
|
else
|
|
res+='\]'
|
|
fi
|
|
continue
|
|
fi
|
|
|
|
# Backslash is an escape char in glob, but not in basic regex,
|
|
# except when followed by a meta character: * { etc
|
|
# Surround next char with "[]"
|
|
if [[ "$c" == '\' ]] ; then
|
|
let ++i
|
|
# backslash at end of string: append a literal '\'
|
|
if [[ $i -ge $len ]] ; then
|
|
c2='\'
|
|
else
|
|
c2="${1:$i:1}"
|
|
fi
|
|
# we can't use this method with '[^]'
|
|
if [[ "$c2" != '^' ]] ; then
|
|
res+="[$c2]"
|
|
continue
|
|
fi
|
|
res+='\^'
|
|
continue
|
|
fi
|
|
|
|
# Escape ^ as \^ -- can't use square brackets
|
|
# because this is a negation character in ranges
|
|
if [[ "$c" == '^' ]] ; then
|
|
res+='\^'
|
|
continue
|
|
fi
|
|
|
|
# Escape these using square brackets:
|
|
# $. - have special meaning in regex
|
|
# /,!|#@ - these are not special, but are frequently
|
|
# used as separators in sed "s" command
|
|
if [[ "$c" == '$' || "$c" == '.' ||
|
|
"$c" == '/' || "$c" == ',' || "$c" == '!' ||
|
|
"$c" == '|' || "$c" == '#' || "$c" == '@' ]] ; then
|
|
res+="[$c]"
|
|
continue
|
|
fi
|
|
|
|
# "?" => "."
|
|
if [[ "$c" == '?' ]] ; then
|
|
res+='.'
|
|
continue
|
|
fi
|
|
|
|
# "*" => ".*"
|
|
if [[ "$c" == '*' ]] ; then
|
|
res+='.*'
|
|
continue
|
|
fi
|
|
|
|
# anything else: append as is
|
|
res+="$c"
|
|
done
|
|
echo "^${res}\$"
|
|
if [[ "$restore_xtrace" == 1 ]] ; then
|
|
set -x # debug output of this "set" is suppressed
|
|
set -x # execute it again, this time with "-x" already on
|
|
: "$FUNCNAME: restored debug trace"
|
|
fi
|
|
}
|
|
|
|
# unit tests
|
|
if [[ "$_GLOB_UTILS_TEST" == 1 ]] ; then
|
|
expect() {
|
|
local glob="$1"
|
|
local expected="$2"
|
|
local actual
|
|
actual="$(glob_to_basic_regex "$1")"
|
|
if [[ "$actual" != "$expected" ]] ; then
|
|
echo "${BASH_SOURCE}:${BASH_LINENO}: glob_to_basic_regex '$glob': expected '$expected' actual '$actual'" >&2
|
|
exit 1
|
|
fi
|
|
}
|
|
expect 'a[0-9]b' '^a[0-9]b$'
|
|
expect 'a[!0-9]b' '^a[^0-9]b$'
|
|
expect 'a?b' '^a.b$'
|
|
expect 'a*b' '^a.*b$'
|
|
expect 'a\*b' '^a[*]b$'
|
|
expect 'a^b$c' '^a\^b[$]c$'
|
|
expect '/foo/*' '^[/]foo[/].*$'
|
|
expect 'a.b' '^a[.]b$'
|
|
expect 'a\[b' '^a[[]b$'
|
|
expect 'a[a-z]b[!A-Z]c[!0-9[]d!^' '^a[a-z]b[^A-Z]c[^0-9[]d[!]\^$'
|
|
expect 'abc\' '^abc[\]$'
|
|
expect 'a\b' '^a[b]$'
|
|
fi
|