Find and replace in same string in bash - regex

Question

I have a string with some words in them, example a=1 b=2 c=3 a=50. Now I want to parse this and create another string a=50 b=2 c=3 which is essentially the same as above except that if the same phrase before the = is encountered for the second time the first one is over written with the latest one, so in the end there are only unique phrases on the left of =. Here is what I got till now:
a="a=1 b=2 c=3 a=50"
o=()
for i in $a
do
reg=${i%=*}
if [[ ${o[*]} == *"$reg"* ]]
then
o=$(echo ${o[*]} | sed -e "s/\$reg=\S/\$i")
else
o+=( $i )
fi
done
What am I doing wrong here?

Answer 1

I'd take an entirely different approach, not based on regular expressions or string rewriting.
declare -A values=( ) # Initialize an associative array ("hash", "map")
while IFS= read -r -d' ' word; do # iterate over input words, separated by spaces
if [[ $word = *=* ]]; then # ignore any word that doesn't have an "=" in it
values[${word%%=*}]=${word#*=} # add everything before the "=" as a key...
fi # ...with everything after the "=" as a value
done
for key in "${!values[#]}"; do # Then iterate over keys we found
value="${values[$key]}" # ...extract the values for each...
printf '%s=%s ' "$key" "$value" # ...and print the pairs.
done
echo # When done iterating, print a newline.
Because the words are being processed first-to-last through the string, updates take effect before the print loop is reached.

Answer 2

Using awk
$ awk -F= -v RS=" |\n" '{a[$1]=$2} END{for (k in a) printf "%s=%s ",k,a[k]}' <<<"a=1 b=2 c=3 a=50"
a=50 b=2 c=3
How it works:
-F=
Set the field separator to be an equal sign.
-v RS=" |\n"
Set the record separator to be either a space or a newline.
a[$1]=$2
Update associative array a with the latest value.
END{for (k in a) printf "%s=%s ",k,a[k]}
In no particular order, print out the final values.
Using bash
Like Charles Duffy's approach, this uses read -d" " to parse the string. This approach, however, uses IFS="=" to separate names and values.
Two loops are required. The first gathers the values. The second reassembles the updated values in the original order:
a="a=1 b=2 c=3 a=50"
declare -A b
while IFS== read -d" " name value
do
b["$name"]="$value"
done <<<"$a "
declare -A seen
while IFS== read -d" " name value
do
[ "${seen[$name]}" ] || o="$o $name=${b["$name"]}"
seen[$name]=1
done <<<"$a "
echo "$o"

Answer 3

Easily done with perl:
echo "a=1 b=2 c=3 a=50" \
| sed "s/ /\n/g" \
| perl -e '
my %hash = ();
while(<>){
$line = $_;
if($line =~ m/(\S+)=(\S+)/) {
$hash{$1} = $2;
}
}
for $key (sort keys %hash) {
print "$key=$hash{$key}\n";
}'
...or, all on one line:
echo "a=1 b=2 c=3 a=50" | sed "s/ /\n/g" | perl -e 'my %hash = (); while(<>){ $line = $_; if($line =~ m/(\S+)=(\S+)/) { $hash{$1} = $2; } } for $key (sort keys %hash) { print "$key=$hash{$key}\n"; }'