Shell-write a file into shell variable - shell

Question

I have a file like this format:
a;b;c
e;d;f
how can I use shell to read the file detail information into variables?
I would have 6 variables to store the data.
more detailed information for this is that as the following shows:
I have written a script:
#!/bin/sh
unset ret
ret=0
if [ "$#" -ne 3 ]; then
logger -p err "Usage: $0 eth_name rule_file table_name"
ret=1
exit ret
fi
OFS=$IFS # store field separator
IFS=";" # define field separator
eth_name=$1 # ethernet device name
rule_file=$2 # input file name
table_name=$3 # lookup table name
logger -p notice "$0 $eth_name $rule_file $table_name"
unset a # reference to line array
unset i j # index
unset m n # dimension
### read route configuration
i=0
while read line
do
a=A$i
unset $a
declare -a $a='($line)'
i=$((i+1))
done < $rule_file
# store number of lines
m=$i
# function for apply route
add_route()
{
if [ "source" = "$1" ]; then
src_address=$(ifconfig $eth_name | sed -n 's/.*inet addr:\([0-9.]\+\)\s.*/\1/p')
ip rule add from $src_address lookup $table_name
ret=$?
logger -p notice "ip rule add from $src_address lookup $table_name $ret"
elif [ "default" = "$1" ]; then
ip route add default via $2 table $table_name
ret=$?
logger -p notice "ip route add default via $2 table $table_name $ret"
else
ipaddress_range=$1
gateway_ipaddress=$2
ip route add $ipaddress_range via $gateway_ipaddress dev $eth_name table $table_name
ret=$?
logger -p notice "ip route add $ipaddress_range via $gateway_ipaddress dev $eth_name table $table_name $ret"
fi
}
### apply route configuration
for ((i=0; i < $m; i++))
do
a=A$i
# get line size
# double escape '\\' for sub shell '``' and 'echo'
p0=`eval echo \\${$a[0]}`
p1=`eval echo \\${$a[1]}`
add_route $p0 $p1
done
IFS=$OFS
the rule file's format is as the following shows:
source;
default_route;172.20.5.192/26
default_gateway;172.20.5.254
172.17.23.64/26;172.20.5.254
172.31.252.0/24;172.20.5.254
172.31.254.0/24;172.20.5.254
10.217.1.0/24;172.20.5.254
10.217.2.0/24;172.20.5.254
this script is working normally under the bash environment, now my linux system is not having bash now, this script is not working now, how to change the script to make the script running?
the function for this script is very simple, write every line into the linux system's ip rule and ip route. need to throw 3 variables to make the script running.

Answer 1

You can easily achieve it with read.
#!/bin/bash
while IFS=";" read -r var1 var2 var3; do
command
done <file
exit 0
where:
IFS is your delimiter
varN are your vars, you can use any name you want
N.B. If you need stdin you need to use a file descriptor:
#!/bin/bash
exec 3<file
while IFS=";" read -r var1 var2 var3 <&3; do
command
done
exec 3>&-
exit 0
Further readings here.
N.B.#2 The command read is not the faster solution, usually after 5k lines lose 100 ms compared to other tools (E.G. awk).

Answer 2

The following parses your original sample text and saves fields to sequentially numbered variables, stripping values out of lines using parameter expansion, and doesn't care how many fields you have per line.
#!/bin/sh
i=0
while read line; do
while [ -n "$line" ]; do
eval this_$((i=i+1))="\${line%%;*}"
last="$line"
line="${line#*;}"
[ "$last" = "$line" ] && break
done
done < input.txt
I've tested this successfully with both bash and FreeBSD's /bin/sh (which is based on ash). (Note that FreeBSD's /bin/sh doesn't seem to like arithmetic expressions like $((i++)), but both shells I tested are fine with the notation above.)
But from the look of the script you updated your question with, this isn't what you need. It seems that you have input data with:
one record per line, and
two fields, separated by semicolons.
But I wonder if you even need to store things in variables. It seems to me you'd be looking more for the following type of structure.
#!/bin/sh
...
while IFS=";" read range gateway; do
case "$range" in
source)
: Note that $gateway is blank
;;
default_route)
: do something
;;
default_gateway)
: do something else
;;
[0-9]*)
ip route add "$range" via "$gateway" dev "$eth_name" table "$table_name"
;;
*)
printf 'ERROR: unknown range in rule file: %s\n' "$range" >&2
;;
esac
done < $rule_file
Additional input validation wouldn't hurt.