CODEWITHSUNDEEP
bashshellscripting
nbohari
nbohari

Reputation: 21

Bash: Reading variables containing single or double words from a file

I am trying to read contents of a file into variables using bash v4.1.x The input file may look like this:

1373232436 785907701 "abc 245" 0 1
1373232436 1048824909 "def pqr" 1 0
1373232486 785907701 "uvw ghn" 0 1
1373232486 1048824909 "1109 xyz" 1 0

If I use 

cat <filename>|while read col1 col2 col3 col4 col5 col6
do
...
...
done

I should get col3 values to be 

"abc 245"
"def pqr"
"uvw ghn"
"1109 xyz"

Upvotes: 2

Views: 189

Answers (4)

chepner
chepner

Reputation: 531335

Assuming that only the third field can be quoted as shown, I would use a regular expression to split each line into columns.

while read -r line; do
    [[ $line =~ ^(.*)\ (.*)\ (\".*\")\ (.*)\ (.*)$ ]] || continue
    col1=${BASH_REMATCH[1]}
    col2=${BASH_REMATCH[2]}
    col3=${BASH_REMATCH[3]}
    col4=${BASH_REMATCH[4]}
    col5=${BASH_REMATCH[5]}
done < file.txt

Upvotes: 2

anishsane
anishsane

Reputation: 20980

You can also use gawk+FPAT

$ gawk 'BEGIN{FPAT="([^ ]*)|\"([^\"]*)\""} {print "\nLine: " NR; for(i=1;i<=NF;i++){print $i}}'  test.csv
Line: 1
1373232436
785907701
"abc 245"
0
1

Line: 2
1373232436
1048824909
"def pqr"
1
0

Line: 3
1373232486
785907701
"uvw ghn"
0
1

Line: 4
1373232486
1048824909
"1109 xyz"
1
0

Note1: FPAT is gawk feature. May not be available with your awk version.
Note2: Just realized, that incidentally, example in the link I mentioned above deals with requirement very similar to yours, though I had written that regex myself. :-)

Upvotes: 1

anubhava
anubhava

Reputation: 785276

You can actually use:

grep -Eo '"[^"]*"|\w+' file

to read each quoted column separately from your input file.

You can use a script like this:

#!/bin/bash

numcols=$(awk -F '"[^"]*"|[^[:blank:]]+' '{print NF-1; exit}' file)

n=1
while read -r w; do
    echo "$w"
    (( (n++ % numcols) )) || echo "<-- End of line $(( (n / numcols) )) -->"
done < <(grep -Eo '"[^"]*"|\w+' file)

For your input file it gives:

1373232436
785907701
"abc 245"
0
1
<-- End of line 1 -->
1373232436
1048824909
"def pqr"
1
0
<-- End of line 2 -->
1373232486
785907701
"uvw ghn"
0
1
<-- End of line 3 -->
1373232486
1048824909
"1109 xyz"
1
0
<-- End of line 4 -->

You can process them individually instead of doing echo "$w".

Upvotes: 0

anishsane
anishsane

Reputation: 20980

I think, your input file is essentially a csv file, with field separator=space.
Then you can use csvtool:

csvtool -t " " cols 1-6 test.csv | while IFS=, read col1 col2 col3 col4 col5 col6; do
    ...
    ...
done

run csvtool --help for more details.

Note: There will NOT be surrounding double quotes around col3 data. So you would get abc 245 & not "abc 245" in the value.

Upvotes: 0

Related Questions