read variable from a file and pass it to sed

Question

I have two files: unknown.csv contains

aaa
bbb
ccc

and another file 01.txt as

;lksdjflk aaa lkdsfjdlk
aaa|xxlkjasd|
    |sadkl;k|lsa;dkl

when I define a variable as j=aaa and try

sed "s/${j}//g" 01.txt

it will find occurrences of aaa and remove it. But if I try to read the j from the unknown.csv as

j=$(sed '1q;d' unknown.csv)

and try

sed "s/${j}//g" 01.txt

nothing happens. echo $j for both cases show same result of aaa.

The goal is to read a variable from the unknow.csv file and use it to replace it in 01.txt file.

Thanks

You should probably have a look at [How to delete all occurrences of a list of words from a text file?](https://unix.stackexchange.com/q/322310/135943) — Wildcard, Aug 06 '18 at 20:01

RalfFriedl · Accepted Answer · 2018-08-06T20:38:56.993

4

Did you create the file unknown.csv on Windows? If so, it probably contains carriage return characters.

Try

(set -x; echo "$j" )

to show the value of $j.

If you want only line 1 and no '\r' at the end, use

sed -nr '1s/\r?$//p' unknown.csv

edited Aug 06 '18 at 20:38

answered Aug 06 '18 at 19:49

RalfFriedl

I created it in ubuntu but there is `\r` carriage character in `j` and it showed up when I used `set -x`. I can remove it using `j=$(sed -e '1q;d' unknown.csv|sed 's/\r//g')`. Is there a better way for that? Thanks – Dan R Aug 06 '18 at 20:00
3

`1q;d` nonsense aside, it's worth mentioning that working with unknown input requires pre-processing LHS (in this case`$j`) so as to escape any reserved characters. – don_crissti Aug 06 '18 at 20:10
@don_crissti - :) – mikeserv Aug 07 '18 at 02:52
@mikeserv - hey Mike, good to see you around again ! – don_crissti Aug 08 '18 at 21:26

1 Answers1