AWK, SED to display a column based on Column name

Question

I am working on a shell script which has a lot of SED and AWK commands in it. Now, I am stuck up in a place where i couldn't proceed further.

For example :- If i run the below command. It prints the first column.

awk 'OFS="\t" {print $1;}' finename.txt

But, Is there a way to print a particular column using column name instead of column number. Guess there is no straight forward way to do it, But there should be some trick to do this.

Sample data(tab seperated)

  itemNumber  Price   Mango   Apple   Bannana
   112201      purchased   need-to-plan    purchased
   112202  55  yet-to-buy  yet-to-buy  purchased
   112202  67  need-to-plan    purchased   purchased
   112203  456 need-to-plan    need-to-plan    need-to-plan
   112203  33  need-to-plan    yet-to-buy  need-to-plan
   112204  456 need-to-plan    yet-to-buy  need-to-plan
   112204      yet-to-buy  purchased   need-to-plan
   112205  77  yet-to-buy  purchased   need-to-plan
   112205  99  yet-to-buy  purchased   yet-to-buy
   112206  0   yet-to-buy  purchased   yet-to-buy

200_OK · Accepted Answer · 2022-04-02 15:07:19Z

4

Something like this?

awk -vcol=Price '(NR==1){colnum=-1;for(i=1;i<=NF;i++)if($(i)==col)colnum=i;}{print $(colnum)}' filename.csv

I should explain:

-v sets a variable, so with -vcol=Price I'm declaring the column that I'm looking for.
The (NR==1) selects the first line and the code after that searches through the column headers for the one we want to print.
Finally we print the column.

For multiple columns the approach is similar:

awk -F'\t' -vcols=fun,coming '(NR==1){n=split(cols,cs,",");for(c=1;c<=n;c++){for(i=1;i<=NF;i++)if($(i)==cs[c])ci[c]=i}}{for(i=1;i<=n;i++)printf "%s" FS,$(ci[i]);printf "\n"}' filename.csv

edited Apr 2, 2022 at 15:07

200_OK

9,5232 gold badges35 silver badges55 bronze badges

answered Aug 2, 2015 at 15:35

Max Murphy

2,0111 gold badge23 silver badges34 bronze badges

Sign up to request clarification or add additional context in comments.

6 Comments

meuh Over a year ago

You might want to add -F'\t' because the data seems to have some empty columns, so the default whitespace separator will not keep the column indexes.

Max Murphy Over a year ago

Yes -F'\t' will do that. As usual you have lots of options. Tab separated input: BEGIN{IFS="\t"} output: BEGIN{OFS="\t"} and both BEGIN{FS="\t"}. But -F is the shortest of the lot so should win :-)

like_to_learn Over a year ago

@MaxMurphy - Thats Awesome. Exactly what i was looking for.

like_to_learn Over a year ago

@meuh - thanks for adding that. I was about to ask :-)

like_to_learn Over a year ago

@MaxMurphy - Any idea how to print multiple columns with column name ?

|

Vincent · Accepted Answer · 2020-03-06 13:19:25Z

0

Thanks @Max Murphy.

Personally needed awk -f' ' (so with space) and | column -t

In full:

awk -F' ' -vcols=pkts,bytes,source,destination '(NR==2){n=split(cols,cs,",");for(c=1;c<=n;c++){for(i=1;i<=NF;i++)if($(i)==cs[c])ci[c]=i}}{for(i=1;i<=n;i++)printf "%s" FS,$(ci[i]);printf "\n"}' | column -t

I used it to sort output of iptables -L -n -v --line-numbers on Debian

(Not a better answer to the question, but I think people appreciate a bit more elaboration.)

answered Mar 6, 2020 at 13:19

Vincent

1671 silver badge6 bronze badges

Collectives™ on Stack Overflow

AWK, SED to display a column based on Column name

2 Answers 2

6 Comments

Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

6 Comments

Comments

Linked

Related