Puzzle for Unix/Linux shell scripters

[dante]

Can the following be done easily with Unix/Linux shell command line?  It's easy in Python/Ruby but I'm curious how much more minimalist it could be (if it can be done at all) in shell:

Take these two files (there could be many more, but for this example just two) and combine them also adding a field and populate depending on file name.

Input

File0001.csv:
FieldB,FieldC
A,0
A,1
A,2

File0002.csv:
FieldB,FieldC
B,0
B,1
B,2

Output

FileCombined.csv
FieldA,FieldB,FieldC
0001,A,0
0001,A,1
0001,A,2
0002,B,0
0002,B,1
0002,B,2

FieldA = is generated on the file it came from.  Hopefully the field names/lines of text don't have to be treated as separate entities but just text.

[DanielT]

Not at my computer to be able to write the code but this is a no-brainer for shell just plain bash and awk.

[abjr]

Dealing with the headers in the input files and for the output makes it a little ugly, but this should work:

Code: [Select]

$ awk 'BEGIN { OFS=","; print "FieldA,FieldB,FieldC" } FNR>1 { print FILENAME, $0 }' File00*.csv | sed -r 's/^File([0-9]+)\.csv/\1/'
FieldA,FieldB,FieldC
0001,A,0
0001,A,1
0001,A,2
0002,B,0
0002,B,1
0002,B,2



[dante]

Thanks... you are absolutely right it's a little ugly.   I just need to dissect it to see what's going on.  Thanks!

[vimx]

Here is a simple one-liner solution with sed.  This assumes that the csv files are named consistently and the same fields are present in each.

Code: [Select]

$ echo "FieldA,FieldB,FieldC" >FileCombined.csv; for F in $(\ls File*[0-9].csv); do A=$(echo "$F" |sed "s/[^0-9]//g"); cat "$F" |tail -n +2 |sed "s/^/$A,/"; done >>FileCombined.csv
Cheers.