brendano / awkspeed
watch download tarball
public
Forks3Right
Watchers4Right
Description: Speed testing for a data munging task
Homepage: http://anyall.org/blog/?p=652
Clone URL: git://github.com/brendano/awkspeed.git
awkspeed / 2num_3line.awk
100644 8 lines (6 sloc) 0.359 kb
raw blame history 
1
2
3
4
5
6
7
8

# USAGE: awk -f 2num.awk file1 file2 file2
# Take (item, feat, value) triplets and numberize them as sparse matrices.
# Each file should have item numbers count from 1, but feat numbers are global.
 
!imap[FILENAME,$1] {imap[FILENAME,$1] = ++I[FILENAME]}
!jmap[$2] {jmap[$2] = ++J; print $2 > "vocab"}
{ print imap[FILENAME,$1], jmap[$2], $3 > (FILENAME "n")}