Blame


1 08708877 2003-11-25 devnull # turn output of mkindex into form needed by dict
2 08708877 2003-11-25 devnull BEGIN {
3 08708877 2003-11-25 devnull if(ARGC != 2) {
4 08708877 2003-11-25 devnull print "Usage: awk -F' ' -f canonind.awk rawindex > index"
5 08708877 2003-11-25 devnull exit 1
6 08708877 2003-11-25 devnull }
7 08708877 2003-11-25 devnull file = ARGV[1]
8 08708877 2003-11-25 devnull ARGV[1] = ""
9 08708877 2003-11-25 devnull while ((getline < file) > 0) {
10 08708877 2003-11-25 devnull for(i = 2; i <= NF; i++) {
11 08708877 2003-11-25 devnull w = $i
12 08708877 2003-11-25 devnull if(length(w) == 0)
13 08708877 2003-11-25 devnull continue
14 08708877 2003-11-25 devnull b = index(w, "(")
15 08708877 2003-11-25 devnull e = index(w, ")")
16 08708877 2003-11-25 devnull if(b && e && b < e) {
17 08708877 2003-11-25 devnull w1 = substr(w, 1, b-1)
18 08708877 2003-11-25 devnull w2 = substr(w, b+1, e-b-1)
19 08708877 2003-11-25 devnull w3 = substr(w, e+1)
20 08708877 2003-11-25 devnull printf "%s%s\t%d\n", w1, w3, $1 > "junk"
21 08708877 2003-11-25 devnull printf "%s%s%s\t%d\n", w1, w2, w3, $1 > "junk"
22 08708877 2003-11-25 devnull } else
23 08708877 2003-11-25 devnull printf "%s\t%d\n", w, $1 > "junk"
24 08708877 2003-11-25 devnull }
25 08708877 2003-11-25 devnull }
26 08708877 2003-11-25 devnull system("sort -u -t' ' +0f -1 +0 -1 +1n -2 < junk")
27 08708877 2003-11-25 devnull system("rm junk")
28 08708877 2003-11-25 devnull exit 0
29 08708877 2003-11-25 devnull }