From aa83d77271f0d2be72067058f78abd1780f3b69e Mon Sep 17 00:00:00 2001 From: wkj Date: Mon, 17 May 2004 03:22:35 +0000 Subject: More troff bits; if you want them elsewhere, feel free to repo copy them. --- src/cmd/index/doclean | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+) create mode 100644 src/cmd/index/doclean (limited to 'src/cmd/index/doclean') diff --git a/src/cmd/index/doclean b/src/cmd/index/doclean new file mode 100644 index 00000000..c1763fd9 --- /dev/null +++ b/src/cmd/index/doclean @@ -0,0 +1,27 @@ +#!/bin/sh + +awk ' # doclean + +# Input: "number tab IX string +# 107 IX self-reference #1186 - +# 281 TL APPENDIX A AMPL Reference Manual #26 - +# Output: string (tab) number +# excess spaces are removed output string +# note reversal of order; rest of programs expect it + +# This contains some special pleading for the AMPL book + +BEGIN { FS = OFS = "\t" } + +/\t(TL|H1|H2|H3|LASTPAGE)/ { next } # zap expected noise + +$0 !~ /[0-9ixv]+\tIX / { + print "doclean: non index line: " $0 | "cat 1>&2"; next +} + +{ sub(/IX +/, "", $2) # zap "IX " + sub(/ +#[0-9]+ .*$/, "", $2) # zap trailing blanks, slug, file + gsub(/ +/, " ", $2) # compress internal blanks + print $2, $1 # item (tab) page number +} +' $* -- cgit v1.2.3