#!/bin/bash
# you can blame this mess on Terry Vessels, aka "grouch"
# released under the terms of the GPL, http://www.gnu.org/licenses/gpl.txt
# an attempt to create a script to generate some anchor tags
# in selected files (transcripts, etc) from groklaw
#
# s_file == source file
# d_file == destination, without anchors
# l_file == destination, with anchor links
s_file=$1
d_file=$2
l_file=$3

#if [ ! -f ${s_file} ]||[ -z ${s_file} ]||[ -z ${d_file} ]
#then
#	echo "Usage: $0 [source file] [destination file] [link file]"
#	exit
#fi
#if [ -e $2 ]
#then
#	echo "destination file $2 exists!"
#	echo "either choose another filename to write to, or rm $2"
#	exit
#fi
#if [ -e $3 ]
#then
#	echo "link file $3 exists!"
#	echo "either choose another filename to write to, or rm $2"
#	exit
#fi

# get everything between <!-- ARTICLE START -->
#   and <!-- ARTICLE END -->
# change all CR (\015) to LF (\012)
# put back LF in front of each < and after each >
# take out blank lines
# awk to insert an <a name="number">[[number]]</a> tag 
#   for each <p except <pre and each <br
art_start=$(grep -n '<!-- ARTICLE START -->' ${s_file} |\
	awk '{line_num=substr($1,0,index($1,":")-1); print line_num}')
art_end=$(grep -n '<!-- ARTICLE END -->' ${s_file} |\
	awk '{line_num=substr($1,0,index($1,":")-1); print line_num}')
art_len=$((1 + ${art_end} - ${art_start}))
head -${art_end} ${s_file} | tail -${art_len} |\
tr '\015' '\012' |\
sed -e 's/</\
</g' |\
sed -e 's/>/>\
/g' |\
grep '[^[:space:]].' | tee $d_file |\
awk '{FS=""}{
	if ($1=="<" && (($2=="p" && $3!="r") || ($2=="b" && $3=="r")))
	{
		print $0 "<a name=\"" NR "\">[["NR"]]</a><br>"
	}
	else
	{
		print $0
	}
}' > $l_file