#!/bin/bash
#
# pak-web-scan -- 2008-09-01, last edit 2008-09-06
#
# discover semi-static ssi include files and provide a list of absolute 
#  filepaths along with the SSI directive tag for inclusion on the  
#  pak-web-site script command line
#
# Copyright (C) 2008 Grant Coady <grant@bugsplatter.id.au> GPLv2
#
# call: pak-web-site ... -v ssi_list="$(pak-web-scan)" ...
#
# list format:
#  path_1;ssi_tag_1:path_2;ssi_tag_2: [...] path_n;ssi_tag_n
#
# output list is sorted and contains no duplicate entries
#
# may be called from below $doc_root when performing a partial site update, 
#  no sanity checking for being called outside of the server document area
#
# all ssi file include directives must have full path from doc_root to 
#  include file, active (ie. required at web delivery time) SSI includes 
#  (eg. netdraw) not to use full paths and/or the .html file extension so 
#  they're not caught by this script (even if they were caught, they'd be 
#  thrown out by pak-web-site due to lack of <!-- ssi-file --> signature)
#

doc_root="/home/web/bugsplatter"

grep -r '<!--#include virtual="' . | grep -v archive | \
	awk -v doc_root="$doc_root" '
	{
		sub(/^[^:]+:/, ""); sub(/" -->.*$/, "")
		sub(/^.*<!--#include virtual="/, "")
		link = "<!--#include virtual=\"" $1 "\" -->"
	}
	/^\// && /\.(html|text)$/ { ++list[doc_root $1 ";" link]; next }
	END {
		n = asorti(list, sorted)
		for (i = 1; i <= n; i++) { s = (s ? s ":" : "") sorted[i] }
		print s
	}'
# end