#!/bin/rc

nl='
';

fn readlinks {
	awk '
	function isurl(s) { return s ~ /^\.\/|^\.\.\/|^http:\/\/|^https:\/\/|^gemini:\/\// }
	
	# Image: <URL> [<X>x<Y>px] (<ALT>) <CAPTION> Source: <SOURCE URL>
	/^(Image|Banner|Icon): [^ ]+ \[[0-9]+x[0-9]+px\]/ && isurl((url = $2)) { print url; next; }
	# Video: <URL> [<X>x<Y>px] (<ALT>)
	/^Video: [^ ]+ \[[0-9]+x[0-9]+px\] \(Preview: [^ )]+\)/ && isurl((url = $2)) && isurl((previewurl = substr($5, 1, length($5) - 1))) { print url; print previewurl; next; }
	# Audio: <URL>
	/^Audio: [^ ]+$/ && isurl((url = $2)) { print url; next; }
	# <ALT>: <URL> (<APPENDIX ALT>: <APPENDIX URL>)
	match($0, /: [^ ]+ \(/) && (altend = RSTART) && (xaltstart = RSTART + RLENGTH) && isurl((url = substr($0, RSTART + 2, RLENGTH - 4))) && match($0, /: [^ ]+\)$/) && (xaltend = RSTART) && isurl((xurl = substr($0, RSTART + 2, RLENGTH - 3))) { print url; print xurl; next; }
	# <PREFIX>: <URL> (<ALT>)
	match($0, /: [^ ]+ \(/) && isurl((url = substr($0, RSTART + 2, RLENGTH - 4))) && /\)$/ { print url; next; }
	# <URL> (<APPENDIX ALT>: <APPENDIX URL>)
	match($0, /^[^ ]+ \(/) && (xaltstart = RLENGTH + 1) && match($0, /: [^ ]+\)$/) && (xaltend = RSTART) && isurl((url = $1)) && isurl((xurl = substr($NF, 1, length($NF) - 1))) { print url; print xurl; next; }
	# <URL> (<ALT>) — <SUFFIX>
	/^[^ ]+ \(/ && match($0, /\) — /) && isurl((url = $1)) { print url; next; }
	# <URL> (<ALT>)
	/^[^ ]+ \(/ && /\)$/ && isurl((url = $1)) { print url; next; }
	# <URL>
	isurl((url = $0)) { print url; next; }
	# See also: <URL>
	# Further reading: <URL>
	# For more details, see: <URL>
	/(See also|Further reading|For more details, see): [^ ]+$/ && isurl((url = $NF)) { print url; next; }
	# - <ALT>: <URL>
	# .* forces last occurrence.
	match($0, /- .*: /) && isurl((url = substr($0, RLENGTH + 1))) { print url; next; }
	# - <URL>
	/^- / && isurl((url = substr($0, 3))) { print url; next; }
	# <ALT>: <URL>
	# .* forces last occurrence.
	match($0, /.*: /) && isurl((url = substr($0, RLENGTH + 1))) { print url; next; }
	# — <AUTHOR>, <URL>
	/^— / && /, [^ ]+$/ && isurl((url = $NF)) { print url; next; }
	' $1
}

echo 'strict graph {';
echo '	layout=fdp'
echo '	node [label="" shape=circle width=0.1 height=0.1 fixedsize=true style=filled fillcolor=white]'

for (dir in `{ walk -d }) {
	if (test -e `$nl{ echo $dir^/index.txt }) echo '	'"$dir/";
	if not echo '	'"$dir/" '[style="filled,dotted"]';
}

for (dir in `{ walk -d }) {
	if (test ! -e `$nl{ echo $dir^/index.txt }) {
		for (subdir in `{ walk -d -n1,1 $dir }) {
			echo '	'"$dir/" '--' "$subdir/" '[style="filled,dotted"]';
		}
	}
}

for (file in `{ walk -f | grep '(^|/)index.txt$' }) {
	for (url in `$nl{ readlinks $file }) {
		if (~ $url ./* ../*) {
			if (~ $url */index.txt */) {
				relurl=`$nl{ cleanname `$nl{ basename -d $file }^/`$nl{ echo $url } };
				if (~ $relurl */index.txt) echo '	'"`{ basename -d $file }^/" '--' "`{ basename -d $relurl }^/";
				if not echo '	'"`{ basename -d $file }^/" '--' "$relurl/" '[style="filled,dotted"]';
			}
		}
	}
}

echo '}'