#!/bin/rc

nl='
';

result=0;
fn error { echo $* >[1=2]; result=-1; }

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 FNR, 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($5) - 1))) { print FNR, url; print FNR, previewurl; next; }
	# Audio: <URL>
	/^Audio: [^ ]+$/ && isurl((url = $2)) { print FNR, 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 FNR, url; print FNR, xurl; next; }
	# <PREFIX>: <URL> (<ALT>)
	match($0, /: [^ ]+ \(/) && isurl((url = substr($0, RSTART + 2, RLENGTH - 4))) && /\)$/ { print FNR, 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 FNR, url; print FNR, xurl; next; }
	# <URL> (<ALT>) — <SUFFIX>
	/^[^ ]+ \(/ && match($0, /\) — /) && isurl((url = $1)) { print FNR, url; next; }
	# <URL> (<ALT>)
	/^[^ ]+ \(/ && /\)$/ && isurl((url = $1)) { print FNR, url; next; }
	# <URL>
	isurl((url = $0)) { print FNR, url; next; }
	# See also: <URL>
	# Further reading: <URL>
	# For more details, see: <URL>
	/(See also|Further reading|For more details, see): [^ ]+$/ && isurl((url = $NF)) { print FNR, url; next; }
	# - <ALT>: <URL>
	# .* forces last occurrence.
	match($0, /- .*: /) && isurl((url = substr($0, RLENGTH + 1))) { print FNR, url; next; }
	# - <URL>
	/^- / && isurl((url = substr($0, 3))) { print FNR, url; next; }
	# <ALT>: <URL>
	# .* forces last occurrence.
	match($0, /.*: /) && isurl((url = substr($0, RLENGTH + 1))) { print FNR, url; next; }
	# — <AUTHOR>, <URL>
	/^— / && /, [^ ]+$/ && isurl((url = $NF)) { print FNR, url; next; }
	' $1
}

fn checksections {
	awk '
	function error(s) {
		printf "%s:%s: %s\n", FILENAME, FNR, s;
		result = 1;
	}
	
	/^[0-9]+\. / {
		s = substr($1, 1, length($1) - 1) + 0;
		if (s != section + 1) error("misnumbered section");
		section = s;
		subsection = 0;
		next;
	}
	/^[0-9]+\.[0-9]+\. / {
		i = index($1, ".");
		s = substr($1, 1, i - 1) + 0;
		ss = substr($1, i + 1, length($1) - i - 1) + 0;
		if (s != section || ss != subsection + 1) error("misnumbered subsection");
		subsection = ss;
		next;
	}
	END {
		exit result;
	}
	' $1 >[1=2]
}

fn read {
	s=$*($#*);
	while (! ~ $#* 2) {
		$1=`{ echo $s | awk '{ print $1 }' };
		s=`{ echo $s | awk 'match($0, / *[^ ]+ */) { print substr($0, RLENGTH) }' };
		shift;
	}
	$1=$s;
}

fn relpath { cleanname `$nl{ basename -d $2 }^/`$nl{ echo $1 } }
fn abspath { echo `{ cleanname `{ pwd }^/$1 } }

fn checkifbacklinks {
	from=$1;
	to=$2;
	absfrom=`{ abspath $1 };
	absto=`{ abspath $2 };
	if (! {
		   { ~ $absto /lib/garden/journal/* }
		|| { ~ $absto /lib/garden/index.txt }
	    || { ~ $absfrom /lib/garden/bootable-media/index.txt && ~ $absto /lib/garden/lily/nixos/index.txt }
		|| { ~ $absfrom /lib/garden/plan-9/wacom/index.txt && ~ $absto /lib/garden/intuos/9/index.txt }
	}) {
		found=no;
		for (link in `$nl{ readlinks $from }) {
			read loc url $link;
			if (~ $url ./* ../* && ~ `{ relpath $url $from } $to) found=yes;
		}
		if (~ $found no) error $from: no backlink to $to;
	}
}

for (file in `{ walk -f | grep '(^|/)index.txt$' }) {
	for (link in `$nl{ readlinks $file }) {
		read loc url $link;
		if (~ $url ./* ../*) {
			otherfile=`$nl{ relpath `$nl{ echo $url } $file };
			if (test -e `$nl{ echo $otherfile }) {
				if (~ $url */index.txt) {
					checkifbacklinks $otherfile $file; # Check if otherfile backlinks to file.
				}
			}
			if not error $file:$loc: $otherfile does not exist;
		}
	}
	if (checksections $file);
	if not result=1;
}

exit $result;