Merge branch 'ready/html5'

3 files changed, 68 insertions, 201 deletions

diff --git a/t/html.t b/t/html.t
deleted file mode 100755
index 84c561fa8..000000000
--- a/t/html.t
+++ /dev/null
@@ -1,30 +0,0 @@
-#!/usr/bin/perl
-use warnings;
-use strict;
-use Test::More;
-
-my @pages;
-
-BEGIN {
-	@pages=qw(index features news plugins/map security);
-	if (system("command -v validate >/dev/null") != 0) {
-		plan skip_all => "html validator not present";
-	}
-	else {
-		plan(tests => int @pages + 2);
-	}
-	use_ok("IkiWiki");
-}
-
-# Have to build the html pages first.
-# Note that just building them like this doesn't exersise all the possible
-# html that can be generated, in particular it misses some of the action
-# links at the top, etc.
-ok(system("make >/dev/null") == 0);
-
-foreach my $page (@pages) {
-        print "# Validating $page\n";
-	ok(system("validate html/$page.html") == 0);
-}
-
-# TODO: validate form output html
diff --git a/t/relativity.t b/t/relativity.t
index 054f8f664..a1f3ad830 100755
--- a/t/relativity.t
+++ b/t/relativity.t
@@ -59,7 +59,6 @@ cgiurl: $args{cgiurl}
 $w3mmodeline
 cgi_wrapper: t/tmp/ikiwiki.cgi
 cgi_wrappermode: 0754
-html5: $args{html5}
 # make it easier to test previewing
 add_plugins:
 - anonok
@@ -148,7 +147,6 @@ sub test_startup {
 
 sub test_site1_perfectly_ordinary_ikiwiki {
 	write_setup_file(
-		html5	=> 0,
 		url	=> "http://example.com/wiki/",
 		cgiurl	=> "http://example.com/cgi-bin/ikiwiki.cgi",
 	);
@@ -157,77 +155,35 @@ sub test_site1_perfectly_ordinary_ikiwiki {
 	# url and cgiurl are on the same host so the cgiurl is host-relative
 	check_generated_content(qr{<a[^>]+href="/cgi-bin/ikiwiki.cgi\?do=prefs"});
 	my %bits = parse_cgi_content(run_cgi());
-	is($bits{basehref}, "http://example.com/wiki/");
+	like($bits{basehref}, qr{^(?:(?:http:)?//example\.com)?/wiki/$});
 	like($bits{stylehref}, qr{^(?:(?:http:)?//example.com)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:/wiki|\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:http:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
 
 	# when accessed via HTTPS, links are secure
 	%bits = parse_cgi_content(run_cgi(is_https => 1));
-	is($bits{basehref}, "https://example.com/wiki/");
+	like($bits{basehref}, qr{^(?:(?:https:)?//example\.com)?/wiki/$});
 	like($bits{stylehref}, qr{^(?:(?:https:)?//example.com)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:/wiki|\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
 
 	# when accessed via a different hostname, links stay on that host
 	%bits = parse_cgi_content(run_cgi(HTTP_HOST => 'staging.example.net'));
-	is($bits{basehref}, "http://staging.example.net/wiki/");
+	like($bits{basehref}, qr{^(?:(?:http:)?//staging\.example\.net)?/wiki/$});
 	like($bits{stylehref}, qr{^(?:(?:http:)?//staging.example.net)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:/wiki|\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:http:)?//staging.example.net)?/cgi-bin/ikiwiki.cgi$});
 
 	# previewing a page
 	%bits = parse_cgi_content(run_cgi(is_preview => 1));
-	is($bits{basehref}, "http://example.com/wiki/a/b/c/");
+	like($bits{basehref}, qr{^(?:(?:http:)?//example\.com)?/wiki/a/b/c/$});
 	like($bits{stylehref}, qr{^(?:(?:http:)?//example.com)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:/wiki|\.\./\.\./\.\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:http:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
-
-	# in html5, the <base> is allowed to be relative, and we take full
-	# advantage of that
-	write_setup_file(
-		html5	=> 1,
-		url	=> "http://example.com/wiki/",
-		cgiurl	=> "http://example.com/cgi-bin/ikiwiki.cgi",
-	);
-	thoroughly_rebuild();
-	check_cgi_mode_bits();
-	# url and cgiurl are on the same host so the cgiurl is host-relative
-	check_generated_content(qr{<a[^>]+href="/cgi-bin/ikiwiki.cgi\?do=prefs"});
-
-	%bits = parse_cgi_content(run_cgi());
-	is($bits{basehref}, "/wiki/");
-	is($bits{stylehref}, "/wiki/style.css");
-	is($bits{tophref}, "/wiki/");
-	is($bits{cgihref}, "/cgi-bin/ikiwiki.cgi");
-
-	# when accessed via HTTPS, links are secure - this is easy because under
-	# html5 they're independent of the URL at which the CGI was accessed
-	%bits = parse_cgi_content(run_cgi(is_https => 1));
-	is($bits{basehref}, "/wiki/");
-	is($bits{stylehref}, "/wiki/style.css");
-	is($bits{tophref}, "/wiki/");
-	is($bits{cgihref}, "/cgi-bin/ikiwiki.cgi");
-
-	# when accessed via a different hostname, links stay on that host -
-	# this is really easy in html5 because we can use relative URLs
-	%bits = parse_cgi_content(run_cgi(HTTP_HOST => 'staging.example.net'));
-	is($bits{basehref}, "/wiki/");
-	is($bits{stylehref}, "/wiki/style.css");
-	is($bits{tophref}, "/wiki/");
-	is($bits{cgihref}, "/cgi-bin/ikiwiki.cgi");
-
-	# previewing a page
-	%bits = parse_cgi_content(run_cgi(is_preview => 1));
-	is($bits{basehref}, "/wiki/a/b/c/");
-	is($bits{stylehref}, "/wiki/style.css");
-	like($bits{tophref}, qr{^(?:/wiki|\.\./\.\./\.\.)/$});
-	is($bits{cgihref}, "/cgi-bin/ikiwiki.cgi");
 }
 
 sub test_site2_static_content_and_cgi_on_different_servers {
 	write_setup_file(
-		html5	=> 0,
 		url	=> "http://static.example.com/",
 		cgiurl	=> "http://cgi.example.com/ikiwiki.cgi",
 	);
@@ -238,14 +194,14 @@ sub test_site2_static_content_and_cgi_on_different_servers {
 	check_generated_content(qr{<a[^>]+href="(?:http:)?//cgi.example.com/ikiwiki.cgi\?do=prefs"});
 
 	my %bits = parse_cgi_content(run_cgi(SCRIPT_NAME => '/ikiwiki.cgi', HTTP_HOST => 'cgi.example.com'));
-	like($bits{basehref}, qr{^http://static.example.com/$});
+	like($bits{basehref}, qr{^(?:(?:http:)?//static.example.com)?/$});
 	like($bits{stylehref}, qr{^(?:(?:http:)?//static.example.com)?/style.css$});
 	like($bits{tophref}, qr{^(?:http:)?//static.example.com/$});
 	like($bits{cgihref}, qr{^(?:(?:http:)?//cgi.example.com)?/ikiwiki.cgi$});
 
 	# when accessed via HTTPS, links are secure
 	%bits = parse_cgi_content(run_cgi(is_https => 1, SCRIPT_NAME => '/ikiwiki.cgi', HTTP_HOST => 'cgi.example.com'));
-	like($bits{basehref}, qr{^https://static.example.com/$});
+	like($bits{basehref}, qr{^(?:https:)?//static\.example\.com/$});
 	like($bits{stylehref}, qr{^(?:(?:https:)?//static.example.com)?/style.css$});
 	like($bits{tophref}, qr{^(?:https:)?//static.example.com/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//cgi.example.com)?/ikiwiki.cgi$});
@@ -253,7 +209,7 @@ sub test_site2_static_content_and_cgi_on_different_servers {
 	# when accessed via a different hostname, links to the CGI (only) should
 	# stay on that host?
 	%bits = parse_cgi_content(run_cgi(is_preview => 1, SCRIPT_NAME => '/ikiwiki.cgi', HTTP_HOST => 'staging.example.net'));
-	like($bits{basehref}, qr{^http://static.example.com/a/b/c/$});
+	like($bits{basehref}, qr{^(?:http:)?//static\.example\.com/a/b/c/$});
 	like($bits{stylehref}, qr{^(?:(?:http:)?//static.example.com|\.\./\.\./\.\.)/style.css$});
 	like($bits{tophref}, qr{^(?:(?:http:)?//static.example.com|\.\./\.\./\.\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:http:)?//(?:staging\.example\.net|cgi\.example\.com))?/ikiwiki.cgi$});
@@ -261,48 +217,10 @@ sub test_site2_static_content_and_cgi_on_different_servers {
 	local $TODO = "use self-referential CGI URL?";
 	like($bits{cgihref}, qr{^(?:(?:http:)?//staging.example.net)?/ikiwiki.cgi$});
 	}
-
-	write_setup_file(
-		html5	=> 1,
-		url	=> "http://static.example.com/",
-		cgiurl	=> "http://cgi.example.com/ikiwiki.cgi",
-	);
-	thoroughly_rebuild();
-	check_cgi_mode_bits();
-	# url and cgiurl are not on the same host so the cgiurl has to be
-	# protocol-relative or absolute
-	check_generated_content(qr{<a[^>]+href="(?:http:)?//cgi.example.com/ikiwiki.cgi\?do=prefs"});
-
-	%bits = parse_cgi_content(run_cgi(SCRIPT_NAME => '/ikiwiki.cgi', HTTP_HOST => 'cgi.example.com'));
-	is($bits{basehref}, "//static.example.com/");
-	is($bits{stylehref}, "//static.example.com/style.css");
-	is($bits{tophref}, "//static.example.com/");
-	is($bits{cgihref}, "//cgi.example.com/ikiwiki.cgi");
-
-	# when accessed via HTTPS, links are secure - in fact they're exactly the
-	# same as when accessed via HTTP
-	%bits = parse_cgi_content(run_cgi(is_https => 1, SCRIPT_NAME => '/ikiwiki.cgi', HTTP_HOST => 'cgi.example.com'));
-	is($bits{basehref}, "//static.example.com/");
-	is($bits{stylehref}, "//static.example.com/style.css");
-	is($bits{tophref}, "//static.example.com/");
-	is($bits{cgihref}, "//cgi.example.com/ikiwiki.cgi");
-
-	# when accessed via a different hostname, links to the CGI (only) should
-	# stay on that host?
-	%bits = parse_cgi_content(run_cgi(is_preview => 1, SCRIPT_NAME => '/ikiwiki.cgi', HTTP_HOST => 'staging.example.net'));
-	is($bits{basehref}, "//static.example.com/a/b/c/");
-	is($bits{stylehref}, "//static.example.com/style.css");
-	is($bits{tophref}, "../../../");
-	like($bits{cgihref}, qr{//(?:staging\.example\.net|cgi\.example\.com)/ikiwiki\.cgi});
-	TODO: {
-	local $TODO = "use self-referential CGI URL maybe?";
-	is($bits{cgihref}, "//staging.example.net/ikiwiki.cgi");
-	}
 }
 
 sub test_site3_we_specifically_want_everything_to_be_secure {
 	write_setup_file(
-		html5	=> 0,
 		url	=> "https://example.com/wiki/",
 		cgiurl	=> "https://example.com/cgi-bin/ikiwiki.cgi",
 	);
@@ -313,7 +231,7 @@ sub test_site3_we_specifically_want_everything_to_be_secure {
 
 	# when accessed via HTTPS, links are secure
 	my %bits = parse_cgi_content(run_cgi(is_https => 1));
-	is($bits{basehref}, "https://example.com/wiki/");
+	like($bits{basehref}, qr{^(?:(?:https:)?//example\.com)?/wiki/$});
 	like($bits{stylehref}, qr{^(?:(?:https:)?//example.com)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:/wiki|\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
@@ -331,25 +249,22 @@ sub test_site3_we_specifically_want_everything_to_be_secure {
 
 	# when accessed via a different hostname, links stay on that host
 	%bits = parse_cgi_content(run_cgi(is_https => 1, HTTP_HOST => 'staging.example.net'));
-	is($bits{basehref}, "https://staging.example.net/wiki/");
+	like($bits{basehref}, qr{^(?:(?:https:)?//staging\.example\.net)?/wiki/$});
 	like($bits{stylehref}, qr{^(?:(?:https:)?//staging.example.net)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:/wiki|\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//staging.example.net)?/cgi-bin/ikiwiki.cgi$});
 
 	# previewing a page
 	%bits = parse_cgi_content(run_cgi(is_preview => 1, is_https => 1));
-	is($bits{basehref}, "https://example.com/wiki/a/b/c/");
+	like($bits{basehref}, qr{^(?:(?:https:)?//example\.com)?/wiki/a/b/c/$});
 	like($bits{stylehref}, qr{^(?:(?:https:)?//example.com)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:/wiki|\.\./\.\./\.\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
-
-	# not testing html5: 0 here because that ends up identical to site 1
 }
 
 sub test_site4_cgi_is_secure_static_content_doesnt_have_to_be {
 	# (NetBSD wiki)
 	write_setup_file(
-		html5	=> 0,
 		url	=> "http://example.com/wiki/",
 		cgiurl	=> "https://example.com/cgi-bin/ikiwiki.cgi",
 	);
@@ -360,14 +275,15 @@ sub test_site4_cgi_is_secure_static_content_doesnt_have_to_be {
 
 	# when accessed via HTTPS, links are secure (to avoid mixed-content)
 	my %bits = parse_cgi_content(run_cgi(is_https => 1));
-	is($bits{basehref}, "https://example.com/wiki/");
+	like($bits{basehref}, qr{^(?:(?:https:)?//example\.com)?/wiki/$});
 	like($bits{stylehref}, qr{^(?:(?:https:)?//example.com)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:/wiki|\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
 
-	# when not accessed via HTTPS, ???
+	# FIXME: when not accessed via HTTPS, should the static content be
+	# forced to https anyway? For now we accept either
 	%bits = parse_cgi_content(run_cgi());
-	like($bits{basehref}, qr{^https?://example.com/wiki/$});
+	like($bits{basehref}, qr{^(?:(?:https?)?://example\.com)?/wiki/$});
 	like($bits{stylehref}, qr{^(?:(?:https?:)?//example.com)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:(?:https?://example.com)?/wiki|\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
@@ -376,7 +292,7 @@ sub test_site4_cgi_is_secure_static_content_doesnt_have_to_be {
 	%bits = parse_cgi_content(run_cgi(is_https => 1, HTTP_HOST => 'staging.example.net'));
 	# because the static and dynamic stuff is on the same server, we assume that
 	# both are also on the staging server
-	like($bits{basehref}, qr{^https://staging.example.net/wiki/$});
+	like($bits{basehref}, qr{^(?:(?:https:)?//staging\.example\.net)?/wiki/$});
 	like($bits{stylehref}, qr{^(?:(?:https:)?//staging.example.net)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:(?:(?:https:)?//staging.example.net)?/wiki|\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//(?:staging\.example\.net|example\.com))?/cgi-bin/ikiwiki.cgi$});
@@ -387,63 +303,15 @@ sub test_site4_cgi_is_secure_static_content_doesnt_have_to_be {
 
 	# previewing a page
 	%bits = parse_cgi_content(run_cgi(is_preview => 1, is_https => 1));
-	is($bits{basehref}, "https://example.com/wiki/a/b/c/");
+	like($bits{basehref}, qr{^(?:(?:https:)?//example\.com)?/wiki/a/b/c/$});
 	like($bits{stylehref}, qr{^(?:(?:https:)?//example.com)?/wiki/style.css$});
 	like($bits{tophref}, qr{^(?:/wiki|\.\./\.\./\.\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
-
-	write_setup_file(
-		html5	=> 1,
-		url	=> "http://example.com/wiki/",
-		cgiurl	=> "https://example.com/cgi-bin/ikiwiki.cgi",
-	);
-	thoroughly_rebuild();
-	check_cgi_mode_bits();
-	# url and cgiurl are on the same host but different schemes
-	check_generated_content(qr{<a[^>]+href="https://example.com/cgi-bin/ikiwiki.cgi\?do=prefs"});
-
-	# when accessed via HTTPS, links are secure (to avoid mixed-content)
-	%bits = parse_cgi_content(run_cgi(is_https => 1));
-	is($bits{basehref}, "/wiki/");
-	is($bits{stylehref}, "/wiki/style.css");
-	is($bits{tophref}, "/wiki/");
-	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
-
-	# when not accessed via HTTPS, ???
-	%bits = parse_cgi_content(run_cgi());
-	like($bits{basehref}, qr{^(?:https?://example.com)?/wiki/$});
-	like($bits{stylehref}, qr{^(?:(?:https?:)?//example.com)?/wiki/style.css$});
-	like($bits{tophref}, qr{^(?:(?:https?://example.com)?/wiki|\.)/$});
-	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
-
-	# when accessed via a different hostname, links stay on that host
-	%bits = parse_cgi_content(run_cgi(is_https => 1, HTTP_HOST => 'staging.example.net'));
-	# because the static and dynamic stuff is on the same server, we assume that
-	# both are also on the staging server
-	is($bits{basehref}, "/wiki/");
-	is($bits{stylehref}, "/wiki/style.css");
-	like($bits{tophref}, qr{^(?:/wiki|\.)/$});
-	like($bits{cgihref}, qr{^(?:(?:https:)?//(?:example\.com|staging\.example\.net))?/cgi-bin/ikiwiki.cgi$});
-	TODO: {
-	local $TODO = "this should really point back to itself but currently points to example.com";
-	like($bits{cgihref}, qr{^(?:(?:https:)?//staging.example.net)?/cgi-bin/ikiwiki.cgi$});
-	}
-
-	# previewing a page
-	%bits = parse_cgi_content(run_cgi(is_preview => 1, is_https => 1));
-	is($bits{basehref}, "/wiki/a/b/c/");
-	is($bits{stylehref}, "/wiki/style.css");
-	like($bits{tophref}, qr{^(?:/wiki|\.\./\.\./\.\.)/$});
-	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
-
-	# Deliberately not testing https static content with http cgiurl,
-	# because that makes remarkably little sense.
 }
 
 sub test_site5_w3mmode {
 	# as documented in [[w3mmode]]
 	write_setup_file(
-		html5	=> 0, 
 		url	=> undef,
 		cgiurl	=> "ikiwiki.cgi",
 		w3mmode	=> 1,
@@ -459,28 +327,10 @@ sub test_site5_w3mmode {
 	like($bits{cgihref}, qr{^(?:file://)?/\$LIB/ikiwiki-w3m.cgi/ikiwiki.cgi$});
 	like($bits{basehref}, qr{^(?:(?:file:)?//)?\Q$pwd\E/t/tmp/out/$});
 	like($bits{stylehref}, qr{^(?:(?:(?:file:)?//)?\Q$pwd\E/t/tmp/out|\.)/style.css$});
-
-	write_setup_file(
-		html5	=> 1,
-		url	=> undef,
-		cgiurl	=> "ikiwiki.cgi",
-		w3mmode	=> 1,
-	);
-	thoroughly_rebuild();
-	check_cgi_mode_bits();
-	# FIXME: does /$LIB/ikiwiki-w3m.cgi work under w3m?
-	check_generated_content(qr{<a[^>]+href="(?:file://)?/\$LIB/ikiwiki-w3m.cgi/ikiwiki.cgi\?do=prefs"});
-
-	%bits = parse_cgi_content(run_cgi(PATH_INFO => '/ikiwiki.cgi', SCRIPT_NAME => '/cgi-bin/ikiwiki-w3m.cgi'));
-	like($bits{tophref}, qr{^(?:\Q$pwd\E/t/tmp/out|\.)/$});
-	like($bits{cgihref}, qr{^(?:file://)?/\$LIB/ikiwiki-w3m.cgi/ikiwiki.cgi$});
-	like($bits{basehref}, qr{^(?:(?:file:)?//)?\Q$pwd\E/t/tmp/out/$});
-	like($bits{stylehref}, qr{^(?:(?:(?:file:)?//)?\Q$pwd\E/t/tmp/out|\.)/style.css$});
 }
 
 sub test_site6_behind_reverse_proxy {
 	write_setup_file(
-		html5	=> 0,
 		url	=> "https://example.com/wiki/",
 		cgiurl	=> "https://example.com/cgi-bin/ikiwiki.cgi",
 		reverse_proxy => 1,
@@ -495,18 +345,15 @@ sub test_site6_behind_reverse_proxy {
 	my %bits = parse_cgi_content(run_cgi(HTTP_HOST => 'localhost'));
 	like($bits{tophref}, qr{^(?:/wiki|\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
-	is($bits{basehref}, "https://example.com/wiki/");
+	like($bits{basehref}, qr{^(?:(?:https:)?//example\.com)?/wiki/$});
 	like($bits{stylehref}, qr{^(?:(?:https:)?//example.com)?/wiki/style.css$});
 
 	# previewing a page
 	%bits = parse_cgi_content(run_cgi(is_preview => 1, HTTP_HOST => 'localhost'));
 	like($bits{tophref}, qr{^(?:/wiki|\.\./\.\./\.\.)/$});
 	like($bits{cgihref}, qr{^(?:(?:https:)?//example.com)?/cgi-bin/ikiwiki.cgi$});
-	is($bits{basehref}, "https://example.com/wiki/a/b/c/");
+	like($bits{basehref}, qr{^(?:(?:https)?://example\.com)?/wiki/a/b/c/$});
 	like($bits{stylehref}, qr{^(?:(?:https:)?//example.com)?/wiki/style.css$});
-
-	# not testing html5: 1 because it would be the same as site 1 -
-	# the reverse_proxy config option is unnecessary under html5
 }
 
 test_startup();
diff --git a/t/wellformed.t b/t/wellformed.t
new file mode 100755
index 000000000..cee0ff3be
--- /dev/null
+++ b/t/wellformed.t
@@ -0,0 +1,50 @@
+#!/usr/bin/perl
+use warnings;
+use strict;
+use Cwd qw();
+use File::Find;
+use Test::More;
+
+plan(skip_all => "XML::Parser not available")
+	unless eval q{use XML::Parser (); 1;};
+
+use IkiWiki;
+
+ok(system("make >/dev/null") == 0);
+
+chdir("html") || die "chdir: $!";
+
+sub wanted {
+	my $file = $_;
+	return if -d $file;
+	$file =~ s{^\./}{};
+	return if $file !~ m/\.html$/;
+	if (eval {
+		XML::Parser->new()->parsefile($file);
+		1;
+	}) {
+		pass($file);
+	}
+	elsif ($file =~ m{^(?:
+			# user-contributed, contains explicit <br>
+			plugins/contrib/gallery |
+			# use templatebody when branchable.com has been upgraded
+			templates/ |
+			# malformed content in <pre> not escaped by discount
+			tips/convert_mediawiki_to_ikiwiki
+			# user-contributed, content is anyone's guess
+			users/ |
+			)}x) {
+		TODO: {
+			local $TODO = $@;
+			fail($file);
+		}
+	}
+}
+
+find({
+	no_chdir => 1,
+	wanted => \&wanted,
+}, '.');
+
+done_testing;