hare

commit a0fdca9f8135c0f7e3f1d266385af10b31f04660
parent daa4a2f9178dfb64ff5e431c37914923850d3064
Author: Drew DeVault <sir@cmpwn.com>
Date:   Sun, 11 Apr 2021 09:20:56 -0400

format::xml: validate closing tags

Diffstat:
M
format/xml/+test.ha
 | 
52
++++++++++++++++++++++++++++++++++++++--------------
M
format/xml/parser.ha
 | 
29
+++++++++++++++++++++++++++--
M
format/xml/types.ha
 | 
3
++-

3 files changed, 67 insertions(+), 17 deletions(-)
diff --git a/format/xml/+test.ha b/format/xml/+test.ha
@@ -6,16 +6,21 @@ use strings;
 	xmltest("<?xml version='1.0' encoding='utf-8' ?>
 <root>
 	<hello name='foobar'>world</hello>
+	<foobar bar='foobaz' />
 </root>", [
 		"root": elementstart,
 		"\n\t": text,
 		"hello": elementstart,
 		("name", "foobar"): attribute,
 		"world": text,
-		elementend,
+		"hello": elementend,
+		"\n\t": text,
+		"foobar": elementstart,
+		("bar", "foobaz"): attribute,
+		"foobar": elementend,
 		"\n": text,
-		elementend,
-	]);
+		"root": elementend,
+	], false);
 };
 
 @test fn comments() void = {
@@ -29,10 +34,10 @@ use strings;
 		"hello": elementstart,
 		("name", "foobar"): attribute,
 		"world": text,
-		elementend,
+		"hello": elementend,
 		"\n": text,
-		elementend,
-	]);
+		"root": elementend,
+	], false);
 };
 
 @test fn entities() void = {
@@ -45,10 +50,10 @@ use strings;
 		"hello": elementstart,
 		("name", "foobar &!!"): attribute,
 		"<world> \"'": text,
-		elementend,
+		"hello": elementend,
 		"\n": text,
-		elementend,
-	]);
+		"root": elementend,
+	], false);
 };
 
 @test fn cdata() void = {
@@ -60,11 +65,23 @@ use strings;
 		"\n\t": text,
 		"Hello world &foo <bar>": text,
 		"\n": text,
-		elementend,
-	]);
+		"root": elementend,
+	], false);
 };
 
-fn xmltest(input: str, expected: []token) void = {
+@test fn errors() void = {
+	xmltest("<?xml version='1.0' encoding='utf-8' ?>
+<root>
+	<hello name='foobar'></world>
+</root>", [
+		"root": elementstart,
+		"\n\t": text,
+		"hello": elementstart,
+		("name", "foobar"): attribute,
+	], true);
+};
+
+fn xmltest(input: str, expected: []token, error: bool) void = {
 	let in = bufio::fixed(strings::toutf8(input), io::mode::READ);
 	defer io::close(in);
 	let parser = parse(in) as *parser;
@@ -87,8 +104,15 @@ fn xmltest(input: str, expected: []token) void = {
 				let ex = expected[i] as text;
 				assert(tx == ex);
 			},
-			elementend => assert(expected[i] is elementend),
+			el: elementend => {
+				let ex = expected[i] as elementend;
+				assert(el == ex);
+			},
 		};
 	};
-	assert(scan(parser) is void);
+	if (error) {
+		assert(scan(parser) is error);
+	} else {
+		assert(scan(parser) is void);
+	};
 };
diff --git a/format/xml/parser.ha b/format/xml/parser.ha
@@ -59,6 +59,10 @@ export fn parser_free(par: *parser) void = {
 	io::close(par.namebuf);
 	io::close(par.entbuf);
 	io::close(par.textbuf);
+	for (let i = 0z; i < len(par.tags); i += 1) {
+		free(par.tags[i]);
+	};
+	free(par.tags);
 	free(par);
 };
 
@@ -108,7 +112,7 @@ export fn scan(par: *parser) (token | void | error) = {
 			if (rn == '/') {
 				want(par, '>')?;
 				par.state = state::ELEMENT;
-				return elementend;
+				return poptag(par, "")?: elementend;
 			} else if (rn == '>') {
 				par.state = state::ELEMENT;
 				return scan(par)?;
@@ -121,6 +125,21 @@ export fn scan(par: *parser) (token | void | error) = {
 	};
 };
 
+fn poptag(par: *parser, expect: str) (str | error) = {
+	if (len(par.tags) == 0) {
+		return syntaxerr;
+	};
+	let pop = par.tags[len(par.tags) - 1];
+	delete(par.tags[len(par.tags) - 1]);
+	defer free(pop);
+	if (expect != "" && expect != pop) {
+		return syntaxerr;
+	};
+	strio::reset(par.namebuf);
+	strio::concat(par.namebuf, pop);
+	return strio::string(par.namebuf);
+};
+
 fn scan_attr(par: *parser) (token | error) = {
 	let name = scan_name(par, par.namebuf)?;
 	want(par, OPTWS, '=', OPTWS);
@@ -243,7 +262,13 @@ fn scan_element(par: *parser) (token | error) = {
 		},
 	};
 	let name = scan_name(par, par.namebuf)?;
-	return if (close) elementend else name: elementstart;
+	if (close) {
+		poptag(par, name)?;
+		return name: elementend;
+	} else {
+		append(par.tags, strings::dup(name));
+		return name: elementstart;
+	};
 };
 
 fn scan_entity(par: *parser) (rune | error) = {
diff --git a/format/xml/types.ha b/format/xml/types.ha
@@ -8,6 +8,7 @@ export type parser = struct {
 	in: *io::stream,
 	buf: [os::BUFSIZ]u8,
 	state: state,
+	tags: []str,
 	// strio buffers:
 	namebuf: *io::stream,
 	entbuf: *io::stream,
@@ -24,7 +25,7 @@ export type state = enum {
 export type elementstart = str;
 
 // The end of an XML element, e.g. /> or </example>
-export type elementend = void;
+export type elementend = str;
 
 // An attribute of an XML element, e.g. foo="bar"
 export type attribute = (str, str);