net: add parse_headers function and handle header line folding (#10936)

Closes https://github.com/vlang/v/issues/10930

vlib/net/http/header.v

@@ -644,6 +644,14 @@ fn is_valid(header string) ? {
 			})
 		}
 	}
+	if header.len == 0 {
+		return IError(HeaderKeyError{
+			msg: "Invalid header key: '$header'"
+			code: 2
+			header: header
+			invalid_char: 0
+		})
+	}
 }
 
 // is_token checks if the byte is valid for a header token
@@ -659,3 +667,34 @@ fn is_token(b byte) bool {
 pub fn (h Header) str() string {
 	return h.render(version: .v1_1)
 }
+
+// parse_headers parses a newline delimited string into a Header struct
+fn parse_headers(s string) ?Header {
+	mut h := new_header()
+	mut last_key := ''
+	mut last_value := ''
+	for line in s.split_into_lines() {
+		if line.len == 0 {
+			break
+		}
+		// handle header fold
+		if line[0] == ` ` || line[0] == `\t` {
+			last_value += ' ${line.trim(' \t')}'
+			continue
+		} else if last_key != '' {
+			h.add_custom(last_key, last_value) ?
+		}
+		last_key, last_value = parse_header(line) ?
+	}
+	h.add_custom(last_key, last_value) ?
+	return h
+}
+
+fn parse_header(s string) ?(string, string) {
+	if !s.contains(':') {
+		return error('missing colon in header')
+	}
+	words := s.split_nth(':', 2)
+	// TODO: parse quoted text according to the RFC
+	return words[0], words[1].trim(' \t')
+}

vlib/net/http/header_test.v

@@ -1,7 +1,7 @@
 module http
 
 fn test_header_new() {
-	h := new_header({ key: .accept, value: 'nothing' },
+	h := new_header(HeaderConfig{ key: .accept, value: 'nothing' },
 		key: .expires
 		value: 'yesterday'
 	)
@@ -37,7 +37,7 @@ fn test_header_get() ? {
 }
 
 fn test_header_set() ? {
-	mut h := new_header({ key: .dnt, value: 'one' },
+	mut h := new_header(HeaderConfig{ key: .dnt, value: 'one' },
 		key: .dnt
 		value: 'two'
 	)
@@ -47,7 +47,7 @@ fn test_header_set() ? {
 }
 
 fn test_header_delete() {
-	mut h := new_header({ key: .dnt, value: 'one' },
+	mut h := new_header(HeaderConfig{ key: .dnt, value: 'one' },
 		key: .dnt
 		value: 'two'
 	)
@@ -323,3 +323,39 @@ fn test_header_join() ? {
 	assert h3.contains_custom('Server')
 	assert h3.contains_custom('foo')
 }
+
+fn parse_headers_test(s string, expected map[string]string) ? {
+	assert parse_headers(s) ? == new_custom_header_from_map(expected) ?
+}
+
+fn test_parse_headers() ? {
+	parse_headers_test('foo: bar', map{
+		'foo': 'bar'
+	}) ?
+	parse_headers_test('foo: \t  bar', map{
+		'foo': 'bar'
+	}) ?
+	parse_headers_test('foo: bar\r\n\tbaz', map{
+		'foo': 'bar baz'
+	}) ?
+	parse_headers_test('foo: bar \r\n\tbaz\r\n   buzz', map{
+		'foo': 'bar baz buzz'
+	}) ?
+	parse_headers_test('foo: bar\r\nbar:baz', map{
+		'foo': 'bar'
+		'bar': 'baz'
+	}) ?
+	parse_headers_test('foo: bar\r\nbar:baz\r\n', map{
+		'foo': 'bar'
+		'bar': 'baz'
+	}) ?
+	parse_headers_test('foo: bar\r\nbar:baz\r\n\r\n', map{
+		'foo': 'bar'
+		'bar': 'baz'
+	}) ?
+	assert parse_headers('foo: bar\r\nfoo:baz') ?.custom_values('foo') == ['bar', 'baz']
+
+	if x := parse_headers(' oops: oh no') {
+		return error('should have errored, but got $x')
+	}
+}

vlib/net/http/request.v

@@ -217,15 +217,6 @@ fn parse_request_line(s string) ?(Method, urllib.URL, Version) {
 	return method, target, version
 }
 
-fn parse_header(s string) ?(string, string) {
-	if !s.contains(':') {
-		return error('missing colon in header')
-	}
-	words := s.split_nth(':', 2)
-	// TODO: parse quoted text according to the RFC
-	return words[0], words[1].trim_left(' \t')
-}
-
 // Parse URL encoded key=value&key=value forms
 fn parse_form(body string) map[string]string {
 	words := body.split('&')

vlib/net/http/response.v

@@ -34,8 +34,7 @@ pub fn (resp Response) bytestr() string {
 }
 
 // Parse a raw HTTP response into a Response object
-pub fn parse_response(resp string) Response {
-	mut header := new_header()
+pub fn parse_response(resp string) ?Response {
 	// TODO: Cookie data type
 	mut cookies := map[string]string{}
 	first_header := resp.all_before('\n')
@@ -44,28 +43,10 @@ pub fn parse_response(resp string) Response {
 		val := first_header.find_between(' ', ' ')
 		status_code = val.int()
 	}
-	mut text := ''
 	// Build resp header map and separate the body
-	mut nl_pos := 3
-	mut i := 1
-	for {
-		old_pos := nl_pos
-		nl_pos = resp.index_after('\n', nl_pos + 1)
-		if nl_pos == -1 {
-			break
-		}
-		h := resp[old_pos + 1..nl_pos]
-		// End of headers
-		if h.len <= 1 {
-			text = resp[nl_pos + 1..]
-			break
-		}
-		i++
-		pos := h.index(':') or { continue }
-		mut key := h[..pos]
-		val := h[pos + 2..].trim_space()
-		header.add_custom(key, val) or { eprintln('$err; skipping header') }
-	}
+	start_idx, end_idx := find_headers_range(resp) ?
+	header := parse_headers(resp.substr(start_idx, end_idx)) ?
+	mut text := resp.substr(end_idx, resp.len)
 	// set cookies
 	for cookie in header.values(.set_cookie) {
 		parts := cookie.split_nth('=', 2)
@@ -81,3 +62,23 @@ pub fn parse_response(resp string) Response {
 		text: text
 	}
 }
+
+// find_headers_range returns the start (inclusive) and end (exclusive)
+// index of the headers in the string, including the trailing newlines. This
+// helper function expects the first line in `data` to be the HTTP status line
+// (HTTP/1.1 200 OK).
+fn find_headers_range(data string) ?(int, int) {
+	start_idx := data.index('\n') or { return error('no start index found') } + 1
+	mut count := 0
+	for i := start_idx; i < data.len; i++ {
+		if data[i] == `\n` {
+			count++
+		} else if data[i] != `\r` {
+			count = 0
+		}
+		if count == 2 {
+			return start_idx, i + 1
+		}
+	}
+	return error('no end index found')
+}