1
// Copyright 2016 The Go Authors. All rights reserved.
2
// Use of this source code is governed by a BSD-style
3
// license that can be found in the LICENSE file.
12
"golang.org/x/net/idna"
15
var isTokenTable = [127]bool{
95
func IsTokenRune(r rune) bool {
97
return i < len(isTokenTable) && isTokenTable[i]
100
func isNotToken(r rune) bool {
101
return !IsTokenRune(r)
104
// HeaderValuesContainsToken reports whether any string in values
105
// contains the provided token, ASCII case-insensitively.
106
func HeaderValuesContainsToken(values []string, token string) bool {
107
for _, v := range values {
108
if headerValueContainsToken(v, token) {
115
// isOWS reports whether b is an optional whitespace byte, as defined
116
// by RFC 7230 section 3.2.3.
117
func isOWS(b byte) bool { return b == ' ' || b == '\t' }
119
// trimOWS returns x with all optional whitespace removes from the
121
func trimOWS(x string) string {
122
// TODO: consider using strings.Trim(x, " \t") instead,
123
// if and when it's fast enough. See issue 10292.
124
// But this ASCII-only code will probably always beat UTF-8
126
for len(x) > 0 && isOWS(x[0]) {
129
for len(x) > 0 && isOWS(x[len(x)-1]) {
135
// headerValueContainsToken reports whether v (assumed to be a
136
// 0#element, in the ABNF extension described in RFC 7230 section 7)
137
// contains token amongst its comma-separated tokens, ASCII
138
// case-insensitively.
139
func headerValueContainsToken(v string, token string) bool {
140
for comma := strings.IndexByte(v, ','); comma != -1; comma = strings.IndexByte(v, ',') {
141
if tokenEqual(trimOWS(v[:comma]), token) {
146
return tokenEqual(trimOWS(v), token)
149
// lowerASCII returns the ASCII lowercase version of b.
150
func lowerASCII(b byte) byte {
151
if 'A' <= b && b <= 'Z' {
152
return b + ('a' - 'A')
157
// tokenEqual reports whether t1 and t2 are equal, ASCII case-insensitively.
158
func tokenEqual(t1, t2 string) bool {
159
if len(t1) != len(t2) {
162
for i, b := range t1 {
163
if b >= utf8.RuneSelf {
164
// No UTF-8 or non-ASCII allowed in tokens.
167
if lowerASCII(byte(b)) != lowerASCII(t2[i]) {
174
// isLWS reports whether b is linear white space, according
175
// to http://www.w3.org/Protocols/rfc2616/rfc2616-sec2.html#sec2.2
177
// LWS = [CRLF] 1*( SP | HT )
178
func isLWS(b byte) bool { return b == ' ' || b == '\t' }
180
// isCTL reports whether b is a control byte, according
181
// to http://www.w3.org/Protocols/rfc2616/rfc2616-sec2.html#sec2.2
183
// CTL = <any US-ASCII control character
184
// (octets 0 - 31) and DEL (127)>
185
func isCTL(b byte) bool {
186
const del = 0x7f // a CTL
187
return b < ' ' || b == del
190
// ValidHeaderFieldName reports whether v is a valid HTTP/1.x header name.
191
// HTTP/2 imposes the additional restriction that uppercase ASCII
192
// letters are not allowed.
196
// header-field = field-name ":" OWS field-value OWS
199
// tchar = "!" / "#" / "$" / "%" / "&" / "'" / "*" / "+" / "-" / "." /
200
// "^" / "_" / "`" / "|" / "~" / DIGIT / ALPHA
201
func ValidHeaderFieldName(v string) bool {
205
for _, r := range v {
213
// ValidHostHeader reports whether h is a valid host header.
214
func ValidHostHeader(h string) bool {
215
// The latest spec is actually this:
217
// http://tools.ietf.org/html/rfc7230#section-5.4
218
// Host = uri-host [ ":" port ]
220
// Where uri-host is:
221
// http://tools.ietf.org/html/rfc3986#section-3.2.2
223
// But we're going to be much more lenient for now and just
224
// search for any byte that's not a valid byte in any of those
226
for i := 0; i < len(h); i++ {
227
if !validHostByte[h[i]] {
234
// See the validHostHeader comment.
235
var validHostByte = [256]bool{
236
'0': true, '1': true, '2': true, '3': true, '4': true, '5': true, '6': true, '7': true,
237
'8': true, '9': true,
239
'a': true, 'b': true, 'c': true, 'd': true, 'e': true, 'f': true, 'g': true, 'h': true,
240
'i': true, 'j': true, 'k': true, 'l': true, 'm': true, 'n': true, 'o': true, 'p': true,
241
'q': true, 'r': true, 's': true, 't': true, 'u': true, 'v': true, 'w': true, 'x': true,
242
'y': true, 'z': true,
244
'A': true, 'B': true, 'C': true, 'D': true, 'E': true, 'F': true, 'G': true, 'H': true,
245
'I': true, 'J': true, 'K': true, 'L': true, 'M': true, 'N': true, 'O': true, 'P': true,
246
'Q': true, 'R': true, 'S': true, 'T': true, 'U': true, 'V': true, 'W': true, 'X': true,
247
'Y': true, 'Z': true,
249
'!': true, // sub-delims
250
'$': true, // sub-delims
251
'%': true, // pct-encoded (and used in IPv6 zones)
252
'&': true, // sub-delims
253
'(': true, // sub-delims
254
')': true, // sub-delims
255
'*': true, // sub-delims
256
'+': true, // sub-delims
257
',': true, // sub-delims
258
'-': true, // unreserved
259
'.': true, // unreserved
260
':': true, // IPv6address + Host expression's optional port
261
';': true, // sub-delims
262
'=': true, // sub-delims
264
'\'': true, // sub-delims
266
'_': true, // unreserved
267
'~': true, // unreserved
270
// ValidHeaderFieldValue reports whether v is a valid "field-value" according to
271
// http://www.w3.org/Protocols/rfc2616/rfc2616-sec4.html#sec4.2 :
273
// message-header = field-name ":" [ field-value ]
274
// field-value = *( field-content | LWS )
275
// field-content = <the OCTETs making up the field-value
276
// and consisting of either *TEXT or combinations
277
// of token, separators, and quoted-string>
279
// http://www.w3.org/Protocols/rfc2616/rfc2616-sec2.html#sec2.2 :
281
// TEXT = <any OCTET except CTLs,
283
// LWS = [CRLF] 1*( SP | HT )
284
// CTL = <any US-ASCII control character
285
// (octets 0 - 31) and DEL (127)>
289
// field-value = *( field-content / obs-fold )
290
// obj-fold = N/A to http2, and deprecated
291
// field-content = field-vchar [ 1*( SP / HTAB ) field-vchar ]
292
// field-vchar = VCHAR / obs-text
294
// VCHAR = "any visible [USASCII] character"
296
// http2 further says: "Similarly, HTTP/2 allows header field values
297
// that are not valid. While most of the values that can be encoded
298
// will not alter header field parsing, carriage return (CR, ASCII
299
// 0xd), line feed (LF, ASCII 0xa), and the zero character (NUL, ASCII
300
// 0x0) might be exploited by an attacker if they are translated
301
// verbatim. Any request or response that contains a character not
302
// permitted in a header field value MUST be treated as malformed
303
// (Section 8.1.2.6). Valid characters are defined by the
304
// field-content ABNF rule in Section 3.2 of [RFC7230]."
306
// This function does not (yet?) properly handle the rejection of
307
// strings that begin or end with SP or HTAB.
308
func ValidHeaderFieldValue(v string) bool {
309
for i := 0; i < len(v); i++ {
311
if isCTL(b) && !isLWS(b) {
318
func isASCII(s string) bool {
319
for i := 0; i < len(s); i++ {
320
if s[i] >= utf8.RuneSelf {
327
// PunycodeHostPort returns the IDNA Punycode version
328
// of the provided "host" or "host:port" string.
329
func PunycodeHostPort(v string) (string, error) {
334
host, port, err := net.SplitHostPort(v)
336
// The input 'v' argument was just a "host" argument,
337
// without a port. This error should not be returned
342
host, err = idna.ToASCII(host)
344
// Non-UTF-8? Not representable in Punycode, in any
351
return net.JoinHostPort(host, port), nil