Black Lives Matter. Support the Equal Justice Initiative.

Source file src/net/url/url.go

Documentation: net/url

     1  // Copyright 2009 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.
     4  
     5  // Package url parses URLs and implements query escaping.
     6  package url
     7  
     8  // See RFC 3986. This package generally follows RFC 3986, except where
     9  // it deviates for compatibility reasons. When sending changes, first
    10  // search old issues for history on decisions. Unit tests should also
    11  // contain references to issue numbers with details.
    12  
    13  import (
    14  	"errors"
    15  	"fmt"
    16  	"sort"
    17  	"strconv"
    18  	"strings"
    19  )
    20  
    21  // Error reports an error and the operation and URL that caused it.
    22  type Error struct {
    23  	Op  string
    24  	URL string
    25  	Err error
    26  }
    27  
    28  func (e *Error) Unwrap() error { return e.Err }
    29  func (e *Error) Error() string { return fmt.Sprintf("%s %q: %s", e.Op, e.URL, e.Err) }
    30  
    31  func (e *Error) Timeout() bool {
    32  	t, ok := e.Err.(interface {
    33  		Timeout() bool
    34  	})
    35  	return ok && t.Timeout()
    36  }
    37  
    38  func (e *Error) Temporary() bool {
    39  	t, ok := e.Err.(interface {
    40  		Temporary() bool
    41  	})
    42  	return ok && t.Temporary()
    43  }
    44  
    45  const upperhex = "0123456789ABCDEF"
    46  
    47  func ishex(c byte) bool {
    48  	switch {
    49  	case '0' <= c && c <= '9':
    50  		return true
    51  	case 'a' <= c && c <= 'f':
    52  		return true
    53  	case 'A' <= c && c <= 'F':
    54  		return true
    55  	}
    56  	return false
    57  }
    58  
    59  func unhex(c byte) byte {
    60  	switch {
    61  	case '0' <= c && c <= '9':
    62  		return c - '0'
    63  	case 'a' <= c && c <= 'f':
    64  		return c - 'a' + 10
    65  	case 'A' <= c && c <= 'F':
    66  		return c - 'A' + 10
    67  	}
    68  	return 0
    69  }
    70  
    71  type encoding int
    72  
    73  const (
    74  	encodePath encoding = 1 + iota
    75  	encodePathSegment
    76  	encodeHost
    77  	encodeZone
    78  	encodeUserPassword
    79  	encodeQueryComponent
    80  	encodeFragment
    81  )
    82  
    83  type EscapeError string
    84  
    85  func (e EscapeError) Error() string {
    86  	return "invalid URL escape " + strconv.Quote(string(e))
    87  }
    88  
    89  type InvalidHostError string
    90  
    91  func (e InvalidHostError) Error() string {
    92  	return "invalid character " + strconv.Quote(string(e)) + " in host name"
    93  }
    94  
    95  // Return true if the specified character should be escaped when
    96  // appearing in a URL string, according to RFC 3986.
    97  //
    98  // Please be informed that for now shouldEscape does not check all
    99  // reserved characters correctly. See golang.org/issue/5684.
   100  func shouldEscape(c byte, mode encoding) bool {
   101  	// §2.3 Unreserved characters (alphanum)
   102  	if 'a' <= c && c <= 'z' || 'A' <= c && c <= 'Z' || '0' <= c && c <= '9' {
   103  		return false
   104  	}
   105  
   106  	if mode == encodeHost || mode == encodeZone {
   107  		// §3.2.2 Host allows
   108  		//	sub-delims = "!" / "$" / "&" / "'" / "(" / ")" / "*" / "+" / "," / ";" / "="
   109  		// as part of reg-name.
   110  		// We add : because we include :port as part of host.
   111  		// We add [ ] because we include [ipv6]:port as part of host.
   112  		// We add < > because they're the only characters left that
   113  		// we could possibly allow, and Parse will reject them if we
   114  		// escape them (because hosts can't use %-encoding for
   115  		// ASCII bytes).
   116  		switch c {
   117  		case '!', '$', '&', '\'', '(', ')', '*', '+', ',', ';', '=', ':', '[', ']', '<', '>', '"':
   118  			return false
   119  		}
   120  	}
   121  
   122  	switch c {
   123  	case '-', '_', '.', '~': // §2.3 Unreserved characters (mark)
   124  		return false
   125  
   126  	case '$', '&', '+', ',', '/', ':', ';', '=', '?', '@': // §2.2 Reserved characters (reserved)
   127  		// Different sections of the URL allow a few of
   128  		// the reserved characters to appear unescaped.
   129  		switch mode {
   130  		case encodePath: // §3.3
   131  			// The RFC allows : @ & = + $ but saves / ; , for assigning
   132  			// meaning to individual path segments. This package
   133  			// only manipulates the path as a whole, so we allow those
   134  			// last three as well. That leaves only ? to escape.
   135  			return c == '?'
   136  
   137  		case encodePathSegment: // §3.3
   138  			// The RFC allows : @ & = + $ but saves / ; , for assigning
   139  			// meaning to individual path segments.
   140  			return c == '/' || c == ';' || c == ',' || c == '?'
   141  
   142  		case encodeUserPassword: // §3.2.1
   143  			// The RFC allows ';', ':', '&', '=', '+', '$', and ',' in
   144  			// userinfo, so we must escape only '@', '/', and '?'.
   145  			// The parsing of userinfo treats ':' as special so we must escape
   146  			// that too.
   147  			return c == '@' || c == '/' || c == '?' || c == ':'
   148  
   149  		case encodeQueryComponent: // §3.4
   150  			// The RFC reserves (so we must escape) everything.
   151  			return true
   152  
   153  		case encodeFragment: // §4.1
   154  			// The RFC text is silent but the grammar allows
   155  			// everything, so escape nothing.
   156  			return false
   157  		}
   158  	}
   159  
   160  	if mode == encodeFragment {
   161  		// RFC 3986 §2.2 allows not escaping sub-delims. A subset of sub-delims are
   162  		// included in reserved from RFC 2396 §2.2. The remaining sub-delims do not
   163  		// need to be escaped. To minimize potential breakage, we apply two restrictions:
   164  		// (1) we always escape sub-delims outside of the fragment, and (2) we always
   165  		// escape single quote to avoid breaking callers that had previously assumed that
   166  		// single quotes would be escaped. See issue #19917.
   167  		switch c {
   168  		case '!', '(', ')', '*':
   169  			return false
   170  		}
   171  	}
   172  
   173  	// Everything else must be escaped.
   174  	return true
   175  }
   176  
   177  // QueryUnescape does the inverse transformation of QueryEscape,
   178  // converting each 3-byte encoded substring of the form "%AB" into the
   179  // hex-decoded byte 0xAB.
   180  // It returns an error if any % is not followed by two hexadecimal
   181  // digits.
   182  func QueryUnescape(s string) (string, error) {
   183  	return unescape(s, encodeQueryComponent)
   184  }
   185  
   186  // PathUnescape does the inverse transformation of PathEscape,
   187  // converting each 3-byte encoded substring of the form "%AB" into the
   188  // hex-decoded byte 0xAB. It returns an error if any % is not followed
   189  // by two hexadecimal digits.
   190  //
   191  // PathUnescape is identical to QueryUnescape except that it does not
   192  // unescape '+' to ' ' (space).
   193  func PathUnescape(s string) (string, error) {
   194  	return unescape(s, encodePathSegment)
   195  }
   196  
   197  // unescape unescapes a string; the mode specifies
   198  // which section of the URL string is being unescaped.
   199  func unescape(s string, mode encoding) (string, error) {
   200  	// Count %, check that they're well-formed.
   201  	n := 0
   202  	hasPlus := false
   203  	for i := 0; i < len(s); {
   204  		switch s[i] {
   205  		case '%':
   206  			n++
   207  			if i+2 >= len(s) || !ishex(s[i+1]) || !ishex(s[i+2]) {
   208  				s = s[i:]
   209  				if len(s) > 3 {
   210  					s = s[:3]
   211  				}
   212  				return "", EscapeError(s)
   213  			}
   214  			// Per https://tools.ietf.org/html/rfc3986#page-21
   215  			// in the host component %-encoding can only be used
   216  			// for non-ASCII bytes.
   217  			// But https://tools.ietf.org/html/rfc6874#section-2
   218  			// introduces %25 being allowed to escape a percent sign
   219  			// in IPv6 scoped-address literals. Yay.
   220  			if mode == encodeHost && unhex(s[i+1]) < 8 && s[i:i+3] != "%25" {
   221  				return "", EscapeError(s[i : i+3])
   222  			}
   223  			if mode == encodeZone {
   224  				// RFC 6874 says basically "anything goes" for zone identifiers
   225  				// and that even non-ASCII can be redundantly escaped,
   226  				// but it seems prudent to restrict %-escaped bytes here to those
   227  				// that are valid host name bytes in their unescaped form.
   228  				// That is, you can use escaping in the zone identifier but not
   229  				// to introduce bytes you couldn't just write directly.
   230  				// But Windows puts spaces here! Yay.
   231  				v := unhex(s[i+1])<<4 | unhex(s[i+2])
   232  				if s[i:i+3] != "%25" && v != ' ' && shouldEscape(v, encodeHost) {
   233  					return "", EscapeError(s[i : i+3])
   234  				}
   235  			}
   236  			i += 3
   237  		case '+':
   238  			hasPlus = mode == encodeQueryComponent
   239  			i++
   240  		default:
   241  			if (mode == encodeHost || mode == encodeZone) && s[i] < 0x80 && shouldEscape(s[i], mode) {
   242  				return "", InvalidHostError(s[i : i+1])
   243  			}
   244  			i++
   245  		}
   246  	}
   247  
   248  	if n == 0 && !hasPlus {
   249  		return s, nil
   250  	}
   251  
   252  	var t strings.Builder
   253  	t.Grow(len(s) - 2*n)
   254  	for i := 0; i < len(s); i++ {
   255  		switch s[i] {
   256  		case '%':
   257  			t.WriteByte(unhex(s[i+1])<<4 | unhex(s[i+2]))
   258  			i += 2
   259  		case '+':
   260  			if mode == encodeQueryComponent {
   261  				t.WriteByte(' ')
   262  			} else {
   263  				t.WriteByte('+')
   264  			}
   265  		default:
   266  			t.WriteByte(s[i])
   267  		}
   268  	}
   269  	return t.String(), nil
   270  }
   271  
   272  // QueryEscape escapes the string so it can be safely placed
   273  // inside a URL query.
   274  func QueryEscape(s string) string {
   275  	return escape(s, encodeQueryComponent)
   276  }
   277  
   278  // PathEscape escapes the string so it can be safely placed inside a URL path segment,
   279  // replacing special characters (including /) with %XX sequences as needed.
   280  func PathEscape(s string) string {
   281  	return escape(s, encodePathSegment)
   282  }
   283  
   284  func escape(s string, mode encoding) string {
   285  	spaceCount, hexCount := 0, 0
   286  	for i := 0; i < len(s); i++ {
   287  		c := s[i]
   288  		if shouldEscape(c, mode) {
   289  			if c == ' ' && mode == encodeQueryComponent {
   290  				spaceCount++
   291  			} else {
   292  				hexCount++
   293  			}
   294  		}
   295  	}
   296  
   297  	if spaceCount == 0 && hexCount == 0 {
   298  		return s
   299  	}
   300  
   301  	var buf [64]byte
   302  	var t []byte
   303  
   304  	required := len(s) + 2*hexCount
   305  	if required <= len(buf) {
   306  		t = buf[:required]
   307  	} else {
   308  		t = make([]byte, required)
   309  	}
   310  
   311  	if hexCount == 0 {
   312  		copy(t, s)
   313  		for i := 0; i < len(s); i++ {
   314  			if s[i] == ' ' {
   315  				t[i] = '+'
   316  			}
   317  		}
   318  		return string(t)
   319  	}
   320  
   321  	j := 0
   322  	for i := 0; i < len(s); i++ {
   323  		switch c := s[i]; {
   324  		case c == ' ' && mode == encodeQueryComponent:
   325  			t[j] = '+'
   326  			j++
   327  		case shouldEscape(c, mode):
   328  			t[j] = '%'
   329  			t[j+1] = upperhex[c>>4]
   330  			t[j+2] = upperhex[c&15]
   331  			j += 3
   332  		default:
   333  			t[j] = s[i]
   334  			j++
   335  		}
   336  	}
   337  	return string(t)
   338  }
   339  
   340  // A URL represents a parsed URL (technically, a URI reference).
   341  //
   342  // The general form represented is:
   343  //
   344  //	[scheme:][//[userinfo@]host][/]path[?query][#fragment]
   345  //
   346  // URLs that do not start with a slash after the scheme are interpreted as:
   347  //
   348  //	scheme:opaque[?query][#fragment]
   349  //
   350  // Note that the Path field is stored in decoded form: /%47%6f%2f becomes /Go/.
   351  // A consequence is that it is impossible to tell which slashes in the Path were
   352  // slashes in the raw URL and which were %2f. This distinction is rarely important,
   353  // but when it is, the code should use RawPath, an optional field which only gets
   354  // set if the default encoding is different from Path.
   355  //
   356  // URL's String method uses the EscapedPath method to obtain the path. See the
   357  // EscapedPath method for more details.
   358  type URL struct {
   359  	Scheme     string
   360  	Opaque     string    // encoded opaque data
   361  	User       *Userinfo // username and password information
   362  	Host       string    // host or host:port
   363  	Path       string    // path (relative paths may omit leading slash)
   364  	RawPath    string    // encoded path hint (see EscapedPath method)
   365  	ForceQuery bool      // append a query ('?') even if RawQuery is empty
   366  	RawQuery   string    // encoded query values, without '?'
   367  	Fragment   string    // fragment for references, without '#'
   368  }
   369  
   370  // User returns a Userinfo containing the provided username
   371  // and no password set.
   372  func User(username string) *Userinfo {
   373  	return &Userinfo{username, "", false}
   374  }
   375  
   376  // UserPassword returns a Userinfo containing the provided username
   377  // and password.
   378  //
   379  // This functionality should only be used with legacy web sites.
   380  // RFC 2396 warns that interpreting Userinfo this way
   381  // ``is NOT RECOMMENDED, because the passing of authentication
   382  // information in clear text (such as URI) has proven to be a
   383  // security risk in almost every case where it has been used.''
   384  func UserPassword(username, password string) *Userinfo {
   385  	return &Userinfo{username, password, true}
   386  }
   387  
   388  // The Userinfo type is an immutable encapsulation of username and
   389  // password details for a URL. An existing Userinfo value is guaranteed
   390  // to have a username set (potentially empty, as allowed by RFC 2396),
   391  // and optionally a password.
   392  type Userinfo struct {
   393  	username    string
   394  	password    string
   395  	passwordSet bool
   396  }
   397  
   398  // Username returns the username.
   399  func (u *Userinfo) Username() string {
   400  	if u == nil {
   401  		return ""
   402  	}
   403  	return u.username
   404  }
   405  
   406  // Password returns the password in case it is set, and whether it is set.
   407  func (u *Userinfo) Password() (string, bool) {
   408  	if u == nil {
   409  		return "", false
   410  	}
   411  	return u.password, u.passwordSet
   412  }
   413  
   414  // String returns the encoded userinfo information in the standard form
   415  // of "username[:password]".
   416  func (u *Userinfo) String() string {
   417  	if u == nil {
   418  		return ""
   419  	}
   420  	s := escape(u.username, encodeUserPassword)
   421  	if u.passwordSet {
   422  		s += ":" + escape(u.password, encodeUserPassword)
   423  	}
   424  	return s
   425  }
   426  
   427  // Maybe rawurl is of the form scheme:path.
   428  // (Scheme must be [a-zA-Z][a-zA-Z0-9+-.]*)
   429  // If so, return scheme, path; else return "", rawurl.
   430  func getscheme(rawurl string) (scheme, path string, err error) {
   431  	for i := 0; i < len(rawurl); i++ {
   432  		c := rawurl[i]
   433  		switch {
   434  		case 'a' <= c && c <= 'z' || 'A' <= c && c <= 'Z':
   435  		// do nothing
   436  		case '0' <= c && c <= '9' || c == '+' || c == '-' || c == '.':
   437  			if i == 0 {
   438  				return "", rawurl, nil
   439  			}
   440  		case c == ':':
   441  			if i == 0 {
   442  				return "", "", errors.New("missing protocol scheme")
   443  			}
   444  			return rawurl[:i], rawurl[i+1:], nil
   445  		default:
   446  			// we have encountered an invalid character,
   447  			// so there is no valid scheme
   448  			return "", rawurl, nil
   449  		}
   450  	}
   451  	return "", rawurl, nil
   452  }
   453  
   454  // split slices s into two substrings separated by the first occurrence of
   455  // sep. If cutc is true then sep is excluded from the second substring.
   456  // If sep does not occur in s then s and the empty string is returned.
   457  func split(s string, sep byte, cutc bool) (string, string) {
   458  	i := strings.IndexByte(s, sep)
   459  	if i < 0 {
   460  		return s, ""
   461  	}
   462  	if cutc {
   463  		return s[:i], s[i+1:]
   464  	}
   465  	return s[:i], s[i:]
   466  }
   467  
   468  // Parse parses rawurl into a URL structure.
   469  //
   470  // The rawurl may be relative (a path, without a host) or absolute
   471  // (starting with a scheme). Trying to parse a hostname and path
   472  // without a scheme is invalid but may not necessarily return an
   473  // error, due to parsing ambiguities.
   474  func Parse(rawurl string) (*URL, error) {
   475  	// Cut off #frag
   476  	u, frag := split(rawurl, '#', true)
   477  	url, err := parse(u, false)
   478  	if err != nil {
   479  		return nil, &Error{"parse", u, err}
   480  	}
   481  	if frag == "" {
   482  		return url, nil
   483  	}
   484  	if url.Fragment, err = unescape(frag, encodeFragment); err != nil {
   485  		return nil, &Error{"parse", rawurl, err}
   486  	}
   487  	return url, nil
   488  }
   489  
   490  // ParseRequestURI parses rawurl into a URL structure. It assumes that
   491  // rawurl was received in an HTTP request, so the rawurl is interpreted
   492  // only as an absolute URI or an absolute path.
   493  // The string rawurl is assumed not to have a #fragment suffix.
   494  // (Web browsers strip #fragment before sending the URL to a web server.)
   495  func ParseRequestURI(rawurl string) (*URL, error) {
   496  	url, err := parse(rawurl, true)
   497  	if err != nil {
   498  		return nil, &Error{"parse", rawurl, err}
   499  	}
   500  	return url, nil
   501  }
   502  
   503  // parse parses a URL from a string in one of two contexts. If
   504  // viaRequest is true, the URL is assumed to have arrived via an HTTP request,
   505  // in which case only absolute URLs or path-absolute relative URLs are allowed.
   506  // If viaRequest is false, all forms of relative URLs are allowed.
   507  func parse(rawurl string, viaRequest bool) (*URL, error) {
   508  	var rest string
   509  	var err error
   510  
   511  	if stringContainsCTLByte(rawurl) {
   512  		return nil, errors.New("net/url: invalid control character in URL")
   513  	}
   514  
   515  	if rawurl == "" && viaRequest {
   516  		return nil, errors.New("empty url")
   517  	}
   518  	url := new(URL)
   519  
   520  	if rawurl == "*" {
   521  		url.Path = "*"
   522  		return url, nil
   523  	}
   524  
   525  	// Split off possible leading "http:", "mailto:", etc.
   526  	// Cannot contain escaped characters.
   527  	if url.Scheme, rest, err = getscheme(rawurl); err != nil {
   528  		return nil, err
   529  	}
   530  	url.Scheme = strings.ToLower(url.Scheme)
   531  
   532  	if strings.HasSuffix(rest, "?") && strings.Count(rest, "?") == 1 {
   533  		url.ForceQuery = true
   534  		rest = rest[:len(rest)-1]
   535  	} else {
   536  		rest, url.RawQuery = split(rest, '?', true)
   537  	}
   538  
   539  	if !strings.HasPrefix(rest, "/") {
   540  		if url.Scheme != "" {
   541  			// We consider rootless paths per RFC 3986 as opaque.
   542  			url.Opaque = rest
   543  			return url, nil
   544  		}
   545  		if viaRequest {
   546  			return nil, errors.New("invalid URI for request")
   547  		}
   548  
   549  		// Avoid confusion with malformed schemes, like cache_object:foo/bar.
   550  		// See golang.org/issue/16822.
   551  		//
   552  		// RFC 3986, §3.3:
   553  		// In addition, a URI reference (Section 4.1) may be a relative-path reference,
   554  		// in which case the first path segment cannot contain a colon (":") character.
   555  		colon := strings.Index(rest, ":")
   556  		slash := strings.Index(rest, "/")
   557  		if colon >= 0 && (slash < 0 || colon < slash) {
   558  			// First path segment has colon. Not allowed in relative URL.
   559  			return nil, errors.New("first path segment in URL cannot contain colon")
   560  		}
   561  	}
   562  
   563  	if (url.Scheme != "" || !viaRequest && !strings.HasPrefix(rest, "///")) && strings.HasPrefix(rest, "//") {
   564  		var authority string
   565  		authority, rest = split(rest[2:], '/', false)
   566  		url.User, url.Host, err = parseAuthority(authority)
   567  		if err != nil {
   568  			return nil, err
   569  		}
   570  	}
   571  	// Set Path and, optionally, RawPath.
   572  	// RawPath is a hint of the encoding of Path. We don't want to set it if
   573  	// the default escaping of Path is equivalent, to help make sure that people
   574  	// don't rely on it in general.
   575  	if err := url.setPath(rest); err != nil {
   576  		return nil, err
   577  	}
   578  	return url, nil
   579  }
   580  
   581  func parseAuthority(authority string) (user *Userinfo, host string, err error) {
   582  	i := strings.LastIndex(authority, "@")
   583  	if i < 0 {
   584  		host, err = parseHost(authority)
   585  	} else {
   586  		host, err = parseHost(authority[i+1:])
   587  	}
   588  	if err != nil {
   589  		return nil, "", err
   590  	}
   591  	if i < 0 {
   592  		return nil, host, nil
   593  	}
   594  	userinfo := authority[:i]
   595  	if !validUserinfo(userinfo) {
   596  		return nil, "", errors.New("net/url: invalid userinfo")
   597  	}
   598  	if !strings.Contains(userinfo, ":") {
   599  		if userinfo, err = unescape(userinfo, encodeUserPassword); err != nil {
   600  			return nil, "", err
   601  		}
   602  		user = User(userinfo)
   603  	} else {
   604  		username, password := split(userinfo, ':', true)
   605  		if username, err = unescape(username, encodeUserPassword); err != nil {
   606  			return nil, "", err
   607  		}
   608  		if password, err = unescape(password, encodeUserPassword); err != nil {
   609  			return nil, "", err
   610  		}
   611  		user = UserPassword(username, password)
   612  	}
   613  	return user, host, nil
   614  }
   615  
   616  // parseHost parses host as an authority without user
   617  // information. That is, as host[:port].
   618  func parseHost(host string) (string, error) {
   619  	if strings.HasPrefix(host, "[") {
   620  		// Parse an IP-Literal in RFC 3986 and RFC 6874.
   621  		// E.g., "[fe80::1]", "[fe80::1%25en0]", "[fe80::1]:80".
   622  		i := strings.LastIndex(host, "]")
   623  		if i < 0 {
   624  			return "", errors.New("missing ']' in host")
   625  		}
   626  		colonPort := host[i+1:]
   627  		if !validOptionalPort(colonPort) {
   628  			return "", fmt.Errorf("invalid port %q after host", colonPort)
   629  		}
   630  
   631  		// RFC 6874 defines that %25 (%-encoded percent) introduces
   632  		// the zone identifier, and the zone identifier can use basically
   633  		// any %-encoding it likes. That's different from the host, which
   634  		// can only %-encode non-ASCII bytes.
   635  		// We do impose some restrictions on the zone, to avoid stupidity
   636  		// like newlines.
   637  		zone := strings.Index(host[:i], "%25")
   638  		if zone >= 0 {
   639  			host1, err := unescape(host[:zone], encodeHost)
   640  			if err != nil {
   641  				return "", err
   642  			}
   643  			host2, err := unescape(host[zone:i], encodeZone)
   644  			if err != nil {
   645  				return "", err
   646  			}
   647  			host3, err := unescape(host[i:], encodeHost)
   648  			if err != nil {
   649  				return "", err
   650  			}
   651  			return host1 + host2 + host3, nil
   652  		}
   653  	} else if i := strings.LastIndex(host, ":"); i != -1 {
   654  		colonPort := host[i:]
   655  		if !validOptionalPort(colonPort) {
   656  			return "", fmt.Errorf("invalid port %q after host", colonPort)
   657  		}
   658  	}
   659  
   660  	var err error
   661  	if host, err = unescape(host, encodeHost); err != nil {
   662  		return "", err
   663  	}
   664  	return host, nil
   665  }
   666  
   667  // setPath sets the Path and RawPath fields of the URL based on the provided
   668  // escaped path p. It maintains the invariant that RawPath is only specified
   669  // when it differs from the default encoding of the path.
   670  // For example:
   671  // - setPath("/foo/bar")   will set Path="/foo/bar" and RawPath=""
   672  // - setPath("/foo%2fbar") will set Path="/foo/bar" and RawPath="/foo%2fbar"
   673  // setPath will return an error only if the provided path contains an invalid
   674  // escaping.
   675  func (u *URL) setPath(p string) error {
   676  	path, err := unescape(p, encodePath)
   677  	if err != nil {
   678  		return err
   679  	}
   680  	u.Path = path
   681  	if escp := escape(path, encodePath); p == escp {
   682  		// Default encoding is fine.
   683  		u.RawPath = ""
   684  	} else {
   685  		u.RawPath = p
   686  	}
   687  	return nil
   688  }
   689  
   690  // EscapedPath returns the escaped form of u.Path.
   691  // In general there are multiple possible escaped forms of any path.
   692  // EscapedPath returns u.RawPath when it is a valid escaping of u.Path.
   693  // Otherwise EscapedPath ignores u.RawPath and computes an escaped
   694  // form on its own.
   695  // The String and RequestURI methods use EscapedPath to construct
   696  // their results.
   697  // In general, code should call EscapedPath instead of
   698  // reading u.RawPath directly.
   699  func (u *URL) EscapedPath() string {
   700  	if u.RawPath != "" && validEncodedPath(u.RawPath) {
   701  		p, err := unescape(u.RawPath, encodePath)
   702  		if err == nil && p == u.Path {
   703  			return u.RawPath
   704  		}
   705  	}
   706  	if u.Path == "*" {
   707  		return "*" // don't escape (Issue 11202)
   708  	}
   709  	return escape(u.Path, encodePath)
   710  }
   711  
   712  // validEncodedPath reports whether s is a valid encoded path.
   713  // It must not contain any bytes that require escaping during path encoding.
   714  func validEncodedPath(s string) bool {
   715  	for i := 0; i < len(s); i++ {
   716  		// RFC 3986, Appendix A.
   717  		// pchar = unreserved / pct-encoded / sub-delims / ":" / "@".
   718  		// shouldEscape is not quite compliant with the RFC,
   719  		// so we check the sub-delims ourselves and let
   720  		// shouldEscape handle the others.
   721  		switch s[i] {
   722  		case '!', '$', '&', '\'', '(', ')', '*', '+', ',', ';', '=', ':', '@':
   723  			// ok
   724  		case '[', ']':
   725  			// ok - not specified in RFC 3986 but left alone by modern browsers
   726  		case '%':
   727  			// ok - percent encoded, will decode
   728  		default:
   729  			if shouldEscape(s[i], encodePath) {
   730  				return false
   731  			}
   732  		}
   733  	}
   734  	return true
   735  }
   736  
   737  // validOptionalPort reports whether port is either an empty string
   738  // or matches /^:\d*$/
   739  func validOptionalPort(port string) bool {
   740  	if port == "" {
   741  		return true
   742  	}
   743  	if port[0] != ':' {
   744  		return false
   745  	}
   746  	for _, b := range port[1:] {
   747  		if b < '0' || b > '9' {
   748  			return false
   749  		}
   750  	}
   751  	return true
   752  }
   753  
   754  // String reassembles the URL into a valid URL string.
   755  // The general form of the result is one of:
   756  //
   757  //	scheme:opaque?query#fragment
   758  //	scheme://userinfo@host/path?query#fragment
   759  //
   760  // If u.Opaque is non-empty, String uses the first form;
   761  // otherwise it uses the second form.
   762  // Any non-ASCII characters in host are escaped.
   763  // To obtain the path, String uses u.EscapedPath().
   764  //
   765  // In the second form, the following rules apply:
   766  //	- if u.Scheme is empty, scheme: is omitted.
   767  //	- if u.User is nil, userinfo@ is omitted.
   768  //	- if u.Host is empty, host/ is omitted.
   769  //	- if u.Scheme and u.Host are empty and u.User is nil,
   770  //	   the entire scheme://userinfo@host/ is omitted.
   771  //	- if u.Host is non-empty and u.Path begins with a /,
   772  //	   the form host/path does not add its own /.
   773  //	- if u.RawQuery is empty, ?query is omitted.
   774  //	- if u.Fragment is empty, #fragment is omitted.
   775  func (u *URL) String() string {
   776  	var buf strings.Builder
   777  	if u.Scheme != "" {
   778  		buf.WriteString(u.Scheme)
   779  		buf.WriteByte(':')
   780  	}
   781  	if u.Opaque != "" {
   782  		buf.WriteString(u.Opaque)
   783  	} else {
   784  		if u.Scheme != "" || u.Host != "" || u.User != nil {
   785  			if u.Host != "" || u.Path != "" || u.User != nil {
   786  				buf.WriteString("//")
   787  			}
   788  			if ui := u.User; ui != nil {
   789  				buf.WriteString(ui.String())
   790  				buf.WriteByte('@')
   791  			}
   792  			if h := u.Host; h != "" {
   793  				buf.WriteString(escape(h, encodeHost))
   794  			}
   795  		}
   796  		path := u.EscapedPath()
   797  		if path != "" && path[0] != '/' && u.Host != "" {
   798  			buf.WriteByte('/')
   799  		}
   800  		if buf.Len() == 0 {
   801  			// RFC 3986 §4.2
   802  			// A path segment that contains a colon character (e.g., "this:that")
   803  			// cannot be used as the first segment of a relative-path reference, as
   804  			// it would be mistaken for a scheme name. Such a segment must be
   805  			// preceded by a dot-segment (e.g., "./this:that") to make a relative-
   806  			// path reference.
   807  			if i := strings.IndexByte(path, ':'); i > -1 && strings.IndexByte(path[:i], '/') == -1 {
   808  				buf.WriteString("./")
   809  			}
   810  		}
   811  		buf.WriteString(path)
   812  	}
   813  	if u.ForceQuery || u.RawQuery != "" {
   814  		buf.WriteByte('?')
   815  		buf.WriteString(u.RawQuery)
   816  	}
   817  	if u.Fragment != "" {
   818  		buf.WriteByte('#')
   819  		buf.WriteString(escape(u.Fragment, encodeFragment))
   820  	}
   821  	return buf.String()
   822  }
   823  
   824  // Values maps a string key to a list of values.
   825  // It is typically used for query parameters and form values.
   826  // Unlike in the http.Header map, the keys in a Values map
   827  // are case-sensitive.
   828  type Values map[string][]string
   829  
   830  // Get gets the first value associated with the given key.
   831  // If there are no values associated with the key, Get returns
   832  // the empty string. To access multiple values, use the map
   833  // directly.
   834  func (v Values) Get(key string) string {
   835  	if v == nil {
   836  		return ""
   837  	}
   838  	vs := v[key]
   839  	if len(vs) == 0 {
   840  		return ""
   841  	}
   842  	return vs[0]
   843  }
   844  
   845  // Set sets the key to value. It replaces any existing
   846  // values.
   847  func (v Values) Set(key, value string) {
   848  	v[key] = []string{value}
   849  }
   850  
   851  // Add adds the value to key. It appends to any existing
   852  // values associated with key.
   853  func (v Values) Add(key, value string) {
   854  	v[key] = append(v[key], value)
   855  }
   856  
   857  // Del deletes the values associated with key.
   858  func (v Values) Del(key string) {
   859  	delete(v, key)
   860  }
   861  
   862  // ParseQuery parses the URL-encoded query string and returns
   863  // a map listing the values specified for each key.
   864  // ParseQuery always returns a non-nil map containing all the
   865  // valid query parameters found; err describes the first decoding error
   866  // encountered, if any.
   867  //
   868  // Query is expected to be a list of key=value settings separated by
   869  // ampersands or semicolons. A setting without an equals sign is
   870  // interpreted as a key set to an empty value.
   871  func ParseQuery(query string) (Values, error) {
   872  	m := make(Values)
   873  	err := parseQuery(m, query)
   874  	return m, err
   875  }
   876  
   877  func parseQuery(m Values, query string) (err error) {
   878  	for query != "" {
   879  		key := query
   880  		if i := strings.IndexAny(key, "&;"); i >= 0 {
   881  			key, query = key[:i], key[i+1:]
   882  		} else {
   883  			query = ""
   884  		}
   885  		if key == "" {
   886  			continue
   887  		}
   888  		value := ""
   889  		if i := strings.Index(key, "="); i >= 0 {
   890  			key, value = key[:i], key[i+1:]
   891  		}
   892  		key, err1 := QueryUnescape(key)
   893  		if err1 != nil {
   894  			if err == nil {
   895  				err = err1
   896  			}
   897  			continue
   898  		}
   899  		value, err1 = QueryUnescape(value)
   900  		if err1 != nil {
   901  			if err == nil {
   902  				err = err1
   903  			}
   904  			continue
   905  		}
   906  		m[key] = append(m[key], value)
   907  	}
   908  	return err
   909  }
   910  
   911  // Encode encodes the values into ``URL encoded'' form
   912  // ("bar=baz&foo=quux") sorted by key.
   913  func (v Values) Encode() string {
   914  	if v == nil {
   915  		return ""
   916  	}
   917  	var buf strings.Builder
   918  	keys := make([]string, 0, len(v))
   919  	for k := range v {
   920  		keys = append(keys, k)
   921  	}
   922  	sort.Strings(keys)
   923  	for _, k := range keys {
   924  		vs := v[k]
   925  		keyEscaped := QueryEscape(k)
   926  		for _, v := range vs {
   927  			if buf.Len() > 0 {
   928  				buf.WriteByte('&')
   929  			}
   930  			buf.WriteString(keyEscaped)
   931  			buf.WriteByte('=')
   932  			buf.WriteString(QueryEscape(v))
   933  		}
   934  	}
   935  	return buf.String()
   936  }
   937  
   938  // resolvePath applies special path segments from refs and applies
   939  // them to base, per RFC 3986.
   940  func resolvePath(base, ref string) string {
   941  	var full string
   942  	if ref == "" {
   943  		full = base
   944  	} else if ref[0] != '/' {
   945  		i := strings.LastIndex(base, "/")
   946  		full = base[:i+1] + ref
   947  	} else {
   948  		full = ref
   949  	}
   950  	if full == "" {
   951  		return ""
   952  	}
   953  	src := strings.Split(full, "/")
   954  	dst := make([]string, 0, len(src))
   955  	for _, elem := range src {
   956  		switch elem {
   957  		case ".":
   958  			// drop
   959  		case "..":
   960  			if len(dst) > 0 {
   961  				dst = dst[:len(dst)-1]
   962  			}
   963  		default:
   964  			dst = append(dst, elem)
   965  		}
   966  	}
   967  	if last := src[len(src)-1]; last == "." || last == ".." {
   968  		// Add final slash to the joined path.
   969  		dst = append(dst, "")
   970  	}
   971  	return "/" + strings.TrimPrefix(strings.Join(dst, "/"), "/")
   972  }
   973  
   974  // IsAbs reports whether the URL is absolute.
   975  // Absolute means that it has a non-empty scheme.
   976  func (u *URL) IsAbs() bool {
   977  	return u.Scheme != ""
   978  }
   979  
   980  // Parse parses a URL in the context of the receiver. The provided URL
   981  // may be relative or absolute. Parse returns nil, err on parse
   982  // failure, otherwise its return value is the same as ResolveReference.
   983  func (u *URL) Parse(ref string) (*URL, error) {
   984  	refurl, err := Parse(ref)
   985  	if err != nil {
   986  		return nil, err
   987  	}
   988  	return u.ResolveReference(refurl), nil
   989  }
   990  
   991  // ResolveReference resolves a URI reference to an absolute URI from
   992  // an absolute base URI u, per RFC 3986 Section 5.2. The URI reference
   993  // may be relative or absolute. ResolveReference always returns a new
   994  // URL instance, even if the returned URL is identical to either the
   995  // base or reference. If ref is an absolute URL, then ResolveReference
   996  // ignores base and returns a copy of ref.
   997  func (u *URL) ResolveReference(ref *URL) *URL {
   998  	url := *ref
   999  	if ref.Scheme == "" {
  1000  		url.Scheme = u.Scheme
  1001  	}
  1002  	if ref.Scheme != "" || ref.Host != "" || ref.User != nil {
  1003  		// The "absoluteURI" or "net_path" cases.
  1004  		// We can ignore the error from setPath since we know we provided a
  1005  		// validly-escaped path.
  1006  		url.setPath(resolvePath(ref.EscapedPath(), ""))
  1007  		return &url
  1008  	}
  1009  	if ref.Opaque != "" {
  1010  		url.User = nil
  1011  		url.Host = ""
  1012  		url.Path = ""
  1013  		return &url
  1014  	}
  1015  	if ref.Path == "" && ref.RawQuery == "" {
  1016  		url.RawQuery = u.RawQuery
  1017  		if ref.Fragment == "" {
  1018  			url.Fragment = u.Fragment
  1019  		}
  1020  	}
  1021  	// The "abs_path" or "rel_path" cases.
  1022  	url.Host = u.Host
  1023  	url.User = u.User
  1024  	url.setPath(resolvePath(u.EscapedPath(), ref.EscapedPath()))
  1025  	return &url
  1026  }
  1027  
  1028  // Query parses RawQuery and returns the corresponding values.
  1029  // It silently discards malformed value pairs.
  1030  // To check errors use ParseQuery.
  1031  func (u *URL) Query() Values {
  1032  	v, _ := ParseQuery(u.RawQuery)
  1033  	return v
  1034  }
  1035  
  1036  // RequestURI returns the encoded path?query or opaque?query
  1037  // string that would be used in an HTTP request for u.
  1038  func (u *URL) RequestURI() string {
  1039  	result := u.Opaque
  1040  	if result == "" {
  1041  		result = u.EscapedPath()
  1042  		if result == "" {
  1043  			result = "/"
  1044  		}
  1045  	} else {
  1046  		if strings.HasPrefix(result, "//") {
  1047  			result = u.Scheme + ":" + result
  1048  		}
  1049  	}
  1050  	if u.ForceQuery || u.RawQuery != "" {
  1051  		result += "?" + u.RawQuery
  1052  	}
  1053  	return result
  1054  }
  1055  
  1056  // Hostname returns u.Host, stripping any valid port number if present.
  1057  //
  1058  // If the result is enclosed in square brackets, as literal IPv6 addresses are,
  1059  // the square brackets are removed from the result.
  1060  func (u *URL) Hostname() string {
  1061  	host, _ := splitHostPort(u.Host)
  1062  	return host
  1063  }
  1064  
  1065  // Port returns the port part of u.Host, without the leading colon.
  1066  //
  1067  // If u.Host doesn't contain a valid numeric port, Port returns an empty string.
  1068  func (u *URL) Port() string {
  1069  	_, port := splitHostPort(u.Host)
  1070  	return port
  1071  }
  1072  
  1073  // splitHostPort separates host and port. If the port is not valid, it returns
  1074  // the entire input as host, and it doesn't check the validity of the host.
  1075  // Unlike net.SplitHostPort, but per RFC 3986, it requires ports to be numeric.
  1076  func splitHostPort(hostport string) (host, port string) {
  1077  	host = hostport
  1078  
  1079  	colon := strings.LastIndexByte(host, ':')
  1080  	if colon != -1 && validOptionalPort(host[colon:]) {
  1081  		host, port = host[:colon], host[colon+1:]
  1082  	}
  1083  
  1084  	if strings.HasPrefix(host, "[") && strings.HasSuffix(host, "]") {
  1085  		host = host[1 : len(host)-1]
  1086  	}
  1087  
  1088  	return
  1089  }
  1090  
  1091  // Marshaling interface implementations.
  1092  // Would like to implement MarshalText/UnmarshalText but that will change the JSON representation of URLs.
  1093  
  1094  func (u *URL) MarshalBinary() (text []byte, err error) {
  1095  	return []byte(u.String()), nil
  1096  }
  1097  
  1098  func (u *URL) UnmarshalBinary(text []byte) error {
  1099  	u1, err := Parse(string(text))
  1100  	if err != nil {
  1101  		return err
  1102  	}
  1103  	*u = *u1
  1104  	return nil
  1105  }
  1106  
  1107  // validUserinfo reports whether s is a valid userinfo string per RFC 3986
  1108  // Section 3.2.1:
  1109  //     userinfo    = *( unreserved / pct-encoded / sub-delims / ":" )
  1110  //     unreserved  = ALPHA / DIGIT / "-" / "." / "_" / "~"
  1111  //     sub-delims  = "!" / "$" / "&" / "'" / "(" / ")"
  1112  //                   / "*" / "+" / "," / ";" / "="
  1113  //
  1114  // It doesn't validate pct-encoded. The caller does that via func unescape.
  1115  func validUserinfo(s string) bool {
  1116  	for _, r := range s {
  1117  		if 'A' <= r && r <= 'Z' {
  1118  			continue
  1119  		}
  1120  		if 'a' <= r && r <= 'z' {
  1121  			continue
  1122  		}
  1123  		if '0' <= r && r <= '9' {
  1124  			continue
  1125  		}
  1126  		switch r {
  1127  		case '-', '.', '_', ':', '~', '!', '$', '&', '\'',
  1128  			'(', ')', '*', '+', ',', ';', '=', '%', '@':
  1129  			continue
  1130  		default:
  1131  			return false
  1132  		}
  1133  	}
  1134  	return true
  1135  }
  1136  
  1137  // stringContainsCTLByte reports whether s contains any ASCII control character.
  1138  func stringContainsCTLByte(s string) bool {
  1139  	for i := 0; i < len(s); i++ {
  1140  		b := s[i]
  1141  		if b < ' ' || b == 0x7f {
  1142  			return true
  1143  		}
  1144  	}
  1145  	return false
  1146  }
  1147  

View as plain text