feat: allow suffix after wildcard in wildcard access URL (#4524)

coderd/httpapi/url.go

@@ -17,21 +17,10 @@ var (
 		// {PORT/APP_NAME}--{AGENT_NAME}--{WORKSPACE_NAME}--{USERNAME}
 		`^(?P<AppName>%[1]s)--(?P<AgentName>%[1]s)--(?P<WorkspaceName>%[1]s)--(?P<Username>%[1]s)$`,
 		nameRegex))
+
+	validHostnameLabelRegex = regexp.MustCompile(`^[a-z0-9]([-a-z0-9]*[a-z0-9])?$`)
 )
 
-// SplitSubdomain splits a subdomain from the rest of the hostname. E.g.:
-//   - "foo.bar.com" becomes "foo", "bar.com"
-//   - "foo.bar.baz.com" becomes "foo", "bar.baz.com"
-//   - "foo" becomes "foo", ""
-func SplitSubdomain(hostname string) (subdomain string, rest string) {
-	toks := strings.SplitN(hostname, ".", 2)
-	if len(toks) < 2 {
-		return toks[0], ""
-	}
-
-	return toks[0], toks[1]
-}
-
 // ApplicationURL is a parsed application URL hostname.
 type ApplicationURL struct {
 	// Only one of AppName or Port will be set.
@@ -111,3 +100,81 @@ func HostnamesMatch(a, b string) bool {
 
 	return strings.EqualFold(aHost, bHost)
 }
+
+// CompileHostnamePattern compiles a hostname pattern into a regular expression.
+// A hostname pattern is a string that may contain a single wildcard character
+// at the beginning. The wildcard character matches any number of hostname-safe
+// characters excluding periods. The pattern is case-insensitive.
+//
+// The supplied pattern:
+//   - must not start or end with a period
+//   - must contain exactly one asterisk at the beginning
+//   - must not contain any other wildcard characters
+//   - must not contain any other characters that are not hostname-safe (including
+//     whitespace)
+//   - must contain at least two hostname labels/segments (i.e. "foo" or "*" are
+//     not valid patterns, but "foo.bar" and "*.bar" are).
+//
+// The returned regular expression will match an entire hostname with optional
+// trailing periods and whitespace. The first submatch will be the wildcard
+// match.
+func CompileHostnamePattern(pattern string) (*regexp.Regexp, error) {
+	pattern = strings.ToLower(pattern)
+	if strings.Contains(pattern, "http:") || strings.Contains(pattern, "https:") {
+		return nil, xerrors.Errorf("hostname pattern must not contain a scheme: %q", pattern)
+	}
+	if strings.Contains(pattern, ":") {
+		return nil, xerrors.Errorf("hostname pattern must not contain a port: %q", pattern)
+	}
+	if strings.HasPrefix(pattern, ".") || strings.HasSuffix(pattern, ".") {
+		return nil, xerrors.Errorf("hostname pattern must not start or end with a period: %q", pattern)
+	}
+	if strings.Count(pattern, ".") < 1 {
+		return nil, xerrors.Errorf("hostname pattern must contain at least two labels/segments: %q", pattern)
+	}
+	if strings.Count(pattern, "*") != 1 {
+		return nil, xerrors.Errorf("hostname pattern must contain exactly one asterisk: %q", pattern)
+	}
+	if !strings.HasPrefix(pattern, "*") {
+		return nil, xerrors.Errorf("hostname pattern must only contain an asterisk at the beginning: %q", pattern)
+	}
+	for i, label := range strings.Split(pattern, ".") {
+		if i == 0 {
+			// We have to allow the asterisk to be a valid hostname label.
+			label = strings.TrimPrefix(label, "*")
+			label = "a" + label
+		}
+		if !validHostnameLabelRegex.MatchString(label) {
+			return nil, xerrors.Errorf("hostname pattern contains invalid label %q: %q", label, pattern)
+		}
+	}
+
+	// Replace periods with escaped periods.
+	regexPattern := strings.ReplaceAll(pattern, ".", "\\.")
+
+	// Capture wildcard match.
+	regexPattern = strings.Replace(regexPattern, "*", "([^.]+)", 1)
+
+	// Allow trailing period.
+	regexPattern = regexPattern + "\\.?"
+
+	// Allow optional port number.
+	regexPattern += "(:\\d+)?"
+
+	// Allow leading and trailing whitespace.
+	regexPattern = `^\s*` + regexPattern + `\s*$`
+
+	return regexp.Compile(regexPattern)
+}
+
+// ExecuteHostnamePattern executes a pattern generated by CompileHostnamePattern
+// and returns the wildcard match. If the pattern does not match the hostname,
+// returns false.
+func ExecuteHostnamePattern(pattern *regexp.Regexp, hostname string) (string, bool) {
+	matches := pattern.FindStringSubmatch(hostname)
+	if len(matches) < 2 {
+		return "", false
+	}
+
+	return matches[1], true
+}