Add RFC 7639 canonical percent codec for ALPN protocol identifiers. (#596)

Preserve token tchar octets (except '%') and enforce uppercase hex.

Author: arturobernalg

httpcore5/src/main/java/org/apache/hc/core5/net/PercentCodec.java

@@ -33,6 +33,8 @@
 import java.nio.charset.StandardCharsets;
 import java.util.BitSet;
 
+import org.apache.hc.core5.annotation.Internal;
+
 /**
  * Percent-encoding.
  *
@@ -113,6 +115,36 @@ public class PercentCodec {
         RFC5987_UNRESERVED.set('~');
     }
 
+    static final BitSet HTTP_TOKEN_UNRESERVED = new BitSet(256);
+
+    static {
+        // HTTP token characters (tchar) minus '%' (percent-encoded per RFC 7639 canonical form)
+        for (int i = 'a'; i <= 'z'; i++) {
+            HTTP_TOKEN_UNRESERVED.set(i);
+        }
+        for (int i = 'A'; i <= 'Z'; i++) {
+            HTTP_TOKEN_UNRESERVED.set(i);
+        }
+        for (int i = '0'; i <= '9'; i++) {
+            HTTP_TOKEN_UNRESERVED.set(i);
+        }
+
+        HTTP_TOKEN_UNRESERVED.set('!');
+        HTTP_TOKEN_UNRESERVED.set('#');
+        HTTP_TOKEN_UNRESERVED.set('$');
+        HTTP_TOKEN_UNRESERVED.set('&');
+        HTTP_TOKEN_UNRESERVED.set('\'');
+        HTTP_TOKEN_UNRESERVED.set('*');
+        HTTP_TOKEN_UNRESERVED.set('+');
+        HTTP_TOKEN_UNRESERVED.set('-');
+        HTTP_TOKEN_UNRESERVED.set('.');
+        HTTP_TOKEN_UNRESERVED.set('^');
+        HTTP_TOKEN_UNRESERVED.set('_');
+        HTTP_TOKEN_UNRESERVED.set('`');
+        HTTP_TOKEN_UNRESERVED.set('|');
+        HTTP_TOKEN_UNRESERVED.set('~');
+    }
+
     static final BitSet PCHAR = new BitSet(256);
     static final BitSet USERINFO = new BitSet(256);
     static final BitSet REG_NAME = new BitSet(256);
@@ -217,10 +249,12 @@ public static String decode(final CharSequence content, final Charset charset) {
 
     public static final PercentCodec RFC3986 = new PercentCodec(UNRESERVED);
     public static final PercentCodec RFC5987 = new PercentCodec(RFC5987_UNRESERVED);
+    public static final PercentCodec HTTP_TOKEN = new PercentCodec(HTTP_TOKEN_UNRESERVED);
 
     private final BitSet unreserved;
 
-    private PercentCodec(final BitSet unreserved) {
+    @Internal
+    public PercentCodec(final BitSet unreserved) {
         this.unreserved = unreserved;
     }
 

httpcore5/src/test/java/org/apache/hc/core5/net/TestPercentCodec.java

@@ -98,4 +98,30 @@ void verifyRfc5987EncodingandDecoding() {
         assertThat(PercentCodec.RFC5987.decode(PercentCodec.RFC5987.encode(s)), CoreMatchers.equalTo(s));
     }
 
+    @Test
+    void testRfc7639CanonicalAlpnTokenEncoding() {
+        // RFC 7639 requires protocol-id to be a token and applies additional canonical constraints:
+        // - Octets not allowed in tokens MUST be percent-encoded (RFC 3986).
+        // - '%' MUST be percent-encoded.
+        // - Octets that are valid token characters MUST NOT be percent-encoded (except '%').
+        // - Uppercase hex digits MUST be used.
+        assertEquals("h2", PercentCodec.HTTP_TOKEN.encode("h2"));
+        assertEquals("http%2F1.1", PercentCodec.HTTP_TOKEN.encode("http/1.1"));
+        assertEquals("%25", PercentCodec.HTTP_TOKEN.encode("%"));
+        assertEquals("foo+bar", PercentCodec.HTTP_TOKEN.encode("foo+bar"));
+        assertEquals("!#$&'*+-.^_`|~", PercentCodec.HTTP_TOKEN.encode("!#$&'*+-.^_`|~"));
+        assertEquals("foo bar", PercentCodec.HTTP_TOKEN.decode("foo%20bar"));
+        assertEquals("ws/é", PercentCodec.HTTP_TOKEN.decode("ws%2F%C3%A9"));
+    }
+
+    @Test
+    void testPercentCodecEncodeIsNotRfc7639Canonical() {
+        // PercentCodec.encode(..) uses RFC 3986 UNRESERVED as the safe set.
+        // This percent-encodes valid RFC 7230 tchar like '+', '*', '!', '|', which RFC 7639 forbids.
+        assertEquals("foo%2Bbar", PercentCodec.encode("foo+bar", StandardCharsets.UTF_8));
+        assertEquals("%2A", PercentCodec.encode("*", StandardCharsets.UTF_8));
+        assertEquals("%21", PercentCodec.encode("!", StandardCharsets.UTF_8));
+        assertEquals("%7C", PercentCodec.encode("|", StandardCharsets.UTF_8));
+    }
+
 }