Added EncodingPolicy enum (ALL_RESERVED, RFC_3986) and switched query and fragment encoding to PercentCodec.FRAGMENT under RFC_3986

Author: arturobernalg

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

@@ -84,16 +84,6 @@ public class PercentCodec {
         URIC.or(UNRESERVED);
     }
 
-    static final BitSet FRAGMENT_SAFE = new BitSet(256);
-    static {
-        FRAGMENT_SAFE.or(UNRESERVED);
-        FRAGMENT_SAFE.or(SUB_DELIMS);
-        FRAGMENT_SAFE.set(':');
-        FRAGMENT_SAFE.set('@');
-        FRAGMENT_SAFE.set('/');
-        FRAGMENT_SAFE.set('?');
-    }
-
     static final BitSet RFC5987_UNRESERVED = new BitSet(256);
 
     static {

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

@@ -87,9 +87,36 @@ public static URIBuilder loopbackAddress() {
     private Charset charset;
     private String fragment;
     private String encodedFragment;
+    private EncodingPolicy encodingPolicy = EncodingPolicy.ALL_RESERVED;
 
     private boolean plusAsBlank;
 
+    /**
+     * Defines the encoding policy for URI components in {@link URIBuilder}.
+     * This enum controls how characters are percent-encoded when constructing a URI,
+     * allowing flexibility between strict encoding and RFC 3986-compliant behavior.
+     *
+     * @since 5.4
+     */
+    public enum EncodingPolicy {
+        /**
+         * Encodes all reserved characters, allowing only unreserved characters
+         * (ALPHA, DIGIT, "-", ".", "_", "~") to remain unencoded. This is a strict
+         * policy suitable for conservative URI production where maximum encoding
+         * is desired.
+         */
+        ALL_RESERVED,
+
+        /**
+         * Follows RFC 3986 component-specific encoding rules. For example, query and
+         * fragment components allow unreserved characters, sub-delimiters ("!", "$",
+         * "&", "'", "(", ")", "*", "+", ",", ";", "="), and additional characters
+         * (":", "@", "/", "?") to remain unencoded, as defined by {@code PercentCodec.FRAGMENT}.
+         * This policy ensures compliance with RFC 3986 while maintaining interoperability.
+         */
+        RFC_3986
+    }
+
     /**
      * Constructs an empty instance.
      */
@@ -175,6 +202,22 @@ public URIBuilder setCharset(final Charset charset) {
         return this;
     }
 
+    /**
+     * Sets the encoding policy for this {@link URIBuilder}.
+     * The encoding policy determines how URI components (e.g., query, fragment) are
+     * percent-encoded when building the URI string. If not set, the default policy
+     * is {@link EncodingPolicy#RFC_3986}.
+     *
+     * @param encodingPolicy the encoding policy to apply, or {@code null} to reset
+     *                       to the default ({@link EncodingPolicy#RFC_3986})
+     * @return this {@link URIBuilder} instance for method chaining
+     * @since 5.4
+     */
+    public URIBuilder setEncodingPolicy(final EncodingPolicy encodingPolicy) {
+        this.encodingPolicy = encodingPolicy;
+        return this;
+    }
+
     /**
      * Gets the authority.
      *
@@ -356,18 +399,22 @@ private String buildString() {
                 } else if (this.userInfo != null) {
                     final int idx = this.userInfo.indexOf(':');
                     if (idx != -1) {
-                        PercentCodec.encode(sb, this.userInfo.substring(0, idx), this.charset, PercentCodec.USERINFO, false);
+                        PercentCodec.encode(sb, this.userInfo.substring(0, idx), this.charset,
+                                encodingPolicy == EncodingPolicy.ALL_RESERVED ? PercentCodec.UNRESERVED : PercentCodec.USERINFO, false);
                         sb.append(':');
-                        PercentCodec.encode(sb, this.userInfo.substring(idx + 1), this.charset, PercentCodec.USERINFO, false);
+                        PercentCodec.encode(sb, this.userInfo.substring(idx + 1), this.charset,
+                                encodingPolicy == EncodingPolicy.ALL_RESERVED ? PercentCodec.UNRESERVED : PercentCodec.USERINFO, false);
                     } else {
-                        PercentCodec.encode(sb, this.userInfo, this.charset, PercentCodec.USERINFO, false);
+                        PercentCodec.encode(sb, this.userInfo, this.charset,
+                                encodingPolicy == EncodingPolicy.ALL_RESERVED ? PercentCodec.UNRESERVED : PercentCodec.USERINFO, false);
                     }
                     sb.append("@");
                 }
                 if (InetAddressUtils.isIPv6(this.host)) {
                     sb.append("[").append(this.host).append("]");
                 } else {
-                    PercentCodec.encode(sb, this.host, this.charset, PercentCodec.REG_NAME, false);
+                    PercentCodec.encode(sb, this.host, this.charset,
+                            encodingPolicy == EncodingPolicy.ALL_RESERVED ? PercentCodec.UNRESERVED : PercentCodec.REG_NAME, false);
                 }
                 if (this.port >= 0) {
                     sb.append(":").append(this.port);
@@ -391,14 +438,16 @@ private String buildString() {
                 formatQuery(sb, this.queryParams, this.charset, false);
             } else if (this.query != null) {
                 sb.append("?");
-                PercentCodec.encode(sb, this.query, this.charset, PercentCodec.QUERY, false);
+                PercentCodec.encode(sb, this.query, this.charset,
+                        encodingPolicy == EncodingPolicy.ALL_RESERVED ? PercentCodec.URIC : PercentCodec.QUERY, false);
             }
         }
         if (this.encodedFragment != null) {
             sb.append("#").append(this.encodedFragment);
         } else if (this.fragment != null) {
             sb.append("#");
-            PercentCodec.encode(sb, this.fragment, this.charset, PercentCodec.FRAGMENT, false);
+            PercentCodec.encode(sb, this.fragment, this.charset,
+                    encodingPolicy == EncodingPolicy.ALL_RESERVED ? PercentCodec.UNRESERVED : PercentCodec.FRAGMENT, false);
         }
         return sb.toString();
     }

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

@@ -171,7 +171,7 @@ void testFormatQuery() {
     @Test
     void testHierarchicalUri() throws Exception {
         final URI uri = new URI("http", "stuff", "localhost", 80, "/some stuff", "param=stuff", "fragment");
-        final URIBuilder uribuilder = new URIBuilder(uri);
+        final URIBuilder uribuilder = new URIBuilder(uri).setEncodingPolicy(URIBuilder.EncodingPolicy.ALL_RESERVED);
         final URI result = uribuilder.build();
         Assertions.assertEquals(new URI("http://stuff@localhost:80/some%20stuff?param=stuff#fragment"), result);
     }
@@ -606,7 +606,7 @@ void testHttpHost() throws Exception {
 
     @Test
     void testSetHostWithReservedChars() throws Exception {
-        final URIBuilder uribuilder = new URIBuilder();
+        final URIBuilder uribuilder = new URIBuilder().setEncodingPolicy(URIBuilder.EncodingPolicy.ALL_RESERVED);
         uribuilder.setScheme("http").setHost("!example!.com");
         Assertions.assertEquals(URI.create("http://%21example%21.com"), uribuilder.build());
     }
@@ -996,6 +996,7 @@ void testFragmentEncoding() throws Exception {
                 .setScheme("http")
                 .setHost("example.com")
                 .setFragment(fragment)
+                .setEncodingPolicy(URIBuilder.EncodingPolicy.RFC_3986)
                 .build();
 
         Assertions.assertEquals(expectedEncodedFragment, uri.getRawFragment());
@@ -1010,6 +1011,7 @@ void testCustomQueryEncoding() throws Exception {
                 .setScheme("http")
                 .setHost("example.com")
                 .setCustomQuery(query)
+                .setEncodingPolicy(URIBuilder.EncodingPolicy.RFC_3986)
                 .build();
 
         Assertions.assertEquals(expectedEncodedQuery, uri.getRawQuery());