Parse rfc6532 addresses

core/pom.xml

@@ -69,6 +69,18 @@
             <groupId>org.eclipse.angus</groupId>
             <artifactId>jakarta.mail</artifactId>
         </dependency>
+        <dependency>
+            <groupId>org.openjdk.jmh</groupId>
+            <artifactId>jmh-core</artifactId>
+            <version>1.37</version>
+            <scope>test</scope>
+        </dependency>
+        <dependency>
+            <groupId>org.openjdk.jmh</groupId>
+            <artifactId>jmh-generator-annprocess</artifactId>
+            <version>1.37</version>
+            <scope>test</scope>
+        </dependency>
         <dependency>
             <groupId>org.reactivestreams</groupId>
             <artifactId>reactive-streams</artifactId>
@@ -79,4 +91,32 @@
         </dependency>
     </dependencies>
 
+    <build>
+        <plugins>
+            <plugin>
+                <groupId>org.apache.maven.plugins</groupId>
+                <artifactId>maven-compiler-plugin</artifactId>
+                <executions>
+                    <execution>
+                        <!-- The JMH benchmarks under src/test need their annotation
+                             processor to emit META-INF/BenchmarkList; without it the
+                             JMH runner aborts with "Unable to find the resource".
+                             Scoped to the test sources so main stays -proc:none. -->
+                        <id>default-testCompile</id>
+                        <configuration>
+                            <proc>full</proc>
+                            <annotationProcessorPaths>
+                                <path>
+                                    <groupId>org.openjdk.jmh</groupId>
+                                    <artifactId>jmh-generator-annprocess</artifactId>
+                                    <version>1.37</version>
+                                </path>
+                            </annotationProcessorPaths>
+                        </configuration>
+                    </execution>
+                </executions>
+            </plugin>
+        </plugins>
+    </build>
+
 </project>

core/src/main/java/org/apache/james/core/Domain.java

@@ -20,6 +20,7 @@
 package org.apache.james.core;
 
 import java.io.Serializable;
+import java.net.IDN;
 import java.util.Locale;
 import java.util.Objects;
 
@@ -42,6 +43,22 @@ public boolean matches(char c) {
     public static final Domain LOCALHOST = Domain.of("localhost");
     public static final int MAXIMUM_DOMAIN_LENGTH = 253;
 
+    /**
+     * Whether {@code s} is pure US-ASCII. Defined here rather than in
+     * {@link MailAddress} -- which exposes it as public API -- because
+     * {@code Domain} must not depend on {@code MailAddress}: the two would
+     * then initialise each other, and {@link #LOCALHOST} is built in this
+     * class's static initialiser.
+     */
+    static boolean isAscii(String s) {
+        for (int i = 0; i < s.length(); i++) {
+            if (s.charAt(i) > 0x7F) {
+                return false;
+            }
+        }
+        return true;
+    }
+
     private static String removeBrackets(String domainName) {
         if (!(domainName.startsWith("[") && domainName.endsWith("]"))) {
             return domainName;
@@ -54,9 +71,37 @@ public static Domain of(String domain) {
         Preconditions.checkArgument(domain.length() <= MAXIMUM_DOMAIN_LENGTH,
             "Domain name length should not exceed %s characters", MAXIMUM_DOMAIN_LENGTH);
 
-        String domainWithoutBrackets = removeBrackets(domain);
+        String withoutBrackets = removeBrackets(domain);
+        String domainWithoutBrackets;
+        if (isAscii(withoutBrackets)) {
+            // RFC 3490 4.1: ToASCII skips both nameprep and Punycode when the
+            // input is already all-ASCII and hands it back unchanged, so IDN
+            // would only cost time here -- and it is on the path of every
+            // address James parses. The empty-label and length checks it also
+            // performs are done below by assertValidPart anyway.
+            domainWithoutBrackets = withoutBrackets;
+        } else {
+            try {
+                domainWithoutBrackets = IDN.toASCII(withoutBrackets, IDN.ALLOW_UNASSIGNED);
+            } catch (IllegalArgumentException e) {
+                // IDN.toASCII's own message can be cryptic ("Empty label is
+                // not a legal name", "A prohibited code point was found in
+                // the input..."). Let's save wear and tear on the poor
+                // developer's brain.
+                throw new IllegalArgumentException(
+                    "Domain '" + domain + "' is invalid according to IDNA: " + e.getMessage(), e);
+            }
+        }
         Preconditions.checkArgument(PART_CHAR_MATCHER.matchesAllOf(domainWithoutBrackets),
-            "Domain parts ASCII chars must be a-z A-Z 0-9 - or _ in %s", domain);
+                                    "Domain parts ASCII chars must be a-z A-Z 0-9 - or _ in %s", domain);
+
+        if (domainWithoutBrackets.startsWith("xn--") ||
+            domainWithoutBrackets.contains(".xn--")) {
+            domainWithoutBrackets = IDN.toUnicode(domainWithoutBrackets);
+            Preconditions.checkArgument(!domainWithoutBrackets.startsWith("xn--") &&
+                                        !domainWithoutBrackets.contains(".xn--"),
+                                        "A-label could not be decoded to Unicode in %s", domain);
+        }
 
         int pos = 0;
         int nextDot = domainWithoutBrackets.indexOf('.');

core/src/main/java/org/apache/james/core/MailAddress.java

@@ -19,6 +19,8 @@
 
 package org.apache.james.core;
 
+import java.net.IDN;
+import java.text.Normalizer;
 import java.util.Locale;
 import java.util.Objects;
 import java.util.Optional;
@@ -172,7 +174,16 @@ private int stripSourceRoute(String address, int pos) {
      * @throws AddressException if the parse failed
      */
     public MailAddress(String address) throws AddressException {
+        // RFC 6532 §3.1 recommends NFC normalisation. Canonically-equivalent
+        // Unicode strings (for example U+00E9 vs U+0065 U+0301 — both render
+        // as "é") then produce equal MailAddress objects with equal hashCode
+        // values, which dedup, alias resolution and routing-table lookups
+        // rely on. NFC is a no-op for pure ASCII, and this runs for every
+        // address James parses, so ASCII skips the normaliser outright.
         address = address.trim();
+        if (!Domain.isAscii(address)) {
+            address = Normalizer.normalize(address, Normalizer.Form.NFC);
+        }
         int pos = 0;
 
         // Test if mail address has source routing information (RFC-821) and get rid of it!!
@@ -397,6 +408,24 @@ public String asString() {
         return localPart + "@" + domain.asString();
     }
 
+    /**
+     * Whether this address is pure US-ASCII, i.e. it can be carried without
+     * RFC 6531 (SMTPUTF8) and reported with the RFC 3798 {@code rfc822}
+     * addr-type rather than the RFC 6533 {@code utf-8} one.
+     */
+    public boolean isAscii() {
+        return isAscii(localPart) && isAscii(domain.asString());
+    }
+
+    /**
+     * Whether {@code s} is pure US-ASCII. Shared by the callers that hold an
+     * address as a raw string rather than as a {@link MailAddress} -- the SMTP
+     * command handlers and the remote-delivery SMTPUTF8 strategy.
+     */
+    public static boolean isAscii(String s) {
+        return Domain.isAscii(s);
+    }
+
     @Override
     public String toString() {
         return localPart + "@" + Optional.ofNullable(domain)
@@ -418,7 +447,7 @@ public Optional<InternetAddress> toInternetAddress() {
         try {
             return Optional.of(new InternetAddress(toString()));
         } catch (AddressException ae) {
-            LOGGER.warn("A valid address '{}' as per James criterial fails to parse as a jakarta.mail InternetAdrress", asString());
+            LOGGER.warn("A valid address '{}' as per James criteria fails to parse as a jakarta.mail InternetAdrress", asString());
             return Optional.empty();
         }
     }
@@ -513,6 +542,24 @@ private int parseQuotedLocalPart(StringBuilder lpSB, String address, int pos) th
                             "characters exception <CR>, <LF>, quote (\"), or backslash (\\) at position " +
                             (pos + 1) + " in '" + address + "'");
                 }
+                // Same surrogate-pair check as in parseUnquotedLocalPart:
+                // unpaired or mis-ordered surrogates would produce
+                // ill-formed UTF-8 on output.
+                if (Character.isLowSurrogate(q)) {
+                    throw new AddressException("Unpaired UTF-16 low surrogate in quoted local-part at position " +
+                            (pos + 1) + " in '" + address + "'");
+                }
+                if (Character.isHighSurrogate(q)) {
+                    if (pos + 1 >= address.length()
+                            || !Character.isLowSurrogate(address.charAt(pos + 1))) {
+                        throw new AddressException("Unpaired UTF-16 high surrogate in quoted local-part at position " +
+                                (pos + 1) + " in '" + address + "'");
+                    }
+                    lpSB.append(q);
+                    lpSB.append(address.charAt(pos + 1));
+                    pos += 2;
+                    continue;
+                }
                 lpSB.append(q);
                 pos++;
             }
@@ -549,18 +596,40 @@ private int parseUnquotedLocalPart(StringBuilder lpSB, String address, int pos)
                 //End of local-part
                 break;
             } else {
-                //<c> ::= any one of the 128 ASCII characters, but not any
-                //    <special> or <SP>
+                //<c> ::= any printable ASCII character, or any non-ASCII
+                //    unicode codepoint, but not <special> or <SP>
                 //<special> ::= "<" | ">" | "(" | ")" | "[" | "]" | "\" | "."
                 //    | "," | ";" | ":" | "@"  """ | the control
-                //    characters (ASCII codes 0 through 31 inclusive and
-                //    127)
+                //    characters (ASCII codes 0 through 31 inclusive,
+                //    127, and the C1 controls 128 through 159)
                 //<SP> ::= the space character (ASCII code 32)
                 char c = address.charAt(pos);
-                if (c <= 31 || c >= 127 || c == ' ') {
+                if (c <= 31 || c == 127 || c == ' ' || (c >= 0x80 && c <= 0x9F)) {
                     throw new AddressException("Invalid character in local-part (user account) at position " +
                             (pos + 1) + " in '" + address + "'", address, pos + 1);
                 }
+                // Java strings are UTF-16, so a supplementary-plane
+                // codepoint (emoji, CJK extension, etc.) appears here as a
+                // high-surrogate followed by a low-surrogate. We must keep
+                // them paired so the address can serialise as well-formed
+                // UTF-8 (RFC 6532 §3.1) — a lone or mis-ordered surrogate
+                // would produce ill-formed UTF-8 octets on output.
+                if (Character.isLowSurrogate(c)) {
+                    throw new AddressException("Unpaired UTF-16 low surrogate in local-part at position " +
+                            (pos + 1) + " in '" + address + "'", address, pos + 1);
+                }
+                if (Character.isHighSurrogate(c)) {
+                    if (pos + 1 >= address.length()
+                            || !Character.isLowSurrogate(address.charAt(pos + 1))) {
+                        throw new AddressException("Unpaired UTF-16 high surrogate in local-part at position " +
+                                (pos + 1) + " in '" + address + "'", address, pos + 1);
+                    }
+                    lpSB.append(c);
+                    lpSB.append(address.charAt(pos + 1));
+                    pos += 2;
+                    lastCharDot = false;
+                    continue;
+                }
                 int i = 0;
                 while (i < SPECIAL.length) {
                     if (c == SPECIAL[i]) {
@@ -688,6 +757,7 @@ private int parseDomain(StringBuilder dSB, String address, int pos) throws Addre
         // in practice though, we should relax this as domain names can start
         // with digits as well as letters.  So only check that doesn't start
         // or end with hyphen.
+        boolean unicode = false;
         while (true) {
             if (pos >= address.length()) {
                 break;
@@ -700,13 +770,31 @@ private int parseDomain(StringBuilder dSB, String address, int pos) throws Addre
                 resultSB.append(ch);
                 pos++;
                 continue;
+            } else if (ch >= 0x0080) {
+                resultSB.append(ch);
+                pos++;
+                unicode = true;
+                continue;
             }
             if (ch == '.') {
                 break;
             }
             throw new AddressException("Invalid character at " + pos + " in '" + address + "'", address, pos);
         }
         String result = resultSB.toString();
+        if (unicode) {
+            try {
+                result = IDN.toASCII(result, IDN.ALLOW_UNASSIGNED);
+            } catch (IllegalArgumentException e) {
+                throw new AddressException("Domain invalid according to IDNA", address);
+            }
+        }
+        if (result.startsWith("xn--") || result.contains(".xn--")) {
+            result = IDN.toUnicode(result);
+            if (result.startsWith("xn--") || result.contains(".xn--")) {
+                throw new AddressException("Domain invalid according to IDNA", address);
+            }
+        }
         if (result.startsWith("-") || result.endsWith("-")) {
             throw new AddressException("Domain name cannot begin or end with a hyphen \"-\" at position " +
                     (pos + 1) + " in '" + address + "'", address, pos + 1);

core/src/test/java/org/apache/james/core/DomainTest.java

@@ -0,0 +1,91 @@
+/****************************************************************
+ * Licensed to the Apache Software Foundation (ASF) under one   *
+ * or more contributor license agreements.  See the NOTICE file *
+ * distributed with this work for additional information        *
+ * regarding copyright ownership.  The ASF licenses this file   *
+ * to you under the Apache License, Version 2.0 (the            *
+ * "License"); you may not use this file except in compliance   *
+ * with the License.  You may obtain a copy of the License at   *
+ *                                                              *
+ *   http://www.apache.org/licenses/LICENSE-2.0                 *
+ *                                                              *
+ * Unless required by applicable law or agreed to in writing,   *
+ * software distributed under the License is distributed on an  *
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY       *
+ * KIND, either express or implied.  See the License for the    *
+ * specific language governing permissions and limitations      *
+ * under the License.                                           *
+ ****************************************************************/
+
+package org.apache.james.core;
+
+import static org.assertj.core.api.Assertions.assertThat;
+import static org.assertj.core.api.Assertions.assertThatThrownBy;
+
+import java.util.stream.Stream;
+
+import org.junit.jupiter.api.Test;
+import org.junit.jupiter.params.ParameterizedTest;
+import org.junit.jupiter.params.provider.Arguments;
+import org.junit.jupiter.params.provider.MethodSource;
+
+
+class DomainTest {
+    @Test
+    void testPlainDomain() {
+        Domain d1 = Domain.of("example.com");
+        assertThat(d1.name().equals(d1.asString())).isTrue();
+        Domain d2 = Domain.of("Example.com");
+        assertThat(d2.name()).isNotEqualTo(d2.asString());
+        assertThat(d1.asString()).isEqualTo(d2.asString());
+    }
+
+    @Test
+    void testIPv4Domain() {
+        Domain d1 = Domain.of("192.0.4.1");
+        assertThat(d1.asString()).isEqualTo("192.0.4.1");
+    }
+
+    @Test
+    void testPunycodeIDN() {
+        Domain d1 = Domain.of("xn--gr-zia.example");
+        assertThat(d1.asString()).isEqualTo("grå.example");
+    }
+
+    @Test
+    void testDevanagariDomain() {
+        Domain d1 = Domain.of("डाटामेल.भारत");
+        assertThat(d1.asString()).isEqualTo(d1.name());
+    }
+
+    private static Stream<Arguments> malformedDomains() {
+        return Stream.of(
+                         "😊☺️.example", // emoji not permitted by IDNA
+                         "#.example", // really and truly not permitted
+                         "\uFEFF.example", // U+FEFF is the byte order mark
+                         "\u200C.example", // U+200C is a zero-width non-joiner
+                         "\u200Eibm.example" // U+200E is left-to-right
+                         )
+            .map(Arguments::of);
+    }
+
+    @ParameterizedTest
+    @MethodSource("malformedDomains")
+    void testMalformedDomains(String malformed) {
+        assertThatThrownBy(() -> Domain.of(malformed))
+            .as("rejecting malformed domain " + malformed)
+            .isInstanceOf(IllegalArgumentException.class);
+    }
+
+    @ParameterizedTest
+    @MethodSource("malformedDomains")
+    void exceptionForMalformedDomainShouldNameTheOffendingInput(String malformed) {
+        // Without the offending input in the message, "Domain invalid
+        // according to IDNA" leaves a future debugger guessing what
+        // string actually triggered it.
+        assertThatThrownBy(() -> Domain.of(malformed))
+            .hasMessageContaining(malformed);
+    }
+}
+
+