#91 Fixed dealing with ambiguous [,;] delimiters (see https://stackoverflow.com/questions/45825426/how-to-tokenize-scan-or-split-this-string-of-email-addresses)

Author: bbottema

src/main/java/org/simplejavamail/internal/util/MiscUtil.java

@@ -87,13 +87,20 @@ public static String readInputStreamToString(@Nonnull final InputStream inputStr
 		}
 		return byteArrayOutputStream.toString(checkNonEmptyArgument(charset, "charset").name());
 	}
-
+	
+	/**
+	 * Recognizes the tails of each address entry, so it can replace the [';] delimiters, thereby disambiguating the delimiters, since they can
+	 * appear in names as well (making it difficult to split on [,;] delimiters.
+	 *
+	 * @param emailAddressList The delimited list of addresses (or single address) optionally including the name.
+	 * @return Array of address entries optionally including the names, trimmed for spaces or trailing delimiters.
+	 */
 	@Nonnull
 	public static String[] extractEmailAddresses(@Nonnull final String emailAddressList) {
-		//noinspection DynamicRegexReplaceableByCompiledPattern
+		// recognize value tails and replace the delimiters there, disambiguating delimiters
 		return checkNonEmptyArgument(emailAddressList, "emailAddressList")
-				.replace(';', ',')
-				.replaceAll("\\s*,\\s*", ",")
-				.split(",");
+				.replaceAll("(@.*?>?)\\s*[,;]", "$1<|>")
+				.replaceAll("<\\|>$", "") // remove trailing delimiter
+				.split("\\s*<\\|>\\s*"); // split on delimiter including surround space
 	}
 }

src/test/java/org/simplejavamail/internal/util/MiscUtilTest.java

@@ -11,78 +11,94 @@ public void checkNotNull() {
 		assertThat(MiscUtil.checkNotNull("blah", null)).isEqualTo("blah");
 		assertThat(MiscUtil.checkNotNull(23523, null)).isEqualTo(23523);
 	}
-
+	
 	@Test(expected = NullPointerException.class)
 	public void checkNotNullWithException() {
 		MiscUtil.checkNotNull(null, null);
 	}
-
+	
 	@Test
 	public void checkArgumentNotEmpty() {
 		assertThat(MiscUtil.checkArgumentNotEmpty("blah", null)).isEqualTo("blah");
 		assertThat(MiscUtil.checkArgumentNotEmpty(234, null)).isEqualTo(234);
 	}
-
+	
 	@Test(expected = IllegalArgumentException.class)
 	public void checkArgumentNotEmptyWithEmptyString() {
 		MiscUtil.checkArgumentNotEmpty("", null);
 	}
-
+	
 	@Test(expected = IllegalArgumentException.class)
 	public void checkArgumentNotEmptyWithNullString() {
 		MiscUtil.checkArgumentNotEmpty(null, null);
 	}
-
+	
 	@Test
 	public void valueNullOrEmpty() {
 		assertThat(MiscUtil.valueNullOrEmpty("")).isEqualTo(true);
 		assertThat(MiscUtil.valueNullOrEmpty(null)).isEqualTo(true);
 		assertThat(MiscUtil.valueNullOrEmpty("blah")).isEqualTo(false);
 		assertThat(MiscUtil.valueNullOrEmpty(2534)).isEqualTo(false);
 	}
-
+	
 	@Test(expected = IllegalArgumentException.class)
 	public void testExtractEmailAddresses_MissingAddress() {
 		MiscUtil.extractEmailAddresses(null);
 	}
-
+	
 	@Test(expected = IllegalArgumentException.class)
 	public void testExtractEmailAddresses_EmptyAddress() {
 		MiscUtil.extractEmailAddresses("");
 	}
-
+	
 	@Test
 	public void testExtractEmailAddresses_SingleAddress() {
 		String[] singleAddressList = MiscUtil.extractEmailAddresses("[email protected]");
 		assertThat(singleAddressList).hasSize(1);
 		assertThat(singleAddressList).contains("[email protected]");
 	}
-
+	
 	@Test
 	public void testExtractEmailAddresses_MultipleAddressesWithCommas() {
 		String[] singleAddressList = MiscUtil.extractEmailAddresses("[email protected],[email protected],[email protected]");
 		assertThat(singleAddressList).hasSize(3);
 		assertThat(singleAddressList).contains("[email protected]", "[email protected]", "[email protected]");
 	}
-
+	
 	@Test
 	public void testExtractEmailAddresses_MultipleAddressesWithSemicolons() {
 		String[] singleAddressList = MiscUtil.extractEmailAddresses("[email protected];[email protected];[email protected]");
 		assertThat(singleAddressList).hasSize(3);
 		assertThat(singleAddressList).contains("[email protected]", "[email protected]", "[email protected]");
 	}
-
+	
 	@Test
 	public void testExtractEmailAddresses_MultipleAddressesMixedCommasAndSemicolons() {
 		String[] singleAddressList = MiscUtil.extractEmailAddresses("[email protected],[email protected];[email protected];[email protected],[email protected]");
 		assertThat(singleAddressList).hasSize(5);
 		assertThat(singleAddressList).contains("[email protected]", "[email protected]", "[email protected]", "[email protected]", "[email protected]");
 	}
-
+	
 	@Test
 	public void testExtractEmailAddresses_MultipleAddressesTralingSpaces() {
 		String[] singleAddressList = MiscUtil.extractEmailAddresses("[email protected], [email protected] ;[email protected];[email protected] , [email protected],[email protected]");
 		assertThat(singleAddressList).hasSize(6);
 		assertThat(singleAddressList).contains("[email protected]", "[email protected]", "[email protected]", "[email protected]", "[email protected]", "[email protected]");
 	}
+	
+	@Test
+	public void testExtractEmailAddresses() {
+		String testInput = "[email protected],Sixpack, \"Joe 1\" <[email protected]>, Sixpack, Joe 2 <[email protected]> ;Sixpack, Joe, 3<name@domain" +
+				".com> , [email protected],[email protected];[email protected]; \" Joe Sixpack 4 \"  <[email protected]>;";
+		assertThat(MiscUtil.extractEmailAddresses(testInput)).containsExactlyInAnyOrder(
+				"[email protected]",
+				"Sixpack, \"Joe 1\" <[email protected]>",
+				"Sixpack, Joe 2 <[email protected]>",
+				"Sixpack, Joe, 3<[email protected]>",
+				"[email protected]",
+				"[email protected]",
+				"[email protected]",
+				"\" Joe Sixpack 4 \"  <[email protected]>"
+		);
+	}
 }