fix encoding of long terms in the subject (#284)

For a long term with no whitespace the term was not getting broken up and would be longer than allowed by some mail systems causing the base64 encoded data to show as the subject. This changes it to just wrap based on the length and ignore whitespace.
Netflix · Nov 19, 2016 · 1dcbf8d · 1dcbf8d
1 parent 79795ac
commit 1dcbf8d
Show file tree

Hide file tree

Showing 5 changed files with 85 additions and 52 deletions.
diff --git a/iep-ses/src/main/java/com/netflix/iep/ses/EncodingUtils.java b/iep-ses/src/main/java/com/netflix/iep/ses/EncodingUtils.java
@@ -100,20 +100,23 @@ static String wrap(String s, int offset, int maxLength) {
   static String wrapBase64(String s, int offset, int maxLength) {
     String prefix = "=?UTF-8?B?";
     String suffix = "?=";
-    String wrapped = wrap(s, offset, maxLength - prefix.length() - suffix.length());
-    String[] lines = wrapped.split("[\r\n]+");
+
+    int length = maxLength - prefix.length() - suffix.length() - 1;
+    int start = 0;
+    int end = length - offset;
     StringBuilder builder = new StringBuilder();
-    builder.append(prefix)
-        .append(base64(lines[0].getBytes(StandardCharsets.UTF_8)))
-        .append(suffix);
-    for (int i = 1; i < lines.length; ++i) {
+    do {
+      String part = s.substring(start, Math.min(end, s.length()));
       builder.append(EncodingUtils.CRLF)
           .append(' ')
           .append(prefix)
-          .append(base64(lines[i].getBytes(StandardCharsets.UTF_8)))
+          .append(base64(part.getBytes(StandardCharsets.UTF_8)))
           .append(suffix);
-    }
-    return builder.toString();
+      start = end;
+      end += length;
+    } while (start < s.length());
+
+    return builder.toString().trim();
   }
 
   /** Base64 encode the input data. */

diff --git a/iep-ses/src/test/java/com/netflix/iep/ses/EmailRequestBuilderTest.java b/iep-ses/src/test/java/com/netflix/iep/ses/EmailRequestBuilderTest.java
@@ -199,6 +199,21 @@ public void tooLongSubject() throws IOException {
         .withTextBody("Body of the email message."));
   }
 
+  @Test
+  public void tooLongSubjectNoSpace() throws IOException {
+    // https://tools.ietf.org/html/rfc5322#section-2.1.1
+    String base = "Too_long_subject_no_space__";
+    String subject = base;
+    while (subject.length() < 256) {
+      subject += base;
+    }
+    checkMessage("tooLongSubjectNoSpace", new EmailRequestBuilder()
+        .withSource(FROM)
+        .withToAddresses(TO)
+        .withSubject(subject)
+        .withTextBody("Body of the email message."));
+  }
+
   @Test
   public void subjectWithUtf8() throws IOException {
     String subject = "警报";

diff --git a/iep-ses/src/test/resources/longSubject b/iep-ses/src/test/resources/longSubject
@@ -3,16 +3,15 @@ Content-Type: multipart/mixed;
   boundary="331239ab-8a31-4cc6-84d6-5557f96ebc3a"
 From: bob@example.com
 To: andrew@example.com
-Subject: =?UTF-8?B?VGVzdCBtZXNzYWdlIHdpdGggYSBsb25n?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRo?=
- =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0?=
- =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQ=?=
- =?UTF-8?B?IHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZw==?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRo?=
- =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0?=
- =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQ=?=
- =?UTF-8?B?IHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZw==?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3Qu?=
+Subject: =?UTF-8?B?VGVzdCBtZXNzYWdlIHdpdGggYSBsb25nIHJlcA==?=
+ =?UTF-8?B?ZWF0ZWQgc3ViamVjdC4gVGVzdCBtZXNzYWdlIHdpdGggYSBsbw==?=
+ =?UTF-8?B?bmcgcmVwZWF0ZWQgc3ViamVjdC4gVGVzdCBtZXNzYWdlIHdpdA==?=
+ =?UTF-8?B?aCBhIGxvbmcgcmVwZWF0ZWQgc3ViamVjdC4gVGVzdCBtZXNzYQ==?=
+ =?UTF-8?B?Z2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQgc3ViamVjdC4gVGVzdA==?=
+ =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQgc3ViamVjdA==?=
+ =?UTF-8?B?LiBUZXN0IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQgcw==?=
+ =?UTF-8?B?dWJqZWN0LiBUZXN0IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZQ==?=
+ =?UTF-8?B?YXRlZCBzdWJqZWN0LiA=?=
 
 --331239ab-8a31-4cc6-84d6-5557f96ebc3a
 Content-Type: text/plain; charset=UTF-8

diff --git a/iep-ses/src/test/resources/tooLongSubject b/iep-ses/src/test/resources/tooLongSubject
@@ -3,39 +3,35 @@ Content-Type: multipart/mixed;
   boundary="331239ab-8a31-4cc6-84d6-5557f96ebc3a"
 From: bob@example.com
 To: andrew@example.com
-Subject: =?UTF-8?B?VGVzdCBtZXNzYWdlIHdpdGggYSBsb25n?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRo?=
- =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0?=
- =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQ=?=
- =?UTF-8?B?IHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZw==?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRo?=
- =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0?=
- =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQ=?=
- =?UTF-8?B?IHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZw==?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRo?=
- =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0?=
- =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQ=?=
- =?UTF-8?B?IHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZw==?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRo?=
- =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0?=
- =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQ=?=
- =?UTF-8?B?IHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZw==?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRo?=
- =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0?=
- =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQ=?=
- =?UTF-8?B?IHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZw==?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRo?=
- =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0?=
- =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQ=?=
- =?UTF-8?B?IHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZw==?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRo?=
- =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0?=
- =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQ=?=
- =?UTF-8?B?IHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZw==?=
- =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRo?=
- =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0?=
- =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQ=?=
- =?UTF-8?B?IHN1YmplY3Qu?=
+Subject: =?UTF-8?B?VGVzdCBtZXNzYWdlIHdpdGggYSBsb25nIHJlcA==?=
+ =?UTF-8?B?ZWF0ZWQgc3ViamVjdC4gVGVzdCBtZXNzYWdlIHdpdGggYSBsbw==?=
+ =?UTF-8?B?bmcgcmVwZWF0ZWQgc3ViamVjdC4gVGVzdCBtZXNzYWdlIHdpdA==?=
+ =?UTF-8?B?aCBhIGxvbmcgcmVwZWF0ZWQgc3ViamVjdC4gVGVzdCBtZXNzYQ==?=
+ =?UTF-8?B?Z2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQgc3ViamVjdC4gVGVzdA==?=
+ =?UTF-8?B?IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQgc3ViamVjdA==?=
+ =?UTF-8?B?LiBUZXN0IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQgcw==?=
+ =?UTF-8?B?dWJqZWN0LiBUZXN0IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZQ==?=
+ =?UTF-8?B?YXRlZCBzdWJqZWN0LiBUZXN0IG1lc3NhZ2Ugd2l0aCBhIGxvbg==?=
+ =?UTF-8?B?ZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0IG1lc3NhZ2Ugd2l0aA==?=
+ =?UTF-8?B?IGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0IG1lc3NhZw==?=
+ =?UTF-8?B?ZSB3aXRoIGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0LiBUZXN0IA==?=
+ =?UTF-8?B?bWVzc2FnZSB3aXRoIGEgbG9uZyByZXBlYXRlZCBzdWJqZWN0Lg==?=
+ =?UTF-8?B?IFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZyByZXBlYXRlZCBzdQ==?=
+ =?UTF-8?B?YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZyByZXBlYQ==?=
+ =?UTF-8?B?dGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIGEgbG9uZw==?=
+ =?UTF-8?B?IHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZSB3aXRoIA==?=
+ =?UTF-8?B?YSBsb25nIHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbWVzc2FnZQ==?=
+ =?UTF-8?B?IHdpdGggYSBsb25nIHJlcGVhdGVkIHN1YmplY3QuIFRlc3QgbQ==?=
+ =?UTF-8?B?ZXNzYWdlIHdpdGggYSBsb25nIHJlcGVhdGVkIHN1YmplY3QuIA==?=
+ =?UTF-8?B?VGVzdCBtZXNzYWdlIHdpdGggYSBsb25nIHJlcGVhdGVkIHN1Yg==?=
+ =?UTF-8?B?amVjdC4gVGVzdCBtZXNzYWdlIHdpdGggYSBsb25nIHJlcGVhdA==?=
+ =?UTF-8?B?ZWQgc3ViamVjdC4gVGVzdCBtZXNzYWdlIHdpdGggYSBsb25nIA==?=
+ =?UTF-8?B?cmVwZWF0ZWQgc3ViamVjdC4gVGVzdCBtZXNzYWdlIHdpdGggYQ==?=
+ =?UTF-8?B?IGxvbmcgcmVwZWF0ZWQgc3ViamVjdC4gVGVzdCBtZXNzYWdlIA==?=
+ =?UTF-8?B?d2l0aCBhIGxvbmcgcmVwZWF0ZWQgc3ViamVjdC4gVGVzdCBtZQ==?=
+ =?UTF-8?B?c3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQgc3ViamVjdC4gVA==?=
+ =?UTF-8?B?ZXN0IG1lc3NhZ2Ugd2l0aCBhIGxvbmcgcmVwZWF0ZWQgc3Viag==?=
+ =?UTF-8?B?ZWN0Lg==?=
 
 --331239ab-8a31-4cc6-84d6-5557f96ebc3a
 Content-Type: text/plain; charset=UTF-8

diff --git a/iep-ses/src/test/resources/tooLongSubjectNoSpace b/iep-ses/src/test/resources/tooLongSubjectNoSpace
@@ -0,0 +1,20 @@
+MIME-Version: 1.0
+Content-Type: multipart/mixed;
+  boundary="331239ab-8a31-4cc6-84d6-5557f96ebc3a"
+From: bob@example.com
+To: andrew@example.com
+Subject: =?UTF-8?B?VG9vX2xvbmdfc3ViamVjdF9ub19zcGFjZV9fVA==?=
+ =?UTF-8?B?b29fbG9uZ19zdWJqZWN0X25vX3NwYWNlX19Ub29fbG9uZ19zdQ==?=
+ =?UTF-8?B?YmplY3Rfbm9fc3BhY2VfX1Rvb19sb25nX3N1YmplY3Rfbm9fcw==?=
+ =?UTF-8?B?cGFjZV9fVG9vX2xvbmdfc3ViamVjdF9ub19zcGFjZV9fVG9vXw==?=
+ =?UTF-8?B?bG9uZ19zdWJqZWN0X25vX3NwYWNlX19Ub29fbG9uZ19zdWJqZQ==?=
+ =?UTF-8?B?Y3Rfbm9fc3BhY2VfX1Rvb19sb25nX3N1YmplY3Rfbm9fc3BhYw==?=
+ =?UTF-8?B?ZV9fVG9vX2xvbmdfc3ViamVjdF9ub19zcGFjZV9fVG9vX2xvbg==?=
+ =?UTF-8?B?Z19zdWJqZWN0X25vX3NwYWNlX18=?=
+
+--331239ab-8a31-4cc6-84d6-5557f96ebc3a
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: base64
+
+Qm9keSBvZiB0aGUgZW1haWwgbWVzc2FnZS4=
+--331239ab-8a31-4cc6-84d6-5557f96ebc3a--