author Ben Bucksch <>
Mon, 21 May 2018 18:44:41 +0200
changeset 23964 96fab4a2b81189101231b12106823748c9a70a94
child 23971 18881dd127e3b0c0d3f97390c9094e309d4dd9c1
permissions -rw-r--r--
Bug 1419417 - Parse HTML to make sure that tags and attributes are properly closed. r=mkmelin,jorgk This fixes the efail <> security bug, which opens a HTML tag or attribute in an HTML MIME part, then puts in a PGP-encrypted part, and then another HTML part with the closing quote or tag. This could be e.g. <img src=' or <form><textarea>, CSS URL or similar features that send out the following text as URL and therefore leak it to the attacker who crafted the email. The PGP part will then be decrypted and leak. The bug was that we just passed HTML through verbatim. The frontend does not have any further precautions, either. The correct solution here is to jail each MIME part into a separate <iframe type="content"> in the UI. However, we don't want one scrollbar for each MIME part, but one scroll for the entire body. <iframe seamless> would allow that, but it was never implemented in Firefox and is now dead. We might later find a workaround, but this is more work and can't be done short term. The fix here in libmime first parses the HTML that we get in the HTML MIME part, and then immediately serialized it again. That ensures that the HTML document is complete, syntactically correct, and all tags and attributes are properly closed, before we start with the next MIME part.

/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, You can obtain one at */

/* Most of this code is copied from mimethsa. If you find a bug here, check that class, too. */

/* This runs the entire HTML document through the Mozilla HTML parser, and
   then outputs it as string again. This ensures that the HTML document is
   syntactically correct and complete and all tags and attributes are closed.

   That prevents "MIME in the middle" attacks like
   The base problem is that we concatenate different MIME parts in the output
   and render them all together as a single HTML document in the display.

   The better solution would be to put each MIME part into its own <iframe type="content">.
   during rendering. Unfortunately, we'd need <iframe seamless> for that.
   That would remove the need for this workaround, and stop even more attack classes.

#include "mimeTextHTMLParsed.h"
#include "prmem.h"
#include "prlog.h"
#include "msgCore.h"
#include "mozilla/dom/DOMParser.h"
#include "nsIDocument.h"
#include "nsIDocumentEncoder.h"
#include "mozilla/ErrorResult.h"
#include "nsIPrefBranch.h"

#define MIME_SUPERCLASS mimeInlineTextHTMLClass
MimeDefClass(MimeInlineTextHTMLParsed, MimeInlineTextHTMLParsedClass,
             mimeInlineTextHTMLParsedClass, &MIME_SUPERCLASS);

static int MimeInlineTextHTMLParsed_parse_line(const char *, int32_t,
                                               MimeObject *);
static int MimeInlineTextHTMLParsed_parse_begin(MimeObject *obj);
static int MimeInlineTextHTMLParsed_parse_eof(MimeObject *, bool);
static void MimeInlineTextHTMLParsed_finalize(MimeObject *obj);

static int
MimeInlineTextHTMLParsedClassInitialize(MimeInlineTextHTMLParsedClass *clazz)
  MimeObjectClass *oclass = (MimeObjectClass *)clazz;
  NS_ASSERTION(!oclass->class_initialized, "problem with superclass");
  oclass->parse_line  = MimeInlineTextHTMLParsed_parse_line;
  oclass->parse_begin = MimeInlineTextHTMLParsed_parse_begin;
  oclass->parse_eof   = MimeInlineTextHTMLParsed_parse_eof;
  oclass->finalize    = MimeInlineTextHTMLParsed_finalize;

  return 0;

static int
MimeInlineTextHTMLParsed_parse_begin(MimeObject *obj)
  MimeInlineTextHTMLParsed *me = (MimeInlineTextHTMLParsed *)obj;
  me->complete_buffer = new nsString();
  int status = ((MimeObjectClass*)&MIME_SUPERCLASS)->parse_begin(obj);
  if (status < 0)
    return status;

  // Dump the charset we get from the mime headers into a HTML <meta http-equiv>.
  char *content_type = obj->headers ?
    MimeHeaders_get(obj->headers, HEADER_CONTENT_TYPE, false, false) : 0;
  if (content_type)
    char* charset = MimeHeaders_get_parameter(content_type,
                                              NULL, NULL);
    if (charset)
      nsAutoCString charsetline(
        "\n<meta http-equiv=\"content-type\" content=\"text/html; charset=");
      charsetline += charset;
      charsetline += "\">\n";
      int status = MimeObject_write(obj,
      if (status < 0)
        return status;
  return 0;

static int
MimeInlineTextHTMLParsed_parse_eof(MimeObject *obj, bool abort_p)

  if (obj->closed_p)
    return 0;
  int status = ((MimeObjectClass*)&MIME_SUPERCLASS)->parse_eof(obj, abort_p);
  if (status < 0)
    return status;
  MimeInlineTextHTMLParsed *me = (MimeInlineTextHTMLParsed *)obj;

  // We have to cache all lines and parse the whole document at once.
  // There's a useful sounding function parseFromStream(), but it only allows XML
  // mimetypes, not HTML. Methinks that's because the HTML soup parser
  // needs the entire doc to make sense of the gibberish that people write.
  if (!me || !me->complete_buffer)
    return 0;

  nsString& rawHTML = *(me->complete_buffer);
  nsString parsed;
  nsresult rv;

  // Parse the HTML source.
  mozilla::ErrorResult rv2;
  RefPtr<mozilla::dom::DOMParser> parser =
  nsCOMPtr<nsIDocument> document = parser->ParseFromString(
    rawHTML, mozilla::dom::SupportedType::Text_html, rv2);
  if (rv2.Failed())
    return -1;

  // Serialize it back to HTML source again.
  nsCOMPtr<nsIDocumentEncoder> encoder = do_CreateInstance(
  uint32_t aFlags = 0;
  rv = encoder->Init(document, NS_LITERAL_STRING("text/html"), aFlags);
  rv = encoder->EncodeToString(parsed);

  // Write it out.
  NS_ConvertUTF16toUTF8 resultCStr(parsed);
  status = ((MimeObjectClass*)&MIME_SUPERCLASS)->parse_line(
    resultCStr.BeginWriting(), resultCStr.Length(), obj);
  return status;

MimeInlineTextHTMLParsed_finalize(MimeObject *obj)
  MimeInlineTextHTMLParsed *me = (MimeInlineTextHTMLParsed *)obj;

  if (me && me->complete_buffer)
    obj->clazz->parse_eof(obj, false);
    delete me->complete_buffer;
    me->complete_buffer = NULL;


static int
MimeInlineTextHTMLParsed_parse_line(const char *line, int32_t length,
                                    MimeObject *obj)
  MimeInlineTextHTMLParsed *me = (MimeInlineTextHTMLParsed *)obj;

  if (!me || !(me->complete_buffer))
    return -1;

  nsCString linestr(line, length);
  NS_ConvertUTF8toUTF16 line_ucs2(linestr.get());
  if (length && line_ucs2.IsEmpty())
    CopyASCIItoUTF16(linestr, line_ucs2);

  return 0;