path: root/netwerk/streamconv/converters/nsUnknownDecoder.cpp

/* -*- Mode: C++; tab-width: 2; 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 http://mozilla.org/MPL/2.0/. */

#include "nsUnknownDecoder.h"
#include "nsIPipe.h"
#include "nsIInputStream.h"
#include "nsIOutputStream.h"
#include "nsMimeTypes.h"
#include "nsIPrefBranch.h"

#include "nsCRT.h"

#include "nsIMIMEService.h"

#include "nsIViewSourceChannel.h"
#include "nsIHttpChannel.h"
#include "nsIForcePendingChannel.h"
#include "nsIEncodedChannel.h"
#include "nsIURI.h"
#include "nsStringStream.h"
#include "nsNetCID.h"
#include "nsNetUtil.h"

#include <algorithm>

#define MAX_BUFFER_SIZE 512u

NS_IMPL_ISUPPORTS(nsUnknownDecoder::ConvertedStreamListener, nsIStreamListener,
                  nsIRequestObserver)

nsUnknownDecoder::ConvertedStreamListener::ConvertedStreamListener(
    nsUnknownDecoder* aDecoder) {
  mDecoder = aDecoder;
}

nsresult nsUnknownDecoder::ConvertedStreamListener::AppendDataToString(
    nsIInputStream* inputStream, void* closure, const char* rawSegment,
    uint32_t toOffset, uint32_t count, uint32_t* writeCount) {
  nsCString* decodedData = static_cast<nsCString*>(closure);
  decodedData->Append(rawSegment, count);
  *writeCount = count;
  return NS_OK;
}

NS_IMETHODIMP
nsUnknownDecoder::ConvertedStreamListener::OnStartRequest(nsIRequest* request) {
  return NS_OK;
}

NS_IMETHODIMP
nsUnknownDecoder::ConvertedStreamListener::OnDataAvailable(
    nsIRequest* request, nsIInputStream* stream, uint64_t offset,
    uint32_t count) {
  uint32_t read;
  nsAutoCString decodedData;
  {
    MutexAutoLock lock(mDecoder->mMutex);
    decodedData = mDecoder->mDecodedData;
  }
  nsresult rv =
      stream->ReadSegments(AppendDataToString, &decodedData, count, &read);
  if (NS_FAILED(rv)) {
    return rv;
  }
  MutexAutoLock lock(mDecoder->mMutex);
  mDecoder->mDecodedData = decodedData;
  return NS_OK;
}

NS_IMETHODIMP
nsUnknownDecoder::ConvertedStreamListener::OnStopRequest(nsIRequest* request,
                                                         nsresult status) {
  return NS_OK;
}

nsUnknownDecoder::nsUnknownDecoder()
    : mBuffer(nullptr),
      mBufferLen(0),
      mRequireHTMLsuffix(false),
      mMutex("nsUnknownDecoder"),
      mDecodedData("") {
  nsCOMPtr<nsIPrefBranch> prefs = do_GetService(NS_PREFSERVICE_CONTRACTID);
  if (prefs) {
    bool val;
    if (NS_SUCCEEDED(prefs->GetBoolPref("security.requireHTMLsuffix", &val)))
      mRequireHTMLsuffix = val;
  }
}

nsUnknownDecoder::~nsUnknownDecoder() {
  if (mBuffer) {
    delete[] mBuffer;
    mBuffer = nullptr;
  }
}

// ----
//
// nsISupports implementation...
//
// ----

NS_IMPL_ADDREF(nsUnknownDecoder)
NS_IMPL_RELEASE(nsUnknownDecoder)

NS_INTERFACE_MAP_BEGIN(nsUnknownDecoder)
  NS_INTERFACE_MAP_ENTRY(nsIStreamConverter)
  NS_INTERFACE_MAP_ENTRY(nsIStreamListener)
  NS_INTERFACE_MAP_ENTRY(nsIRequestObserver)
  NS_INTERFACE_MAP_ENTRY(nsIContentSniffer)
  NS_INTERFACE_MAP_ENTRY(nsIThreadRetargetableStreamListener)
  NS_INTERFACE_MAP_ENTRY_AMBIGUOUS(nsISupports, nsIStreamListener)
NS_INTERFACE_MAP_END

// ----
//
// nsIStreamConverter methods...
//
// ----

NS_IMETHODIMP
nsUnknownDecoder::Convert(nsIInputStream* aFromStream, const char* aFromType,
                          const char* aToType, nsISupports* aCtxt,
                          nsIInputStream** aResultStream) {
  return NS_ERROR_NOT_IMPLEMENTED;
}

NS_IMETHODIMP
nsUnknownDecoder::AsyncConvertData(const char* aFromType, const char* aToType,
                                   nsIStreamListener* aListener,
                                   nsISupports* aCtxt) {
  NS_ASSERTION(aListener && aFromType && aToType,
               "null pointer passed into multi mixed converter");
  // hook up our final listener. this guy gets the various On*() calls we want
  // to throw at him.
  //

  MutexAutoLock lock(mMutex);
  mNextListener = aListener;
  return (aListener) ? NS_OK : NS_ERROR_FAILURE;
}

NS_IMETHODIMP
nsUnknownDecoder::GetConvertedType(const nsACString& aFromType,
                                   nsIChannel* aChannel, nsACString& aToType) {
  return NS_ERROR_NOT_IMPLEMENTED;
}

// ----
//
// nsIStreamListener methods...
//
// ----

NS_IMETHODIMP
nsUnknownDecoder::OnDataAvailable(nsIRequest* request, nsIInputStream* aStream,
                                  uint64_t aSourceOffset, uint32_t aCount) {
  nsresult rv = NS_OK;

  bool contentTypeEmpty;
  {
    MutexAutoLock lock(mMutex);
    if (!mNextListener) return NS_ERROR_FAILURE;

    contentTypeEmpty = mContentType.IsEmpty();
  }

  if (contentTypeEmpty) {
    uint32_t count, len;

    // If the buffer has not been allocated by now, just fail...
    if (!mBuffer) return NS_ERROR_OUT_OF_MEMORY;

    //
    // Determine how much of the stream should be read to fill up the
    // sniffer buffer...
    //
    if (mBufferLen + aCount >= MAX_BUFFER_SIZE) {
      count = MAX_BUFFER_SIZE - mBufferLen;
    } else {
      count = aCount;
    }

    // Read the data into the buffer...
    rv = aStream->Read((mBuffer + mBufferLen), count, &len);
    if (NS_FAILED(rv)) return rv;

    mBufferLen += len;
    aCount -= len;

    if (aCount) {
      //
      // Adjust the source offset...  The call to FireListenerNotifications(...)
      // will make the first OnDataAvailable(...) call with an offset of 0.
      // So, this offset needs to be adjusted to reflect that...
      //
      aSourceOffset += mBufferLen;

      DetermineContentType(request);

      rv = FireListenerNotifications(request, nullptr);
    }
  }

  // Must not fire ODA again if it failed once
  if (aCount && NS_SUCCEEDED(rv)) {
#ifdef DEBUG
    {
      MutexAutoLock lock(mMutex);
      NS_ASSERTION(!mContentType.IsEmpty(),
                   "Content type should be known by now.");
    }
#endif

    nsCOMPtr<nsIStreamListener> listener;
    {
      MutexAutoLock lock(mMutex);
      listener = mNextListener;
    }
    rv = listener->OnDataAvailable(request, aStream, aSourceOffset, aCount);
  }

  return rv;
}

// ----
//
// nsIRequestObserver methods...
//
// ----

NS_IMETHODIMP
nsUnknownDecoder::OnStartRequest(nsIRequest* request) {
  nsresult rv = NS_OK;

  {
    MutexAutoLock lock(mMutex);
    if (!mNextListener) return NS_ERROR_FAILURE;
  }

  // Allocate the sniffer buffer...
  if (NS_SUCCEEDED(rv) && !mBuffer) {
    mBuffer = new char[MAX_BUFFER_SIZE];

    if (!mBuffer) {
      rv = NS_ERROR_OUT_OF_MEMORY;
    }
  }

  // Do not pass the OnStartRequest on to the next listener (yet)...
  return rv;
}

NS_IMETHODIMP
nsUnknownDecoder::OnStopRequest(nsIRequest* request, nsresult aStatus) {
  nsresult rv = NS_OK;

  bool contentTypeEmpty;
  {
    MutexAutoLock lock(mMutex);
    if (!mNextListener) return NS_ERROR_FAILURE;

    contentTypeEmpty = mContentType.IsEmpty();
  }

  //
  // The total amount of data is less than the size of the sniffer buffer.
  // Analyze the buffer now...
  //
  if (contentTypeEmpty) {
    DetermineContentType(request);

    // Make sure channel listeners see channel as pending while we call
    // OnStartRequest/OnDataAvailable, even though the underlying channel
    // has already hit OnStopRequest.
    nsCOMPtr<nsIForcePendingChannel> forcePendingChannel =
        do_QueryInterface(request);
    if (forcePendingChannel) {
      forcePendingChannel->ForcePending(true);
    }

    rv = FireListenerNotifications(request, nullptr);

    if (NS_FAILED(rv)) {
      aStatus = rv;
    }

    // now we need to set pending state to false before calling OnStopRequest
    if (forcePendingChannel) {
      forcePendingChannel->ForcePending(false);
    }
  }

  nsCOMPtr<nsIStreamListener> listener;
  {
    MutexAutoLock lock(mMutex);
    listener = mNextListener;
    mNextListener = nullptr;
  }
  rv = listener->OnStopRequest(request, aStatus);

  return rv;
}

// ----
//
// nsIContentSniffer methods...
//
// ----
NS_IMETHODIMP
nsUnknownDecoder::GetMIMETypeFromContent(nsIRequest* aRequest,
                                         const uint8_t* aData, uint32_t aLength,
                                         nsACString& type) {
  // This is only used by sniffer, therefore we do not need to lock anything
  // here.
  nsCOMPtr<nsIChannel> channel(do_QueryInterface(aRequest));
  if (channel) {
    nsCOMPtr<nsILoadInfo> loadInfo = channel->LoadInfo();
    if (loadInfo->GetSkipContentSniffing()) {
      return NS_ERROR_NOT_AVAILABLE;
    }
  }

  mBuffer = const_cast<char*>(reinterpret_cast<const char*>(aData));
  mBufferLen = aLength;
  DetermineContentType(aRequest);
  mBuffer = nullptr;
  mBufferLen = 0;
  type.Assign(mContentType);
  mContentType.Truncate();
  return type.IsEmpty() ? NS_ERROR_NOT_AVAILABLE : NS_OK;
}

// Actual sniffing code

bool nsUnknownDecoder::AllowSniffing(nsIRequest* aRequest) {
  if (!mRequireHTMLsuffix) {
    return true;
  }

  nsCOMPtr<nsIChannel> channel = do_QueryInterface(aRequest);
  if (!channel) {
    NS_ERROR("QI failed");
    return false;
  }

  nsCOMPtr<nsIURI> uri;
  if (NS_FAILED(channel->GetURI(getter_AddRefs(uri))) || !uri) {
    return false;
  }

  nsCOMPtr<nsILoadInfo> loadInfo = channel->LoadInfo();
  if (loadInfo->GetSkipContentSniffing()) {
    return false;
  }

  return !uri->SchemeIs("file");
}

/**
 * This is the array of sniffer entries that depend on "magic numbers"
 * in the file.  Each entry has either a type associated with it (set
 * these with the SNIFFER_ENTRY macro) or a function to be executed
 * (set these with the SNIFFER_ENTRY_WITH_FUNC macro).  The function
 * should take a single nsIRequest* and returns bool -- true if
 * it sets mContentType, false otherwise
 */
nsUnknownDecoder::nsSnifferEntry nsUnknownDecoder::sSnifferEntries[] = {
    SNIFFER_ENTRY("%PDF-", APPLICATION_PDF),

    SNIFFER_ENTRY("%!PS-Adobe-", APPLICATION_POSTSCRIPT),

    // Files that start with mailbox delimiters let's provisionally call
    // text/plain
    SNIFFER_ENTRY("From", TEXT_PLAIN), SNIFFER_ENTRY(">From", TEXT_PLAIN),

    // If the buffer begins with "#!" or "%!" then it is a script of
    // some sort...  "Scripts" can include arbitrary data to be passed
    // to an interpreter, so we need to decide whether we can call this
    // text or whether it's data.
    SNIFFER_ENTRY_WITH_FUNC("#!", &nsUnknownDecoder::LastDitchSniff),

    // XXXbz should (and can) we also include the various ways that <?xml can
    // appear as UTF-16 and such?  See http://www.w3.org/TR/REC-xml#sec-guessing
    SNIFFER_ENTRY_WITH_FUNC("<?xml", &nsUnknownDecoder::SniffForXML)};

uint32_t nsUnknownDecoder::sSnifferEntryNum =
    sizeof(nsUnknownDecoder::sSnifferEntries) /
    sizeof(nsUnknownDecoder::nsSnifferEntry);

void nsUnknownDecoder::DetermineContentType(nsIRequest* aRequest) {
  {
    MutexAutoLock lock(mMutex);
    NS_ASSERTION(mContentType.IsEmpty(), "Content type is already known.");
    if (!mContentType.IsEmpty()) return;
  }

  nsCOMPtr<nsIChannel> channel(do_QueryInterface(aRequest));
  if (channel) {
    nsCOMPtr<nsILoadInfo> loadInfo = channel->LoadInfo();
    if (loadInfo->GetSkipContentSniffing()) {
      /*
       * If we did not get a useful Content-Type from the server
       * but also have sniffing disabled, just determine whether
       * to use text/plain or octetstream and log an error to the Console
       */
      LastDitchSniff(aRequest);

      nsCOMPtr<nsIHttpChannel> httpChannel(do_QueryInterface(aRequest));
      if (httpChannel) {
        nsAutoCString type;
        httpChannel->GetContentType(type);
        nsCOMPtr<nsIURI> requestUri;
        httpChannel->GetURI(getter_AddRefs(requestUri));
        nsAutoCString spec;
        requestUri->GetSpec(spec);
        if (spec.Length() > 50) {
          spec.Truncate(50);
          spec.AppendLiteral("...");
        }
        httpChannel->LogMimeTypeMismatch(
            "XTCOWithMIMEValueMissing"_ns, false, NS_ConvertUTF8toUTF16(spec),
            // Type is not used in the Error Message but required
            NS_ConvertUTF8toUTF16(type));
      }
      return;
    }
  }

  const char* testData = mBuffer;
  uint32_t testDataLen = mBufferLen;
  // Check if data are compressed.
  nsAutoCString decodedData;

  if (channel) {
    // ConvertEncodedData is always called only on a single thread for each
    // instance of an object.
    nsresult rv = ConvertEncodedData(aRequest, mBuffer, mBufferLen);
    if (NS_SUCCEEDED(rv)) {
      MutexAutoLock lock(mMutex);
      decodedData = mDecodedData;
    }
    if (!decodedData.IsEmpty()) {
      testData = decodedData.get();
      testDataLen = std::min(decodedData.Length(), MAX_BUFFER_SIZE);
    }
  }

  // First, run through all the types we can detect reliably based on
  // magic numbers
  uint32_t i;
  for (i = 0; i < sSnifferEntryNum; ++i) {
    if (testDataLen >= sSnifferEntries[i].mByteLen &&  // enough data
        memcmp(testData, sSnifferEntries[i].mBytes,
               sSnifferEntries[i].mByteLen) == 0) {  // and type matches
      NS_ASSERTION(
          sSnifferEntries[i].mMimeType ||
              sSnifferEntries[i].mContentTypeSniffer,
          "Must have either a type string or a function to set the type");
      NS_ASSERTION(!sSnifferEntries[i].mMimeType ||
                       !sSnifferEntries[i].mContentTypeSniffer,
                   "Both a type string and a type sniffing function set;"
                   " using type string");
      if (sSnifferEntries[i].mMimeType) {
        MutexAutoLock lock(mMutex);
        mContentType = sSnifferEntries[i].mMimeType;
        NS_ASSERTION(!mContentType.IsEmpty(),
                     "Content type should be known by now.");
        return;
      }
      if ((this->*(sSnifferEntries[i].mContentTypeSniffer))(aRequest)) {
#ifdef DEBUG
        MutexAutoLock lock(mMutex);
        NS_ASSERTION(!mContentType.IsEmpty(),
                     "Content type should be known by now.");
#endif
        return;
      }
    }
  }

  nsAutoCString sniffedType;
  NS_SniffContent(NS_DATA_SNIFFER_CATEGORY, aRequest, (const uint8_t*)testData,
                  testDataLen, sniffedType);
  {
    MutexAutoLock lock(mMutex);
    mContentType = sniffedType;
    if (!mContentType.IsEmpty()) {
      return;
    }
  }

  if (SniffForHTML(aRequest)) {
#ifdef DEBUG
    MutexAutoLock lock(mMutex);
    NS_ASSERTION(!mContentType.IsEmpty(),
                 "Content type should be known by now.");
#endif
    return;
  }

  // We don't know what this is yet.  Before we just give up, try
  // the URI from the request.
  if (SniffURI(aRequest)) {
#ifdef DEBUG
    MutexAutoLock lock(mMutex);
    NS_ASSERTION(!mContentType.IsEmpty(),
                 "Content type should be known by now.");
#endif
    return;
  }

  LastDitchSniff(aRequest);
#ifdef DEBUG
  MutexAutoLock lock(mMutex);
  NS_ASSERTION(!mContentType.IsEmpty(), "Content type should be known by now.");
#endif
}

bool nsUnknownDecoder::SniffForHTML(nsIRequest* aRequest) {
  /*
   * To prevent a possible attack, we will not consider this to be
   * html content if it comes from the local file system and our prefs
   * are set right
   */
  if (!AllowSniffing(aRequest)) {
    return false;
  }

  MutexAutoLock lock(mMutex);

  // Now look for HTML.
  const char* str;
  const char* end;
  if (mDecodedData.IsEmpty()) {
    str = mBuffer;
    end = mBuffer + mBufferLen;
  } else {
    str = mDecodedData.get();
    end = mDecodedData.get() + std::min(mDecodedData.Length(), MAX_BUFFER_SIZE);
  }

  // skip leading whitespace
  while (str != end && nsCRT::IsAsciiSpace(*str)) {
    ++str;
  }

  // did we find something like a start tag?
  if (str == end || *str != '<' || ++str == end) {
    return false;
  }

  // If we seem to be SGML or XML and we got down here, just pretend we're HTML
  if (*str == '!' || *str == '?') {
    mContentType = TEXT_HTML;
    return true;
  }

  uint32_t bufSize = end - str;
  // We use sizeof(_tagstr) below because that's the length of _tagstr
  // with the one char " " or ">" appended.
#define MATCHES_TAG(_tagstr)                                  \
  (bufSize >= sizeof(_tagstr) &&                              \
   (PL_strncasecmp(str, _tagstr " ", sizeof(_tagstr)) == 0 || \
    PL_strncasecmp(str, _tagstr ">", sizeof(_tagstr)) == 0))

  if (MATCHES_TAG("html") || MATCHES_TAG("frameset") || MATCHES_TAG("body") ||
      MATCHES_TAG("head") || MATCHES_TAG("script") || MATCHES_TAG("iframe") ||
      MATCHES_TAG("a") || MATCHES_TAG("img") || MATCHES_TAG("table") ||
      MATCHES_TAG("title") || MATCHES_TAG("link") || MATCHES_TAG("base") ||
      MATCHES_TAG("style") || MATCHES_TAG("div") || MATCHES_TAG("p") ||
      MATCHES_TAG("font") || MATCHES_TAG("applet") || MATCHES_TAG("meta") ||
      MATCHES_TAG("center") || MATCHES_TAG("form") || MATCHES_TAG("isindex") ||
      MATCHES_TAG("h1") || MATCHES_TAG("h2") || MATCHES_TAG("h3") ||
      MATCHES_TAG("h4") || MATCHES_TAG("h5") || MATCHES_TAG("h6") ||
      MATCHES_TAG("b") || MATCHES_TAG("pre")) {
    mContentType = TEXT_HTML;
    return true;
  }

#undef MATCHES_TAG

  return false;
}

bool nsUnknownDecoder::SniffForXML(nsIRequest* aRequest) {
  // Just like HTML, this should be able to be shut off.
  if (!AllowSniffing(aRequest)) {
    return false;
  }

  // First see whether we can glean anything from the uri...
  if (!SniffURI(aRequest)) {
    // Oh well; just generic XML will have to do
    MutexAutoLock lock(mMutex);
    mContentType = TEXT_XML;
  }

  return true;
}

bool nsUnknownDecoder::SniffURI(nsIRequest* aRequest) {
  nsCOMPtr<nsIChannel> channel(do_QueryInterface(aRequest));
  nsCOMPtr<nsILoadInfo> loadInfo = channel->LoadInfo();
  if (loadInfo->GetSkipContentSniffing()) {
    return false;
  }
  nsCOMPtr<nsIMIMEService> mimeService(do_GetService("@mozilla.org/mime;1"));
  if (mimeService) {
    nsCOMPtr<nsIChannel> channel = do_QueryInterface(aRequest);
    if (channel) {
      nsCOMPtr<nsIURI> uri;
      nsresult result = channel->GetURI(getter_AddRefs(uri));
      if (NS_SUCCEEDED(result) && uri) {
        nsAutoCString type;
        result = mimeService->GetTypeFromURI(uri, type);
        if (NS_SUCCEEDED(result)) {
          MutexAutoLock lock(mMutex);
          mContentType = type;
          return true;
        }
      }
    }
  }

  return false;
}

// This macro is based on RFC 2046 Section 4.1.2.  Treat any char 0-31
// except the 9-13 range (\t, \n, \v, \f, \r) and char 27 (used by
// encodings like Shift_JIS) as non-text
#define IS_TEXT_CHAR(ch) \
  (((unsigned char)(ch)) > 31 || (9 <= (ch) && (ch) <= 13) || (ch) == 27)

bool nsUnknownDecoder::LastDitchSniff(nsIRequest* aRequest) {
  // All we can do now is try to guess whether this is text/plain or
  // application/octet-stream

  MutexAutoLock lock(mMutex);

  const char* testData;
  uint32_t testDataLen;
  if (mDecodedData.IsEmpty()) {
    testData = mBuffer;
    // Since some legacy text files end with 0x1A, reading the entire buffer
    // will lead misdetection.
    testDataLen = std::min<uint32_t>(mBufferLen, MAX_BUFFER_SIZE);
  } else {
    testData = mDecodedData.get();
    testDataLen = std::min(mDecodedData.Length(), MAX_BUFFER_SIZE);
  }

  // First, check for a BOM.  If we see one, assume this is text/plain
  // in whatever encoding.  If there is a BOM _and_ text we will
  // always have at least 4 bytes in the buffer (since the 2-byte BOMs
  // are for 2-byte encodings and the UTF-8 BOM is 3 bytes).
  if (testDataLen >= 4) {
    const unsigned char* buf = (const unsigned char*)testData;
    if ((buf[0] == 0xFE && buf[1] == 0xFF) ||  // UTF-16, Big Endian
        (buf[0] == 0xFF && buf[1] == 0xFE) ||  // UTF-16 or UCS-4, Little Endian
        (buf[0] == 0xEF && buf[1] == 0xBB && buf[2] == 0xBF) ||  // UTF-8
        (buf[0] == 0 && buf[1] == 0 && buf[2] == 0xFE &&
         buf[3] == 0xFF)) {  // UCS-4, Big Endian

      mContentType = TEXT_PLAIN;
      return true;
    }
  }

  // Now see whether the buffer has any non-text chars.  If not, then let's
  // just call it text/plain...
  //
  uint32_t i;
  for (i = 0; i < testDataLen && IS_TEXT_CHAR(testData[i]); i++) {
  }

  if (i == testDataLen) {
    mContentType = TEXT_PLAIN;
  } else {
    mContentType = APPLICATION_OCTET_STREAM;
  }

  return true;
}

nsresult nsUnknownDecoder::FireListenerNotifications(nsIRequest* request,
                                                     nsISupports* aCtxt) {
  nsresult rv = NS_OK;

  nsCOMPtr<nsIStreamListener> listener;
  nsAutoCString contentType;
  {
    MutexAutoLock lock(mMutex);
    if (!mNextListener) return NS_ERROR_FAILURE;

    listener = mNextListener;
    contentType = mContentType;
  }

  if (!contentType.IsEmpty()) {
    nsCOMPtr<nsIViewSourceChannel> viewSourceChannel =
        do_QueryInterface(request);
    if (viewSourceChannel) {
      rv = viewSourceChannel->SetOriginalContentType(contentType);
    } else {
      nsCOMPtr<nsIChannel> channel = do_QueryInterface(request, &rv);
      if (NS_SUCCEEDED(rv)) {
        // Set the new content type on the channel...
        rv = channel->SetContentType(contentType);
      }
    }

    NS_ASSERTION(NS_SUCCEEDED(rv), "Unable to set content type on channel!");

    if (NS_FAILED(rv)) {
      // Cancel the request to make sure it has the correct status if
      // mNextListener looks at it.
      request->Cancel(rv);
      listener->OnStartRequest(request);
      return rv;
    }
  }

  // Fire the OnStartRequest(...)
  rv = listener->OnStartRequest(request);

  if (NS_SUCCEEDED(rv)) {
    // install stream converter if required
    nsCOMPtr<nsIEncodedChannel> encodedChannel = do_QueryInterface(request);
    if (encodedChannel) {
      nsCOMPtr<nsIStreamListener> listenerNew;
      rv = encodedChannel->DoApplyContentConversions(
          listener, getter_AddRefs(listenerNew), aCtxt);
      if (NS_SUCCEEDED(rv) && listenerNew) {
        MutexAutoLock lock(mMutex);
        mNextListener = listenerNew;
        listener = listenerNew;
      }
    }
  }

  if (!mBuffer) return NS_ERROR_OUT_OF_MEMORY;

  // If the request was canceled, then we need to treat that equivalently
  // to an error returned by OnStartRequest.
  if (NS_SUCCEEDED(rv)) request->GetStatus(&rv);

  // Fire the first OnDataAvailable for the data that was read from the
  // stream into the sniffer buffer...
  if (NS_SUCCEEDED(rv) && (mBufferLen > 0)) {
    uint32_t len = 0;
    nsCOMPtr<nsIInputStream> in;
    nsCOMPtr<nsIOutputStream> out;

    // Create a pipe and fill it with the data from the sniffer buffer.
    rv = NS_NewPipe(getter_AddRefs(in), getter_AddRefs(out), MAX_BUFFER_SIZE,
                    MAX_BUFFER_SIZE);

    if (NS_SUCCEEDED(rv)) {
      rv = out->Write(mBuffer, mBufferLen, &len);
      if (NS_SUCCEEDED(rv)) {
        if (len == mBufferLen) {
          rv = listener->OnDataAvailable(request, in, 0, len);
        } else {
          NS_ERROR("Unable to write all the data into the pipe.");
          rv = NS_ERROR_FAILURE;
        }
      }
    }
  }

  delete[] mBuffer;
  mBuffer = nullptr;
  mBufferLen = 0;

  return rv;
}

nsresult nsUnknownDecoder::ConvertEncodedData(nsIRequest* request,
                                              const char* data,
                                              uint32_t length) {
  nsresult rv = NS_OK;

  {
    MutexAutoLock lock(mMutex);
    mDecodedData = "";
  }
  nsCOMPtr<nsIEncodedChannel> encodedChannel(do_QueryInterface(request));
  if (encodedChannel) {
    RefPtr<ConvertedStreamListener> strListener =
        new ConvertedStreamListener(this);

    nsCOMPtr<nsIStreamListener> listener;
    rv = encodedChannel->DoApplyContentConversions(
        strListener, getter_AddRefs(listener), nullptr);

    if (NS_FAILED(rv)) {
      return rv;
    }

    if (listener) {
      listener->OnStartRequest(request);

      if (length) {
        nsCOMPtr<nsIStringInputStream> rawStream =
            do_CreateInstance(NS_STRINGINPUTSTREAM_CONTRACTID);
        if (!rawStream) return NS_ERROR_FAILURE;

        rv = rawStream->SetData((const char*)data, length);
        NS_ENSURE_SUCCESS(rv, rv);

        rv = listener->OnDataAvailable(request, rawStream, 0, length);
        NS_ENSURE_SUCCESS(rv, rv);
      }

      listener->OnStopRequest(request, NS_OK);
    }
  }
  return rv;
}

//
// nsIThreadRetargetableStreamListener methods
//
NS_IMETHODIMP
nsUnknownDecoder::CheckListenerChain() {
  nsCOMPtr<nsIThreadRetargetableStreamListener> listener;
  {
    MutexAutoLock lock(mMutex);
    listener = do_QueryInterface(mNextListener);
  }
  if (!listener) {
    return NS_ERROR_NO_INTERFACE;
  }

  return listener->CheckListenerChain();
}

void nsBinaryDetector::DetermineContentType(nsIRequest* aRequest) {
  nsCOMPtr<nsIHttpChannel> httpChannel = do_QueryInterface(aRequest);
  if (!httpChannel) {
    return;
  }

  nsCOMPtr<nsILoadInfo> loadInfo = httpChannel->LoadInfo();
  if (loadInfo->GetSkipContentSniffing()) {
    LastDitchSniff(aRequest);
    return;
  }
  // It's an HTTP channel.  Check for the text/plain mess
  nsAutoCString contentTypeHdr;
  Unused << httpChannel->GetResponseHeader("Content-Type"_ns, contentTypeHdr);
  nsAutoCString contentType;
  httpChannel->GetContentType(contentType);

  // Make sure to do a case-sensitive exact match comparison here.  Apache
  // 1.x just sends text/plain for "unknown", while Apache 2.x sends
  // text/plain with a ISO-8859-1 charset.  Debian's Apache version, just to
  // be different, sends text/plain with iso-8859-1 charset.  For extra fun,
  // FC7, RHEL4, and Ubuntu Feisty send charset=UTF-8.  Don't do general
  // case-insensitive comparison, since we really want to apply this crap as
  // rarely as we can.
  if (!contentType.EqualsLiteral("text/plain") ||
      (!contentTypeHdr.EqualsLiteral("text/plain") &&
       !contentTypeHdr.EqualsLiteral("text/plain; charset=ISO-8859-1") &&
       !contentTypeHdr.EqualsLiteral("text/plain; charset=iso-8859-1") &&
       !contentTypeHdr.EqualsLiteral("text/plain; charset=UTF-8"))) {
    return;
  }

  // Check whether we have content-encoding.  If we do, don't try to
  // detect the type.
  // XXXbz we could improve this by doing a local decompress if we
  // wanted, I'm sure.
  nsAutoCString contentEncoding;
  Unused << httpChannel->GetResponseHeader("Content-Encoding"_ns,
                                           contentEncoding);
  if (!contentEncoding.IsEmpty()) {
    return;
  }

  LastDitchSniff(aRequest);
  MutexAutoLock lock(mMutex);
  if (mContentType.EqualsLiteral(APPLICATION_OCTET_STREAM)) {
    // We want to guess at it instead
    mContentType = APPLICATION_GUESS_FROM_EXT;
  } else {
    // Let the text/plain type we already have be, so that other content
    // sniffers can also get a shot at this data.
    mContentType.Truncate();
  }
}