Add SafeBrowsing support for checking downloaded zip files that contain executables.

chrome/browser/safe_browsing/download_protection_service.cc

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "chrome/browser/safe_browsing/download_protection_service.h"
 
 #include "base/bind.h"
 #include "base/compiler_specific.h"
 #include "base/format_macros.h"
 #include "base/memory/scoped_ptr.h"
 #include "base/memory/weak_ptr.h"
 #include "base/message_loop_helpers.h"
 #include "base/metrics/histogram.h"
 #include "base/stl_util.h"
 #include "base/string_number_conversions.h"
 #include "base/string_util.h"
 #include "base/stringprintf.h"
 #include "base/threading/sequenced_worker_pool.h"
 #include "base/time.h"
 #include "chrome/browser/safe_browsing/safe_browsing_service.h"
 #include "chrome/browser/safe_browsing/signature_util.h"
 #include "chrome/browser/ui/browser_list.h"
 #include "chrome/common/safe_browsing/csd.pb.h"
 #include "chrome/common/url_constants.h"
+#include "chrome/common/zip_reader.h"
 #include "content/public/browser/browser_thread.h"
 #include "content/public/browser/download_item.h"
 #include "content/public/browser/page_navigator.h"
 #include "content/public/common/url_fetcher.h"
 #include "content/public/common/url_fetcher_delegate.h"
 #include "net/base/load_flags.h"
 #include "net/base/x509_cert_types.h"
 #include "net/base/x509_certificate.h"
 #include "net/http/http_status_code.h"
 #include "net/url_request/url_request_context_getter.h"
 #include "net/url_request/url_request_status.h"
 
 using content::BrowserThread;
 
 namespace {
 static const int64 kDownloadRequestTimeoutMs = 3000;
 }  // namespace
 
 namespace safe_browsing {
 
 const char DownloadProtectionService::kDownloadRequestUrl[] =
     "https://sb-ssl.google.com/safebrowsing/clientreport/download";
 
 namespace {
+bool IsArchiveFile(const FilePath& file) {
+  return file.MatchesExtension(FILE_PATH_LITERAL(".zip"));
+}
+
 bool IsBinaryFile(const FilePath& file) {
   return (file.MatchesExtension(FILE_PATH_LITERAL(".exe")) ||
           file.MatchesExtension(FILE_PATH_LITERAL(".cab")) ||
           file.MatchesExtension(FILE_PATH_LITERAL(".msi")) ||
           file.MatchesExtension(FILE_PATH_LITERAL(".crx")) ||
-          file.MatchesExtension(FILE_PATH_LITERAL(".apk")));
+          file.MatchesExtension(FILE_PATH_LITERAL(".apk")) ||
+          // Archives _may_ contain binaries, we'll check in
+          // ExtractFileFeatures.
+          IsArchiveFile(file));
 }
 
 ClientDownloadRequest::DownloadType GetDownloadType(const FilePath& file) {
   DCHECK(IsBinaryFile(file));
   if (file.MatchesExtension(FILE_PATH_LITERAL(".apk")))
     return ClientDownloadRequest::ANDROID_APK;
   else if (file.MatchesExtension(FILE_PATH_LITERAL(".crx")))
     return ClientDownloadRequest::CHROME_EXTENSION;
+  // For zip files, we use the ZIPPED_EXECUTABLE type since we will only send
+  // the pingback if we find an executable inside the zip archive.
+  else if (file.MatchesExtension(FILE_PATH_LITERAL(".zip")))
+    return ClientDownloadRequest::ZIPPED_EXECUTABLE;
   return ClientDownloadRequest::WIN_EXECUTABLE;
 }
 
 // List of extensions for which we track some UMA stats.
 enum MaliciousExtensionType {
   EXTENSION_EXE,
   EXTENSION_MSI,
   EXTENSION_CAB,
   EXTENSION_SYS,
   EXTENSION_SCR,
@@ skipping 57 matching lines @@
   DOWNLOAD_HASH_CHECKS_TOTAL,
   DOWNLOAD_HASH_CHECKS_MALWARE,
 
   // Memory space for histograms is determined by the max.
   // ALWAYS ADD NEW VALUES BEFORE THIS ONE.
   DOWNLOAD_CHECKS_MAX
 };
 }  // namespace
 
 DownloadProtectionService::DownloadInfo::DownloadInfo()
-    : total_bytes(0), user_initiated(false) {}
+    : total_bytes(0), user_initiated(false), zipped_executable(false) {}
 
 DownloadProtectionService::DownloadInfo::~DownloadInfo() {}
 
 std::string DownloadProtectionService::DownloadInfo::DebugString() const {
   std::string chain;
   for (size_t i = 0; i < download_url_chain.size(); ++i) {
     chain += download_url_chain[i].spec();
     if (i < download_url_chain.size() - 1) {
       chain += " -> ";
     }
@@ skipping 317 matching lines @@
 
  private:
   friend struct BrowserThread::DeleteOnThread<BrowserThread::UI>;
   friend class base::DeleteHelper<CheckClientDownloadRequest>;
 
   virtual ~CheckClientDownloadRequest() {
     DCHECK(BrowserThread::CurrentlyOn(BrowserThread::UI));
   }
 
   void ExtractFileFeatures() {
+    // If we're checking an archive file, look to see if there are any
+    // executables inside.  If not, we will skip the pingback for this
+    // download.
+    if (info_.target_file.MatchesExtension(FILE_PATH_LITERAL(".zip"))) {
+      ExtractZipFeatures();
+      if (!info_.zipped_executable) {
+        RecordImprovedProtectionStats(REASON_ARCHIVE_WITHOUT_BINARIES);
+        PostFinishTask(SAFE);
+        return;
+      }
+    } else {
+      DCHECK(!IsArchiveFile(info_.target_file));
+      ExtractSignatureFeatures();
+    }
+
+    // TODO(noelutz): DownloadInfo should also contain the IP address of
+    // every URL in the redirect chain.  We also should check whether the
+    // download URL is hosted on the internal network.
+    BrowserThread::PostTask(
+        BrowserThread::IO,
+        FROM_HERE,
+        base::Bind(&CheckClientDownloadRequest::CheckWhitelists, this));
+  }
+
+  void ExtractSignatureFeatures() {
+    base::TimeTicks start_time = base::TimeTicks::Now();
     signature_util_->CheckSignature(info_.local_file, &signature_info_);
     bool is_signed = (signature_info_.certificate_chain_size() > 0);
     if (is_signed) {
       VLOG(2) << "Downloaded a signed binary: " << info_.local_file.value();
     } else {
       VLOG(2) << "Downloaded an unsigned binary: " << info_.local_file.value();
     }
     UMA_HISTOGRAM_BOOLEAN("SBClientDownload.SignedBinaryDownload", is_signed);
+    UMA_HISTOGRAM_TIMES("SBClientDownload.ExtractSignatureFeaturesTime",
+                        base::TimeTicks::Now() - start_time);
+  }
 
-    // TODO(noelutz): DownloadInfo should also contain the IP address of every
-    // URL in the redirect chain.  We also should check whether the download
-    // URL is hosted on the internal network.
-    BrowserThread::PostTask(
-        BrowserThread::IO,
-        FROM_HERE,
-        base::Bind(&CheckClientDownloadRequest::CheckWhitelists, this));
+  void ExtractZipFeatures() {
+    base::TimeTicks start_time = base::TimeTicks::Now();
+    zip::ZipReader reader;
+    if (reader.Open(info_.local_file)) {
+      for (; reader.HasMore(); reader.AdvanceToNextEntry()) {
+        if (!reader.OpenCurrentEntryInZip()) {
+          VLOG(1) << "Failed to open current entry in zip file: "
+                  << info_.local_file.value();
+          continue;
+        }
+        const FilePath& file = reader.current_entry_info()->file_path();
+        // Don't consider an archived archive to be executable.

[mattm, 2012/05/12 00:47:06]

maybe add a histogram for archived archives too?

[Brian Ryner, 2012/05/15 00:58:34]

Done.

+        if (IsBinaryFile(file) && !IsArchiveFile(file)) {
+          VLOG(2) << "Downloaded a zipped executable: "
+                  << info_.local_file.value();
+          info_.zipped_executable = true;
+          break;
+        } else {
+          VLOG(3) << "Ignoring non-binary file: " << file.value();
+        }
+      }
+    } else {
+      VLOG(1) << "Failed to open zip file: " << info_.local_file.value();
+    }
+    UMA_HISTOGRAM_BOOLEAN("SBClientDownload.ZipFileHasExecutable",
+                          info_.zipped_executable);
+    UMA_HISTOGRAM_TIMES("SBClientDownload.ExtractZipFeaturesTime",
+                        base::TimeTicks::Now() - start_time);
   }
 
   void CheckWhitelists() {
     DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
     DownloadCheckResultReason reason = REASON_MAX;
     if (!sb_service_.get()) {
       reason = REASON_SB_DISABLED;
     } else {
       for (size_t i = 0; i < info_.download_url_chain.size(); ++i) {
         const GURL& url = info_.download_url_chain[i];
@@ skipping 239 matching lines @@
   // Currently, the UI only works on Windows.  On Linux and Mac we still
   // want to show the dangerous file type warning if the file is possibly
   // dangerous which means we have to always return false here.
 #if defined(OS_WIN)
   DownloadCheckResultReason reason = REASON_MAX;
   ClientDownloadRequest::DownloadType type =
       ClientDownloadRequest::WIN_EXECUTABLE;
   return (CheckClientDownloadRequest::IsSupportedDownload(info,
                                                           &reason,
                                                           &type) &&
-          ClientDownloadRequest::WIN_EXECUTABLE == type);
+          ClientDownloadRequest::WIN_EXECUTABLE == type ||
+          ClientDownloadRequest::ZIPPED_EXECUTABLE == type);
 #else
   return false;
 #endif
 }
 
 void DownloadProtectionService::CancelPendingRequests() {
   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::UI));
   for (std::set<scoped_refptr<CheckClientDownloadRequest> >::iterator it =
            download_requests_.begin();
        it != download_requests_.end();) {
@@ skipping 93 matching lines @@
 
   std::string issuer_fp = base::HexEncode(issuer.fingerprint().data,
                                           sizeof(issuer.fingerprint().data));
   for (std::set<std::string>::iterator it = paths_to_check.begin();
        it != paths_to_check.end(); ++it) {
     whitelist_strings->push_back("cert/" + issuer_fp + *it);
   }
 }
 
 }  // namespace safe_browsing