[SafeBrowsing] Consolidate extension lookup logic and extension lists.

chrome/common/safe_browsing/download_protection_util.cc

Index: chrome/common/safe_browsing/download_protection_util.cc
diff --git a/chrome/common/safe_browsing/download_protection_util.cc b/chrome/common/safe_browsing/download_protection_util.cc
index 7fccf6fbed9eba05ea8a5719955e303df1250ae2..fd56472a0db63695637c0e11b00f023f52fb84cf 100644
--- a/chrome/common/safe_browsing/download_protection_util.cc
+++ b/chrome/common/safe_browsing/download_protection_util.cc
@@ -4,102 +4,203 @@
 
 #include "chrome/common/safe_browsing/download_protection_util.h"
 
+#include <algorithm>
+
 #include "base/files/file_path.h"
 #include "base/logging.h"
 
 namespace safe_browsing {
 namespace download_protection_util {
 
-bool IsArchiveFile(const base::FilePath& file) {
-  // List of interesting archive file formats. These are by no means exhaustive,
-  // but are currently file types that Safe Browsing would like to see pings for
-  // due to the possibility of them being used as wrapper formats for malicious
-  // payloads.
-  const base::FilePath::CharType* kArchiveFileTypes[] = {
-    FILE_PATH_LITERAL(".zip"),
-    FILE_PATH_LITERAL(".rar"),
-    FILE_PATH_LITERAL(".7z"),
-    FILE_PATH_LITERAL(".cab"),
-    FILE_PATH_LITERAL(".xz"),
-    FILE_PATH_LITERAL(".gz"),
-    FILE_PATH_LITERAL(".tgz"),
-    FILE_PATH_LITERAL(".bz2"),
-    FILE_PATH_LITERAL(".tar"),
-    FILE_PATH_LITERAL(".arj"),
-    FILE_PATH_LITERAL(".lzh"),
-    FILE_PATH_LITERAL(".lha"),
-    FILE_PATH_LITERAL(".wim"),
-    FILE_PATH_LITERAL(".z"),
-    FILE_PATH_LITERAL(".lzma"),
-    FILE_PATH_LITERAL(".cpio"),
+namespace {
+
+// This enum matches the SBClientDownloadExtensions enum in histograms.xml. If
+// you are adding a value here, you should also add to the enum definition in
+// histograms.xml. Additions only at the end just in front of EXTENSION_MAX,
+// natch.
+enum SBClientDownloadExtensions {
+  EXTENSION_EXE,
+  EXTENSION_MSI,
+  EXTENSION_CAB,
+  EXTENSION_SYS,
+  EXTENSION_SCR,
+  EXTENSION_DRV,
+  EXTENSION_BAT,
+  EXTENSION_ZIP,
+  EXTENSION_RAR,
+  EXTENSION_DLL,
+  EXTENSION_PIF,
+  EXTENSION_COM,
+  EXTENSION_JAR,
+  EXTENSION_CLASS,
+  EXTENSION_PDF,
+  EXTENSION_VB,
+  EXTENSION_REG,
+  EXTENSION_GRP,
+  EXTENSION_OTHER,  // The "other" bucket. This is in the middle of the enum
+                    // due to historical reasons.
+  EXTENSION_CRX,
+  EXTENSION_APK,
+  EXTENSION_DMG,
+  EXTENSION_PKG,
+  EXTENSION_TORRENT,
+  EXTENSION_WEBSITE,
+  EXTENSION_URL,
+  EXTENSION_VBE,
+  EXTENSION_VBS,
+  EXTENSION_JS,
+  EXTENSION_JSE,
+  EXTENSION_MHT,
+  EXTENSION_MHTML,
+  EXTENSION_MSC,
+  EXTENSION_MSP,
+  EXTENSION_MST,
+  EXTENSION_BAS,
+  EXTENSION_HTA,
+  EXTENSION_MSH,
+  EXTENSION_MSH1,
+  EXTENSION_MSH1XML,
+  EXTENSION_MSH2,
+  EXTENSION_MSH2XML,
+  EXTENSION_MSHXML,
+  EXTENSION_PS1,
+  EXTENSION_PS1XML,
+  EXTENSION_PS2,
+  EXTENSION_PS2XML,
+  EXTENSION_PSC1,
+  EXTENSION_PSC2,
+  EXTENSION_SCF,
+  EXTENSION_SCT,
+  EXTENSION_WSF,
+  EXTENSION_7Z,
+  EXTENSION_XZ,
+  EXTENSION_GZ,
+  EXTENSION_TGZ,
+  EXTENSION_BZ2,
+  EXTENSION_TAR,
+  EXTENSION_ARJ,
+  EXTENSION_LZH,
+  EXTENSION_LHA,
+  EXTENSION_WIM,
+  EXTENSION_Z,
+  EXTENSION_LZMA,
+  EXTENSION_CPIO,
+  EXTENSION_CMD,
+  EXTENSION_APP,
+  EXTENSION_OSX,
+
+  // New values go above this one.
+  EXTENSION_MAX
+};
+
+struct SafeBrowsingFiletype {
+  const base::FilePath::CharType* const extension;
+  int uma_value;
+  bool is_supported_binary;
+  bool is_archive;
+};
+
+const SafeBrowsingFiletype kSafeBrowsingFileTypes[] = {
+    // KEEP THIS LIST SORTED!
+    {FILE_PATH_LITERAL(".7z"), EXTENSION_7Z, false, true},
+    {FILE_PATH_LITERAL(".apk"), EXTENSION_APK, true, false},
+    {FILE_PATH_LITERAL(".app"), EXTENSION_APP, true, false},
+    {FILE_PATH_LITERAL(".arj"), EXTENSION_ARJ, false, true},
+    {FILE_PATH_LITERAL(".bas"), EXTENSION_BAS, true, false},
+    {FILE_PATH_LITERAL(".bat"), EXTENSION_BAT, true, false},
+    {FILE_PATH_LITERAL(".bz2"), EXTENSION_BZ2, false, true},
+    {FILE_PATH_LITERAL(".cab"), EXTENSION_CAB, false, true},
+    {FILE_PATH_LITERAL(".cmd"), EXTENSION_CMD, true, false},
+    {FILE_PATH_LITERAL(".com"), EXTENSION_COM, true, false},
+    {FILE_PATH_LITERAL(".cpio"), EXTENSION_CPIO, false, true},
+    {FILE_PATH_LITERAL(".crx"), EXTENSION_CRX, true, false},
+    {FILE_PATH_LITERAL(".dll"), EXTENSION_DLL, true, false},
+    {FILE_PATH_LITERAL(".dmg"), EXTENSION_DMG, true, false},
+    {FILE_PATH_LITERAL(".exe"), EXTENSION_EXE, true, false},
+    {FILE_PATH_LITERAL(".gz"), EXTENSION_GZ, false, true},
+    {FILE_PATH_LITERAL(".hta"), EXTENSION_HTA, true, false},
+    {FILE_PATH_LITERAL(".js"), EXTENSION_JS, true, false},
+    {FILE_PATH_LITERAL(".jse"), EXTENSION_JSE, true, false},
+    {FILE_PATH_LITERAL(".lha"), EXTENSION_LHA, false, true},
+    {FILE_PATH_LITERAL(".lzh"), EXTENSION_LZH, false, true},
+    {FILE_PATH_LITERAL(".lzma"), EXTENSION_LZMA, false, true},
+    {FILE_PATH_LITERAL(".mht"), EXTENSION_MHT, true, false},
+    {FILE_PATH_LITERAL(".mhtml"), EXTENSION_MHTML, true, false},
+    {FILE_PATH_LITERAL(".msc"), EXTENSION_MSC, true, false},
+    {FILE_PATH_LITERAL(".msh"), EXTENSION_MSH, true, false},
+    {FILE_PATH_LITERAL(".msh1"), EXTENSION_MSH1, true, false},
+    {FILE_PATH_LITERAL(".msh1xml"), EXTENSION_MSH1XML, true, false},
+    {FILE_PATH_LITERAL(".msh2"), EXTENSION_MSH2, true, false},
+    {FILE_PATH_LITERAL(".msh2xml"), EXTENSION_MSH2XML, true, false},
+    {FILE_PATH_LITERAL(".mshxml"), EXTENSION_MSHXML, true, false},
+    {FILE_PATH_LITERAL(".msi"), EXTENSION_MSI, true, false},
+    {FILE_PATH_LITERAL(".msp"), EXTENSION_MSP, true, false},
+    {FILE_PATH_LITERAL(".mst"), EXTENSION_MST, true, false},
+    {FILE_PATH_LITERAL(".osx"), EXTENSION_OSX, true, false},
+    {FILE_PATH_LITERAL(".pif"), EXTENSION_PIF, true, false},
+    {FILE_PATH_LITERAL(".pkg"), EXTENSION_PKG, true, false},
+    {FILE_PATH_LITERAL(".ps1"), EXTENSION_PS1, true, false},
+    {FILE_PATH_LITERAL(".ps1xml"), EXTENSION_PS1XML, true, false},
+    {FILE_PATH_LITERAL(".ps2"), EXTENSION_PS2, true, false},
+    {FILE_PATH_LITERAL(".ps2xml"), EXTENSION_PS2XML, true, false},
+    {FILE_PATH_LITERAL(".psc1"), EXTENSION_PSC1, true, false},
+    {FILE_PATH_LITERAL(".psc2"), EXTENSION_PSC2, true, false},
+    {FILE_PATH_LITERAL(".rar"), EXTENSION_RAR, false, true},
+    {FILE_PATH_LITERAL(".reg"), EXTENSION_REG, true, false},
+    {FILE_PATH_LITERAL(".scf"), EXTENSION_SCF, true, false},
+    {FILE_PATH_LITERAL(".scr"), EXTENSION_SCR, true, false},
+    {FILE_PATH_LITERAL(".sct"), EXTENSION_SCT, true, false},
+    {FILE_PATH_LITERAL(".tar"), EXTENSION_TAR, false, true},
+    {FILE_PATH_LITERAL(".tgz"), EXTENSION_TGZ, false, true},
+    {FILE_PATH_LITERAL(".url"), EXTENSION_URL, true, false},
+    {FILE_PATH_LITERAL(".vb"), EXTENSION_VB, true, false},
+    {FILE_PATH_LITERAL(".vbe"), EXTENSION_VBE, true, false},
+    {FILE_PATH_LITERAL(".vbs"), EXTENSION_VBS, true, false},
+    {FILE_PATH_LITERAL(".website"), EXTENSION_WEBSITE, true, false},
+    {FILE_PATH_LITERAL(".wim"), EXTENSION_WIM, false, true},
+    {FILE_PATH_LITERAL(".wsf"), EXTENSION_WSF, true, false},
+    {FILE_PATH_LITERAL(".xz"), EXTENSION_XZ, false, true},
+    {FILE_PATH_LITERAL(".z"), EXTENSION_Z, false, true},
+    {FILE_PATH_LITERAL(".zip"), EXTENSION_ZIP, true, true},
+};
+
+const SafeBrowsingFiletype& GetFileType(const base::FilePath& file) {
+  static const SafeBrowsingFiletype kOther = {
+    nullptr, EXTENSION_OTHER, false, false
   };
-  for (const auto& extension : kArchiveFileTypes) {
-    if (file.MatchesExtension(extension))
-      return true;
-  }
-  // TODO(mattm): should .dmg be checked here instead of IsSupportedBinaryFile?
-  return false;
+
+  base::FilePath::StringType extension = file.FinalExtension();
+  SafeBrowsingFiletype needle = {extension.c_str()};
+
+  const auto begin = kSafeBrowsingFileTypes;
+  const auto end = kSafeBrowsingFileTypes + arraysize(kSafeBrowsingFileTypes);
+  const auto result = std::lower_bound(
+      begin, end, needle,
+      [](const SafeBrowsingFiletype& left, const SafeBrowsingFiletype& right) {
+        return base::FilePath::CompareLessIgnoreCase(left.extension,
+                                                     right.extension);
+      });
+  if (result == end ||
+      !base::FilePath::CompareEqualIgnoreCase(needle.extension,
+                                              result->extension))
+    return kOther;
+  return *result;
 }
 
-bool IsSupportedBinaryFile(const base::FilePath& file) {
-  const base::FilePath::CharType* kSupportedBinaryFileTypes[] = {
-      // Executable extensions for MS Windows.
-      FILE_PATH_LITERAL(".cmd"),
-      FILE_PATH_LITERAL(".com"),
-      FILE_PATH_LITERAL(".dll"),
-      FILE_PATH_LITERAL(".exe"),
-      FILE_PATH_LITERAL(".msc"),
-      FILE_PATH_LITERAL(".msi"),
-      FILE_PATH_LITERAL(".msp"),
-      FILE_PATH_LITERAL(".mst"),
-      FILE_PATH_LITERAL(".pif"),
-      FILE_PATH_LITERAL(".scr"),
-      // Not binary, but still contain executable code, or can be used to launch
-      // other executables.
-      FILE_PATH_LITERAL(".bas"),
-      FILE_PATH_LITERAL(".bat"),
-      FILE_PATH_LITERAL(".hta"),
-      FILE_PATH_LITERAL(".js"),
-      FILE_PATH_LITERAL(".jse"),
-      FILE_PATH_LITERAL(".mht"),
-      FILE_PATH_LITERAL(".mhtml"),
-      FILE_PATH_LITERAL(".msh"),
-      FILE_PATH_LITERAL(".msh1"),
-      FILE_PATH_LITERAL(".msh1xml"),
-      FILE_PATH_LITERAL(".msh2"),
-      FILE_PATH_LITERAL(".msh2xml"),
-      FILE_PATH_LITERAL(".mshxml"),
-      FILE_PATH_LITERAL(".ps1"),
-      FILE_PATH_LITERAL(".ps1xml"),
-      FILE_PATH_LITERAL(".ps2"),
-      FILE_PATH_LITERAL(".ps2xml"),
-      FILE_PATH_LITERAL(".psc1"),
-      FILE_PATH_LITERAL(".psc2"),
-      FILE_PATH_LITERAL(".reg"),
-      FILE_PATH_LITERAL(".scf"),
-      FILE_PATH_LITERAL(".sct"),
-      FILE_PATH_LITERAL(".url"),
-      FILE_PATH_LITERAL(".vb"),
-      FILE_PATH_LITERAL(".vbe"),
-      FILE_PATH_LITERAL(".vbs"),
-      FILE_PATH_LITERAL(".website"),
-      FILE_PATH_LITERAL(".wsf"),
-      // Chrome extensions and android APKs are also reported.
-      FILE_PATH_LITERAL(".apk"),
-      FILE_PATH_LITERAL(".crx"),
-      // Mac extensions.
-      FILE_PATH_LITERAL(".app"),
-      FILE_PATH_LITERAL(".dmg"),
-      FILE_PATH_LITERAL(".osx"),
-      FILE_PATH_LITERAL(".pkg"),
-  };
-  for (const auto& extension : kSupportedBinaryFileTypes)
-    if (file.MatchesExtension(extension))
-      return true;
+} // namespace
+
+const int kSBClientDownloadExtensionsMax = EXTENSION_MAX;
 
-  // .zip files are examined for any executables or other archives they may
-  // contain. Currently no other archive formats are supported.
-  return file.MatchesExtension(FILE_PATH_LITERAL(".zip"));
+bool IsArchiveFile(const base::FilePath& file) {
+  // List of interesting archive file formats in kSafeBrowsingFileTypes is by no
+  // means exhaustive, but are currently file types that Safe Browsing would
+  // like to see pings for due to the possibility of them being used as wrapper
+  // formats for malicious payloads.
+  return GetFileType(file).is_archive;
+}
+
+bool IsSupportedBinaryFile(const base::FilePath& file) {
+  return GetFileType(file).is_supported_binary;
 }
 
 ClientDownloadRequest::DownloadType GetDownloadType(
@@ -125,5 +226,9 @@ ClientDownloadRequest::DownloadType GetDownloadType(
   return ClientDownloadRequest::WIN_EXECUTABLE;
 }
 
+int GetSBClientDownloadExtensionValueForUMA(const base::FilePath& file) {
+  return GetFileType(file).uma_value;
+}
+
 }  // namespace download_protection_util
 }  // namespace safe_browsing