Download filename encoding fix [try2]:

Fall back on other charsets when using the URL to suggest a filename. The URL should be escaped UTF-8, but may use another encoding instead.

Check for empty string in DecodeWord (avoids crash).

BUG=44301
TEST=see bug. Also, net_unittests.

Review URL: http://codereview.chromium.org/2733005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@49475 0039d316-1c4b-4281-b951-d872f2087c98

2 files changed, 37 insertions, 4 deletions

diff --git a/net/base/net_util.cc b/net/base/net_util.cc
index 60f626d4..5eb2e68 100644
--- a/net/base/net_util.cc
+++ b/net/base/net_util.cc
@@ -272,6 +272,11 @@ bool DecodeWord(const std::string& encoded_word,
                 const std::string& referrer_charset,
                 bool* is_rfc2047,
                 std::string* output) {
+  *is_rfc2047 = false;
+  output->clear();
+  if (encoded_word.empty())
+    return true;
+
   if (!IsStringASCII(encoded_word)) {
     // Try UTF-8, referrer_charset and the native OS default charset in turn.
     if (IsStringUTF8(encoded_word)) {
@@ -287,7 +292,7 @@ bool DecodeWord(const std::string& encoded_word,
         *output = WideToUTF8(base::SysNativeMBToWide(encoded_word));
       }
     }
-    *is_rfc2047 = false;
+
     return true;
   }
 
@@ -1081,7 +1086,7 @@ FilePath GetSuggestedFilename(const GURL& url,
   }
 
   const std::string filename_from_cd = GetFileNameFromCD(content_disposition,
-                                                          referrer_charset);
+                                                         referrer_charset);
 #if defined(OS_WIN)
   FilePath::StringType filename = UTF8ToWide(filename_from_cd);
 #elif defined(OS_POSIX)
@@ -1102,10 +1107,21 @@ FilePath GetSuggestedFilename(const GURL& url,
       const std::string unescaped_url_filename = UnescapeURLComponent(
           url.ExtractFileName(),
           UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS);
+
+      // The URL's path should be escaped UTF-8, but may not be.
+      std::string decoded_filename = unescaped_url_filename;
+      if (!IsStringASCII(decoded_filename)) {
+        bool ignore;
+        // TODO(jshin): this is probably not robust enough. To be sure, we
+        // need encoding detection.
+        DecodeWord(unescaped_url_filename, referrer_charset, &ignore,
+                   &decoded_filename);
+      }
+
 #if defined(OS_WIN)
-      filename = UTF8ToWide(unescaped_url_filename);
+      filename = UTF8ToWide(decoded_filename);
 #elif defined(OS_POSIX)
-      filename = unescaped_url_filename;
+      filename = decoded_filename;
 #endif
     }
   }
diff --git a/net/base/net_util_unittest.cc b/net/base/net_util_unittest.cc
index dff3d1c..9bc5d58 100644
--- a/net/base/net_util_unittest.cc
+++ b/net/base/net_util_unittest.cc
@@ -1060,6 +1060,23 @@ TEST(NetUtilTest, GetSuggestedFilename) {
     "",
     L"",
     L"test"},
+    // The filename encoding is specified by the referrer charset.
+    {"http://example.com/V%FDvojov%E1%20psychologie.doc",
+     "",
+     "iso-8859-1",
+     L"",
+     L"V\u00fdvojov\u00e1 psychologie.doc"},
+    // The filename encoding doesn't match the referrer charset, the
+    // system charset, or UTF-8.
+    // TODO(jshin): we need to handle this case.
+#if 0
+    {"http://example.com/V%FDvojov%E1%20psychologie.doc",
+     "",
+     "utf-8",
+     L"",
+     L"V\u00fdvojov\u00e1 psychologie.doc",
+    },
+#endif
   };
   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
 #if defined(OS_WIN)