FTP: Fix Windows directory listing parser for empty file names.

BUG=66097
TEST=net_unittests

Review URL: http://codereview.chromium.org/5688005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69558 0039d316-1c4b-4281-b951-d872f2087c98

2 files changed, 45 insertions, 7 deletions

diff --git a/net/ftp/ftp_directory_listing_parser_windows.cc b/net/ftp/ftp_directory_listing_parser_windows.cc
index 7bce556..ef733d5 100644
--- a/net/ftp/ftp_directory_listing_parser_windows.cc
+++ b/net/ftp/ftp_directory_listing_parser_windows.cc
@@ -15,7 +15,7 @@ namespace {
 
 bool WindowsDateListingToTime(const std::vector<string16>& columns,
                               base::Time* time) {
-  DCHECK_LE(4U, columns.size());
+  DCHECK_LE(3U, columns.size());
 
   base::Time::Exploded time_exploded = { 0 };
 
@@ -84,15 +84,18 @@ bool FtpDirectoryListingParserWindows::ConsumeLine(const string16& line) {
   std::vector<string16> columns;
   base::SplitString(CollapseWhitespace(line, false), ' ', &columns);
 
-  // We may receive file names containing spaces, which can make the number of
-  // columns arbitrarily large. We will handle that later. For now just make
-  // sure we have all the columns that should normally be there.
-  if (columns.size() < 4)
+  // Every line of the listing consists of the following:
+  //
+  //   1. date
+  //   2. time
+  //   3. size in bytes (or "<DIR>" for directories)
+  //   4. filename (may be empty or contain spaces)
+  //
+  // For now, make sure we have 1-3, and handle 4 later.
+  if (columns.size() < 3)
     return false;
 
   FtpDirectoryListingEntry entry;
-  entry.name = FtpUtil::GetStringPartAfterColumns(line, 3);
-
   if (EqualsASCII(columns[2], "<DIR>")) {
     entry.type = FtpDirectoryListingEntry::DIRECTORY;
     entry.size = -1;
@@ -107,6 +110,14 @@ bool FtpDirectoryListingParserWindows::ConsumeLine(const string16& line) {
   if (!WindowsDateListingToTime(columns, &entry.last_modified))
     return false;
 
+  entry.name = FtpUtil::GetStringPartAfterColumns(line, 3);
+  if (entry.name.empty()) {
+    // Some FTP servers send listing entries with empty names. It's not obvious
+    // how to display such an entry, so we ignore them. We don't want to make
+    // the parsing fail at this point though. Other entries can still be useful.
+    return true;
+  }
+
   entries_.push(entry);
   return true;
 }
diff --git a/net/ftp/ftp_directory_listing_parser_windows_unittest.cc b/net/ftp/ftp_directory_listing_parser_windows_unittest.cc
index 60309cb..aeb64e7 100644
--- a/net/ftp/ftp_directory_listing_parser_windows_unittest.cc
+++ b/net/ftp/ftp_directory_listing_parser_windows_unittest.cc
@@ -64,15 +64,42 @@ TEST_F(FtpDirectoryListingParserWindowsTest, Good) {
   }
 }
 
+TEST_F(FtpDirectoryListingParserWindowsTest, Ignored) {
+  const char* ignored_cases[] = {
+    "12-07-10  12:05AM       <DIR>    ",  // http://crbug.com/66097
+    "12-07-10  12:05AM       1234    ",
+  };
+  for (size_t i = 0; i < arraysize(ignored_cases); i++) {
+    SCOPED_TRACE(base::StringPrintf("Test[%" PRIuS "]: %s", i,
+                                    ignored_cases[i]));
+
+    net::FtpDirectoryListingParserWindows parser;
+    EXPECT_TRUE(parser.ConsumeLine(UTF8ToUTF16(ignored_cases[i])));
+    EXPECT_FALSE(parser.EntryAvailable());
+    EXPECT_TRUE(parser.OnEndOfInput());
+    EXPECT_FALSE(parser.EntryAvailable());
+  }
+}
+
 TEST_F(FtpDirectoryListingParserWindowsTest, Bad) {
   const char* bad_cases[] = {
     "",
     "garbage",
+    "11-02-09  05:32PM       <GARBAGE>",
     "11-02-09  05:32PM       <GARBAGE>      NT",
+    "11-02-09  05:32         <DIR>",
+    "11-FEB-09 05:32PM       <DIR>",
+    "11-02     05:32PM       <DIR>",
+    "11-02-09  05:32PM                 -1",
     "11-02-09  05:32         <DIR>          NT",
     "11-FEB-09 05:32PM       <DIR>          NT",
     "11-02     05:32PM       <DIR>          NT",
     "11-02-09  05:32PM                 -1   NT",
+    "99-25-10  12:00AM                  0",
+    "12-99-10  12:00AM                  0",
+    "12-25-10  99:00AM                  0",
+    "12-25-10  12:99AM                  0",
+    "12-25-10  12:00ZM                  0",
     "99-25-10  12:00AM                  0   months out of range",
     "12-99-10  12:00AM                  0   days out of range",
     "12-25-10  99:00AM                  0   hours out of range",