FTP: correctly handle newlines in file names

net/ftp/ftp_directory_listing_parser.cc

 // Copyright (c) 2011 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 "net/ftp/ftp_directory_listing_parser.h"
 
 #include "base/bind.h"
 #include "base/callback.h"
 #include "base/i18n/icu_encoding_detection.h"
 #include "base/i18n/icu_string_conversions.h"
 #include "base/stl_util.h"
 #include "base/string_split.h"
 #include "base/string_util.h"
+#include "base/utf_string_conversions.h"
 #include "net/base/net_errors.h"
 #include "net/ftp/ftp_directory_listing_parser_ls.h"
 #include "net/ftp/ftp_directory_listing_parser_netware.h"
 #include "net/ftp/ftp_directory_listing_parser_os2.h"
 #include "net/ftp/ftp_directory_listing_parser_vms.h"
 #include "net/ftp/ftp_directory_listing_parser_windows.h"
 #include "net/ftp/ftp_server_type_histograms.h"
 
 namespace net {
 
@@ skipping 10 matching lines @@
       return ERR_ENCODING_CONVERSION_FAILED;
     }
   }
 
   return OK;
 }
 
 // Parses |text| as an FTP directory listing. Fills in |entries|
 // and |server_type| and returns network error code.
 int ParseListing(const string16& text,
+                 const string16& newline_separator,
                  const std::string& encoding,
                  const base::Time& current_time,
                  std::vector<FtpDirectoryListingEntry>* entries,
                  FtpServerType* server_type) {
   std::vector<string16> lines;
-  base::SplitString(text, '\n', &lines);
+  base::SplitStringUsingSubstr(text, newline_separator, &lines);
 
   struct {
     base::Callback<bool(void)> callback;
     FtpServerType server_type;
   } parsers[] = {
     {
       base::Bind(&ParseFtpDirectoryListingLs, lines, current_time, entries),
       SERVER_LS
     },
     {
@@ skipping 26 matching lines @@
   entries->clear();
   return ERR_UNRECOGNIZED_FTP_DIRECTORY_LISTING_FORMAT;
 }
 
 // Detects encoding of |text| and parses it as an FTP directory listing.
 // Fills in |entries| and |server_type| and returns network error code.
 int DecodeAndParse(const std::string& text,
                    const base::Time& current_time,
                    std::vector<FtpDirectoryListingEntry>* entries,
                    FtpServerType* server_type) {
+  const char* kNewlineSeparators[] = { "\n", "\r\n", "\r" };

[mmenke, 2012/12/12 17:20:44]

SplitString trims whitespace, so if |text| only has "\r\n" line breaks, it
returns identical results with "\n" and "\r\n".

Suppose servers could use CRLF line breaks, and have LFs in file names, but that
seems unlikely.

[mmenke, 2012/12/12 17:20:44]

If we run into a server that uses "\r"-style linebreaks, with no "total blah"
prefix, and try to parse it with "\n" style linebreaks, won't we just treat it
as one file with a really long name?

I'd like to see a test with multiple lines using "\r" line breaks.

[Paweł Hajdan Jr., 2012/12/12 18:26:19]

That's exactly what the bug is about. :)

[Paweł Hajdan Jr., 2012/12/12 18:26:19]

Good point. I just removed "\r" then. I haven't yet seen a server sending these
line endings (and I've done research to check as many servers as possible). We
can worry about that when it happens, not in advance.

[mmenke, 2012/12/12 18:36:12]

Oops - didn't see the bug.

+
   std::vector<std::string> encodings;
   if (!base::DetectAllEncodings(text, &encodings))
     return ERR_ENCODING_DETECTION_FAILED;
 
   // Use first encoding that can be used to decode the text.
   for (size_t i = 0; i < encodings.size(); i++) {
     string16 converted_text;
     if (base::CodepageToUTF16(text,
                               encodings[i].c_str(),
                               base::OnStringConversionError::FAIL,
                               &converted_text)) {
-      int rv = ParseListing(converted_text,
-                            encodings[i],
-                            current_time,
-                            entries,
-                            server_type);
-      if (rv == OK)
-        return rv;
+      for (size_t j = 0; j < arraysize(kNewlineSeparators); j++) {
+        int rv = ParseListing(converted_text,
+                              ASCIIToUTF16(kNewlineSeparators[j]),
+                              encodings[i],
+                              current_time,
+                              entries,
+                              server_type);
+        if (rv == OK)
+          return rv;
+      }
     }
   }
 
   entries->clear();
   *server_type = SERVER_UNKNOWN;
   return ERR_UNRECOGNIZED_FTP_DIRECTORY_LISTING_FORMAT;
 }
 
 }  // namespace
 
 FtpDirectoryListingEntry::FtpDirectoryListingEntry()
     : type(UNKNOWN),
       size(-1) {
 }
 
 int ParseFtpDirectoryListing(const std::string& text,
                              const base::Time& current_time,
                              std::vector<FtpDirectoryListingEntry>* entries) {
   FtpServerType server_type = SERVER_UNKNOWN;
   int rv = DecodeAndParse(text, current_time, entries, &server_type);
   UpdateFtpServerTypeHistograms(server_type);
   return rv;
 }
 
 }  // namespace net