Implement parser for Netware-style FTP LIST response listing.

TEST=Covered by net_unittests.
BUG=25520

Review URL: http://codereview.chromium.org/465059

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@33978 0039d316-1c4b-4281-b951-d872f2087c98

11 files changed, 335 insertions, 57 deletions

diff --git a/net/ftp/ftp_directory_listing_buffer.cc b/net/ftp/ftp_directory_listing_buffer.cc
index a74b7fe..644f59e 100644
--- a/net/ftp/ftp_directory_listing_buffer.cc
+++ b/net/ftp/ftp_directory_listing_buffer.cc
@@ -9,6 +9,7 @@
 #include "base/string_util.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_vms.h"
 #include "net/ftp/ftp_directory_listing_parser_windows.h"
 #include "unicode/ucsdet.h"
@@ -44,6 +45,7 @@ namespace net {
 FtpDirectoryListingBuffer::FtpDirectoryListingBuffer()
     : current_parser_(NULL) {
   parsers_.insert(new FtpDirectoryListingParserLs());
+  parsers_.insert(new FtpDirectoryListingParserNetware());
   parsers_.insert(new FtpDirectoryListingParserVms());
   parsers_.insert(new FtpDirectoryListingParserWindows());
 }
@@ -80,7 +82,6 @@ int FtpDirectoryListingBuffer::ProcessRemainingData() {
   if (rv != OK)
     return rv;
 
-  DCHECK(current_parser_);
   return OK;
 }
 
@@ -169,7 +170,18 @@ int FtpDirectoryListingBuffer::OnEndOfInput() {
   }
 
   if (parsers_.size() != 1) {
-    current_parser_ = NULL;
+    DCHECK(!current_parser_);
+
+    // We may hit an ambiguity in case of listings which have no entries. That's
+    // fine, as long as all remaining parsers agree that the listing is empty.
+    bool all_listings_empty = true;
+    for (ParserSet::iterator i = parsers_.begin(); i != parsers_.end(); ++i) {
+      if ((*i)->EntryAvailable())
+        all_listings_empty = false;
+    }
+    if (all_listings_empty)
+      return OK;
+
     return ERR_UNRECOGNIZED_FTP_DIRECTORY_LISTING_FORMAT;
   }
 
diff --git a/net/ftp/ftp_directory_listing_buffer_unittest.cc b/net/ftp/ftp_directory_listing_buffer_unittest.cc
index d5063c2..cf107b9 100644
--- a/net/ftp/ftp_directory_listing_buffer_unittest.cc
+++ b/net/ftp/ftp_directory_listing_buffer_unittest.cc
@@ -28,13 +28,15 @@ TEST(FtpDirectoryListingBufferTest, Parse) {
     "dir-listing-ls-8",
     "dir-listing-ls-9",
     "dir-listing-ls-10",
-    "dir-listing-windows-1",
-    "dir-listing-windows-2",
+    "dir-listing-netware-1",
+    "dir-listing-netware-2",
     "dir-listing-vms-1",
     "dir-listing-vms-2",
     "dir-listing-vms-3",
     "dir-listing-vms-4",
     "dir-listing-vms-5",
+    "dir-listing-windows-1",
+    "dir-listing-windows-2",
   };
 
   FilePath test_dir;
diff --git a/net/ftp/ftp_directory_listing_parser_ls.cc b/net/ftp/ftp_directory_listing_parser_ls.cc
index 195f4a4..64feeed 100644
--- a/net/ftp/ftp_directory_listing_parser_ls.cc
+++ b/net/ftp/ftp_directory_listing_parser_ls.cc
@@ -57,66 +57,16 @@ string16 GetStringPartAfterColumns(const string16& text, int columns) {
   return result;
 }
 
-bool UnixDateListingToTime(const std::vector<string16>& columns,
-                           base::Time* time) {
-  DCHECK_LE(9U, columns.size());
-
-  base::Time::Exploded time_exploded = { 0 };
-
-  if (!net::FtpUtil::ThreeLetterMonthToNumber(columns[5], &time_exploded.month))
-    return false;
-
-  if (!StringToInt(columns[6], &time_exploded.day_of_month))
-    return false;
-
-  if (!StringToInt(columns[7], &time_exploded.year)) {
-    // Maybe it's time. Does it look like time (MM:HH)?
-    if (columns[7].length() != 5 || columns[7][2] != ':')
-      return false;
-
-    if (!StringToInt(columns[7].substr(0, 2), &time_exploded.hour))
-      return false;
-
-    if (!StringToInt(columns[7].substr(3, 2), &time_exploded.minute))
-      return false;
-
-    // Use current year.
-    base::Time::Exploded now_exploded;
-    base::Time::Now().LocalExplode(&now_exploded);
-    time_exploded.year = now_exploded.year;
-  }
-
-  // We don't know the time zone of the server, so just use local time.
-  *time = base::Time::FromLocalExploded(time_exploded);
-  return true;
-}
-
 }  // namespace
 
 namespace net {
 
 FtpDirectoryListingParserLs::FtpDirectoryListingParserLs()
-    : received_nonempty_line_(false) {
+    : received_nonempty_line_(false),
+      received_total_line_(false) {
 }
 
 bool FtpDirectoryListingParserLs::ConsumeLine(const string16& line) {
-  if (StartsWith(line, ASCIIToUTF16("total "), true) ||
-      StartsWith(line, ASCIIToUTF16("Gesamt "), true)) {
-    // Some FTP servers put a "total n" line at the beginning of the listing
-    // (n is an integer). Allow such a line, but only once, and only if it's
-    // the first non-empty line.
-    //
-    // Note: "Gesamt" is a German word for "total". The case is important here:
-    // for "ls -l" style listings, "total" will be lowercase, and Gesamt will be
-    // capitalized. This helps us distinguish that from a VMS-style listing,
-    // which would use "Total" (note the uppercase first letter).
-
-    if (received_nonempty_line_)
-      return false;
-
-    received_nonempty_line_ = true;
-    return true;
-  }
   if (line.empty() && !received_nonempty_line_) {
     // Allow empty lines only at the beginning of the listing. For example VMS
     // systems in Unix emulation mode add an empty line before the first listing
@@ -128,6 +78,23 @@ bool FtpDirectoryListingParserLs::ConsumeLine(const string16& line) {
   std::vector<string16> columns;
   SplitString(CollapseWhitespace(line, false), ' ', &columns);
 
+  // Some FTP servers put a "total n" line at the beginning of the listing
+  // (n is an integer). Allow such a line, but only once, and only if it's
+  // the first non-empty line. Do not match the word exactly, because it may be
+  // in different languages (at least English and German have been seen in the
+  // field).
+  if (columns.size() == 2 && !received_total_line_) {
+    received_total_line_ = true;
+
+    int total_number;
+    if (!StringToInt(columns[1], &total_number))
+      return false;
+    if (total_number < 0)
+      return false;
+
+    return true;
+  }
+
   // 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.
@@ -159,8 +126,10 @@ bool FtpDirectoryListingParserLs::ConsumeLine(const string16& line) {
   if (entry.type != FtpDirectoryListingEntry::FILE)
     entry.size = -1;
 
-  if (!UnixDateListingToTime(columns, &entry.last_modified))
+  if (!FtpUtil::LsDateListingToTime(columns[5], columns[6], columns[7],
+                                    &entry.last_modified)) {
     return false;
+  }
 
   entry.name = GetStringPartAfterColumns(line, 8);
   if (entry.type == FtpDirectoryListingEntry::SYMLINK) {
diff --git a/net/ftp/ftp_directory_listing_parser_ls.h b/net/ftp/ftp_directory_listing_parser_ls.h
index 689fb7e..3fd0d32 100644
--- a/net/ftp/ftp_directory_listing_parser_ls.h
+++ b/net/ftp/ftp_directory_listing_parser_ls.h
@@ -26,6 +26,10 @@ class FtpDirectoryListingParserLs : public FtpDirectoryListingParser {
  private:
   bool received_nonempty_line_;
 
+  // True after we have received a "total n" listing header, where n is an
+  // integer. Only one such header is allowed per listing.
+  bool received_total_line_;
+
   std::queue<FtpDirectoryListingEntry> entries_;
 
   DISALLOW_COPY_AND_ASSIGN(FtpDirectoryListingParserLs);
diff --git a/net/ftp/ftp_directory_listing_parser_netware.cc b/net/ftp/ftp_directory_listing_parser_netware.cc
new file mode 100644
index 0000000..20d2b25
--- /dev/null
+++ b/net/ftp/ftp_directory_listing_parser_netware.cc
@@ -0,0 +1,102 @@
+// Copyright (c) 2009 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_netware.h"
+
+#include <vector>
+
+#include "base/string_util.h"
+#include "net/ftp/ftp_util.h"
+
+namespace {
+
+bool LooksLikeNetwarePermissionsListing(const string16& text) {
+  if (text.length() != 10)
+    return false;
+
+  if (text[0] != '[' || text[9] != ']')
+    return false;
+  return (text[1] == 'R' || text[1] == '-') &&
+         (text[2] == 'W' || text[2] == '-') &&
+         (text[3] == 'C' || text[3] == '-') &&
+         (text[4] == 'E' || text[4] == '-') &&
+         (text[5] == 'A' || text[5] == '-') &&
+         (text[6] == 'F' || text[6] == '-') &&
+         (text[7] == 'M' || text[7] == '-') &&
+         (text[8] == 'S' || text[8] == '-');
+}
+
+}  // namespace
+
+namespace net {
+
+FtpDirectoryListingParserNetware::FtpDirectoryListingParserNetware()
+    : received_first_line_(false) {
+}
+
+bool FtpDirectoryListingParserNetware::ConsumeLine(const string16& line) {
+  if (!received_first_line_) {
+    received_first_line_ = true;
+
+    return StartsWith(line, ASCIIToUTF16("total "), true);
+  }
+
+  std::vector<string16> columns;
+  SplitString(CollapseWhitespace(line, false), ' ', &columns);
+
+  if (columns.size() != 8)
+    return false;
+
+  FtpDirectoryListingEntry entry;
+
+  if (columns[0].length() != 1)
+    return false;
+  if (columns[0][0] == 'd') {
+    entry.type = FtpDirectoryListingEntry::DIRECTORY;
+  } else if (columns[0][0] == '-') {
+    entry.type = FtpDirectoryListingEntry::FILE;
+  } else {
+    return false;
+  }
+
+  // Note: on older Netware systems the permissions listing is in the same
+  // column as the entry type (just there is no space between them). We do not
+  // support the older format here for simplicity.
+  if (!LooksLikeNetwarePermissionsListing(columns[1]))
+    return false;
+
+  if (!StringToInt64(columns[3], &entry.size))
+    return false;
+  if (entry.size < 0)
+    return false;
+  if (entry.type != FtpDirectoryListingEntry::FILE)
+    entry.size = -1;
+
+  // Netware uses the same date listing format as Unix "ls -l".
+  if (!FtpUtil::LsDateListingToTime(columns[4], columns[5], columns[6],
+                                    &entry.last_modified)) {
+    return false;
+  }
+
+  entry.name = columns[7];
+
+  entries_.push(entry);
+  return true;
+}
+
+bool FtpDirectoryListingParserNetware::OnEndOfInput() {
+  return true;
+}
+
+bool FtpDirectoryListingParserNetware::EntryAvailable() const {
+  return !entries_.empty();
+}
+
+FtpDirectoryListingEntry FtpDirectoryListingParserNetware::PopEntry() {
+  FtpDirectoryListingEntry entry = entries_.front();
+  entries_.pop();
+  return entry;
+}
+
+}  // namespace net
diff --git a/net/ftp/ftp_directory_listing_parser_netware.h b/net/ftp/ftp_directory_listing_parser_netware.h
new file mode 100644
index 0000000..a3b2cef
--- /dev/null
+++ b/net/ftp/ftp_directory_listing_parser_netware.h
@@ -0,0 +1,37 @@
+// Copyright (c) 2009 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.
+
+#ifndef NET_FTP_FTP_DIRECTORY_LISTING_PARSER_NETWARE_H_
+#define NET_FTP_FTP_DIRECTORY_LISTING_PARSER_NETWARE_H_
+
+#include <queue>
+
+#include "net/ftp/ftp_directory_listing_parser.h"
+
+namespace net {
+
+// Parser for Netware-style directory listing.
+class FtpDirectoryListingParserNetware : public FtpDirectoryListingParser {
+ public:
+  FtpDirectoryListingParserNetware();
+
+  // FtpDirectoryListingParser methods:
+  virtual FtpServerType GetServerType() const { return SERVER_NETWARE; }
+  virtual bool ConsumeLine(const string16& line);
+  virtual bool OnEndOfInput();
+  virtual bool EntryAvailable() const;
+  virtual FtpDirectoryListingEntry PopEntry();
+
+ private:
+  // True after we have received the first line of input.
+  bool received_first_line_;
+
+  std::queue<FtpDirectoryListingEntry> entries_;
+
+  DISALLOW_COPY_AND_ASSIGN(FtpDirectoryListingParserNetware);
+};
+
+}  // namespace net
+
+#endif  // NET_FTP_FTP_DIRECTORY_LISTING_PARSER_NETWARE_H_
diff --git a/net/ftp/ftp_directory_listing_parser_netware_unittest.cc b/net/ftp/ftp_directory_listing_parser_netware_unittest.cc
new file mode 100644
index 0000000..7076a3f
--- /dev/null
+++ b/net/ftp/ftp_directory_listing_parser_netware_unittest.cc
@@ -0,0 +1,53 @@
+// Copyright (c) 2009 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_unittest.h"
+
+#include "base/format_macros.h"
+#include "net/ftp/ftp_directory_listing_parser_netware.h"
+
+namespace {
+
+typedef net::FtpDirectoryListingParserTest FtpDirectoryListingParserNetwareTest;
+
+TEST_F(FtpDirectoryListingParserNetwareTest, Good) {
+  base::Time::Exploded now_exploded;
+  base::Time::Now().LocalExplode(&now_exploded);
+
+  const struct SingleLineTestData good_cases[] = {
+    { "d [RWCEAFMS] ftpadmin 512 Jan 29  2004 pub",
+      net::FtpDirectoryListingEntry::DIRECTORY, "pub", -1,
+      2004, 1, 29, 0, 0 },
+    { "- [RW------] ftpadmin 123 Nov 11  18:25 afile",
+      net::FtpDirectoryListingEntry::FILE, "afile", 123,
+      now_exploded.year, 11, 11, 18, 25 },
+  };
+  for (size_t i = 0; i < arraysize(good_cases); i++) {
+    SCOPED_TRACE(StringPrintf("Test[%" PRIuS "]: %s", i, good_cases[i].input));
+
+    net::FtpDirectoryListingParserNetware parser;
+    // The parser requires a "total n" like before accepting regular input.
+    ASSERT_TRUE(parser.ConsumeLine(UTF8ToUTF16("total 1")));
+    RunSingleLineTestCase(&parser, good_cases[i]);
+  }
+}
+
+TEST_F(FtpDirectoryListingParserNetwareTest, Bad) {
+  const char* bad_cases[] = {
+    "garbage",
+    "d [] ftpadmin 512 Jan 29  2004 pub",
+    "d [XGARBAGE] ftpadmin 512 Jan 29  2004 pub",
+    "d [RWCEAFMS] 512 Jan 29  2004 pub",
+    "d [RWCEAFMS] ftpadmin -1 Jan 29  2004 pub",
+    "l [RW------] ftpadmin 512 Jan 29  2004 pub",
+  };
+  for (size_t i = 0; i < arraysize(bad_cases); i++) {
+    net::FtpDirectoryListingParserNetware parser;
+    // The parser requires a "total n" like before accepting regular input.
+    ASSERT_TRUE(parser.ConsumeLine(UTF8ToUTF16("total 1")));
+    EXPECT_FALSE(parser.ConsumeLine(UTF8ToUTF16(bad_cases[i]))) << bad_cases[i];
+  }
+}
+
+}  // namespace
diff --git a/net/ftp/ftp_server_type_histograms.h b/net/ftp/ftp_server_type_histograms.h
index fef9bb2..ca9f6e4 100644
--- a/net/ftp/ftp_server_type_histograms.h
+++ b/net/ftp/ftp_server_type_histograms.h
@@ -35,6 +35,7 @@ enum FtpServerType {
   SERVER_LS = 9,        // Server using /bin/ls -l listing style.
   SERVER_WINDOWS = 10,  // Server using Windows listing style.
   SERVER_VMS = 11,      // Server using VMS listing style.
+  SERVER_NETWARE = 12,  // Server using Netware listing style.
 
   NUM_OF_SERVER_TYPES
 };
diff --git a/net/ftp/ftp_util.cc b/net/ftp/ftp_util.cc
index 5d4cd37..a0a47cf 100644
--- a/net/ftp/ftp_util.cc
+++ b/net/ftp/ftp_util.cc
@@ -9,6 +9,7 @@
 #include "base/logging.h"
 #include "base/string_tokenizer.h"
 #include "base/string_util.h"
+#include "base/time.h"
 
 // For examples of Unix<->VMS path conversions, see the unit test file. On VMS
 // a path looks differently depending on whether it's a file or directory.
@@ -143,5 +144,37 @@ bool FtpUtil::ThreeLetterMonthToNumber(const string16& text, int* number) {
   return false;
 }
 
+// static
+bool FtpUtil::LsDateListingToTime(const string16& month, const string16& day,
+                                  const string16& rest, base::Time* time) {
+  base::Time::Exploded time_exploded = { 0 };
+
+  if (!ThreeLetterMonthToNumber(month, &time_exploded.month))
+    return false;
+
+  if (!StringToInt(day, &time_exploded.day_of_month))
+    return false;
+
+  if (!StringToInt(rest, &time_exploded.year)) {
+    // Maybe it's time. Does it look like time (MM:HH)?
+    if (rest.length() != 5 || rest[2] != ':')
+      return false;
+
+    if (!StringToInt(rest.substr(0, 2), &time_exploded.hour))
+      return false;
+
+    if (!StringToInt(rest.substr(3, 2), &time_exploded.minute))
+      return false;
+
+    // Use current year.
+    base::Time::Exploded now_exploded;
+    base::Time::Now().LocalExplode(&now_exploded);
+    time_exploded.year = now_exploded.year;
+  }
+
+  // We don't know the time zone of the listing, so just use local time.
+  *time = base::Time::FromLocalExploded(time_exploded);
+  return true;
+}
 
 }  // namespace
diff --git a/net/ftp/ftp_util.h b/net/ftp/ftp_util.h
index 2a73c68..518801b 100644
--- a/net/ftp/ftp_util.h
+++ b/net/ftp/ftp_util.h
@@ -9,6 +9,10 @@
 
 #include "base/string16.h"
 
+namespace base {
+class Time;
+}
+
 namespace net {
 
 class FtpUtil {
@@ -25,6 +29,15 @@ class FtpUtil {
   // Convert three-letter month abbreviation (like Nov) to its number (in range
   // 1-12).
   static bool ThreeLetterMonthToNumber(const string16& text, int* number);
+
+  // Convert a "ls -l" date listing to time. The listing comes in three columns.
+  // The first one contains month, the second one contains day of month.
+  // The first one is either a time (and then the current year is assumed),
+  // or is a year (and then we don't know the time).
+  static bool LsDateListingToTime(const string16& month,
+                                  const string16& day,
+                                  const string16& rest,
+                                  base::Time* time);
 };
 
 }  // namespace net
diff --git a/net/ftp/ftp_util_unittest.cc b/net/ftp/ftp_util_unittest.cc
index 2ccb75b..46db873d 100644
--- a/net/ftp/ftp_util_unittest.cc
+++ b/net/ftp/ftp_util_unittest.cc
@@ -5,6 +5,9 @@
 #include "net/ftp/ftp_util.h"
 
 #include "base/basictypes.h"
+#include "base/format_macros.h"
+#include "base/string_util.h"
+#include "base/time.h"
 #include "testing/gtest/include/gtest/gtest.h"
 
 namespace {
@@ -98,4 +101,53 @@ TEST(FtpUtilTest, VMSPathToUnix) {
   }
 }
 
+TEST(FtpUtilTest, LsDateListingToTime) {
+  base::Time::Exploded now_exploded;
+  base::Time::Now().LocalExplode(&now_exploded);
+
+  const struct {
+    // Input.
+    const char* month;
+    const char* day;
+    const char* rest;
+
+    // Expected output.
+    int expected_year;
+    int expected_month;
+    int expected_day_of_month;
+    int expected_hour;
+    int expected_minute;
+  } kTestCases[] = {
+    { "Nov", "01", "2007", 2007, 11, 1, 0, 0 },
+    { "Jul", "25", "13:37", now_exploded.year, 7, 25, 13, 37 },
+
+    // Test date listings in German, we should support them for FTP servers
+    // giving localized listings.
+    { "M\xc3\xa4r", "13", "2009", 2009, 3, 13, 0, 0 },
+    { "Mai", "1", "10:10", now_exploded.year, 5, 1, 10, 10 },
+    { "Okt", "14", "21:18", now_exploded.year, 10, 14, 21, 18 },
+    { "Dez", "25", "2008", 2008, 12, 25, 0, 0 },
+  };
+  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(kTestCases); i++) {
+    SCOPED_TRACE(StringPrintf("Test[%" PRIuS "]: %s %s %s", i,
+                              kTestCases[i].month, kTestCases[i].day,
+                              kTestCases[i].rest));
+
+    base::Time time;
+    ASSERT_TRUE(net::FtpUtil::LsDateListingToTime(
+        UTF8ToUTF16(kTestCases[i].month), UTF8ToUTF16(kTestCases[i].day),
+        UTF8ToUTF16(kTestCases[i].rest), &time));
+
+    base::Time::Exploded time_exploded;
+    time.LocalExplode(&time_exploded);
+    EXPECT_EQ(kTestCases[i].expected_year, time_exploded.year);
+    EXPECT_EQ(kTestCases[i].expected_month, time_exploded.month);
+    EXPECT_EQ(kTestCases[i].expected_day_of_month, time_exploded.day_of_month);
+    EXPECT_EQ(kTestCases[i].expected_hour, time_exploded.hour);
+    EXPECT_EQ(kTestCases[i].expected_minute, time_exploded.minute);
+    EXPECT_EQ(0, time_exploded.second);
+    EXPECT_EQ(0, time_exploded.millisecond);
+  }
+}
+
 }  // namespace