net/ftp/ftp_directory_listing_parser_ls.cc

   1 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "net/ftp/ftp_directory_listing_parser_ls.h"
   6
   7 #include <vector>
   8
   9 #include "base/strings/string_number_conversions.h"
  10 #include "base/strings/string_split.h"
  11 #include "base/strings/string_util.h"
  12 #include "base/strings/utf_string_conversions.h"
  13 #include "base/time/time.h"
  14 #include "net/ftp/ftp_directory_listing_parser.h"
  15 #include "net/ftp/ftp_util.h"
  16
  17 namespace {
  18
  19 bool TwoColumnDateListingToTime(const base::string16& date,
  20                                 const base::string16& time,
  21                                 base::Time* result) {
  22   base::Time::Exploded time_exploded = { 0 };
  23
  24   // Date should be in format YYYY-MM-DD.
  25   std::vector<base::string16> date_parts;
  26   base::SplitString(date, '-', &date_parts);
  27   if (date_parts.size() != 3)
  28     return false;
  29   if (!base::StringToInt(date_parts[0], &time_exploded.year))
  30     return false;
  31   if (!base::StringToInt(date_parts[1], &time_exploded.month))
  32     return false;
  33   if (!base::StringToInt(date_parts[2], &time_exploded.day_of_month))
  34     return false;
  35
  36   // Time should be in format HH:MM
  37   if (time.length() != 5)
  38     return false;
  39
  40   std::vector<base::string16> time_parts;
  41   base::SplitString(time, ':', &time_parts);
  42   if (time_parts.size() != 2)
  43     return false;
  44   if (!base::StringToInt(time_parts[0], &time_exploded.hour))
  45     return false;
  46   if (!base::StringToInt(time_parts[1], &time_exploded.minute))
  47     return false;
  48   if (!time_exploded.HasValidValues())
  49     return false;
  50
  51   // We don't know the time zone of the server, so just use local time.
  52   *result = base::Time::FromLocalExploded(time_exploded);
  53   return true;
  54 }
  55
  56 // Returns the column index of the end of the date listing and detected
  57 // last modification time.
  58 bool DetectColumnOffsetSizeAndModificationTime(
  59     const std::vector<base::string16>& columns,
  60     const base::Time& current_time,
  61     size_t* offset,
  62     base::string16* size,
  63     base::Time* modification_time) {
  64   // The column offset can be arbitrarily large if some fields
  65   // like owner or group name contain spaces. Try offsets from left to right
  66   // and use the first one that matches a date listing.
  67   //
  68   // Here is how a listing line should look like. A star ("*") indicates
  69   // a required field:
  70   //
  71   //  * 1. permission listing
  72   //    2. number of links (optional)
  73   //  * 3. owner name (may contain spaces)
  74   //    4. group name (optional, may contain spaces)
  75   //  * 5. size in bytes
  76   //  * 6. month
  77   //  * 7. day of month
  78   //  * 8. year or time <-- column_offset will be the index of this column
  79   //    9. file name (optional, may contain spaces)
  80   for (size_t i = 5U; i < columns.size(); i++) {
  81     if (net::FtpUtil::LsDateListingToTime(columns[i - 2],
  82                                           columns[i - 1],
  83                                           columns[i],
  84                                           current_time,
  85                                           modification_time)) {
  86       *size = columns[i - 3];
  87       *offset = i;
  88       return true;
  89     }
  90   }
  91
  92   // Some FTP listings have swapped the "month" and "day of month" columns
  93   // (for example Russian listings). We try to recognize them only after making
  94   // sure no column offset works above (this is a more strict way).
  95   for (size_t i = 5U; i < columns.size(); i++) {
  96     if (net::FtpUtil::LsDateListingToTime(columns[i - 1],
  97                                           columns[i - 2],
  98                                           columns[i],
  99                                           current_time,
 100                                           modification_time)) {
 101       *size = columns[i - 3];
 102       *offset = i;
 103       return true;
 104     }
 105   }
 106
 107   // Some FTP listings use a different date format.
 108   for (size_t i = 5U; i < columns.size(); i++) {
 109     if (TwoColumnDateListingToTime(columns[i - 1],
 110                                    columns[i],
 111                                    modification_time)) {
 112       *size = columns[i - 2];
 113       *offset = i;
 114       return true;
 115     }
 116   }
 117
 118   return false;
 119 }
 120
 121 }  // namespace
 122
 123 namespace net {
 124
 125 bool ParseFtpDirectoryListingLs(
 126     const std::vector<base::string16>& lines,
 127     const base::Time& current_time,
 128     std::vector<FtpDirectoryListingEntry>* entries) {
 129   // True after we have received a "total n" listing header, where n is an
 130   // integer. Only one such header is allowed per listing.
 131   bool received_total_line = false;
 132
 133   for (size_t i = 0; i < lines.size(); i++) {
 134     if (lines[i].empty())
 135       continue;
 136
 137     std::vector<base::string16> columns;
 138     base::SplitString(base::CollapseWhitespace(lines[i], false), ' ', &columns);
 139
 140     // Some FTP servers put a "total n" line at the beginning of the listing
 141     // (n is an integer). Allow such a line, but only once, and only if it's
 142     // the first non-empty line. Do not match the word exactly, because it may
 143     // be in different languages (at least English and German have been seen
 144     // in the field).
 145     if (columns.size() == 2 && !received_total_line) {
 146       received_total_line = true;
 147
 148       int64 total_number;
 149       if (!base::StringToInt64(columns[1], &total_number))
 150         return false;
 151       if (total_number < 0)
 152         return false;
 153
 154       continue;
 155     }
 156
 157     FtpDirectoryListingEntry entry;
 158
 159     size_t column_offset;
 160     base::string16 size;
 161     if (!DetectColumnOffsetSizeAndModificationTime(columns,
 162                                                    current_time,
 163                                                    &column_offset,
 164                                                    &size,
 165                                                    &entry.last_modified)) {
 166       // Some servers send a message in one of the first few lines.
 167       // All those messages have in common is the string ".:",
 168       // where "." means the current directory, and ":" separates it
 169       // from the rest of the message, which may be empty.
 170       if (lines[i].find(base::ASCIIToUTF16(".:")) != base::string16::npos)
 171         continue;
 172
 173       return false;
 174     }
 175
 176     // Do not check "validity" of the permission listing. It's quirky,
 177     // and some servers send garbage here while other parts of the line are OK.
 178
 179     if (!columns[0].empty() && columns[0][0] == 'l') {
 180       entry.type = FtpDirectoryListingEntry::SYMLINK;
 181     } else if (!columns[0].empty() && columns[0][0] == 'd') {
 182       entry.type = FtpDirectoryListingEntry::DIRECTORY;
 183     } else {
 184       entry.type = FtpDirectoryListingEntry::FILE;
 185     }
 186
 187     if (!base::StringToInt64(size, &entry.size)) {
 188       // Some FTP servers do not separate owning group name from file size,
 189       // like "group1234". We still want to display the file name for that
 190       // entry, but can't really get the size (What if the group is named
 191       // "group1", and the size is in fact 234? We can't distinguish between
 192       // that and "group" with size 1234). Use a dummy value for the size.
 193       // TODO(phajdan.jr): Use a value that means "unknown" instead of 0 bytes.
 194       entry.size = 0;
 195     }
 196     if (entry.size < 0) {
 197       // Some FTP servers have bugs that cause them to display the file size
 198       // as negative. They're most likely big files like DVD ISO images.
 199       // We still want to display them, so just say the real file size
 200       // is unknown.
 201       entry.size = -1;
 202     }
 203     if (entry.type != FtpDirectoryListingEntry::FILE)
 204       entry.size = -1;
 205
 206     if (column_offset == columns.size() - 1) {
 207       // If the end of the date listing is the last column, there is no file
 208       // name. Some FTP servers send listing entries with empty names.
 209       // It's not obvious how to display such an entry, so we ignore them.
 210       // We don't want to make the parsing fail at this point though.
 211       // Other entries can still be useful.
 212       continue;
 213     }
 214
 215     entry.name = FtpUtil::GetStringPartAfterColumns(lines[i],
 216                                                     column_offset + 1);
 217
 218     if (entry.type == FtpDirectoryListingEntry::SYMLINK) {
 219       base::string16::size_type pos =
 220           entry.name.rfind(base::ASCIIToUTF16(" -> "));
 221
 222       // We don't require the " -> " to be present. Some FTP servers don't send
 223       // the symlink target, possibly for security reasons.
 224       if (pos != base::string16::npos)
 225         entry.name = entry.name.substr(0, pos);
 226     }
 227
 228     entries->push_back(entry);
 229   }
 230
 231   return true;
 232 }
 233
 234 }  // namespace net