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