net/ftp/ftp_directory_listing_parser_vms.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_vms.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 // Converts the filename component in listing to the filename we can display.
  22 // Returns true on success.
  23 bool ParseVmsFilename(const base::string16& raw_filename,
  24                       base::string16* parsed_filename,
  25                       FtpDirectoryListingEntry::Type* type) {
  26   // On VMS, the files and directories are versioned. The version number is
  27   // separated from the file name by a semicolon. Example: ANNOUNCE.TXT;2.
  28   std::vector<base::string16> listing_parts;
  29   base::SplitString(raw_filename, ';', &listing_parts);
  30   if (listing_parts.size() != 2)
  31     return false;
  32   int version_number;
  33   if (!base::StringToInt(listing_parts[1], &version_number))
  34     return false;
  35   if (version_number < 0)
  36     return false;
  37
  38   // Even directories have extensions in the listings. Don't display extensions
  39   // for directories; it's awkward for non-VMS users. Also, VMS is
  40   // case-insensitive, but generally uses uppercase characters. This may look
  41   // awkward, so we convert them to lower case.
  42   std::vector<base::string16> filename_parts;
  43   base::SplitString(listing_parts[0], '.', &filename_parts);
  44   if (filename_parts.size() != 2)
  45     return false;
  46   if (EqualsASCII(filename_parts[1], "DIR")) {
  47     *parsed_filename = base::StringToLowerASCII(filename_parts[0]);
  48     *type = FtpDirectoryListingEntry::DIRECTORY;
  49   } else {
  50     *parsed_filename = base::StringToLowerASCII(listing_parts[0]);
  51     *type = FtpDirectoryListingEntry::FILE;
  52   }
  53   return true;
  54 }
  55
  56 bool ParseVmsFilesize(const base::string16& input, int64* size) {
  57   if (base::ContainsOnlyChars(input, base::ASCIIToUTF16("*"))) {
  58     // Response consisting of asterisks means unknown size.
  59     *size = -1;
  60     return true;
  61   }
  62
  63   // VMS's directory listing gives us file size in blocks. We assume that
  64   // the block size is 512 bytes. It doesn't give accurate file size, but is the
  65   // best information we have.
  66   const int kBlockSize = 512;
  67
  68   if (base::StringToInt64(input, size)) {
  69     if (*size < 0)
  70       return false;
  71     *size *= kBlockSize;
  72     return true;
  73   }
  74
  75   std::vector<base::string16> parts;
  76   base::SplitString(input, '/', &parts);
  77   if (parts.size() != 2)
  78     return false;
  79
  80   int64 blocks_used, blocks_allocated;
  81   if (!base::StringToInt64(parts[0], &blocks_used))
  82     return false;
  83   if (!base::StringToInt64(parts[1], &blocks_allocated))
  84     return false;
  85   if (blocks_used > blocks_allocated)
  86     return false;
  87   if (blocks_used < 0 || blocks_allocated < 0)
  88     return false;
  89
  90   *size = blocks_used * kBlockSize;
  91   return true;
  92 }
  93
  94 bool LooksLikeVmsFileProtectionListingPart(const base::string16& input) {
  95   if (input.length() > 4)
  96     return false;
  97
  98   // On VMS there are four different permission bits: Read, Write, Execute,
  99   // and Delete. They appear in that order in the permission listing.
 100   std::string pattern("RWED");
 101   base::string16 match(input);
 102   while (!match.empty() && !pattern.empty()) {
 103     if (match[0] == pattern[0])
 104       match = match.substr(1);
 105     pattern = pattern.substr(1);
 106   }
 107   return match.empty();
 108 }
 109
 110 bool LooksLikeVmsFileProtectionListing(const base::string16& input) {
 111   if (input.length() < 2)
 112     return false;
 113   if (input[0] != '(' || input[input.length() - 1] != ')')
 114     return false;
 115
 116   // We expect four parts of the file protection listing: for System, Owner,
 117   // Group, and World.
 118   std::vector<base::string16> parts;
 119   base::SplitString(input.substr(1, input.length() - 2), ',', &parts);
 120   if (parts.size() != 4)
 121     return false;
 122
 123   return LooksLikeVmsFileProtectionListingPart(parts[0]) &&
 124       LooksLikeVmsFileProtectionListingPart(parts[1]) &&
 125       LooksLikeVmsFileProtectionListingPart(parts[2]) &&
 126       LooksLikeVmsFileProtectionListingPart(parts[3]);
 127 }
 128
 129 bool LooksLikeVmsUserIdentificationCode(const base::string16& input) {
 130   if (input.length() < 2)
 131     return false;
 132   return input[0] == '[' && input[input.length() - 1] == ']';
 133 }
 134
 135 bool LooksLikeVMSError(const base::string16& text) {
 136   static const char* kPermissionDeniedMessages[] = {
 137     "%RMS-E-FNF",  // File not found.
 138     "%RMS-E-PRV",  // Access denied.
 139     "%SYSTEM-F-NOPRIV",
 140     "privilege",
 141   };
 142
 143   for (size_t i = 0; i < arraysize(kPermissionDeniedMessages); i++) {
 144     if (text.find(base::ASCIIToUTF16(kPermissionDeniedMessages[i])) !=
 145         base::string16::npos)
 146       return true;
 147   }
 148
 149   return false;
 150 }
 151
 152 bool VmsDateListingToTime(const std::vector<base::string16>& columns,
 153                           base::Time* time) {
 154   DCHECK_EQ(4U, columns.size());
 155
 156   base::Time::Exploded time_exploded = { 0 };
 157
 158   // Date should be in format DD-MMM-YYYY.
 159   std::vector<base::string16> date_parts;
 160   base::SplitString(columns[2], '-', &date_parts);
 161   if (date_parts.size() != 3)
 162     return false;
 163   if (!base::StringToInt(date_parts[0], &time_exploded.day_of_month))
 164     return false;
 165   if (!FtpUtil::AbbreviatedMonthToNumber(date_parts[1],
 166                                          &time_exploded.month))
 167     return false;
 168   if (!base::StringToInt(date_parts[2], &time_exploded.year))
 169     return false;
 170
 171   // Time can be in format HH:MM, HH:MM:SS, or HH:MM:SS.mm. Try to recognize the
 172   // last type first. Do not parse the seconds, they will be ignored anyway.
 173   base::string16 time_column(columns[3]);
 174   if (time_column.length() == 11 && time_column[8] == '.')
 175     time_column = time_column.substr(0, 8);
 176   if (time_column.length() == 8 && time_column[5] == ':')
 177     time_column = time_column.substr(0, 5);
 178   if (time_column.length() != 5)
 179     return false;
 180   std::vector<base::string16> time_parts;
 181   base::SplitString(time_column, ':', &time_parts);
 182   if (time_parts.size() != 2)
 183     return false;
 184   if (!base::StringToInt(time_parts[0], &time_exploded.hour))
 185     return false;
 186   if (!base::StringToInt(time_parts[1], &time_exploded.minute))
 187     return false;
 188
 189   // We don't know the time zone of the server, so just use local time.
 190   *time = base::Time::FromLocalExploded(time_exploded);
 191   return true;
 192 }
 193
 194 }  // namespace
 195
 196 bool ParseFtpDirectoryListingVms(
 197     const std::vector<base::string16>& lines,
 198     std::vector<FtpDirectoryListingEntry>* entries) {
 199   // The first non-empty line is the listing header. It often
 200   // starts with "Directory ", but not always. We set a flag after
 201   // seing the header.
 202   bool seen_header = false;
 203
 204   // Sometimes the listing doesn't end with a "Total" line, but
 205   // it's only okay when it contains some errors (it's needed
 206   // to distinguish it from "ls -l" format).
 207   bool seen_error = false;
 208
 209   for (size_t i = 0; i < lines.size(); i++) {
 210     if (lines[i].empty())
 211       continue;
 212
 213     if (StartsWith(lines[i], base::ASCIIToUTF16("Total of "), true)) {
 214       // After the "total" line, all following lines must be empty.
 215       for (size_t j = i + 1; j < lines.size(); j++)
 216         if (!lines[j].empty())
 217           return false;
 218
 219       return true;
 220     }
 221
 222     if (!seen_header) {
 223       seen_header = true;
 224       continue;
 225     }
 226
 227     if (LooksLikeVMSError(lines[i])) {
 228       seen_error = true;
 229       continue;
 230     }
 231
 232     std::vector<base::string16> columns;
 233     base::SplitString(base::CollapseWhitespace(lines[i], false), ' ', &columns);
 234
 235     if (columns.size() == 1) {
 236       // There can be no continuation if the current line is the last one.
 237       if (i == lines.size() - 1)
 238         return false;
 239
 240       // Skip the next line.
 241       i++;
 242
 243       // This refers to the continuation line.
 244       if (LooksLikeVMSError(lines[i])) {
 245         seen_error = true;
 246         continue;
 247       }
 248
 249       // Join the current and next line and split them into columns.
 250       base::SplitString(
 251           base::CollapseWhitespace(
 252               lines[i - 1] + base::ASCIIToUTF16(" ") + lines[i], false),
 253           ' ',
 254           &columns);
 255     }
 256
 257     FtpDirectoryListingEntry entry;
 258     if (!ParseVmsFilename(columns[0], &entry.name, &entry.type))
 259       return false;
 260
 261     // There are different variants of a VMS listing. Some display
 262     // the protection listing and user identification code, some do not.
 263     if (columns.size() == 6) {
 264       if (!LooksLikeVmsFileProtectionListing(columns[5]))
 265         return false;
 266       if (!LooksLikeVmsUserIdentificationCode(columns[4]))
 267         return false;
 268
 269       // Drop the unneeded data, so that the following code can always expect
 270       // just four columns.
 271       columns.resize(4);
 272     }
 273
 274     if (columns.size() != 4)
 275       return false;
 276
 277     if (!ParseVmsFilesize(columns[1], &entry.size))
 278       return false;
 279     if (entry.type != FtpDirectoryListingEntry::FILE)
 280       entry.size = -1;
 281     if (!VmsDateListingToTime(columns, &entry.last_modified))
 282       return false;
 283
 284     entries->push_back(entry);
 285   }
 286
 287   // The only place where we return true is after receiving the "Total" line,
 288   // that should be present in every VMS listing. Alternatively, if the listing
 289   // contains error messages, it's OK not to have the "Total" line.
 290   return seen_error;
 291 }
 292
 293 }  // namespace net