nucleus/library/filesystem/huge_file.cpp

   1 /*****************************************************************************\
   2 *                                                                             *
   3 *  Name   : huge_file                                                         *
   4 *  Author : Chris Koeritz                                                     *
   5 *                                                                             *
   6 *******************************************************************************
   7 * Copyright (c) 2007-$now By Author.  This program is free software; you can  *
   8 * redistribute it and/or modify it under the terms of the GNU General Public  *
   9 * License as published by the Free Software Foundation; either version 2 of   *
  10 * the License or (at your option) any later version.  This is online at:      *
  11 *     http://www.fsf.org/copyleft/gpl.html                                    *
  12 * Please send any updates to: fred@gruntose.com                               *
  13 \*****************************************************************************/
  14
  15 #include "byte_filer.h"
  16 #include "huge_file.h"
  17
  18 #include <basis/byte_array.h>
  19 #include <basis/functions.h>
  20 #include <basis/guards.h>
  21
  22 #include <stdio.h>
  23 #ifndef __WIN32__
  24   #include <sys/time.h>
  25 #else
  26   #include <time.h>
  27 #endif
  28
  29 #undef LOG
  30 #define LOG(to_print) printf("%s::%s: %s\n", static_class_name(), func, astring(to_print).s())
  31
  32 //#define DEBUG_HUGE_FILE
  33   // uncomment for noisy version.
  34
  35 using namespace basis;
  36
  37 namespace filesystem {
  38
  39 huge_file::huge_file(const astring &filename, const astring &permissions)
  40 : _real_file(new byte_filer(filename, permissions)),
  41   _file_pointer(0)
  42 {
  43 }
  44
  45 huge_file::~huge_file()
  46 {
  47   WHACK(_real_file);
  48 }
  49
  50 const astring &huge_file::name() const { return _real_file->name(); }
  51
  52 void huge_file::flush() { _real_file->flush(); }
  53
  54 bool huge_file::truncate() { return _real_file->truncate(); }
  55
  56 double huge_file::length()
  57 {
  58   FUNCDEF("length");
  59
  60 //trying to read to see if we're past endpoint.
  61 //  if this approach works, length may want to close and reopen file for
  62 //  reading, since we can't add any bytes to it for writing just to find
  63 //  the length out.
  64
  65
  66   double save_posn = _file_pointer;
  67   // skip to the beginning of the file so we can try to find the end.
  68   _file_pointer = 0;
  69   _real_file->seek(0, byte_filer::FROM_START);
  70   size_t naive_size = _real_file->length();
  71   if (naive_size < _real_file->file_size_limit()) {
  72     // lucked out; we are within normal file size limitations.
  73     seek(save_posn, byte_filer::FROM_START);
  74     return double(naive_size);
  75   }
  76
  77   double best_highest = 0.0;  // the maximum we've safely seeked to.
  78
  79   size_t big_jump = byte_filer::file_size_limit();
  80     // try with the largest possible seek at first.
  81
  82   while (true) {
  83 #ifdef DEBUG_HUGE_FILE
  84     LOG(a_sprintf("best highest=%.0f", best_highest));
  85 #endif
  86     // iterate until we reach our exit condition, which seems like it must
  87     // always occur eventually unless the file is being monkeyed with.
  88     bool seek_ret = _real_file->seek(int(big_jump), byte_filer::FROM_CURRENT);
  89 #ifdef DEBUG_HUGE_FILE
  90     LOG(a_sprintf("  seek ret=%d", int(seek_ret)));
  91 #endif
  92     byte_array temp_bytes;
  93     int bytes_read = _real_file->read(temp_bytes, 1);
  94     if (bytes_read < 1)
  95       seek_ret = false;
  96 #ifdef DEBUG_HUGE_FILE
  97     LOG(a_sprintf("  read %d bytes", bytes_read));
  98 #endif
  99     bool at_eof = _real_file->eof();
 100 #ifdef DEBUG_HUGE_FILE
 101     LOG(a_sprintf("  at_eof=%d", int(at_eof)));
 102 #endif
 103     if (seek_ret && !at_eof) {
 104 #ifdef DEBUG_HUGE_FILE
 105       LOG("seek worked, incrementing best highest and trying same jump again");
 106 #endif
 107       // the seek worked, so we'll just jump forward again.
 108       best_highest += double(big_jump);
 109       _file_pointer += double(big_jump);
 110       continue;
 111     } else if (seek_ret && at_eof) {
 112 #ifdef DEBUG_HUGE_FILE
 113       LOG("seek worked but found eof exactly.");
 114 #endif
 115       // the seek did worked, but apparently we've also found the end point.
 116       best_highest += double(big_jump);
 117       _file_pointer += double(big_jump);
 118       break;
 119     } else {
 120       // that seek was too large, so we need to back down and try a smaller
 121       // seek size.
 122 #ifdef DEBUG_HUGE_FILE
 123       LOG("seek failed, going back to best highest and trying same jump again");
 124 #endif
 125       _file_pointer = 0;
 126       _real_file->seek(0, byte_filer::FROM_START);
 127       outcome worked = seek(best_highest, byte_filer::FROM_START);
 128         // this uses our version to position at large sizes.
 129       if (worked != OKAY) {
 130         // this is a bad failure; it says that the file size changed or
 131         // something malfunctioned.  we should always be able to get back to
 132         // the last good size we found if the file is static.
 133         LOG(a_sprintf("failed to seek back to best highest %.0f on ",
 134             best_highest) + _real_file->name());
 135         // try to repair our ideas about the file by starting the process
 136         // over.
 137 //hmmm: count the number of times restarted and bail after N.
 138         seek_ret = _real_file->seek(0, byte_filer::FROM_START);
 139         _file_pointer = 0;
 140         if (!seek_ret) {
 141           // the heck with this.  we can't even go back to the start.  this
 142           // file seems to be screwed up now.
 143           LOG(astring("failed to seek back to start of file!  on ")
 144               + _real_file->name());
 145           return 0;
 146         }
 147         // reset the rest of the positions for our failed attempt to return
 148         // to what we already thought was good.
 149         _file_pointer = 0;
 150         big_jump = byte_filer::file_size_limit();
 151         best_highest = 0;
 152         continue;
 153       }
 154       // okay, nothing bad happened when we went back to our last good point.
 155       if (big_jump <= 0) {
 156         // success in finding the smallest place that we can't seek between.
 157 #ifdef DEBUG_HUGE_FILE
 158         LOG("got down to smallest big jump, 0!");
 159 #endif
 160         break;
 161       }
 162       // formula expects that the maximum file size is a power of 2.
 163       big_jump /= 2;
 164 #ifdef DEBUG_HUGE_FILE
 165       LOG(a_sprintf("restraining big jump down to %u.", big_jump));
 166 #endif
 167       continue;
 168     }
 169   }
 170
 171   // go back to where we started out.
 172   seek(0, byte_filer::FROM_START);
 173   seek(save_posn, byte_filer::FROM_CURRENT);
 174 #ifdef DEBUG_HUGE_FILE
 175   LOG(a_sprintf("saying file len is %.0f.", best_highest + 1.0));
 176 #endif
 177   return best_highest + 1.0;
 178 }
 179
 180 bool huge_file::good() const { return _real_file->good(); }
 181
 182 bool huge_file::eof() const { return _real_file->eof(); }
 183
 184 outcome huge_file::move_to(double absolute_posn)
 185 {
 186 #ifdef DEBUG_HUGE_FILE
 187   FUNCDEF("move_to");
 188 #endif
 189   double difference = absolute_posn - _file_pointer;
 190     // calculate the size we want to offset.
 191 #ifdef DEBUG_HUGE_FILE
 192   LOG(a_sprintf("abs_pos=%.0f difference=%.0f old_filepoint=%.0f",
 193       absolute_posn, difference, _file_pointer));
 194 #endif
 195   // if we're at the same place, we don't have to do anything.
 196   if (difference < 0.000001) {
 197 #ifdef DEBUG_HUGE_FILE
 198     LOG("difference was minimal, saying we're done.");
 199 #endif
 200     return OKAY;
 201   }
 202   while (absolute_value(difference) > 0.000001) {
 203     double seek_size = minimum(double(byte_filer::file_size_limit() - 1),
 204         absolute_value(difference));
 205     if (difference < 0)
 206       seek_size *= -1.0;  // flip sign of seek.
 207 #ifdef DEBUG_HUGE_FILE
 208     LOG(a_sprintf("  seeksize=%d", int(seek_size)));
 209 #endif
 210     bool seek_ret = _real_file->seek(int(seek_size),
 211         byte_filer::FROM_CURRENT);
 212     if (!seek_ret) {
 213 #ifdef DEBUG_HUGE_FILE
 214       LOG(a_sprintf("failed to seek %d from current", int(seek_size)));
 215 #endif
 216       return FAILURE;  // seek failed somehow.
 217     }
 218     _file_pointer += seek_size;
 219 #ifdef DEBUG_HUGE_FILE
 220     LOG(a_sprintf("  now_filepoint=%.0f", _file_pointer));
 221 #endif
 222     difference = absolute_posn - _file_pointer;
 223 #ifdef DEBUG_HUGE_FILE
 224     LOG(a_sprintf("  now_difference=%.0f", difference));
 225 #endif
 226   }
 227   return OKAY;
 228 }
 229
 230 outcome huge_file::seek(double new_position, byte_filer::origins origin)
 231 {
 232 #ifdef DEBUG_HUGE_FILE
 233   FUNCDEF("seek");
 234 #endif
 235   if (origin == byte_filer::FROM_CURRENT) {
 236     return move_to(_file_pointer + new_position);
 237   } else if (origin == byte_filer::FROM_START) {
 238     _file_pointer = 0;
 239     if (!_real_file->seek(0, byte_filer::FROM_START))
 240       return FAILURE;
 241     return move_to(new_position);
 242   } else if (origin == byte_filer::FROM_END) {
 243 #ifdef DEBUG_HUGE_FILE
 244     LOG("into precarious FROM_END case.");
 245 #endif
 246     double file_len = length();  // could take a scary long time possibly.
 247 #ifdef DEBUG_HUGE_FILE
 248     LOG(a_sprintf("  FROM_END got len %.0f.", file_len));
 249 #endif
 250     _file_pointer = file_len;
 251       // it's safe, although not efficient, for us to call the length()
 252       // method here.  our current version of length() uses the byte_filer's
 253       // seek method directly and only FROM_CURRENT and FROM_START from this
 254       // class's seek method.
 255     _real_file->seek(0, byte_filer::FROM_END);
 256     return move_to(_file_pointer - new_position);
 257   }
 258   // unknown origin.
 259   return BAD_INPUT;
 260 }
 261
 262 outcome huge_file::read(byte_array &to_fill, int desired_size, int &size_read)
 263 {
 264   FUNCDEF("read");
 265   size_read = 0;
 266   int ret = _real_file->read(to_fill, desired_size);
 267   if (ret < 0)
 268     return FAILURE;  // couldn't read the bytes.
 269   _file_pointer += double(size_read);
 270   size_read = ret;
 271   return OKAY;
 272 }
 273
 274 outcome huge_file::write(const byte_array &to_write, int &size_written)
 275 {
 276   FUNCDEF("write");
 277   size_written = 0;
 278   int ret = _real_file->write(to_write);
 279   if (ret < 0)
 280     return FAILURE;  // couldn't write the bytes.
 281   _file_pointer += double(size_written);
 282   size_written = ret;
 283   return OKAY;
 284 }
 285
 286 basis::outcome huge_file::touch()
 287 {
 288   FUNCDEF("touch")
 289   if (filename(_real_file->name()).exists()) {
 290     // file exists, so just update time.
 291 #ifndef __WIN32__
 292     int ret = utimes(_real_file->name().observe(), NULL_POINTER);
 293     if (ret != 0)
 294       return FAILURE;
 295 #else
 296     // open the file, although the function says create in its name...
 297     HANDLE f = CreateFile(_real_file->name().observe(),
 298         GENERIC_READ | GENERIC_WRITE, FILE_SHARE_READ | FILE_SHARE_WRITE,
 299         NULL_POINTER, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL_POINTER);
 300     if (!f) {
 301       LOG(a_sprintf("failed to open file %s", _real_file->name().observe()));
 302       return FAILURE;
 303     }
 304     // get current system time in UTC.
 305     SYSTEMTIME *st = new SYSTEMTIME;
 306     GetSystemTime(st);
 307     // convert system time into file time.
 308     FILETIME *t = new FILETIME;
 309     SystemTimeToFileTime(st, t);
 310     // set the file's time.
 311     SetFileTime(f, NULL_POINTER, t, t);
 312 #endif
 313   } else {
 314     // file doesn't exist yet.
 315     byte_array junk(1);
 316     int written;
 317     outcome ret = write(junk, written);
 318     if (ret != OKAY) ret;
 319     if (!truncate())
 320       return FAILURE;
 321   }
 322   return OKAY;
 323 }
 324
 325 } //namespace.
 326