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