nucleus/library/filesystem/heavy_file_ops.cpp

   1 /*****************************************************************************\
   2 *                                                                             *
   3 *  Name   : heavy file operations                                             *
   4 *  Author : Chris Koeritz                                                     *
   5 *                                                                             *
   6 *******************************************************************************
   7 * Copyright (c) 2005-$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 "directory.h"
  16 #include "filename.h"
  17 #include "filename_list.h"
  18 #include "heavy_file_ops.h"
  19 #include "huge_file.h"
  20
  21 #include <basis/functions.h>
  22 #include <basis/guards.h>
  23 #include <structures/object_packers.h>
  24
  25 using namespace basis;
  26 using namespace structures;
  27
  28 namespace filesystem {
  29
  30 #define DEBUG_HEAVY_FILE_OPS
  31   // uncomment for noisier debugging.
  32
  33 #undef LOG
  34 #ifdef DEBUG_HEAVY_FILE_OPS
  35   #include <stdio.h>
  36   #define LOG(to_print) printf("%s::%s: %s\n", static_class_name(), func, astring(to_print).s())
  37 #else
  38   #define LOG(s) {if(!!s){}}
  39 #endif
  40
  41 //////////////
  42
  43 // the smallest we let the packing area's available space get before we stop filling it.
  44 const int MINIMUM_ARRAY_SIZE = 1024;
  45
  46 //////////////
  47
  48 file_transfer_header::file_transfer_header(const file_time &time_stamp)
  49 : _filename(),
  50   _byte_start(0),
  51   _length(0),
  52   _time(time_stamp)
  53 {
  54 }
  55
  56 astring file_transfer_header::text_form() const
  57 {
  58   astring time_text;
  59   _time.text_form(time_text);
  60   return astring("file=") + _filename
  61       + a_sprintf(" start=%d len=%d stamp=", _byte_start, _length)
  62       + time_text;
  63 }
  64
  65 astring file_transfer_header::readable_text_form() const
  66 {
  67   astring time_text;
  68   _time.readable_text_form(time_text);
  69   return _filename
  70       + a_sprintf(" [%d bytes, mod ", _length)
  71       + time_text + "]";
  72 }
  73
  74 void file_transfer_header::pack(byte_array &packed_form) const
  75 {
  76   _filename.pack(packed_form);
  77   attach(packed_form, _byte_start);
  78   attach(packed_form, _length);
  79   _time.pack(packed_form);
  80 }
  81
  82 bool file_transfer_header::unpack(byte_array &packed_form)
  83 {
  84   if (!_filename.unpack(packed_form)) return false;
  85   if (!detach(packed_form, _byte_start)) return false;
  86   if (!detach(packed_form, _length)) return false;
  87   if (!_time.unpack(packed_form)) return false;
  88   return true;
  89 }
  90
  91 int file_transfer_header::packed_size() const
  92 {
  93 byte_array temp;
  94 attach(temp, _byte_start);
  95 //hmmm: really ugly above; we should get a more exact way to know the size of
  96 //      packed doubles.
  97   return _filename.length() + 1
  98       + temp.length()
  99       + sizeof(int)
 100       + _time.packed_size();
 101 }
 102
 103 //////////////
 104
 105 const size_t heavy_file_operations::COPY_CHUNK_FACTOR = 1 * MEGABYTE;
 106
 107 size_t heavy_file_operations::copy_chunk_factor()
 108 { return COPY_CHUNK_FACTOR; }
 109
 110 heavy_file_operations::~heavy_file_operations() {}
 111   // we only need this due to our use of the root_object class_name support.
 112
 113 const char *heavy_file_operations::outcome_name(const outcome &to_name)
 114 {
 115   switch (to_name.value()) {
 116     case SOURCE_MISSING: return "SOURCE_MISSING";
 117     case TARGET_ACCESS_ERROR: return "TARGET_ACCESS_ERROR";
 118     case TARGET_DIR_ERROR: return "TARGET_DIR_ERROR";
 119     default: return common::outcome_name(to_name);
 120   }
 121 }
 122
 123 outcome heavy_file_operations::copy_file(const astring &source,
 124     const astring &destination, int copy_chunk_factor)
 125 {
 126 #ifdef DEBUG_HEAVY_FILE_OPS
 127   FUNCDEF("copy_file");
 128 #endif
 129   // check that the source exists...
 130   filename source_path(source);
 131   if (!source_path.exists()) return SOURCE_MISSING;
 132   file_time source_time(source_path);  // get the time on the source.
 133
 134   // make sure the target directory exists...
 135   filename target_path(destination);
 136   filename targ_dir = target_path.dirname();
 137   if (!directory::recursive_create(targ_dir.raw())) return TARGET_DIR_ERROR;
 138
 139   // open the source for reading.
 140   huge_file source_file(source, "rb");
 141   if (!source_file.good()) return SOURCE_MISSING;
 142 //hmmm: could be source is not accessible instead.
 143
 144   // open target file for writing.
 145   huge_file target_file(destination, "wb");
 146   if (!target_file.good()) return TARGET_ACCESS_ERROR;
 147
 148   byte_array chunk;
 149   int bytes_read = 0;
 150   outcome ret;
 151   while ( (ret = source_file.read(chunk, copy_chunk_factor, bytes_read))
 152       == huge_file::OKAY) {
 153     int bytes_stored;
 154     ret = target_file.write(chunk, bytes_stored);
 155     if (bytes_stored != bytes_read) return TARGET_ACCESS_ERROR;
 156     if (source_file.eof()) break;  // time to escape.
 157   }
 158
 159   // set the time on the target file from the source's time.
 160   source_time.set_time(target_path);
 161
 162 #ifdef DEBUG_HEAVY_FILE_OPS
 163   astring time;
 164   source_time.text_form(time);
 165   LOG(astring("setting file time for ") + source + " to " + time);
 166 #endif
 167
 168   return OKAY;
 169 }
 170
 171 outcome heavy_file_operations::write_file_chunk(const astring &target,
 172     double byte_start, const byte_array &chunk, bool truncate,
 173     int copy_chunk_factor)
 174 {
 175   FUNCDEF("write_file_chunk");
 176   if (byte_start < 0) return BAD_INPUT;
 177
 178   filename targ_name(target);
 179   if (!directory::recursive_create(targ_name.dirname().raw()))
 180     return TARGET_DIR_ERROR;
 181
 182   if (!targ_name.exists()) {
 183     huge_file target_file(target, "w");
 184   }
 185
 186   huge_file target_file(target, "r+b");
 187     // open the file for updating (either read or write).
 188   if (!target_file.good()) return TARGET_ACCESS_ERROR;
 189   double curr_len = target_file.length();
 190
 191   if (curr_len < byte_start) {
 192     byte_array new_chunk;
 193     while (curr_len < byte_start) {
 194       target_file.seek(0, byte_filer::FROM_END);  // go to the end of the file.
 195       new_chunk.reset(minimum(copy_chunk_factor,
 196           int(curr_len - byte_start + 1)));
 197       int written;
 198       outcome ret = target_file.write(new_chunk, written);
 199       if (written < new_chunk.length()) return TARGET_ACCESS_ERROR;
 200       curr_len = target_file.length();
 201     }
 202   }
 203   target_file.seek(byte_start, byte_filer::FROM_START);
 204     // jump to the proper location in the file.
 205   int wrote;
 206   outcome ret = target_file.write(chunk, wrote);
 207   if (wrote != chunk.length()) return TARGET_ACCESS_ERROR;
 208   if (truncate) {
 209     target_file.truncate();
 210   }
 211   return OKAY;
 212 }
 213
 214 basis::outcome heavy_file_operations::advance(const filename_list &to_transfer,
 215     file_transfer_header &last_action)
 216 {
 217   FUNCDEF("advance");
 218   int indy = to_transfer.locate(last_action._filename);
 219   if (negative(indy)) return BAD_INPUT;  // error, file not found in list.
 220   if (indy >= to_transfer.elements() - 1) return FINISHED;  // done.
 221   const file_info *currfile = to_transfer.get(indy + 1);
 222   last_action._filename = currfile->raw();
 223   last_action._time = currfile->_time;
 224
 225 #ifdef DEBUG_HEAVY_FILE_OPS
 226   if (currfile->_time == file_time(time_t(0)))
 227     LOG(astring("failed for ") + currfile->raw() + " -- has zero file time");
 228 #endif
 229
 230   last_action._byte_start = 0;
 231   last_action._length = 0;
 232   return OKAY;
 233 }
 234
 235 outcome heavy_file_operations::buffer_files(const astring &source_root,
 236     const filename_list &to_transfer, file_transfer_header &last_action,
 237     byte_array &storage, int maximum_bytes)
 238 {
 239   FUNCDEF("buffer_files");
 240   storage.reset();  // clear out the current contents.
 241
 242   if (!to_transfer.elements()) {
 243     // we seem to be done.
 244     return FINISHED;
 245   }
 246
 247   outcome to_return = OKAY;
 248
 249   // start filling the array with bytes from the files.
 250   while (storage.length() < maximum_bytes) {
 251     double remaining_in_array = maximum_bytes - storage.length()
 252         - last_action.packed_size();
 253     if (remaining_in_array < MINIMUM_ARRAY_SIZE) {
 254       // ensure that we at least have a reasonable amount of space left
 255       // for storing into the array.
 256       break;
 257     }
 258
 259     // find the current file we're at, as provided in record.
 260     if (!last_action._filename) {
 261       // no filename yet.  assume this is the first thing we've done.
 262       const file_info *currfile = to_transfer.get(0);
 263       last_action._filename = currfile->raw();
 264       last_action._time = currfile->_time;
 265       last_action._byte_start = 0;
 266       last_action._length = 0;
 267     }
 268
 269     const file_info *found = to_transfer.find(last_action._filename);
 270     if (!found) {
 271       // they have referenced a file that we don't have.  that's bad news.
 272       return BAD_INPUT;
 273     }
 274
 275     astring full_file = source_root + "/" + last_action._filename;
 276     huge_file current(full_file, "rb");
 277     if (!current.good()) {
 278       // we need to skip this file.
 279       LOG(astring("skipping bad file: ") + full_file);
 280       to_return = advance(to_transfer, last_action);
 281       if (to_return != OKAY) break;
 282       continue;
 283     }
 284
 285     if (last_action._byte_start + last_action._length >= current.length()) {
 286       // this file is done now.  go to the next one.
 287       LOG(astring("finished stuffing file: ") + full_file);
 288       to_return = advance(to_transfer, last_action);
 289       if (to_return != OKAY) break;
 290       continue;
 291     }
 292
 293     // calculate the largest piece remaining of that file that will fit in the
 294     // allotted space.
 295     double new_start = last_action._byte_start + last_action._length;
 296     double remaining_in_file = current.length() - new_start;
 297     if (remaining_in_file < 0) remaining_in_file = 0;
 298     double new_len = minimum(remaining_in_file, remaining_in_array);
 299
 300     // pack this new piece of the file.
 301     current.seek(new_start, byte_filer::FROM_START);
 302     byte_array new_chunk;
 303     int bytes_read = 0;
 304     outcome ret = current.read(new_chunk, int(new_len), bytes_read);
 305     if (bytes_read != new_len) {
 306       if (!bytes_read) {
 307         // some kind of problem reading the file.
 308         to_return = advance(to_transfer, last_action);
 309         if (to_return != OKAY) break;
 310         continue;
 311       }
 312 //why would this happen?  just complain, i guess.
 313     }
 314
 315     // update the record since it seems we're successful here.
 316     last_action._byte_start = new_start;
 317     last_action._length = int(new_len);
 318
 319     // add in this next new chunk of file.
 320     last_action.pack(storage);  // add the header.
 321     storage += new_chunk;  // add the new stuff.
 322
 323     if (!current.length()) {
 324       // ensure we don't get stuck redoing zero length files, which we allowed
 325       // to go past their end above (since otherwise we'd never see them).
 326       to_return = advance(to_transfer, last_action);
 327       if (to_return != OKAY) break;
 328       continue;
 329     }
 330
 331     // just keep going, if there's space...
 332   }
 333
 334   return to_return;
 335 }
 336
 337 } //namespace.
 338