framework/main/classes/index/file/class_BaseFileIndex.php

   1 <?php
   2 // Own namespace
   3 namespace Org\Mxchange\CoreFramework\Index\File;
   4
   5 // Import framework stuff
   6 use Org\Mxchange\CoreFramework\EntryPoint\ApplicationEntryPoint;
   7 use Org\Mxchange\CoreFramework\Factory\Object\ObjectFactory;
   8 use Org\Mxchange\CoreFramework\Filesystem\File\BaseBinaryFile;
   9 use Org\Mxchange\CoreFramework\Filesystem\File\BinaryFile;
  10 use Org\Mxchange\CoreFramework\Index\BaseIndex;
  11 use Org\Mxchange\CoreFramework\Index\Indexable;
  12 use Org\Mxchange\CoreFramework\Utils\Strings\StringUtils;
  13
  14 // Import SPL stuff
  15 use \OutOfBoundsException;
  16 use \SplFileInfo;
  17 use \UnexpectedValueException;
  18
  19 /**
  20  * A general file-based index class
  21  *
  22  * @author              Roland Haeder <webmaster@ship-simu.org>
  23  * @version             0.0.0
  24  * @copyright   Copyright (c) 2007, 2008 Roland Haeder, 2009 - 2021 Core Developer Team
  25  * @license             GNU GPL 3.0 or any newer version
  26  * @link                http://www.ship-simu.org
  27  *
  28  * This program is free software: you can redistribute it and/or modify
  29  * it under the terms of the GNU General Public License as published by
  30  * the Free Software Foundation, either version 3 of the License, or
  31  * (at your option) any later version.
  32  *
  33  * This program is distributed in the hope that it will be useful,
  34  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  35  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  36  * GNU General Public License for more details.
  37  *
  38  * You should have received a copy of the GNU General Public License
  39  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  40  */
  41 abstract class BaseFileIndex extends BaseIndex implements FileIndexer {
  42         /**
  43          * Minimum block length
  44          */
  45         private static $minimumBlockLength = 0;
  46
  47         /**
  48          * Protected constructor
  49          *
  50          * @param       $className      Name of class
  51          * @return      void
  52          */
  53         protected function __construct (string $className) {
  54                 // Call parent constructor
  55                 parent::__construct($className);
  56         }
  57
  58         /**
  59          * Reads the file header
  60          *
  61          * @return      void
  62          * @throws      UnexpectedValueException        If header length or count of elements is invalid
  63          */
  64         public function readIndexHeader () {
  65                 // First rewind to beginning as the header sits at the beginning ...
  66                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
  67                 $this->getIteratorInstance()->rewind();
  68
  69                 // Get header size
  70                 $headerSize = $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize();
  71
  72                 // Then read it (see constructor for calculation)
  73                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: headerSize=%d', $headerSize));
  74                 $data = $this->getIteratorInstance()->getBinaryFileInstance()->read($headerSize);
  75
  76                 // Have all requested bytes been read?
  77                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Read %d bytes (%d wanted).', strlen($data), $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize()));
  78                 if (strlen($data) != $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize()) {
  79                         // Invalid header length
  80                         throw new UnexpectedValueException(sprintf('data(%d)=%s is not expected length %d',
  81                                 strlen($data),
  82                                 $data,
  83                                 $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize()
  84                         ));
  85                 } elseif (empty(trim($data, chr(0)))) {
  86                         // Empty file header
  87                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: File header is empty - EXIT!');
  88                         return;
  89                 } elseif (substr($data, -1, 1) != chr(BinaryFile::SEPARATOR_HEADER_ENTRIES)) {
  90                         // Bad last character
  91                         throw new UnexpectedValueException(sprintf('data=%s does not end with "%s"',
  92                                 $data,
  93                                 chr(BinaryFile::SEPARATOR_HEADER_ENTRIES)
  94                         ));
  95                 }
  96
  97                 // Okay, then remove it
  98                 $data = substr($data, 0, -1);
  99
 100                 // And update seek position
 101                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->updateSeekPosition() ...');
 102                 $this->getIteratorInstance()->getBinaryFileInstance()->updateSeekPosition();
 103
 104                 /*
 105                  * Now split it:
 106                  *
 107                  * 0 => magic
 108                  * 1 => total entries
 109                  */
 110                 $header = explode(chr(BinaryFile::SEPARATOR_HEADER_DATA), $data);
 111
 112                 // Check if the array has only 3 elements
 113                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: header()=%d', count($header)));
 114                 //* PRINTR-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: header(%d)=%s', count($header), print_r($header, true)));
 115                 if (count($header) != 2) {
 116                         // Bad header
 117                         throw new UnexpectedValueException(sprintf('header()=%d is not expected value 2', count($header)));
 118                 } elseif ($header[0] !== Indexable::INDEX_MAGIC) {
 119                         // Magic must be in first element
 120                         throw new UnexpectedValueException(sprintf('header[0]=%s is not the expected magic (%s)', $header[0], Indexable::INDEX_MAGIC));
 121                 } elseif (strlen($header[1]) != BinaryFile::LENGTH_COUNT) {
 122                         // Length of total entries not matching
 123                         throw new UnexpectedValueException(sprintf('header[1](%d)=%s does not have expected length %d', strlen($header[1]), $header[1], BinaryFile::LENGTH_COUNT));
 124                 }
 125
 126                 // Decode count
 127                 $header[1] = hex2bin($header[1]);
 128
 129                 // Set it here
 130                 $this->getIteratorInstance()->getBinaryFileInstance()->setHeader($header);
 131
 132                 // Trace message
 133                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 134         }
 135
 136         /**
 137          * Flushes the file header
 138          *
 139          * @return      void
 140          */
 141         public function flushFileHeader () {
 142                 // Put all informations together
 143                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 144                 $header = sprintf('%s%s%s%s',
 145                         // Magic
 146                         Indexable::INDEX_MAGIC,
 147
 148                         // Separator header data
 149                         chr(BinaryFile::SEPARATOR_HEADER_DATA),
 150
 151                         // Total entries
 152                         str_pad(StringUtils::dec2hex($this->getIteratorInstance()->getBinaryFileInstance()->getCounter()), BinaryFile::LENGTH_COUNT, '0', STR_PAD_LEFT),
 153
 154                         // Separator header<->entries
 155                         chr(BinaryFile::SEPARATOR_HEADER_ENTRIES)
 156                 );
 157
 158                 // Write it to disk (header is always at seek position 0)
 159                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Calling this->iteratorInstance->writeAtPosition(0, header=%s) ...', $header));
 160                 $this->getIteratorInstance()->getBinaryFileInstance()->writeAtPosition(0, $header);
 161
 162                 // Trace message
 163                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 164         }
 165
 166         /**
 167          * Initializes this file-based index
 168          *
 169          * @param       $fileInfoInstance       An instance of a SplFileInfo class
 170          * @return      void
 171          * @todo        Currently the index file is not cached, please implement a memory-handling class and if enough RAM is found, cache the whole index file.
 172          */
 173         protected function initFileIndex (SplFileInfo $fileInfoInstance) {
 174                 // Get a file i/o pointer instance for index file
 175                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: fileInfoInstance[%s]=%s - CALLED!', get_class($fileInfoInstance), $fileInfoInstance));
 176                 $fileInstance = ObjectFactory::createObjectByConfiguredName('index_file_class', array($fileInfoInstance, $this));
 177
 178                 // Get iterator instance
 179                 $iteratorInstance = ObjectFactory::createObjectByConfiguredName('file_iterator_class', [$fileInstance]);
 180
 181                 // Set iterator here
 182                 $this->setIteratorInstance($iteratorInstance);
 183
 184                 // Calculate header size
 185                 $headerSize = (
 186                         strlen(Indexable::INDEX_MAGIC) +
 187                         strlen(chr(BinaryFile::SEPARATOR_HEADER_DATA)) +
 188                         BinaryFile::LENGTH_COUNT +
 189                         strlen(chr(BinaryFile::SEPARATOR_HEADER_ENTRIES))
 190                 );
 191
 192                 // Set it
 193                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Setting headerSize=%d ...', $headerSize));
 194                 $this->getIteratorInstance()->getBinaryFileInstance()->setHeaderSize($headerSize);
 195
 196                 // Init counters and gaps array
 197                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->initCountersGapsArray() ...');
 198                 $this->getIteratorInstance()->getBinaryFileInstance()->initCountersGapsArray();
 199
 200                 // Default is not created
 201                 $created = false;
 202
 203                 // Is the file's header initialized?
 204                 if (!$this->getIteratorInstance()->getBinaryFileInstance()->isFileHeaderInitialized()) {
 205                         // First pre-allocate a bit
 206                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->preAllocateFile(index) ...');
 207                         $this->getIteratorInstance()->getBinaryFileInstance()->preAllocateFile('index');
 208
 209                         // Then write file header
 210                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->createFileHeader() ...');
 211                         $this->getIteratorInstance()->getBinaryFileInstance()->createFileHeader();
 212
 213                         // Mark as freshly created
 214                         $created = true;
 215                 }
 216
 217                 // Load the file header
 218                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->readIndexHeader() ...');
 219                 $this->readIndexHeader();
 220
 221                 // Freshly created?
 222                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: created=%d', intval($created)));
 223                 if (!$created) {
 224                         // Analyze file structure
 225                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->analyzeFileStructure() ...');
 226                         $this->getIteratorInstance()->getBinaryFileInstance()->analyzeFileStructure();
 227                 }
 228
 229                 // Trace message
 230                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 231         }
 232
 233         /**
 234          * Calculates minimum length for one entry/block
 235          *
 236          * @return      $length         Minimum length for one entry/block
 237          */
 238         public function calculateMinimumBlockLength () {
 239                 // Is it "cached"?
 240                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 241                 if (self::$minimumBlockLength == 0) {
 242                         // Calulcate it
 243                         //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calculating ...');
 244                         self::$minimumBlockLength = (
 245                                 // Type
 246                                 BinaryFile::LENGTH_TYPE + strlen(chr(BinaryFile::SEPARATOR_TYPE_POSITION)) +
 247                                 // Position
 248                                 BinaryFile::LENGTH_POSITION + strlen(chr(BinaryFile::SEPARATOR_ENTRIES))
 249                         );
 250                 }
 251
 252                 // Return it
 253                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: self::minimumBlockLength=%d - EXIT!', self::$minimumBlockLength));
 254                 return self::$minimumBlockLength;
 255         }
 256
 257         /**
 258          * "Getter" for file size
 259          *
 260          * @return      $fileSize       Size of currently loaded file
 261          */
 262         public function getFileSize () {
 263                 // Call iterator's method
 264                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 265                 $fileSize = $this->getIteratorInstance()->getBinaryFileInstance()->getFileSize();
 266
 267                 // Return it
 268                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: fileSize=%d - EXIT!', $fileSize));
 269                 return $fileSize;
 270         }
 271
 272         /**
 273          * Searches for next suitable gap the given length of data can fit in
 274          * including padding bytes.
 275          *
 276          * @param       $length                 Length of raw data
 277          * @return      $seekPosition   Found next gap's seek position
 278          * @throws      InvalidArgumentException        If the parameter is not valid
 279          * @todo        Unfinished work
 280          */
 281         public function searchNextGap (int $length) {
 282                 // Validate parameter
 283                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: length=%d - CALLED!', $length));
 284                 if ($length <= 0) {
 285                         // Throw IAE
 286                         throw new InvalidArgumentException(sprintf('length=%d is not valid', $length));
 287                 }
 288
 289                 // Debug message
 290                 /* DEBUG-DIE: */ ApplicationEntryPoint::exitApplication(sprintf('[%s:%d]: length=%d,this=%s', __METHOD__, __LINE__, $length, print_r($this, true)));
 291         }
 292
 293         /**
 294          * Writes at given position by seeking to it.
 295          *
 296          * @param       $seekPosition   Seek position in file
 297          * @param       $dataStream             Data to be written
 298          * @return      mixed                   Number of writes bytes or false on error
 299          * @throws      OutOfBoundsException    If the position is not seekable
 300          * @throws      InvalidArgumentException        If a parameter is not valid
 301          */
 302         public function writeAtPosition (int $seekPosition, string $dataStream) {
 303                 // Validate parameter
 304                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: seekPosition=%d,dataStream(%d)=%s - CALLED!', $seekPosition, strlen($dataStream), $dataStream));
 305                 if ($seekPosition < 0) {
 306                         // Invalid seek position
 307                         throw new OutOfBoundsException(sprintf('seekPosition=%d is not valid.', $seekPosition));
 308                 } elseif (empty($dataStream)) {
 309                         // Empty dataStream
 310                         throw new InvalidArgumentException('Parameter "dataStream" is empty');
 311                 }
 312
 313                 // Call iterated object's method
 314                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Calling this->iteratorInstance->writeAtPosition(%d, %s) ...', $seekPosition, $dataStream));
 315                 $status = $this->getIteratorInstance()->getBinaryFileInstance()->writeAtPosition($seekPosition, $dataStream);
 316
 317                 // Return status
 318                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: status[%s]=%d - EXIT!', gettype($status), $status));
 319                 return $status;
 320         }
 321
 322         /**
 323          * Checks if this index has been fully and properly loaded.
 324          *
 325          * @return      $isLoaded       Whether this index has been loaded
 326          */
 327         public function isIndexLoaded () {
 328                 // Is the file gaps-only?
 329                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 330                 if ($this->getIteratorInstance()->getBinaryFileInstance()->isFileGapsOnly()) {
 331                         // Then skip below code as this implies the file has been fully analyzed and "loaded"
 332                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Underlaying file is gaps-only: Returning TRUE ... - EXIT!');
 333                         return TRUE;
 334                 }
 335
 336                 // Debug message
 337                 /* DEBUG-DIE: */ ApplicationEntryPoint::exitApplication(sprintf('[%s:%d]: this=%s', __METHOD__, __LINE__, print_r($this, true)));
 338         }
 339
 340 }