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\Arrays\ArrayUtils;
  13 use Org\Mxchange\CoreFramework\Utils\Strings\StringUtils;
  14
  15 // Import SPL stuff
  16 use \OutOfBoundsException;
  17 use \SplFileInfo;
  18 use \UnexpectedValueException;
  19
  20 /**
  21  * A general file-based index class
  22  *
  23  * @author              Roland Haeder <webmaster@ship-simu.org>
  24  * @version             0.0.0
  25  * @copyright   Copyright (c) 2007, 2008 Roland Haeder, 2009 - 2021 Core Developer Team
  26  * @license             GNU GPL 3.0 or any newer version
  27  * @link                http://www.ship-simu.org
  28  *
  29  * This program is free software: you can redistribute it and/or modify
  30  * it under the terms of the GNU General Public License as published by
  31  * the Free Software Foundation, either version 3 of the License, or
  32  * (at your option) any later version.
  33  *
  34  * This program is distributed in the hope that it will be useful,
  35  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  36  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  37  * GNU General Public License for more details.
  38  *
  39  * You should have received a copy of the GNU General Public License
  40  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  41  */
  42 abstract class BaseFileIndex extends BaseIndex implements FileIndexer {
  43         /**
  44          * Minimum block length
  45          */
  46         private static $minimumBlockLength = 0;
  47
  48         /**
  49          * Protected constructor
  50          *
  51          * @param       $className      Name of class
  52          * @return      void
  53          */
  54         protected function __construct (string $className) {
  55                 // Call parent constructor
  56                 parent::__construct($className);
  57         }
  58
  59         /**
  60          * Reads the file header
  61          *
  62          * @return      void
  63          * @throws      UnexpectedValueException        If header length or count of elements is invalid
  64          */
  65         public function readIndexHeader () {
  66                 // First rewind to beginning as the header sits at the beginning ...
  67                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
  68                 $this->getIteratorInstance()->rewind();
  69
  70                 // Get header size
  71                 $headerSize = $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize();
  72
  73                 // Then read it (see constructor for calculation)
  74                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: headerSize=%d', $headerSize));
  75                 $data = $this->getIteratorInstance()->getBinaryFileInstance()->read($headerSize);
  76
  77                 // Have all requested bytes been read?
  78                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Read %d bytes (%d wanted).', strlen($data), $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize()));
  79                 if (strlen($data) != $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize()) {
  80                         // Invalid header length
  81                         throw new UnexpectedValueException(sprintf('data(%d)=%s is not expected length %d',
  82                                 strlen($data),
  83                                 $data,
  84                                 $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize()
  85                         ));
  86                 } elseif (empty(trim($data, chr(0)))) {
  87                         // Empty file header
  88                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: File header is empty - EXIT!');
  89                         return;
  90                 } elseif (substr($data, -1, 1) != chr(BinaryFile::SEPARATOR_HEADER_ENTRIES)) {
  91                         // Bad last character
  92                         throw new UnexpectedValueException(sprintf('data=%s does not end with "%s"',
  93                                 $data,
  94                                 chr(BinaryFile::SEPARATOR_HEADER_ENTRIES)
  95                         ));
  96                 }
  97
  98                 // Okay, then remove it
  99                 $data = substr($data, 0, -1);
 100
 101                 // And update seek position
 102                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->updateSeekPosition() ...');
 103                 $this->getIteratorInstance()->getBinaryFileInstance()->updateSeekPosition();
 104
 105                 /*
 106                  * Now split it:
 107                  *
 108                  * 0 => magic
 109                  * 1 => total entries
 110                  */
 111                 $header = explode(chr(BinaryFile::SEPARATOR_HEADER_DATA), $data);
 112
 113                 // Map numeric entries to associative (alpha-numeric) elements
 114                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: HEADER_INDEX_ELEMENT_COUNT=%d,header()=%d', BinaryFile::HEADER_INDEX_ELEMENT_COUNT, count($header)));
 115                 //* PRINTR-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: header(%d)=%s', count($header), print_r($header, true)));
 116                 $header = ArrayUtils::mapNumericKeysToAssociative($header, [
 117                         BinaryFile::HEADER_NAME_MAGIC,
 118                         BinaryFile::HEADER_NAME_TOTAL_ENTRIES,
 119                 ]);
 120
 121                 // Check if the array has only 2 elements
 122                 if (count($header) != BinaryFile::HEADER_INDEX_ELEMENT_COUNT) {
 123                         // Bad header
 124                         throw new UnexpectedValueException(sprintf('header()=%d is not expected value %d', count($header), BinaryFile::HEADER_INDEX_ELEMENT_COUNT));
 125                 } elseif ($header[BinaryFile::HEADER_NAME_MAGIC] !== Indexable::INDEX_MAGIC) {
 126                         // Magic must be in first element
 127                         throw new UnexpectedValueException(sprintf('header[%s]=%s is not the expected magic (%s)', BinaryFile::HEADER_NAME_MAGIC, $header[BinaryFile::HEADER_NAME_MAGIC], Indexable::INDEX_MAGIC));
 128                 } elseif (strlen($header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES]) != BinaryFile::LENGTH_COUNT) {
 129                         // Length of total entries not matching
 130                         throw new UnexpectedValueException(sprintf('header[%s](%d)=%s does not have expected length %d', BinaryFile::HEADER_NAME_TOTAL_ENTRIES, strlen($header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES]), $header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES], BinaryFile::LENGTH_COUNT));
 131                 }
 132
 133                 // Decode count
 134                 $header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES] = hex2bin($header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES]);
 135
 136                 // Set it here
 137                 $this->getIteratorInstance()->getBinaryFileInstance()->setHeader($header);
 138
 139                 // Trace message
 140                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 141         }
 142
 143         /**
 144          * Flushes the file header
 145          *
 146          * @return      void
 147          */
 148         public function flushFileHeader () {
 149                 // Put all informations together
 150                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 151                 $header = sprintf('%s%s%s%s',
 152                         // Magic
 153                         Indexable::INDEX_MAGIC,
 154
 155                         // Separator header data
 156                         chr(BinaryFile::SEPARATOR_HEADER_DATA),
 157
 158                         // Total entries
 159                         str_pad(StringUtils::dec2hex($this->getIteratorInstance()->getBinaryFileInstance()->getCounter()), BinaryFile::LENGTH_COUNT, '0', STR_PAD_LEFT),
 160
 161                         // Separator header<->entries
 162                         chr(BinaryFile::SEPARATOR_HEADER_ENTRIES)
 163                 );
 164
 165                 // Write it to disk (header is always at seek position 0)
 166                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->writeAtPosition(0, header=%s) ...', $header));
 167                 $this->getIteratorInstance()->getBinaryFileInstance()->writeAtPosition(0, $header);
 168
 169                 // Trace message
 170                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 171         }
 172
 173         /**
 174          * Initializes this file-based index
 175          *
 176          * @param       $fileInfoInstance       An instance of a SplFileInfo class
 177          * @return      void
 178          * @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.
 179          */
 180         protected function initFileIndex (SplFileInfo $fileInfoInstance) {
 181                 // Get a file i/o pointer instance for index file
 182                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: fileInfoInstance[%s]=%s - CALLED!', get_class($fileInfoInstance), $fileInfoInstance));
 183                 $fileInstance = ObjectFactory::createObjectByConfiguredName('index_file_class', array($fileInfoInstance, $this));
 184
 185                 // Get iterator instance
 186                 $iteratorInstance = ObjectFactory::createObjectByConfiguredName('file_iterator_class', [$fileInstance]);
 187
 188                 // Set iterator here
 189                 $this->setIteratorInstance($iteratorInstance);
 190
 191                 // Calculate header size
 192                 $headerSize = (
 193                         strlen(Indexable::INDEX_MAGIC) +
 194                         strlen(chr(BinaryFile::SEPARATOR_HEADER_DATA)) +
 195                         BinaryFile::LENGTH_COUNT +
 196                         strlen(chr(BinaryFile::SEPARATOR_HEADER_ENTRIES))
 197                 );
 198
 199                 // Set it
 200                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Setting headerSize=%d ...', $headerSize));
 201                 $this->getIteratorInstance()->getBinaryFileInstance()->setHeaderSize($headerSize);
 202
 203                 // Init counters and gaps array
 204                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->initCountersGapsArray() ...');
 205                 $this->getIteratorInstance()->getBinaryFileInstance()->initCountersGapsArray();
 206
 207                 // Default is not created
 208                 $created = false;
 209
 210                 // Is the file's header initialized?
 211                 if (!$this->getIteratorInstance()->getBinaryFileInstance()->isFileHeaderInitialized()) {
 212                         // First pre-allocate a bit
 213                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->preAllocateFile(index) ...');
 214                         $this->getIteratorInstance()->getBinaryFileInstance()->preAllocateFile('index');
 215
 216                         // Then write file header
 217                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->createFileHeader() ...');
 218                         $this->getIteratorInstance()->getBinaryFileInstance()->createFileHeader();
 219
 220                         // Mark as freshly created
 221                         $created = true;
 222                 }
 223
 224                 // Load the file header
 225                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->readIndexHeader() ...');
 226                 $this->readIndexHeader();
 227
 228                 // Freshly created?
 229                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: created=%d', intval($created)));
 230                 if (!$created) {
 231                         // Analyze file structure
 232                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->analyzeFileStructure() ...');
 233                         $this->getIteratorInstance()->getBinaryFileInstance()->analyzeFileStructure();
 234                 }
 235
 236                 // Trace message
 237                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 238         }
 239
 240         /**
 241          * Calculates minimum length for one entry/block
 242          *
 243          * @return      $length         Minimum length for one entry/block
 244          */
 245         public function calculateMinimumBlockLength () {
 246                 // Is it "cached"?
 247                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 248                 if (self::$minimumBlockLength == 0) {
 249                         // Calulcate it
 250                         //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calculating ...');
 251                         self::$minimumBlockLength = (
 252                                 // Type
 253                                 BinaryFile::LENGTH_TYPE + strlen(chr(BinaryFile::SEPARATOR_TYPE_POSITION)) +
 254                                 // Position
 255                                 BinaryFile::LENGTH_POSITION + strlen(chr(BinaryFile::SEPARATOR_ENTRIES))
 256                         );
 257                 }
 258
 259                 // Return it
 260                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: self::minimumBlockLength=%d - EXIT!', self::$minimumBlockLength));
 261                 return self::$minimumBlockLength;
 262         }
 263
 264         /**
 265          * "Getter" for file size
 266          *
 267          * @return      $fileSize       Size of currently loaded file
 268          */
 269         public function getFileSize () {
 270                 // Call iterator's method
 271                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 272                 $fileSize = $this->getIteratorInstance()->getBinaryFileInstance()->getFileSize();
 273
 274                 // Return it
 275                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: fileSize=%d - EXIT!', $fileSize));
 276                 return $fileSize;
 277         }
 278
 279         /**
 280          * Searches for next suitable gap the given length of data can fit in
 281          * including padding bytes.
 282          *
 283          * @param       $length                 Length of raw data
 284          * @return      $seekPosition   Found next gap's seek position
 285          * @throws      InvalidArgumentException        If the parameter is not valid
 286          * @todo        Unfinished work
 287          */
 288         public function searchNextGap (int $length) {
 289                 // Validate parameter
 290                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: length=%d - CALLED!', $length));
 291                 if ($length <= 0) {
 292                         // Throw IAE
 293                         throw new InvalidArgumentException(sprintf('length=%d is not valid', $length));
 294                 }
 295
 296                 // Debug message
 297                 /* DEBUG-DIE: */ ApplicationEntryPoint::exitApplication(sprintf('[%s:%d]: length=%d,this=%s', __METHOD__, __LINE__, $length, print_r($this, true)));
 298         }
 299
 300         /**
 301          * Writes at given position by seeking to it.
 302          *
 303          * @param       $seekPosition   Seek position in file
 304          * @param       $dataStream             Data to be written
 305          * @return      mixed                   Number of writes bytes or false on error
 306          * @throws      OutOfBoundsException    If the position is not seekable
 307          * @throws      InvalidArgumentException        If a parameter is not valid
 308          */
 309         public function writeAtPosition (int $seekPosition, string $dataStream) {
 310                 // Validate parameter
 311                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: seekPosition=%d,dataStream(%d)=%s - CALLED!', $seekPosition, strlen($dataStream), $dataStream));
 312                 if ($seekPosition < 0) {
 313                         // Invalid seek position
 314                         throw new OutOfBoundsException(sprintf('seekPosition=%d is not valid.', $seekPosition));
 315                 } elseif (empty($dataStream)) {
 316                         // Empty dataStream
 317                         throw new InvalidArgumentException('Parameter "dataStream" is empty');
 318                 }
 319
 320                 // Call iterated object's method
 321                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->writeAtPosition(%d, %s) ...', $seekPosition, $dataStream));
 322                 $status = $this->getIteratorInstance()->getBinaryFileInstance()->writeAtPosition($seekPosition, $dataStream);
 323
 324                 // Return status
 325                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: status[%s]=%d - EXIT!', gettype($status), $status));
 326                 return $status;
 327         }
 328
 329         /**
 330          * Checks if this index has been fully and properly loaded.
 331          *
 332          * @return      $isLoaded       Whether this index has been loaded
 333          */
 334         public function isIndexLoaded () {
 335                 // Is the file gaps-only?
 336                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 337                 if ($this->getIteratorInstance()->getBinaryFileInstance()->isFileGapsOnly()) {
 338                         // Then skip below code as this implies the file has been fully analyzed and "loaded"
 339                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Underlaying file is gaps-only: Returning TRUE ... - EXIT!');
 340                         return TRUE;
 341                 }
 342
 343                 // Debug message
 344                 /* DEBUG-DIE: */ ApplicationEntryPoint::exitApplication(sprintf('[%s:%d]: this=%s', __METHOD__, __LINE__, print_r($this, true)));
 345         }
 346
 347 }