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