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), $data);
 112
 113                 // Map numeric entries to associative (alpha-numeric) elements
 114                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Calling ArrayUtils::mapNumericKeysToAssociative(%d) ...', count($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                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: HEADER_INDEX_ELEMENT_COUNT=%d,header()=%d', BinaryFile::HEADER_INDEX_ELEMENT_COUNT, count($header)));
 123                 if (count($header) != BinaryFile::HEADER_INDEX_ELEMENT_COUNT) {
 124                         // Bad header
 125                         throw new UnexpectedValueException(sprintf('header()=%d is not expected value %d', count($header), BinaryFile::HEADER_INDEX_ELEMENT_COUNT));
 126                 } elseif ($header[BinaryFile::HEADER_NAME_MAGIC] !== Indexable::INDEX_MAGIC) {
 127                         // Magic must be in first element
 128                         throw new UnexpectedValueException(sprintf('header[%s]=%s is not the expected magic (%s)', BinaryFile::HEADER_NAME_MAGIC, $header[BinaryFile::HEADER_NAME_MAGIC], Indexable::INDEX_MAGIC));
 129                 } elseif (strlen($header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES]) != BinaryFile::LENGTH_COUNT) {
 130                         // Length of total entries not matching
 131                         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));
 132                 }
 133
 134                 // Decode count
 135                 $header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES] = hex2bin($header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES]);
 136
 137                 // Set it here
 138                 $this->getIteratorInstance()->getBinaryFileInstance()->setHeader($header);
 139
 140                 // Trace message
 141                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 142         }
 143
 144         /**
 145          * Flushes the file header
 146          *
 147          * @return      void
 148          */
 149         public function flushFileHeader () {
 150                 // Put all informations together
 151                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 152                 $header = sprintf('%s%s%s%s',
 153                         // Magic
 154                         Indexable::INDEX_MAGIC,
 155
 156                         // Separator header data
 157                         chr(BinaryFile::SEPARATOR_HEADER_DATA),
 158
 159                         // Total entries
 160                         str_pad(StringUtils::dec2hex($this->getIteratorInstance()->getBinaryFileInstance()->getCounter()), BinaryFile::LENGTH_COUNT, '0', STR_PAD_LEFT),
 161
 162                         // Separator header<->entries
 163                         chr(BinaryFile::SEPARATOR_HEADER_ENTRIES)
 164                 );
 165
 166                 // Write it to disk (header is always at seek position 0)
 167                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->writeAtPosition(0, header=%s) ...', $header));
 168                 $this->getIteratorInstance()->getBinaryFileInstance()->writeAtPosition(0, $header);
 169
 170                 // Trace message
 171                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 172         }
 173
 174         /**
 175          * Initializes this file-based index
 176          *
 177          * @param       $fileInfoInstance       An instance of a SplFileInfo class
 178          * @return      void
 179          * @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.
 180          */
 181         protected function initFileIndex (SplFileInfo $fileInfoInstance) {
 182                 // Get a file i/o pointer instance for index file
 183                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: fileInfoInstance[%s]=%s - CALLED!', get_class($fileInfoInstance), $fileInfoInstance));
 184                 $fileInstance = ObjectFactory::createObjectByConfiguredName('index_file_class', array($fileInfoInstance, $this));
 185
 186                 // Get iterator instance
 187                 $iteratorInstance = ObjectFactory::createObjectByConfiguredName('file_iterator_class', [$fileInstance]);
 188
 189                 // Set iterator here
 190                 $this->setIteratorInstance($iteratorInstance);
 191
 192                 // Calculate header size
 193                 $headerSize = (
 194                         strlen(Indexable::INDEX_MAGIC) +
 195                         strlen(chr(BinaryFile::SEPARATOR_HEADER_DATA)) +
 196                         BinaryFile::LENGTH_COUNT +
 197                         strlen(chr(BinaryFile::SEPARATOR_HEADER_ENTRIES))
 198                 );
 199
 200                 // Set it
 201                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Setting headerSize=%d ...', $headerSize));
 202                 $this->getIteratorInstance()->getBinaryFileInstance()->setHeaderSize($headerSize);
 203
 204                 // Init counters and gaps array
 205                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->initCountersGapsArray() ...');
 206                 $this->getIteratorInstance()->getBinaryFileInstance()->initCountersGapsArray();
 207
 208                 // Default is not created
 209                 $created = false;
 210
 211                 // Is the file's header initialized?
 212                 if (!$this->getIteratorInstance()->getBinaryFileInstance()->isFileHeaderInitialized()) {
 213                         // First pre-allocate a bit
 214                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->preAllocateFile(index) ...');
 215                         $this->getIteratorInstance()->getBinaryFileInstance()->preAllocateFile('index');
 216
 217                         // Then write file header
 218                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->createFileHeader() ...');
 219                         $this->getIteratorInstance()->getBinaryFileInstance()->createFileHeader();
 220
 221                         // Mark as freshly created
 222                         $created = true;
 223                 }
 224
 225                 // Load the file header
 226                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->readIndexHeader() ...');
 227                 $this->readIndexHeader();
 228
 229                 // Freshly created?
 230                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: created=%d', intval($created)));
 231                 if (!$created) {
 232                         // Analyze file structure
 233                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->analyzeFileStructure() ...');
 234                         $this->getIteratorInstance()->getBinaryFileInstance()->analyzeFileStructure();
 235                 }
 236
 237                 // Trace message
 238                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 239         }
 240
 241         /**
 242          * Calculates minimum length for one entry/block
 243          *
 244          * @return      $length         Minimum length for one entry/block
 245          */
 246         public function calculateMinimumBlockLength () {
 247                 // Is it "cached"?
 248                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 249                 if (self::$minimumBlockLength == 0) {
 250                         // Calulcate it
 251                         //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calculating ...');
 252                         self::$minimumBlockLength = (
 253                                 // Type
 254                                 BinaryFile::LENGTH_TYPE + strlen(chr(BinaryFile::SEPARATOR_TYPE_POSITION)) +
 255                                 // Position
 256                                 BinaryFile::LENGTH_POSITION + strlen(chr(BinaryFile::SEPARATOR_ENTRIES))
 257                         );
 258                 }
 259
 260                 // Return it
 261                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: self::minimumBlockLength=%d - EXIT!', self::$minimumBlockLength));
 262                 return self::$minimumBlockLength;
 263         }
 264
 265         /**
 266          * "Getter" for file size
 267          *
 268          * @return      $fileSize       Size of currently loaded file
 269          */
 270         public function getFileSize () {
 271                 // Call iterator's method
 272                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 273                 $fileSize = $this->getIteratorInstance()->getBinaryFileInstance()->getFileSize();
 274
 275                 // Return it
 276                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: fileSize=%d - EXIT!', $fileSize));
 277                 return $fileSize;
 278         }
 279
 280         /**
 281          * Searches for next suitable gap the given length of data can fit in
 282          * including padding bytes.
 283          *
 284          * @param       $length                 Length of raw data
 285          * @return      $seekPosition   Found next gap's seek position
 286          * @throws      InvalidArgumentException        If the parameter is not valid
 287          * @todo        Unfinished work
 288          */
 289         public function searchNextGap (int $length) {
 290                 // Validate parameter
 291                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: length=%d - CALLED!', $length));
 292                 if ($length <= 0) {
 293                         // Throw IAE
 294                         throw new InvalidArgumentException(sprintf('length=%d is not valid', $length));
 295                 }
 296
 297                 // Debug message
 298                 /* DEBUG-DIE: */ ApplicationEntryPoint::exitApplication(sprintf('[%s:%d]: length=%d,this=%s', __METHOD__, __LINE__, $length, print_r($this, true)));
 299         }
 300
 301         /**
 302          * Writes at given position by seeking to it.
 303          *
 304          * @param       $seekPosition   Seek position in file
 305          * @param       $dataStream             Data to be written
 306          * @return      mixed                   Number of writes bytes or false on error
 307          * @throws      OutOfBoundsException    If the position is not seekable
 308          * @throws      InvalidArgumentException        If a parameter is not valid
 309          */
 310         public function writeAtPosition (int $seekPosition, string $dataStream) {
 311                 // Validate parameter
 312                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: seekPosition=%d,dataStream(%d)=%s - CALLED!', $seekPosition, strlen($dataStream), $dataStream));
 313                 if ($seekPosition < 0) {
 314                         // Invalid seek position
 315                         throw new OutOfBoundsException(sprintf('seekPosition=%d is not valid.', $seekPosition));
 316                 } elseif (empty($dataStream)) {
 317                         // Empty dataStream
 318                         throw new InvalidArgumentException('Parameter "dataStream" is empty');
 319                 }
 320
 321                 // Call iterated object's method
 322                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Calling this->iteratorInstance->binaryFileInstance->writeAtPosition(%d, %s) ...', $seekPosition, $dataStream));
 323                 $status = $this->getIteratorInstance()->getBinaryFileInstance()->writeAtPosition($seekPosition, $dataStream);
 324
 325                 // Return status
 326                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: status[%s]=%d - EXIT!', gettype($status), $status));
 327                 return $status;
 328         }
 329
 330         /**
 331          * Checks if this index has been fully and properly loaded.
 332          *
 333          * @return      $isLoaded       Whether this index has been loaded
 334          */
 335         public function isIndexLoaded () {
 336                 // Is the file gaps-only?
 337                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 338                 if ($this->getIteratorInstance()->getBinaryFileInstance()->isFileGapsOnly()) {
 339                         // Then skip below code as this implies the file has been fully analyzed and "loaded"
 340                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Underlaying file is gaps-only: Returning TRUE ... - EXIT!');
 341                         return TRUE;
 342                 }
 343
 344                 // Debug message
 345                 /* DEBUG-DIE: */ ApplicationEntryPoint::exitApplication(sprintf('[%s:%d]: this=%s', __METHOD__, __LINE__, print_r($this, true)));
 346         }
 347
 348 }