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\Generic\FrameworkInterface;
  11 use Org\Mxchange\CoreFramework\Index\BaseIndex;
  12 use Org\Mxchange\CoreFramework\Index\Indexable;
  13 use Org\Mxchange\CoreFramework\Utils\Arrays\ArrayUtils;
  14 use Org\Mxchange\CoreFramework\Utils\Strings\StringUtils;
  15
  16 // Import SPL stuff
  17 use \OutOfBoundsException;
  18 use \SplFileInfo;
  19 use \UnexpectedValueException;
  20
  21 /**
  22  * A general file-based index class
  23  *
  24  * @author              Roland Haeder <webmaster@ship-simu.org>
  25  * @version             0.0.0
  26  * @copyright   Copyright (c) 2007, 2008 Roland Haeder, 2009 - 2022 Core Developer Team
  27  * @license             GNU GPL 3.0 or any newer version
  28  * @link                http://www.ship-simu.org
  29  *
  30  * This program is free software: you can redistribute it and/or modify
  31  * it under the terms of the GNU General Public License as published by
  32  * the Free Software Foundation, either version 3 of the License, or
  33  * (at your option) any later version.
  34  *
  35  * This program is distributed in the hope that it will be useful,
  36  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  37  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  38  * GNU General Public License for more details.
  39  *
  40  * You should have received a copy of the GNU General Public License
  41  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  42  */
  43 abstract class BaseFileIndex extends BaseIndex implements FileIndexer {
  44         /**
  45          * Minimum block length
  46          */
  47         private static $minimumBlockLength = 0;
  48
  49         /**
  50          * Protected constructor
  51          *
  52          * @param       $className      Name of class
  53          * @return      void
  54          */
  55         protected function __construct (string $className) {
  56                 // Call parent constructor
  57                 parent::__construct($className);
  58         }
  59
  60         /**
  61          * Reads the file header
  62          *
  63          * @return      void
  64          * @throws      UnexpectedValueException        If header length or count of elements is invalid
  65          */
  66         public function readIndexHeader () {
  67                 // First rewind to beginning as the header sits at the beginning ...
  68                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
  69                 $this->getIteratorInstance()->rewind();
  70
  71                 // Get header size
  72                 $headerSize = $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize();
  73
  74                 // Then read it (see constructor for calculation)
  75                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: headerSize=%d', $headerSize));
  76                 $data = $this->getIteratorInstance()->getBinaryFileInstance()->read($headerSize);
  77
  78                 // Have all requested bytes been read?
  79                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Read %d bytes (%d wanted).', strlen($data), $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize()));
  80                 if (strlen($data) != $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize()) {
  81                         // Invalid header length
  82                         throw new UnexpectedValueException(sprintf('data(%d)=%s is not expected length %d',
  83                                 strlen($data),
  84                                 $data,
  85                                 $this->getIteratorInstance()->getBinaryFileInstance()->getHeaderSize()
  86                         ));
  87                 } elseif (empty(trim($data, chr(0)))) {
  88                         // Empty file header
  89                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: File header is empty - EXIT!');
  90                         return;
  91                 } elseif (substr($data, -1, 1) != chr(BinaryFile::SEPARATOR_HEADER_ENTRIES)) {
  92                         // Bad last character
  93                         throw new UnexpectedValueException(sprintf('data=%s does not end with "%s"',
  94                                 $data,
  95                                 chr(BinaryFile::SEPARATOR_HEADER_ENTRIES)
  96                         ));
  97                 }
  98
  99                 // Okay, then remove it
 100                 $data = substr($data, 0, -1);
 101
 102                 // And update seek position
 103                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Invoking this->iteratorInstance->binaryFileInstance->updateSeekPosition() ...');
 104                 $this->getIteratorInstance()->getBinaryFileInstance()->updateSeekPosition();
 105
 106                 /*
 107                  * Now split it:
 108                  *
 109                  * 0 => magic
 110                  * 1 => total entries
 111                  */
 112                 $header = explode(chr(BinaryFile::SEPARATOR_HEADER_DATA), $data);
 113
 114                 // Map numeric entries to associative (alpha-numeric) elements
 115                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Invoking ArrayUtils::mapNumericKeysToAssociative(%d) ...', count($header)));
 116                 //* PRINTR-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: header(%d)=%s', count($header), print_r($header, true)));
 117                 $header = ArrayUtils::mapNumericKeysToAssociative($header, [
 118                         BinaryFile::HEADER_NAME_MAGIC,
 119                         BinaryFile::HEADER_NAME_TOTAL_ENTRIES,
 120                 ]);
 121
 122                 // Check if the array has only 2 elements
 123                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: HEADER_INDEX_ELEMENT_COUNT=%d,header()=%d', BinaryFile::HEADER_INDEX_ELEMENT_COUNT, count($header)));
 124                 if (count($header) != BinaryFile::HEADER_INDEX_ELEMENT_COUNT) {
 125                         // Bad header
 126                         throw new UnexpectedValueException(sprintf('header()=%d is not expected value %d', count($header), BinaryFile::HEADER_INDEX_ELEMENT_COUNT));
 127                 } elseif ($header[BinaryFile::HEADER_NAME_MAGIC] !== Indexable::INDEX_MAGIC) {
 128                         // Magic must be in first element
 129                         throw new UnexpectedValueException(sprintf('header[%s]=%s is not the expected magic (%s)', BinaryFile::HEADER_NAME_MAGIC, $header[BinaryFile::HEADER_NAME_MAGIC], Indexable::INDEX_MAGIC));
 130                 } elseif (strlen($header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES]) != BinaryFile::LENGTH_COUNT) {
 131                         // Length of total entries not matching
 132                         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));
 133                 }
 134
 135                 // Decode count
 136                 $header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES] = hex2bin($header[BinaryFile::HEADER_NAME_TOTAL_ENTRIES]);
 137
 138                 // Set it here
 139                 $this->getIteratorInstance()->getBinaryFileInstance()->setHeader($header);
 140
 141                 // Trace message
 142                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 143         }
 144
 145         /**
 146          * Flushes the file header
 147          *
 148          * @return      void
 149          */
 150         public function flushFileHeader () {
 151                 // Put all informations together
 152                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 153                 $header = sprintf('%s%s%s%s',
 154                         // Magic
 155                         Indexable::INDEX_MAGIC,
 156
 157                         // Separator header data
 158                         chr(BinaryFile::SEPARATOR_HEADER_DATA),
 159
 160                         // Total entries
 161                         str_pad(StringUtils::dec2hex($this->getIteratorInstance()->getBinaryFileInstance()->getCounter()), BinaryFile::LENGTH_COUNT, '0', STR_PAD_LEFT),
 162
 163                         // Separator header<->entries
 164                         chr(BinaryFile::SEPARATOR_HEADER_ENTRIES)
 165                 );
 166
 167                 // Write it to disk (header is always at seek position 0)
 168                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Invoking this->iteratorInstance->binaryFileInstance->writeAtPosition(0, header=%s) ...', $header));
 169                 $this->getIteratorInstance()->getBinaryFileInstance()->writeAtPosition(0, $header);
 170
 171                 // Trace message
 172                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 173         }
 174
 175         /**
 176          * Initializes this file-based index
 177          *
 178          * @param       $fileInfoInstance       An instance of a SplFileInfo class
 179          * @return      void
 180          * @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.
 181          */
 182         protected function initFileIndex (SplFileInfo $fileInfoInstance) {
 183                 // Get a file i/o pointer instance for index file
 184                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: fileInfoInstance[%s]=%s - CALLED!', get_class($fileInfoInstance), $fileInfoInstance));
 185                 $fileInstance = ObjectFactory::createObjectByConfiguredName('index_file_class', array($fileInfoInstance, $this));
 186
 187                 // Get iterator instance
 188                 $iteratorInstance = ObjectFactory::createObjectByConfiguredName('file_iterator_class', [$fileInstance]);
 189
 190                 // Set iterator here
 191                 $this->setIteratorInstance($iteratorInstance);
 192
 193                 // Calculate header size
 194                 $headerSize = (
 195                         strlen(Indexable::INDEX_MAGIC) +
 196                         strlen(chr(BinaryFile::SEPARATOR_HEADER_DATA)) +
 197                         BinaryFile::LENGTH_COUNT +
 198                         strlen(chr(BinaryFile::SEPARATOR_HEADER_ENTRIES))
 199                 );
 200
 201                 // Set it
 202                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Setting headerSize=%d ...', $headerSize));
 203                 $this->getIteratorInstance()->getBinaryFileInstance()->setHeaderSize($headerSize);
 204
 205                 // Init counters and gaps array
 206                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Invoking this->iteratorInstance->binaryFileInstance->initCountersGapsArray() ...');
 207                 $this->getIteratorInstance()->getBinaryFileInstance()->initCountersGapsArray();
 208
 209                 // Default is not created
 210                 $created = false;
 211
 212                 // Is the file's header initialized?
 213                 if (!$this->getIteratorInstance()->getBinaryFileInstance()->isFileHeaderInitialized()) {
 214                         // First pre-allocate a bit
 215                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Invoking this->iteratorInstance->binaryFileInstance->preAllocateFile(index) ...');
 216                         $this->getIteratorInstance()->getBinaryFileInstance()->preAllocateFile('index');
 217
 218                         // Then write file header
 219                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Invoking this->iteratorInstance->binaryFileInstance->createFileHeader() ...');
 220                         $this->getIteratorInstance()->getBinaryFileInstance()->createFileHeader();
 221
 222                         // Mark as freshly created
 223                         $created = true;
 224                 }
 225
 226                 // Load the file header
 227                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Invoking this->readIndexHeader() ...');
 228                 $this->readIndexHeader();
 229
 230                 // Freshly created?
 231                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: created=%d', intval($created)));
 232                 if (!$created) {
 233                         // Analyze file structure
 234                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Invoking this->iteratorInstance->binaryFileInstance->analyzeFileStructure() ...');
 235                         $this->getIteratorInstance()->getBinaryFileInstance()->analyzeFileStructure();
 236                 }
 237
 238                 // Trace message
 239                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 240         }
 241
 242         /**
 243          * Calculates minimum length for one entry/block
 244          *
 245          * @return      $length         Minimum length for one entry/block
 246          */
 247         public function calculateMinimumBlockLength () {
 248                 // Is it "cached"?
 249                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 250                 if (self::$minimumBlockLength == 0) {
 251                         // Calulcate it
 252                         //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calculating ...');
 253                         self::$minimumBlockLength = (
 254                                 // Type
 255                                 BinaryFile::LENGTH_TYPE + strlen(chr(BinaryFile::SEPARATOR_TYPE_POSITION)) +
 256                                 // Position
 257                                 BinaryFile::LENGTH_POSITION + strlen(chr(BinaryFile::SEPARATOR_ENTRIES))
 258                         );
 259                 }
 260
 261                 // Return it
 262                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: self::minimumBlockLength=%d - EXIT!', self::$minimumBlockLength));
 263                 return self::$minimumBlockLength;
 264         }
 265
 266         /**
 267          * "Getter" for file size
 268          *
 269          * @return      $fileSize       Size of currently loaded file
 270          */
 271         public function getFileSize () {
 272                 // Call iterator's method
 273                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 274                 $fileSize = $this->getIteratorInstance()->getBinaryFileInstance()->getFileSize();
 275
 276                 // Return it
 277                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: fileSize=%d - EXIT!', $fileSize));
 278                 return $fileSize;
 279         }
 280
 281         /**
 282          * Searches for next suitable gap the given length of data can fit in
 283          * including padding bytes.
 284          *
 285          * @param       $length                 Length of raw data
 286          * @return      $seekPosition   Found next gap's seek position
 287          * @throws      InvalidArgumentException        If the parameter is not valid
 288          * @todo        Unfinished work
 289          */
 290         public function searchNextGap (int $length) {
 291                 // Validate parameter
 292                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: length=%d - CALLED!', $length));
 293                 if ($length <= 0) {
 294                         // Throw IAE
 295                         throw new InvalidArgumentException(sprintf('length=%d is not valid', $length));
 296                 }
 297
 298                 // Debug message
 299                 /* DEBUG-DIE: */ ApplicationEntryPoint::exitApplication(sprintf('[%s:%d]: length=%d,this=%s', __METHOD__, __LINE__, $length, print_r($this, true)));
 300         }
 301
 302         /**
 303          * Writes at given position by seeking to it.
 304          *
 305          * @param       $seekPosition   Seek position in file
 306          * @param       $dataStream             Data to be written
 307          * @return      mixed                   Number of writes bytes or false on error
 308          * @throws      OutOfBoundsException    If the position is not seekable
 309          * @throws      InvalidArgumentException        If a parameter is not valid
 310          */
 311         public function writeAtPosition (int $seekPosition, string $dataStream) {
 312                 // Validate parameter
 313                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: seekPosition=%d,dataStream(%d)=%s - CALLED!', $seekPosition, strlen($dataStream), $dataStream));
 314                 if ($seekPosition < 0) {
 315                         // Invalid seek position
 316                         throw new OutOfBoundsException(sprintf('seekPosition=%d is not valid.', $seekPosition));
 317                 } elseif (empty($dataStream)) {
 318                         // Empty dataStream
 319                         throw new InvalidArgumentException('Parameter "dataStream" is empty', FrameworkInterface::EXCEPTION_INVALID_ARGUMENT);
 320                 }
 321
 322                 // Call iterated object's method
 323                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Invoking this->iteratorInstance->binaryFileInstance->writeAtPosition(%d, %s) ...', $seekPosition, $dataStream));
 324                 $status = $this->getIteratorInstance()->getBinaryFileInstance()->writeAtPosition($seekPosition, $dataStream);
 325
 326                 // Return status
 327                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: status[%s]=%d - EXIT!', gettype($status), $status));
 328                 return $status;
 329         }
 330
 331         /**
 332          * Checks if this index has been fully and properly loaded.
 333          *
 334          * @return      $isLoaded       Whether this index has been loaded
 335          */
 336         public function isIndexLoaded () {
 337                 // Is the file gaps-only?
 338                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 339                 if ($this->getIteratorInstance()->getBinaryFileInstance()->isFileGapsOnly()) {
 340                         // Then skip below code as this implies the file has been fully analyzed and "loaded"
 341                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Underlaying file is gaps-only: Returning TRUE ... - EXIT!');
 342                         return TRUE;
 343                 }
 344
 345                 // Debug message
 346                 /* DEBUG-DIE: */ ApplicationEntryPoint::exitApplication(sprintf('[%s:%d]: this=%s', __METHOD__, __LINE__, print_r($this, true)));
 347         }
 348
 349 }