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 - 2020 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                 // Then read it (see constructor for calculation)
  69                 $data = $this->getIteratorInstance()->read($this->getIteratorInstance()->getHeaderSize());
  70
  71                 // Have all requested bytes been read?
  72                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Read %d bytes (%d wanted).', strlen($data), $this->getIteratorInstance()->getHeaderSize()));
  73                 if (strlen($data) != $this->getIteratorInstance()->getHeaderSize()) {
  74                         // Invalid header length
  75                         throw new UnexpectedValueException(sprintf('data(%d)=%s is not expected length %d',
  76                                 strlen($data),
  77                                 $data,
  78                                 $this->getIteratorInstance()->getHeaderSize()
  79                         ));
  80                 } elseif (empty(trim($data, chr(0)))) {
  81                         // Empty file header
  82                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: File header is empty - EXIT!');
  83                         return;
  84                 } elseif (substr($data, -1, 1) != chr(BaseBinaryFile::SEPARATOR_HEADER_ENTRIES)) {
  85                         // Bad last character
  86                         throw new UnexpectedValueException(sprintf('data=%s does not end with "%s"',
  87                                 $data,
  88                                 chr(BaseBinaryFile::SEPARATOR_HEADER_ENTRIES)
  89                         ));
  90                 }
  91
  92                 // Okay, then remove it
  93                 $data = substr($data, 0, -1);
  94
  95                 // And update seek position
  96                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->updateSeekPosition() ...');
  97                 $this->getIteratorInstance()->updateSeekPosition();
  98
  99                 /*
 100                  * Now split it:
 101                  *
 102                  * 0 => magic
 103                  * 1 => total entries
 104                  */
 105                 $header = explode(chr(BaseBinaryFile::SEPARATOR_HEADER_DATA), $data);
 106
 107                 // Check if the array has only 3 elements
 108                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: header()=%d', count($header)));
 109                 //* PRINTR-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: header(%d)=%s', count($header), print_r($header, true)));
 110                 if (count($header) != 2) {
 111                         // Bad header
 112                         throw new UnexpectedValueException(sprintf('header()=%d is not expected value 2', count($header)));
 113                 } elseif ($header[0] !== Indexable::INDEX_MAGIC) {
 114                         // Magic must be in first element
 115                         throw new UnexpectedValueException(sprintf('header[0]=%s is not the expected magic (%s)', $header[0], Indexable::INDEX_MAGIC));
 116                 } elseif (strlen($header[1]) != BaseBinaryFile::LENGTH_COUNT) {
 117                         // Length of total entries not matching
 118                         throw new UnexpectedValueException(sprintf('header[1](%d)=%s does not have expected length %d', strlen($header[1]), $header[1], BaseBinaryFile::LENGTH_COUNT));
 119                 }
 120
 121                 // Decode count
 122                 $header[1] = hex2bin($header[1]);
 123
 124                 // Set it here
 125                 $this->getIteratorInstance()->setHeader($header);
 126
 127                 // Trace message
 128                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 129         }
 130
 131         /**
 132          * Flushes the file header
 133          *
 134          * @return      void
 135          */
 136         public function flushFileHeader () {
 137                 // Put all informations together
 138                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 139                 $header = sprintf('%s%s%s%s',
 140                         // Magic
 141                         Indexable::INDEX_MAGIC,
 142
 143                         // Separator header data
 144                         chr(BaseBinaryFile::SEPARATOR_HEADER_DATA),
 145
 146                         // Total entries
 147                         str_pad(StringUtils::dec2hex($this->getIteratorInstance()->getCounter()), BaseBinaryFile::LENGTH_COUNT, '0', STR_PAD_LEFT),
 148
 149                         // Separator header<->entries
 150                         chr(BaseBinaryFile::SEPARATOR_HEADER_ENTRIES)
 151                 );
 152
 153                 // Write it to disk (header is always at seek position 0)
 154                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Calling this->iteratorInstance->writeAtPosition(0, header=%s) ...', $header));
 155                 $this->getIteratorInstance()->writeAtPosition(0, $header);
 156
 157                 // Trace message
 158                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 159         }
 160
 161         /**
 162          * Initializes this index
 163          *
 164          * @param       $fileInfoInstance       An instance of a SplFileInfo class
 165          * @return      void
 166          * @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.
 167          */
 168         protected function initIndex (SplFileInfo $fileInfoInstance) {
 169                 // Get a file i/o pointer instance for index file
 170                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: fileInfoInstance[%s]=%s - CALLED!', get_class($fileInfoInstance), $fileInfoInstance));
 171                 $fileInstance = ObjectFactory::createObjectByConfiguredName('index_file_class', array($fileInfoInstance, $this));
 172
 173                 // Get iterator instance
 174                 $iteratorInstance = ObjectFactory::createObjectByConfiguredName('file_iterator_class', [$fileInstance]);
 175
 176                 // Set iterator here
 177                 $this->setIteratorInstance($iteratorInstance);
 178
 179                 // Calculate header size
 180                 $headerSize = (
 181                         strlen(Indexable::INDEX_MAGIC) +
 182                         strlen(chr(BaseBinaryFile::SEPARATOR_HEADER_DATA)) +
 183                         BaseBinaryFile::LENGTH_COUNT +
 184                         strlen(chr(BaseBinaryFile::SEPARATOR_HEADER_ENTRIES))
 185                 );
 186
 187                 // Set it
 188                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Setting headerSize=%d ...', $headerSize));
 189                 $this->getIteratorInstance()->setHeaderSize($headerSize);
 190
 191                 // Init counters and gaps array
 192                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->initCountersGapsArray() ...');
 193                 $this->getIteratorInstance()->initCountersGapsArray();
 194
 195                 // Default is not created
 196                 $created = false;
 197
 198                 // Is the file's header initialized?
 199                 if (!$this->getIteratorInstance()->isFileHeaderInitialized()) {
 200                         // First pre-allocate a bit
 201                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->preAllocateFile(index) ...');
 202                         $this->getIteratorInstance()->preAllocateFile('index');
 203
 204                         // Then write file header
 205                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->createFileHeader() ...');
 206                         $this->getIteratorInstance()->createFileHeader();
 207
 208                         // Mark as freshly created
 209                         $created = true;
 210                 }
 211
 212                 // Load the file header
 213                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->readIndexHeader() ...');
 214                 $this->readIndexHeader();
 215
 216                 // Freshly created?
 217                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: created=%d', intval($created)));
 218                 if (!$created) {
 219                         // Analyze file structure
 220                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calling this->iteratorInstance->analyzeFileStructure() ...');
 221                         $this->getIteratorInstance()->analyzeFileStructure();
 222                 }
 223
 224                 // Trace message
 225                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: EXIT!');
 226         }
 227
 228         /**
 229          * Calculates minimum length for one entry/block
 230          *
 231          * @return      $length         Minimum length for one entry/block
 232          */
 233         public function calculateMinimumBlockLength () {
 234                 // Is it "cached"?
 235                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: CALLED!');
 236                 if (self::$minimumBlockLength == 0) {
 237                         // Calulcate it
 238                         //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-FILE-INDEX: Calculating ...');
 239                         self::$minimumBlockLength = (
 240                                 // Type
 241                                 BaseBinaryFile::LENGTH_TYPE + strlen(chr(BaseBinaryFile::SEPARATOR_TYPE_POSITION)) +
 242                                 // Position
 243                                 BaseBinaryFile::LENGTH_POSITION + strlen(chr(BaseBinaryFile::SEPARATOR_ENTRIES))
 244                         );
 245                 }
 246
 247                 // Return it
 248                 //* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: self::minimumBlockLength=%d - EXIT!', self::$minimumBlockLength));
 249                 return self::$minimumBlockLength;
 250         }
 251
 252         /**
 253          * "Getter" for file size
 254          *
 255          * @return      $fileSize       Size of currently loaded file
 256          */
 257         public function getFileSize () {
 258                 // Call iterator's method
 259                 return $this->getIteratorInstance()->getFileSize();
 260         }
 261
 262         /**
 263          * Searches for next suitable gap the given length of data can fit in
 264          * including padding bytes.
 265          *
 266          * @param       $length                 Length of raw data
 267          * @return      $seekPosition   Found next gap's seek position
 268          * @throws      InvalidArgumentException        If the parameter is not valid
 269          * @todo        Unfinished work
 270          */
 271         public function searchNextGap (int $length) {
 272                 // Validate parameter
 273                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: length=%d - CALLED!', $length));
 274                 if ($length <= 0) {
 275                         // Throw IAE
 276                         throw new InvalidArgumentException(sprintf('length=%d is not valid', $length));
 277                 }
 278
 279                 // Partial stub!
 280                 $this->partialStub('length=' . $length);
 281         }
 282
 283         /**
 284          * Writes at given position by seeking to it.
 285          *
 286          * @param       $seekPosition   Seek position in file
 287          * @param       $dataStream             Data to be written
 288          * @return      mixed                   Number of writes bytes or false on error
 289          * @throws      OutOfBoundsException    If the position is not seekable
 290          * @throws      InvalidArgumentException        If a parameter is not valid
 291          */
 292         public function writeAtPosition (int $seekPosition, string $dataStream) {
 293                 // Validate parameter
 294                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: seekPosition=%d,dataStream(%d)=%s - CALLED!', $seekPosition, strlen($dataStream), $dataStream));
 295                 if ($seekPosition < 0) {
 296                         // Invalid seek position
 297                         throw new OutOfBoundsException(sprintf('seekPosition=%d is not valid.', $seekPosition));
 298                 } elseif (empty($dataStream)) {
 299                         // Empty dataStream
 300                         throw new InvalidArgumentException('Parameter "dataStream" is empty');
 301                 }
 302
 303                 // Call iterated object's method
 304                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: Calling this->iteratorInstance->writeAtPosition(%d, %s) ...', $seekPosition, $dataStream));
 305                 $status = $this->getIteratorInstance()->writeAtPosition($seekPosition, $dataStream);
 306
 307                 // Return status
 308                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput(sprintf('BASE-FILE-INDEX: status[%s]=%d - EXIT!', gettype($status), $status));
 309                 return $status;
 310         }
 311
 312         /**
 313          * Checks if this index has been fully and properly loaded.
 314          *
 315          * @return      $isLoaded       Whether this index has been loaded
 316          */
 317         public function isIndexLoaded () {
 318                 // Is the file gaps-only?
 319                 /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-INDEX: CALLED!');
 320                 if ($this->getIteratorInstance()->isFileGapsOnly()) {
 321                         // Then skip below code as this implies the file has been fully analyzed and "loaded"
 322                         /* NOISY-DEBUG: */ self::createDebugInstance(__CLASS__, __LINE__)->debugOutput('BASE-INDEX: Underlaying file is gaps-only: Returning TRUE ... - EXIT!');
 323                         return TRUE;
 324                 }
 325
 326                 // Debug message
 327                 /* DEBUG-DIE: */ ApplicationEntryPoint::exitApplication(sprintf('[%s:%d]: this=%s', __METHOD__, __LINE__, print_r($this, true)));
 328         }
 329
 330 }