function File::decode in Bibliography Module 7.2

Same name and namespace in other branches

1. 6.2 modules/marcParse/php-marc.php \File::decode()
2. 6 marcParse/php-marc.php \File::decode()
3. 7 modules/marcParse/php-marc.php \File::decode()

Decode a given raw MARC record

"Port" of Andy Lesters MARC::File::USMARC->decode() function into PHP. Ideas and "rules" have been used from USMARC::decode().

Parameters

string Raw MARC record:

Return value

Record Decoded MARC Record object

1 call to File::decode()

File::next in modules/marcParse/php-marc.php

Return next Record-object

File

modules/marcParse/php-marc.php, line 234

Class

File

Class File Class to read MARC records from file(s)

Code

function decode($text) {
  if (!preg_match("/^\\d{5}/", $text, $matches)) {
    $this
      ->_croak('Record length "' . substr($text, 0, 5) . '" is not numeric');
  }
  $marc = new Record();

  // Store record length
  $reclen = $matches[0];
  if ($reclen != strlen($text)) {
    $this
      ->_croak("Invalid record length: Leader says {$reclen} bytes, but it's actually " . strlen($text));
  }
  if (substr($text, -1, 1) != END_OF_RECORD) {
    $this
      ->_croak("Invalid record terminator");
  }

  // Store leader
  $marc
    ->leader(substr($text, 0, LEADER_LEN));

  // bytes 12 - 16 of leader give offset to the body of the record
  $data_start = 0 + substr($text, 12, 5);

  // immediately after the leader comes the directory (no separator)
  $dir = substr($text, LEADER_LEN, $data_start - LEADER_LEN - 1);

  // -1 to allow for \x1e at end of directory
  // character after the directory must be \x1e
  if (substr($text, $data_start - 1, 1) != END_OF_FIELD) {
    $this
      ->_croak("No directory found");
  }

  // All directory entries 12 bytes long, so length % 12 must be 0
  if (strlen($dir) % DIRECTORY_ENTRY_LEN != 0) {
    $this
      ->_croak("Invalid directory length");
  }

  // go through all the fields
  $nfields = strlen($dir) / DIRECTORY_ENTRY_LEN;
  for ($n = 0; $n < $nfields; $n++) {

    // As pack returns to key 1, leave place 0 in list empty
    list(, $tagno) = unpack("A3", substr($dir, $n * DIRECTORY_ENTRY_LEN, DIRECTORY_ENTRY_LEN));
    list(, $len) = unpack("A3/A4", substr($dir, $n * DIRECTORY_ENTRY_LEN, DIRECTORY_ENTRY_LEN));
    list(, $offset) = unpack("A3/A4/A5", substr($dir, $n * DIRECTORY_ENTRY_LEN, DIRECTORY_ENTRY_LEN));

    // Check directory validity
    if (!preg_match("/^[0-9A-Za-z]{3}\$/", $tagno)) {
      $this
        ->_croak("Invalid tag in directory: \"{$tagno}\"");
    }
    if (!preg_match("/^\\d{4}\$/", $len)) {
      $this
        ->_croak("Invalid length in directory, tag {$tagno}: \"{$len}\"");
    }
    if (!preg_match("/^\\d{5}\$/", $offset)) {
      $this
        ->_croak("Invalid offset in directory, tag {$tagno}: \"{$offset}\"");
    }
    if ($offset + $len > $reclen) {
      $this
        ->_croak("Directory entry runs off the end of the record tag {$tagno}");
    }
    $tagdata = substr($text, $data_start + $offset, $len);
    if (substr($tagdata, -1, 1) == END_OF_FIELD) {

      # get rid of the end-of-tag character
      $tagdata = substr($tagdata, 0, -1);
      --$len;
    }
    else {
      $this
        ->_croak("field does not end in end of field character in tag {$tagno}");
    }
    if (preg_match("/^\\d+\$/", $tagno) && $tagno < 10) {
      $marc
        ->append_fields(new Field($tagno, $tagdata));
    }
    else {
      $subfields = explode(SUBFIELD_INDICATOR, $tagdata);
      $indicators = array_shift($subfields);
      if (strlen($indicators) > 2 || strlen($indicators) == 0) {
        $this
          ->_warn("Invalid indicators \"{$indicators}\" forced to blanks for tag {$tagno}\n");
        list($ind1, $ind2) = array(
          " ",
          " ",
        );
      }
      else {
        $ind1 = substr($indicators, 0, 1);
        $ind2 = substr($indicators, 1, 1);
      }

      // Split the subfield data into subfield name and data pairs
      $subfield_data = array();
      foreach ($subfields as $subfield) {
        if (strlen($subfield) > 0) {
          $subfield_data[substr($subfield, 0, 1)][] = substr($subfield, 1);
        }
        else {
          $this
            ->_warn("Entirely empty subfield found in tag {$tagno}");
        }
      }
      if (!isset($subfield_data)) {
        $this
          ->_warn("No subfield data found {$location} for tag {$tagno}");
      }
      $marc
        ->append_fields(new Field($tagno, $ind1, $ind2, $subfield_data));
    }
  }
  return $marc;
}