/** * Decode a given raw MARC record * * Port of Andy Lesters MARC::File::USMARC->decode() Perl function into PHP. * * @param string $text Raw MARC record * * @return File_MARC_Record Decoded File_MARC_Record object */ private function _decode($text) { $marc = new File_MARC_Record($this); $matches = array(); if (!preg_match("/^(\\d{5})/", $text, $matches)) { $marc->addWarning(File_MARC_Exception::formatError(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_NONNUMERIC_LENGTH], array("record_length" => substr($text, 0, 5)))); } // Store record length $record_length = $matches[1]; if ($record_length != strlen($text)) { $marc->addWarning(File_MARC_Exception::formatError(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_INCORRECT_LENGTH], array("record_length" => $record_length, "actual" => strlen($text)))); // give up and set the record length to the actual byte length $record_length = strlen($text); } if (substr($text, -1, 1) != File_MARC::END_OF_RECORD) { throw new File_MARC_Exception(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_INVALID_TERMINATOR], File_MARC_Exception::ERROR_INVALID_TERMINATOR); } // Store leader $marc->setLeader(substr($text, 0, File_MARC::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, File_MARC::LEADER_LEN, $data_start - File_MARC::LEADER_LEN - 1); // -1 to allow for \x1e at end of directory // character after the directory must be \x1e if (substr($text, $data_start - 1, 1) != File_MARC::END_OF_FIELD) { $marc->addWarning(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_NO_DIRECTORY]); } // All directory entries 12 bytes long, so length % 12 must be 0 if (strlen($dir) % File_MARC::DIRECTORY_ENTRY_LEN != 0) { $marc->addWarning(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_INVALID_DIRECTORY_LENGTH]); } // go through all the fields $nfields = strlen($dir) / File_MARC::DIRECTORY_ENTRY_LEN; for ($n = 0; $n < $nfields; $n++) { // As pack returns to key 1, leave place 0 in list empty list(, $tag) = unpack("A3", substr($dir, $n * File_MARC::DIRECTORY_ENTRY_LEN, File_MARC::DIRECTORY_ENTRY_LEN)); list(, $len) = unpack("A3/A4", substr($dir, $n * File_MARC::DIRECTORY_ENTRY_LEN, File_MARC::DIRECTORY_ENTRY_LEN)); list(, $offset) = unpack("A3/A4/A5", substr($dir, $n * File_MARC::DIRECTORY_ENTRY_LEN, File_MARC::DIRECTORY_ENTRY_LEN)); // Check directory validity if (!preg_match("/^[0-9A-Za-z]{3}\$/", $tag)) { $marc->addWarning(File_MARC_Exception::formatError(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_INVALID_DIRECTORY_TAG], array("tag" => $tag))); } if (!preg_match("/^\\d{4}\$/", $len)) { $marc->addWarning(File_MARC_Exception::formatError(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_INVALID_DIRECTORY_TAG_LENGTH], array("tag" => $tag, "len" => $len))); } if (!preg_match("/^\\d{5}\$/", $offset)) { $marc->addWarning(File_MARC_Exception::formatError(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_INVALID_DIRECTORY_OFFSET], array("tag" => $tag, "offset" => $offset))); } if ($offset + $len > $record_length) { $marc->addWarning(File_MARC_Exception::formatError(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_INVALID_DIRECTORY], array("tag" => $tag))); } $tag_data = substr($text, $data_start + $offset, $len); if (substr($tag_data, -1, 1) == File_MARC::END_OF_FIELD) { /* get rid of the end-of-tag character */ $tag_data = substr($tag_data, 0, -1); $len--; } else { $marc->addWarning(File_MARC_Exception::formatError(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_FIELD_EOF], array("tag" => $tag))); } if (preg_match("/^\\d+\$/", $tag) and $tag < 10) { $marc->appendField(new File_MARC_Control_Field($tag, $tag_data)); } else { $subfields = explode(File_MARC::SUBFIELD_INDICATOR, $tag_data); $indicators = array_shift($subfields); if (strlen($indicators) != 2) { $errorMessage = File_MARC_Exception::formatError(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_INVALID_INDICATORS], array("tag" => $tag, "indicators" => $indicators)); $marc->addWarning($errorMessage); // Do the best with the indicators we've got if (strlen($indicators) == 1) { $ind1 = $indicators; $ind2 = " "; } else { list($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) { $subfield_data[] = new File_MARC_Subfield(substr($subfield, 0, 1), substr($subfield, 1)); } else { $errorMessage = File_MARC_Exception::formatError(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_EMPTY_SUBFIELD], array("tag" => $tag)); $marc->addWarning($errorMessage); } } if (!isset($subfield_data)) { $errorMessage = File_MARC_Exception::formatError(File_MARC_Exception::$messages[File_MARC_Exception::ERROR_EMPTY_SUBFIELD], array("tag" => $tag)); $marc->addWarning($errorMessage); } // If the data is invalid, let's just ignore the one field try { $new_field = new File_MARC_Data_Field($tag, $subfield_data, $ind1, $ind2); $marc->appendField($new_field); } catch (Exception $e) { $marc->addWarning($e->getMessage()); } } } return $marc; }
/** * Decode a given MARCXML record * * @param string $text MARCXML record element * * @return File_MARC_Record Decoded File_MARC_Record object */ private function _decode($text) { $marc = new File_MARC_Record($this); // Store leader $marc->setLeader($text->leader); // go through all the control fields foreach ($text->controlfield as $controlfield) { $controlfieldattributes = $controlfield->attributes(); $marc->appendField(new File_MARC_Control_Field((string) $controlfieldattributes['tag'], $controlfield)); } // go through all the data fields foreach ($text->datafield as $datafield) { $datafieldattributes = $datafield->attributes(); $subfield_data = array(); foreach ($datafield->subfield as $subfield) { $subfieldattributes = $subfield->attributes(); $subfield_data[] = new File_MARC_Subfield((string) $subfieldattributes['code'], $subfield); } // If the data is invalid, let's just ignore the one field try { $new_field = new File_MARC_Data_Field((string) $datafieldattributes['tag'], $subfield_data, $datafieldattributes['ind1'], $datafieldattributes['ind2']); $marc->appendField($new_field); } catch (Exception $e) { $marc->addWarning($e->getMessage()); } } return $marc; }