Simple interface for working with MARC records using the File_MARC package
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.

Build Status Coverage StyleCI Scrutinizer Code Quality Code Climate Latest Stable Version Total Downloads


A small PHP package providing a simple interface to work with MARC21 records on top of the excellent File_MARC package.

Works with both Binary MARC and MARCXML (namespaced or not), but not the various Line mode MARC formats. Records can be edited using the editing capabilities of File_MARC.

Note that version 0.3.0 introduced a few breaking changes. See releases for more information.

Installation using Composer:

If you have Composer installed, the package can be installed by running

composer require scriptotek/marc

Reading records

Use Collection::fromFile or Collection::fromString to read one or more MARC records from a file or string. The methods autodetect the data format (Binary XML or MARCXML) and whether the XML is namespaced or not.

use Scriptotek\Marc\Collection;

$collection = Collection::fromFile($someFileName);
foreach ($collection as $record) {
    echo $record->getField('250')->getSubfield('a')->getData() . "\n";

The $collection object is an iterator. If you rather want a normal array, for instance in order to count the number of records, you can get that from $collection->toArray().

The loader can extract MARC records from any container XML, so you can pass in an SRU or OAI-PMH response directly:

$response = file_get_contents('' . http_build_query([
    'operation'      => 'searchRetrieve',
    'recordSchema'   => 'marcxml',
    'version'        => '1.1',
    'maximumRecords' => '10',
    'query'          => 'bath.isbn=0761532692',

$records = Collection::fromString($response);
foreach ($records as $record) {

If you only have a single record, you can also use Record::fromFile or Record::fromString. These use the Collection methods under the hood, but returns a single Record object.

use Scriptotek\Marc\Record;

$record = Record::fromFile($someFileName);

Querying with MARCspec

Use the Record::query() method to query a record using the MARCspec language as implemented in the php-marc-spec package package. The method returns a QueryResult object, which is a small wrapper around File_MARC_Reference.

Example: To loop over all 650 fields having $2 noubomn:

foreach ($record->query('650{$2=\noubomn}') as $field) {
   echo $field->getSubfield('a')->getData();

or we could reference the subfield directly, like so:

foreach ($record->query('650$a{$2=\noubomn}') as $subfield) {
   echo $subfield->getData();

You can retrieve single results using first(), which returns the first match, or null if no matches were found:


In the same way, text() returns the data content of the first match, or null if no matches were found:


Convenience methods on the Record class

The Record class extends File_MARC_Record with a few convenience methods to get data from commonly used fields. Each of these methods, except getType(), returns an object or an array of objects of one of the field classes (located in src/Fields). For instance getIsbns() returns an array of Scriptotek\Marc\Isbn objects. All the field classes implements at minimum a __toString() method so you easily can get a string representation of the field for presentation purpose.

Note that all the get methods can also be accessed as attributes thanks to a little PHP magic (__get). So instead of calling $record->getId(), you can use the shorthand variant $record->id.


$record->getType() or $record->type returns either 'Bibliographic', 'Authority' or 'Holdings' based on the value of the sixth character in the leader. See Marc21.php for supporting constants.

if ($record->type == Marc21::BIBLIOGRAPHIC) {
    // ...


$record->getCatalogingForm() or $record->catalogingForm returns the value of LDR/18. See Marc21.php for supporting constants.


$record->getId() or $record->id returns the record id from 001 control field.


$record->getIsbns() or $record->isbns returns an array of Isbn objects from 020 fields.

use Scriptotek\Marc\Record;

$record = Record::fromString('<?xml version="1.0" encoding="UTF-8" ?>
  <record xmlns="""">
    <leader>99999cam a2299999 u 4500</leader>
    <controlfield tag="001">98218834x</controlfield>
    <datafield tag="020" ind1=" " ind2=" ">
      <subfield code="a">8200424421</subfield>
      <subfield code="q">h.</subfield>
      <subfield code="c">Nkr 98.00</subfield>
$isbn = $record->isbns[0];

// Get the string representation of the field:
echo $isbn . "\n";  // '8200424421'

// Get the value of $q using the standard FILE_MARC interface:
echo $isbn->getSubfield('q')->getData() . "\n";  // 'h.'

// or using the shorthand `sf()` method from the Field class:
echo $isbn->sf('q') . "\n";  // 'h.'


$record->getTitle() or $record->title returns a Title objects from 245 field, or null if no such field is present.

Beware that the default string representation may or may not fit your needs. It's currently a concatenation of $a (title), $b (remainder of title), $n(part number) and $p (part title). For the remaining subfields like $f, $g and $k, I haven't decided whether to handle them or not.

Parallel titles are unfortunately encoded in such a way that there's no way I'm aware of to identify them in a secure manner, meaning there's also no secure way to remove them if you don't want to include them.1

I'm trimming off any final '/' ISBD marker. I would have loved to be able to also trim off final dots, but that's not trivial for the same reason identifying parallel titles is not1 – there's just no safe way to tell if the final dot is an ISBD marker or part of the title.2 Since explicit ISBD markers are included in records catalogued in the American tradition, but not in records catalogued in the British tradition, a mix of records from both traditions will look silly.


$record->getSubjects($vocabulary, $tag) or $record->subjects returns an array of Subject and UncontrolledSubject objects from all the 6XX fields. The getSubjects() method have two optional arguments you can use to limit by vocabulary and/or tag.

foreach ($record->getSubjects('mesh', Subject::TOPICAL_TERM) as $subject) {
    echo "{$subject->vocabulary} {$subject->type} {$subject}";

Static options:

  • Subject::glue (default: :) defines what string is used to glue the subfields together in the string representation. For instance, 650 $aPhysics $xHistory $yHistory becomes Physics : History : 20th century when using : as glue, or Physics--History--20th century with '--'.
  • Subject::chopPunctuation (default: true) defines if ending punctuation (.:,;/) is to be chopped off at the end of subjects. Usually, any ending punctuation is an ISBD character that can be safely chopped off, but it might also indicate an abbreviation, and unfortunately there is no way to know.


It's unfortunately easy to err when trying to present data from MARC records in end user applications. A developer learning by example might for instance assume that 300 $a is a subfield for "number of pages".3 A quick glance at e.g. LC's MARC documentation would be enough to prove that wrong, but in other cases it's harder to avoid making false assumptions without deep familiarity with cataloguing rules and practices.

1 That might change in the future. But even if I decide to remove parallel titles, I'm not really sure how to do it in a safe way. Parallel titles are identified by a leading = ISBD marker. If the marker is at the end of subfield $a, we can be certain it's an ISBD marker, but since the $a and $c subfields are not repeatable, multiple titles are just added to the $c subfield. So if we encounter an = sign in the middle middle of $c somewhere, how can we tell if it's an ISBD marker or just an equal sign part of the title (like in the fictive book "$aEating the right way : The 2 + 2 = 5 diet")? Some kind of escaping would have made that clear, but the ISBD principles doesn't seem to call for that, leaving us completely in the dark. That is seriously annoying 😩

2 According to ISBD principles "field 245 ends with a period, even when another mark of punctuation is present, unless the last word in the field is an abbreviation, initial/letter, or data that ends with final punctuation." Determining if something is "an abbreviation, initial/letter, or data that ends with final punctuation" is certainly not an easy task for anything but humans and AI.

3 Our old OPAC used to output something like "Number of pages: One video disc (DVD)…" for DVDs – the developers had apparently just assumed that the content of 300 $a could be represented as "number of pages" in all cases. While that sounds silly, getting the number of pages (for documents that actually have pages) from MARC records can be ridiculously hard; you can safely extract the number from strings like 149 p. (English), 149 s. (Norwegian), etc., but you must ignore the numbers in strings like 10 boxes, 11 v. (volumes) etc. So for a start you need a list of valid abbreviations for "pages" in all relevant languages. Then there's the more complicated cases like 1 score (16 p.) – at first sight it looks like we can tokenize that into (number, unit) pairs, like ("1 score", "16 p.") and only accept the item(s) having an allowed unit (like p.). But then suddenly comes a case like "74 p. of ill., 15 p.", which we would turn into ("74 p. of ill.", "15 p."), accepting 15 p., not the correct 74 p.. So we bite into the grass and start writing rules; if a valid match is found as the start of the string, then accept it, else if …, else try tokenization, etc... it quickly becomes messy and it will certainly fail in some cases. Sad to say, after a few years in the library, I still haven't figured out a general way to extract the number of pages a document have using library data.