/**
* Check of articles is based on code from Ian Hamilton. This version is more
* limited in that it focuses on English, Spanish, French, Italian and German
* articles. Certain possible articles have been removed if they are valid
* English non-articles. This version also disregards 008_language/041 codes
* and just uses the list of articles to provide warnings/suggestions.
*
* source for articles = <http://www.loc.gov/marc/bibliographic/bdapp-e.html>
*
* Should work with fields 130, 240, 245, 440, 630, 730, and 830. Reports error
* if another field is passed in.
*
* @param File_MARC_Field $field Field to check
*
* @return void
*/
protected function checkArticle($field)
{
// add articles here as needed
// Some omitted due to similarity with valid words (e.g. the German 'die').
static $article = array('a' => 'eng glg hun por', 'an' => 'eng', 'das' => 'ger', 'dem' => 'ger', 'der' => 'ger', 'ein' => 'ger', 'eine' => 'ger', 'einem' => 'ger', 'einen' => 'ger', 'einer' => 'ger', 'eines' => 'ger', 'el' => 'spa', 'en' => 'cat dan nor swe', 'gl' => 'ita', 'gli' => 'ita', 'il' => 'ita mlt', 'l' => 'cat fre ita mlt', 'la' => 'cat fre ita spa', 'las' => 'spa', 'le' => 'fre ita', 'les' => 'cat fre', 'lo' => 'ita spa', 'los' => 'spa', 'os' => 'por', 'the' => 'eng', 'um' => 'por', 'uma' => 'por', 'un' => 'cat spa fre ita', 'una' => 'cat spa ita', 'une' => 'fre', 'uno' => 'ita');
// add exceptions here as needed
// may want to make keys lowercase
static $exceptions = array('A & E', 'A & ', 'A-', 'A+', 'A is ', 'A isn\'t ', 'A l\'', 'A la ', 'A posteriori', 'A priori', 'A to ', 'El Nino', 'El Salvador', 'L is ', 'L-', 'La Salle', 'Las Vegas', 'Lo mein', 'Los Alamos', 'Los Angeles');
// get tagno to determine which indicator to check and for reporting
$tagNo = $field->getTag();
// retrieve tagno from subfield 6 if 880 field
if ($tagNo == '880' && ($sub6 = $field->getSubfield('6'))) {
$tagNo = substr($sub6->getData(), 0, 3);
}
// $ind holds nonfiling character indicator value
$ind = '';
// $first_or_second holds which indicator is for nonfiling char value
$first_or_second = '';
if (in_array($tagNo, array(130, 630, 730))) {
$ind = $field->getIndicator(1);
$first_or_second = '1st';
} else {
if (in_array($tagNo, array(240, 245, 440, 830))) {
$ind = $field->getIndicator(2);
$first_or_second = '2nd';
} else {
$this->warn('Internal error: ' . $tagNo . " is not a valid field for article checking\n");
return;
}
}
if (!is_numeric($ind)) {
$this->warn($tagNo . ": Non-filing indicator is non-numeric");
return;
}
// get subfield 'a' of the title field
$titleField = $field->getSubfield('a');
$title = $titleField ? $titleField->getData() : '';
$char1_notalphanum = 0;
// check for apostrophe, quote, bracket, or parenthesis, before first word
// remove if found and add to non-word counter
while (preg_match('/^["\'\\[\\(*]/', $title)) {
$char1_notalphanum++;
$title = preg_replace('/^["\'\\[\\(*]/', '', $title);
}
// split title into first word + rest on space, parens, bracket, apostrophe,
// quote, or hyphen
preg_match('/^([^ \\(\\)\\[\\]\'"\\-]+)([ \\(\\)\\[\\]\'"\\-])?(.*)/i', $title, $hits);
$firstword = isset($hits[1]) ? $hits[1] : '';
$separator = isset($hits[2]) ? $hits[2] : '';
$etc = isset($hits[3]) ? $hits[3] : '';
// get length of first word plus the number of chars removed above plus one
// for the separator
$nonfilingchars = strlen($firstword) + $char1_notalphanum + 1;
// check to see if first word is an exception
$isan_exception = false;
foreach ($exceptions as $current) {
if (substr($title, 0, strlen($current)) == $current) {
$isan_exception = true;
break;
}
}
// lowercase chars of $firstword for comparison with article list
$firstword = strtolower($firstword);
// see if first word is in the list of articles and not an exception
$isan_article = !$isan_exception && isset($article[$firstword]);
// if article then $nonfilingchars should match $ind
if ($isan_article) {
// account for quotes, apostrophes, parens, or brackets before 2nd word
if (strlen($separator) && preg_match('/^[ \\(\\)\\[\\]\'"\\-]+/', $etc)) {
while (preg_match('/^[ "\'\\[\\]\\(\\)*]/', $etc)) {
$nonfilingchars++;
$etc = preg_replace('/^[ "\'\\[\\]\\(\\)*]/', '', $etc);
}
}
if ($nonfilingchars != $ind) {
$this->warn($tagNo . ": First word, {$firstword}, may be an article, check " . "{$first_or_second} indicator ({$ind}).");
}
} else {
// not an article so warn if $ind is not 0
if ($ind != '0') {
$this->warn($tagNo . ": First word, {$firstword}, does not appear to be an " . "article, check {$first_or_second} indicator ({$ind}).");
}
}
}
/**
* Destroys the control field
*/
function __destruct()
{
$this->data = null;
parent::__destruct();
}
/**
* Destroys the data field
*/
function __destruct()
{
$this->subfields = null;
$this->ind1 = null;
$this->ind2 = null;
parent::__destruct();
}
