assert_options(ASSERT_BAIL, 1);
debug_off();
$dbResult = mysql_query("select * from lexems where lexem_id not in " . "(select LexemId from LexemDefinitionMap, Definition " . "where DefinitionId = Definition.Id " . "and SourceId not in (6, 8) and Status = 0)");
$seen = 0;
$split = 0;
while (($dbRow = mysql_fetch_assoc($dbResult)) != null) {
$l = Lexem::createFromDbRow($dbRow);
$defs = Definition::loadByLexemId($l->id);
if (!count($defs)) {
continue;
}
$seen++;
// Remove the -ul accent where possible.
if (text_endsWith($l->form, 'ul')) {
$form = substr($l->form, 0, strlen($l->form) - 2);
$otherLexems = Lexem::loadByForm($form);
if (count($otherLexems)) {
print "REMOVING -UL FROM: {$l->form}\n";
foreach ($otherLexems as $otherLexem) {
foreach ($defs as $def) {
LexemDefinitionMap::associate($otherLexem->id, $def->id);
}
}
$l->delete();
$split++;
continue;
}
}
// Split the word in two, if it leads to two existing lexems, and if both
// have at least three letters.
if (mb_strlen($l->unaccented) >= 8 && (text_endsWith($l->form, 'lui') || text_endsWith($l->form, 'ei') || text_endsWith($l->form, 'ii') || text_endsWith($l->form, 'elor') || text_endsWith($l->form, 'ilor') || text_endsWith($l->form, 'asă') || text_endsWith($l->form, 'scă'))) {
