#TODO: strip from body contents
return $meta[1];
}
+
if (preg_match('{
- </h2> (?: \s+ | <p\sclass="nav\b.*?</p> | <div[^>]*> )* <p> (.*?) </p>
+ </h2> (?: \s+ | <p\sclass="nav\b.*?</p> | <div[^>]*> )* <p> \s* (.*?) </p>
}sx', $this->raw, $bodyp, PREG_OFFSET_CAPTURE)) {
# fallback paragraph contents following the page header
if ($bodyp[1][1] < 256) {
return $bodyp[1][0];
}
}
+
+ # starting paragraph for documents without title (assumed simple/partial)
+ if (strpos($this->raw, '<h2') === FALSE and preg_match('{
+ \A <p> \s* (.*?) </p>
+ }sx', $this->raw, $bodyp)) {
+ return $bodyp[1];
+ }
}
function img()