public function __construct($html, $encoding, $pageNumber, $paragraphsPerPage = HTMLPager::PARAGRAPH_LIMIT)
{
$dom = new DOMDocument();
libxml_use_internal_errors(true);
libxml_clear_errors();
// clean up any errors belonging to other operations
$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', $encoding));
foreach (libxml_get_errors() as $error) {
Kurogo::log(LOG_WARNING, "HTMLPager got loadHTML warning (line {$error->line}; column {$error->column}) {$error->message}", 'data');
}
libxml_clear_errors();
// free up memory associated with the errors
libxml_use_internal_errors(false);
$body = $dom->getElementsByTagName("body")->item(0);
$currentPage = NULL;
$pages = array();
$currentParagraphCount = 0;
foreach ($body->childNodes as $node) {
if ($currentPage == NULL) {
// need to start a new page
if ($node->nodeName == "#text" && trim($node->nodeValue) == "") {
continue;
// this node is blank so do not start a new page yet
}
$currentPage = new HTMLPage();
$pages[] = $currentPage;
}
$currentPage->addNode($node);
if ($node->nodeName == "p") {
$currentParagraphCount++;
}
if ($currentParagraphCount == $paragraphsPerPage) {
$currentPage = NULL;
$currentParagraphCount = 0;
}
}
$this->pages = $pages;
$this->pageCount = count($pages);
if ($pageNumber >= 0 && $pageNumber < $this->pageCount) {
$this->pageNumber = $pageNumber;
}
}