/**
* Accessor for content reduced to visible text. Acts
* like a text mode browser, normalising space and
* reducing images to their alt text.
* @return string Content as plain text.
* @access public
*/
function getText()
{
return SimplePage::normalise($this->content);
}
function testHtmlEntityTranslation()
{
$this->assertEqual(SimplePage::normalise('<>"&''), '<>"&\'');
}
/**
* Visits the given node and all children
* @param object $node Tidy XML node.
*/
private function walkForm($node, $form, $enclosing_label = '')
{
if ($node->name == 'a') {
$this->page->addLink($this->tags()->createTag($node->name, (array) $node->attribute)->addContent($this->innerHtml($node)));
} elseif (in_array($node->name, array('input', 'button', 'textarea', 'select'))) {
$this->addWidgetToForm($node, $form, $enclosing_label);
} elseif ($node->name == 'label') {
$this->labels[] = $this->tags()->createTag($node->name, (array) $node->attribute)->addContent($this->innerHtml($node));
if ($node->hasChildren()) {
foreach ($node->child as $child) {
$this->walkForm($child, $form, SimplePage::normalise($this->innerHtml($node)));
}
}
} elseif ($node->hasChildren()) {
foreach ($node->child as $child) {
$this->walkForm($child, $form);
}
}
return $form;
}
/**
* Accessor for plain text of page as a text browser
* would see it.
* @return string Plain text of page.
* @access public
*/
function getText()
{
if (!$this->text) {
$this->text = SimplePage::normalise($this->raw);
}
return $this->text;
}
function testUtf8WhitespaceNbsp()
{
$this->assertEqual(SimplePage::normalise(' § ' . html_entity_decode(' ', ENT_COMPAT, 'UTF-8') . '729', 'UTF-8'), '§ 729');
$this->assertEqual(SimplePage::normalise(html_entity_decode(' ', ENT_COMPAT, 'UTF-8'), 'UTF-8'), '');
$this->assertEqual(strlen(SimplePage::normalise(html_entity_decode('a b', ENT_COMPAT, 'UTF-8'), 'UTF-8')), 3);
$this->assertEqual(strlen(SimplePage::normalise(html_entity_decode('a b', ENT_COMPAT, 'ISO-8859-1'), 'ISO-8859-1')), 3);
$this->assertEqual(SimplePage::normalise('» »', 'UTF-8'), '» »');
$this->assertEqual(SimplePage::normalise('» »', 'ISO-8859-1'), utf8_decode('» »'));
# latin1 strings should not get converted to utf8
$this->assertEqual(strlen(SimplePage::normalise(utf8_decode('ä'))), 1);
}
