function ProcessElements($reader)
{
for ($element = $reader->Next(); $element != null; $element = $reader->Next()) {
switch ($element->GetType()) {
case Element::e_path:
$data = $element->GetPathData();
$points = $data->GetPoints();
break;
case Element::e_text:
$data = $element->GetTextString();
echo nl2br($data . "\n");
break;
case Element::e_form:
$reader->FormBegin();
ProcessElements($reader);
$reader->End();
break;
}
}
}
echo nl2br("Document has a StructTree root.\n");
for ($i = 0; $i < $tree->GetNumKids(); ++$i) {
// Recursively get structure info for all child elements.
ProcessStructElement($tree->GetKid($i), 0);
}
} else {
echo nl2br("This document does not contain any logical structure.\n");
}
echo nl2br("\nDone 1.\n");
echo nl2br("____________________________________________________________\n");
echo nl2br("Sample 2 - Get parent logical structure elements from\n");
echo nl2br("layout elements.\n");
$reader = new ElementReader();
for ($itr = $doc->GetPageIterator(); $itr->HasNext(); $itr->Next()) {
$reader->Begin($itr->Current());
ProcessElements($reader);
$reader->End();
}
echo nl2br("\nDone 2.\n");
echo nl2br("____________________________________________________________\n");
echo nl2br("Sample 3 - 'XML style' extraction of PDF logical structure and page content.\n");
$mcid_doc_map = array();
$reader = new ElementReader();
for ($itr = $doc->GetPageIterator(); $itr->HasNext(); $itr->Next()) {
$reader->Begin($itr->Current());
$mcid_doc_map[$itr->Current()->GetIndex()] = array();
ProcessElements2($reader, $mcid_doc_map[$itr->Current()->GetIndex()]);
$reader->End();
}
$tree = $doc->GetStructTree();
if ($tree->IsValid()) {
function ProcessElements($reader)
{
while (($element = $reader->Next()) != NULL) {
switch ($element->GetType()) {
case Element::e_path:
ProcessPath($reader, $element);
break;
case Element::e_text_begin:
ProcessText($reader);
break;
case Element::e_form:
$reader->FormBegin();
ProcessElements($reader);
$reader->End();
break;
case Element::e_image:
ProcessImage($element);
break;
}
}
}
$writer = new ElementWriter();
$reader = new ElementReader();
$itr = $doc->GetPageIterator();
while ($itr->HasNext()) {
$page = $itr->Current();
$reader->Begin($page);
$writer->Begin($page, ElementWriter::e_replacement, false);
$map1 = array();
ProcessElements($reader, $writer, $map1);
$writer->End();
$reader->End();
$map2 = array();
while (!(empty($map1) && empty($map2))) {
foreach ($map1 as $k => $v) {
$obj = $v;
$writer->Begin($obj);
$reader->Begin($obj, $page->GetResourceDict());
ProcessElements($reader, $writer, $map2);
$reader->End();
$writer->End();
unset($map1[$k]);
}
if (empty($map1) && !empty($map2)) {
$map1 = $map1 + $map2;
$map2 = array();
}
}
$itr->Next();
}
$doc->Save($output_path . $output_filename, SDFDoc::e_remove_unused);
echo nl2br("Done. Result saved in " . $output_filename . "...\n");
