function read_from_stream($stream, $pdf)
{
$tmp = fread($stream, 2);
if ($tmp != '<<') {
die("Error reading PDF: dictionary read error.");
}
$data = array();
while (true) {
$tok = read_non_whitespace($stream);
if ($tok == '>') {
fread($stream, 1);
break;
}
fseek($stream, -1, 1);
$key = read_object($stream, $pdf);
$tok = read_non_whitespace($stream);
fseek($stream, -1, 1);
$value = read_object($stream, $pdf);
if (in_array($key, array_keys($data))) {
die("Error reading PDF: multiple definitions in dictionary.");
}
$data[$key] = $value;
}
$pos = ftell($stream);
$s = read_non_whitespace($stream);
if ($s == 's' && fread($stream, 5) == 'tream') {
$eol = fread($stream, 1);
while ($eol == ' ') {
$eol = fread($stream, 1);
}
assert($eol == "\n" || $eol == "\r");
if ($eol == "\r") {
fread($stream, 1);
}
$length = $data['/Length'];
if (is_a($length, 'IndirectObject')) {
$t = ftell($stream);
$length = $pdf->get_object($length);
fseek($stream, $t, 0);
}
$data['__streamdata__'] = fread($stream, $length);
$e = read_non_whitespace($stream);
$ndstream = fread($stream, 8);
if ($e + $ndstream != "endstream") {
$pos = ftell($stream);
fseek($stream, -10, 1);
$end = fread($stream, 9);
if ($end == "endstream") {
$data['__streamdata__'] = substr($data['__streamdata__'], 0, -1);
} else {
fseek($stream, $pos, 0);
die("Error reading PDF: Unable to find 'endstream' marker after stream.");
}
}
} else {
fseek($stream, $pos, 0);
}
if (in_array('__streamdata__', array_keys($data))) {
return StreamObject::init_from_dict($data);
} else {
return $data;
}
}
function read($stream)
{
fseek($stream, -1, 2);
$line = '';
while (!$line) {
$line = $this->read_next_end_line($stream);
}
if (substr($line, 0, 5) != '%%EOF') {
die("Error reading PDF: EOF marker not found.");
}
$line = $this->read_next_end_line($stream);
$startxref = (int) $line;
$line = $this->read_next_end_line($stream);
if (substr($line, 0, 9) != 'startxref') {
die("Error reading PDF: startxref not found.");
}
$this->xref = array();
$this->xref_obj_stm = array();
$this->trailer = array();
while (1) {
fseek($stream, $startxref, 0);
$x = fread($stream, 1);
if ($x == "x") {
$ref = fread($stream, 4);
if (substr($ref, 0, 3) != 'ref') {
die("Error reading PDF: xref table read error.");
}
read_non_whitespace($stream);
fseek($stream, -1, 1);
while (1) {
$num = read_object($stream, $this);
read_non_whitespace($stream);
fseek($stream, -1, 1);
$size = read_object($stream, $this);
read_non_whitespace($stream);
fseek($stream, -1, 1);
$cnt = 0;
while ($cnt < $size) {
$line = fread($stream, 20);
if (in_array(substr($line, -1, 1), array('0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 't'))) {
fseek($stream, -1, 1);
}
$tmp = explode(' ', substr($line, 0, 16));
$offset = (int) $tmp[0];
$generation = (int) $tmp[1];
if (!in_array($generation, array_keys($this->xref))) {
$this->xref[$generation] = array();
}
if (!$this->xref[$generation][$num]) {
$this->xref[$generation][$num] = $offset;
}
$cnt += 1;
$num += 1;
}
read_non_whitespace($stream);
fseek($stream, -1, 1);
$trailertag = fread($stream, 7);
if ($trailertag != 'trailer') {
fseek($stream, -7, 1);
} else {
break;
}
}
read_non_whitespace($stream);
fseek($stream, -1, 1);
$new_trailer = read_object($stream, $this);
foreach ($new_trailer as $key => $value) {
if (!in_array($key, array_keys($this->trailer))) {
$this->trailer[$key] = $value;
}
}
if (in_array('/Prev', array_keys($new_trailer))) {
$startxref = $new_trailer['/Prev'];
} else {
break;
}
} else {
if (ctype_digit($x)) {
fseek($stream, -1, 1);
$hdr = $this->read_object_header($stream);
$idnum = $hdr[0];
$generation = $hdr[1];
$xrefstream = read_object($stream, $this);
assert($xrefstream->data['/Type'] == '/XRef');
$this->cache_indirect_object($generation, $idnum, $xrefstream);
$stream_data = $xrefstream->get_data();
$cursor = 0;
$idx_pairs = $xrefstream->data['/Index'];
if (!$idx_pairs) {
$idx_pairs = array(0, $xrefstream->data['/Size']);
}
$entry_sizes = $xrefstream->data['/W'];
foreach ($this->_pairs($idx_pairs) as $pair) {
$num = $pair[0];
$size = $pair[1];
$cnt = 0;
while ($cnt < $size) {
for ($i = 0; $i < count($entry_sizes); $i++) {
$d = substr($stream_data, $cursor, $entry_sizes[$i]);
$cursor += $entry_sizes[$i];
$di = convert_to_int($d, $entry_sizes[$i]);
if ($i == 0) {
$xref_type = $di;
} else {
if ($i == 1) {
if ($xref_type == 0) {
$next_free_object = $di;
} else {
if ($xref_type == 1) {
$byte_offset = $di;
} else {
if ($xref_type == 2) {
$objstr_num = $di;
}
}
}
} else {
if ($i == 2) {
if ($xref_type == 0) {
$next_generation = $di;
} else {
if ($xref_type == 1) {
$generation = $di;
} else {
if ($xref_type == 2) {
$objstr_idx = $di;
}
}
}
}
}
}
}
if ($xref_type == 0) {
} else {
if ($xref_type == 1) {
if (!in_array($generation, array_keys($this->xref))) {
$this->xref[$generation] = array();
}
if (!in_array($num, array_keys($this->xref[$generation]))) {
$this->xref[$generation][$num] = $byte_offset;
}
} else {
if ($xref_type == 2) {
if (!in_array($num, array_keys($this->xref_obj_stm))) {
$this->xref_obj_stm[$num] = array($objstr_num, $objstr_idx);
}
}
}
}
$cnt += 1;
$num += 1;
}
}
$trailer_keys = array('/Root', '/Info', '/ID');
foreach ($trailer_keys as $key) {
if (in_array($key, array_keys($xrefstream->data)) and !in_array($key, array_keys($this->trailer))) {
$this->trailer[$key] = $xrefstream->data[$key];
}
}
if (in_array('/Prev', array_keys($xrefstream->data))) {
$startxref = $xrefstream->data['/Prev'];
} else {
break;
}
} else {
fseek($stream, -11, 1);
$tmp = fread($stream, 20);
$xref_loc = strpos($tmp, 'xref');
if ($xref_loc !== -1) {
$startxref -= 10 - $xref_loc;
continue;
} else {
assert(false);
break;
}
}
}
}
// var_dump($this->get_xmp_metadata());
}
