public function extractFromString($string)
{
$body_text = array();
if ($string != '') {
$o_HtmlParser = new HtmlParser();
$a_DataParser = $o_HtmlParser->parser($string);
if (!$a_DataParser) {
return 'Cannot read your data !';
}
/* do smart binary array */
$i = 0;
$j = 0;
$k = 0;
$endcode = array();
$idx = 0;
$token = array();
foreach ($a_DataParser['tokenized_data'] as $index => $sz_tokenizedData) {
/* If is tag */
if ($this->isTag($a_DataParser, $index)) {
$binary_token[$i] = 1;
} elseif ($this->isTagClose($a_DataParser, $index)) {
$binary_token[$i] = 0;
} else {
$binary_token[$i] = -1;
$token[$idx] = $sz_tokenizedData;
$idx++;
}
$i++;
}
for ($k = 0; $k < $i; $k++) {
$x = $binary_token[$k];
/* Add an index for close tag. We will use it when calculate body text */
if ($x == 0) {
$j++;
$endcode[$j] = 0;
$j++;
continue;
}
if (abs($x + $endcode[$j]) < abs($endcode[$j])) {
$j++;
}
$endcode[$j] += $x;
}
/* Extract body text */
$i_max = 0;
$j_max = 0;
$max = 0;
for ($i = 0; $i < count($endcode) - 1; $i++) {
if ($endcode[$i] >= 0) {
continue;
}
for ($j = $i; $j < count($endcode); $j++) {
if ($endcode[$j] >= 0) {
continue;
}
/* Calculate max in range [i .. j] */
$S = $this->i_TagBefore($endcode, $i) + $this->i_fTagAfter($endcode, $j) + $this->i_fTextBetween($endcode, $i, $j);
if ($S > $max) {
$max = $S;
$i_max = $i;
$j_max = $j;
}
}
}
/* Calculate start and end point */
$start = 0;
$end = 0;
for ($i = 0; $i < $i_max; $i++) {
if ($endcode[$i] == 0) {
$start++;
} else {
$start += abs($endcode[$i]);
}
}
for ($i = 0; $i < $j_max; $i++) {
if ($endcode[$i] == 0) {
$end++;
} else {
$end += abs($endcode[$i]);
}
}
$return_text = array();
/* Calculate body text */
//for($i = $start - 1; $i <= $end -1; $i++)
for ($i = 0; $i <= $end - 1; $i++) {
$body_text[] = $token[$i];
}
}
$body_text = implode(' ', $body_text);
return $body_text;
}
