/**
* Initiates an new PHPCrawlerResponseHeader.
*
* @param string $header_string A complete response-header as it was send by the server
* @param string $source_url The URL of the website the header was recevied from.
* @internal
*/
public function __construct($header_string, $source_url)
{
$this->header_raw = $header_string;
$this->source_url = $source_url;
$this->http_status_code = PHPCrawlerUtils::getHTTPStatusCode($header_string);
$this->content_type = strtolower(PHPCrawlerUtils::getHeaderValue($header_string, "content-type"));
$this->content_length = strtolower(PHPCrawlerUtils::getHeaderValue($header_string, "content-length"));
$this->cookies = PHPCrawlerUtils::getCookiesFromHeader($header_string, $source_url);
}
function handleDocumentInfo($DocInfo)
{
echo "<table class=intbl>";
// Loop over the output-array and print info if wanted
@reset($this->output_array);
while (list($key) = @each($this->output_array)) {
if ($key == "requested_url") {
$str = '<a href="' . $DocInfo->url . '" target=blank>' . $DocInfo->url . '</a>';
echo "<tr><td width=130><nobr>Page requested:</nobr></td><td width=470>" . $str . "</td></tr>";
}
if ($key == "http_status_code") {
if ($DocInfo->http_status_code) {
$str = $DocInfo->http_status_code;
} else {
$str = "-";
}
echo "<tr><td>HTTP-Status:</td><td>" . $str . "</td></tr>";
}
if ($key == "content_type") {
if ($DocInfo->content_type) {
$str = $DocInfo->content_type;
} else {
$str = "-";
}
echo "<tr><td>Content-Type:</td><td>" . $str . "</td></tr>";
}
if ($key == "content_size") {
$str = PHPCrawlerUtils::getHeaderValue($DocInfo->header, "content-length");
if (trim($str) == "") {
$str = "??";
}
echo "<tr><td>Content-Size:</td><td >" . $str . " bytes</td></tr>";
}
if ($key == "content_received") {
if ($DocInfo->received == true) {
$str = "Yes";
} else {
$str = "No";
}
echo "<tr><td>Content received:</td><td >" . $str . "</td></tr>";
}
if ($key == "content_received_completely") {
if ($DocInfo->received_completely == true) {
$str = "Yes";
} else {
$str = "No";
}
echo "<tr><td><nobr>Received completely:</nobr></td><td >" . $str . "</td></tr>";
}
if ($key == "bytes_received") {
echo "<tr><td>Bytes received:</td><td>" . $DocInfo->bytes_received . " bytes</td></tr>";
}
if ($key == "referer_url") {
if ($DocInfo->referer_url == "") {
$str = "-";
} else {
$str =& $page_data["referer_url"];
}
echo "<tr><td><nobr>Refering URL</nobr>:</td><td >" . $str . "</td></tr>";
}
if ($key == "refering_linkcode") {
if ($DocInfo->refering_linkcode == "") {
$str = "-";
} else {
$str = htmlentities($DocInfo->refering_linkcode);
$str = str_replace("\n", "<br>", $str);
}
echo "<tr><td valign=top><nobr>Refering linkcode:</nobr></td><td >" . $str . "</td></tr>";
}
if ($key == "refering_link_raw") {
if ($DocInfo->refering_link_raw == "") {
$str = "-";
} else {
$str = $DocInfo->refering_link_raw;
}
echo "<tr><td><nobr>Refering Link RAW: </nobr></td><td >" . $str . "</td></tr>";
}
if ($key == "refering_linktext") {
if ($DocInfo->refering_linktext == "") {
$str = "-";
} else {
$str = $DocInfo->refering_linktext;
$str = htmlentities($str);
$str = str_replace("\n", "<br>", $str);
}
echo "<tr><td valign=top><nobr>Refering linktext</nobr>:</td><td >" . $str . "</td></tr>";
}
if ($key == "header_send") {
if ($DocInfo->header_send) {
$str = str_replace("\n", "<br>", trim($DocInfo->header_send));
} else {
$str = "-";
}
echo "<tr><td valign=top>Send header:</td><td >" . $str . "</td></tr>";
}
if ($key == "header") {
if ($DocInfo->header) {
$str = str_replace("\n", "<br>", trim($DocInfo->header));
} else {
$str = "-";
}
echo "<tr><td valign=top>Received header:</td><td >" . $str . "</td></tr>";
}
if ($key == "nr_found_links") {
$str = count($DocInfo->links_found);
echo "<tr><td valign=top>Links found:</td><td >" . $str . "</td></tr>";
}
if ($key == "all_found_links") {
echo "<tr><td valign=top>List of found links:</td>";
echo "<td>";
for ($x = 0; $x < count($DocInfo->links_found_url_descriptors); $x++) {
echo $DocInfo->links_found_url_descriptors[$x]->url_rebuild . "<br>";
}
if (count($DocInfo->links_found_url_descriptors) == 0) {
echo "-";
}
echo "</td>";
echo "</tr>";
}
if ($key == "received_to_file") {
if ($DocInfo->received_to_file) {
$str = "Yes";
} else {
$str = "No";
}
echo "<tr><td valign=top>Received to TMP-file:</td><td >" . $str . "</td></tr>";
}
if ($key == "tmpfile_name_size") {
if ($DocInfo->content_tmp_file) {
$str = $DocInfo->content_tmp_file . " (" . filesize($DocInfo->content_tmp_file) . " bytes)";
} else {
$str = "-";
}
echo "<tr><td valign=top>Content TMP-file:</td><td >" . $str . "</td></tr>";
}
if ($key == "received_to_memory") {
if ($DocInfo->received_to_memory) {
$str = "Yes";
} else {
$str = "No";
}
echo "<tr><td valign=top>Received to memory:</td><td >" . $str . "</td></tr>";
}
if ($key == "memory_content_size") {
echo "<tr><td valign=top>Memory-content-size:</td><td >" . strlen($DocInfo->source) . " bytes</td></tr>";
}
}
// Output error if theres one
if ($DocInfo->error_occured) {
echo "<tr>\n <td class=red>Error:</td>\n <td class=red>" . $DocInfo->error_string . "</td>\n </tr>";
}
echo "</table> <br>";
$this->flushOutput();
}
/**
* Checks whether the content of this page/file should be streamed directly to file.
*
* @param string $response_header The response-header
* @return bool TRUE if the content should be streamed to TMP-file
*/
protected function decideStreamToFile($response_header)
{
if (count($this->receive_to_file_content_types) == 0) {
return false;
}
// Get Content-Type from header
$content_type = PHPCrawlerUtils::getHeaderValue($response_header, "content-type");
// No Content-Type given
if ($content_type == null) {
return false;
}
// Check against the given rules
$receive = PHPCrawlerUtils::checkStringAgainstRegexArray($content_type, $this->receive_to_file_content_types);
return $receive;
}