public function test_construct_and_index()
{
$filename = 'data/big_granito_1.html';
$tmr = self::timer();
$mmr = self::memer();
$html = self::file_get_contents($filename);
$mem = self::memer($mmr);
$exe = self::timer($tmr);
self::log("Loaded " . self::fmtNumber(strlen($html) / 1024, 2) . "Kb\tin\t{$exe}\t{$mem} RAM");
$tmr = self::timer();
$mmr = self::memer();
$doc = new hQuery($html, false);
$mem = self::memer($mmr);
$exe = self::timer($tmr);
self::log("Construct " . self::fmtNumber($doc->size / 1024, 2) . "Kb\tin\t{$exe}\t{$mem} RAM");
$doc->location(self::fn($filename));
$tmr = self::timer();
$mmr = self::memer();
$tags = $doc->index();
$mem = self::memer($mmr);
$exe = self::timer($tmr);
$this->assertLessThan(6000000, self::timer($tmr, false), 'should index 3Mb in less then 3 sec');
$count = self::fmtNumber(self::listSumCounts($tags));
self::log("Indexed {$count} tags\tin\t{$exe}\t{$mem} RAM");
self::log("Original Charset: {$doc->charset}");
$tags = array_map('count', $tags);
$counts = NULL;
foreach ($tags as $k => $v) {
$counts[$v] = (empty($counts[$v]) ? '' : $counts[$v] . ', ') . $k;
}
krsort($counts);
// self::log('Tag counts:', $counts);
return array($doc);
}
public static function setUpBeforeClass()
{
hQuery::$_mockup_class = 'TestHQueryTests';
self::$inst = TestHQueryTests::fromHTML('<doctype html>' . '<html>' . '<head>' . '<meta charset="ISO-8859-2" />' . '<title>Sample HTML Doc</title>' . '</head>' . '<body class="test-class">' . '<div id="test-div" class="test-class test-div">' . 'This is some text' . '<a href="/path">' . 'This is a link' . '</a>' . ' between tags' . '<span id="aSpan">Span text</span>' . '</div>' . 'Contents...' . '</body>' . '</html>', self::$baseUrl . 'index.html');
self::log(get_class(self::$inst));
}
/**
* Executes a HTTP write-read session.
*
* @param string $host - IP/HOST address or URL
* @param array $head - list off HTTP headers to be sent along with the request to $host
* @param mixed $body - data to be sent as the contents of the request. If is array or object, a http query is built.
* @param array $options - list of option as key-value:
* timeout - connection timeout in seconds
* host - goes in headers, overrides $host (ex. $host == '127.0.0.1', $options['host'] == 'www.example.com')
* scheme - http, ssl, tls, udp, ...
* close - whether to close connection o not
*
* @return array [contents, headers, http-status-code, http-status-message]
*
* @author Dumitru Uzun
*
*/
public static function http_wr($host, $head = NULL, $body = NULL, $options = NULL)
{
self::$last_http_result = $ret = new stdClass();
empty($options) and $options = array();
if ($p = strpos($host, '://') and $p < 7) {
$ret->url = $host;
$p = parse_url($host);
if (!$p) {
throw new Exception('Wrong host specified');
}
// error
$host = $p['host'];
$path = @$p['path'];
if (isset($p['query'])) {
$path .= '?' . $p['query'];
}
if (isset($p['port'])) {
$port = $p['port'];
}
unset($p['path'], $p['query']);
$options += $p;
} else {
$p = explode('/', $host, 2);
list($host, $path) = $p;
$p = explode(':', $host, 2);
list($host, $port) = $p;
}
if (strncmp($path, '/', 1)) {
$path = '/' . $path;
}
// isset($path) or $path = '/';
if (!isset($port)) {
if (isset($options['port'])) {
$port = $options['port'];
} else {
switch ($options['scheme']) {
case 'tls':
case 'ssl':
case 'https':
$port = 443;
break;
case 'ftp':
$port = 21;
break;
case 'sftp':
$port = 22;
break;
case 'http':
default:
$port = 80;
}
}
}
$ret->host = $host;
$_h = array('host' => isset($options['host']) ? $options['host'] : $host, 'accept' => 'text/html,application/xhtml+xml,application/xml;q =0.9,*/*;q=0.8');
if (!empty($options['scheme'])) {
switch ($p['scheme']) {
case 'http':
case 'ftp':
break;
case 'https':
$host = 'tls://' . $host;
break;
default:
$host = $options['scheme'] . '://' . $host;
}
}
static $boundary = "\r\n\r\n";
$blen = strlen($boundary);
if ($body) {
if (is_array($body) || is_object($body)) {
$body = http_build_query($body);
$_h['content-type'] = 'application/x-www-form-urlencoded';
}
$body = (string) $body;
$_h['content-length'] = strlen($body);
$body .= $boundary;
empty($options['method']) and $options['method'] = 'POST';
} else {
$body = NULL;
}
$meth = @$options['method'] and $meth = strtoupper($meth) or $meth = 'GET';
if ($head) {
if (!is_array($head)) {
$head = explode("\r\n", $head);
}
foreach ($head as $i => $v) {
if (is_int($i)) {
$v = explode(':', $v, 2);
if (count($v) != 2) {
continue;
}
// Invalid header
list($i, $v) = $v;
}
$i = strtolower(strtr($i, ' _', '--'));
$_h[$i] = trim($v);
}
}
if (@$options['decode'] == 'gzip') {
// if(function_exists('gzdecode')) {
$_h['accept-encoding'] = 'gzip';
// }
// else {
// $options['decode'] = NULL;
// }
}
if (!isset($options['close']) || @$options['close']) {
$_h['connection'] = 'close';
} else {
$_h['connection'] = 'keep-alive';
}
$prot = empty($options['protocol']) ? 'HTTP/1.1' : $options['protocol'];
$head = array("{$meth} {$path} {$prot}");
foreach ($_h as $i => $v) {
$i = explode('-', $i);
foreach ($i as &$j) {
$j = ucfirst($j);
}
$i = implode('-', $i);
$head[] = $i . ': ' . $v;
}
$rqst = implode("\r\n", $head) . $boundary . $body;
$head = NULL;
// free mem
$timeout = isset($options['timeout']) ? $options['timeout'] : @ini_get("default_socket_timeout");
$ret->options = $options;
// ------------------- Connection and data transfer -------------------
$errno = $errstr = $rsps = '';
$h = $_rh = NULL;
$fs = @fsockopen($host, $port, $errno, $errstr, $timeout);
if (!$fs) {
throw new Exception('unable to create socket "' . $host . ':' . $port . '"' . $errstr, $errno);
}
if (!fwrite($fs, $rqst)) {
throw new Exception("unable to write");
} else {
$l = $blen - 1;
// read headers
while ($open = !feof($fs) && ($p = @fgets($fs, 1024))) {
if ($p == "\r\n") {
break;
}
$rsps .= $p;
}
if ($rsps) {
$h = explode("\r\n", rtrim($rsps));
list($rprot, $rcode, $rmsg) = explode(' ', array_shift($h), 3);
foreach ($h as $v) {
$v = explode(':', $v, 2);
$_rh[strtoupper(strtr($v[0], '- ', '__'))] = isset($v[1]) ? trim($v[1]) : NULL;
}
$rsps = NULL;
switch ($rcode) {
case 301:
case 302:
case 303:
case 307:
// repeat request using the same method and post data
if (@$options['redirects'] > 0 && ($loc = @$_rh['LOCATION'])) {
$loc = self::abs_url($loc, (empty($options['scheme']) ? '' : $options['scheme'] . '//') . $host . ':' . $port . (empty($options['path']) ? '' : $options['path']));
unset($_h['host'], $options['host'], $options['port'], $options['scheme'], $options['method']);
--$options['redirects'];
// ??? could save cookies for redirect
return self::http_wr($loc, $_h, $body, $options);
}
break;
}
// Detect body length
if (@(!$open) || $rcode < 200 || $rcode == 204 || $rcode == 304 || $meth == 'HEAD') {
$te = 1;
} elseif (isset($_rh['TRANSFER_ENCODING']) && strtolower($_rh['TRANSFER_ENCODING']) === 'chunked') {
$te = 3;
} elseif (isset($_rh['CONTENT_LENGTH'])) {
$bl = (int) $_rh['CONTENT_LENGTH'];
$te = 2;
}
switch ($te) {
case 1:
break;
case 2:
while ($bl > 0 and $open &= !feof($fs) && ($p = @fread($fs, $bl))) {
$rsps .= $p;
$bl -= strlen($p);
}
break;
case 3:
while ($open &= !feof($fs) && ($p = @fgets($fs, 1024))) {
$_re = explode(';', rtrim($p));
$cs = reset($_re);
$bl = hexdec($cs);
if (!$bl) {
break;
}
// empty chunk
while ($bl > 0 and $open &= !feof($fs) && ($p = @fread($fs, $bl))) {
$rsps .= $p;
$bl -= strlen($p);
}
@fgets($fs, 3);
// \r\n
}
if ($open &= !feof($fs) && ($p = @fgets($fs, 1024))) {
if ($p = rtrim($p)) {
// ??? Trailer Header
$v = explode(':', $p, 2);
$_rh[strtoupper(strtr($v[0], '- ', '__'))] = isset($v[1]) ? trim($v[1]) : NULL;
@fgets($fs, 3);
// \r\n
}
}
break;
default:
while ($open &= !feof($fs) && ($p = @fread($fs, 1024))) {
// ???
$rsps .= $p;
}
break;
}
if ($rsps != '' && @$options['decode'] == 'gzip' && @$_rh['CONTENT_ENCODING'] == 'gzip') {
$r = self::gzdecode($rsps);
if ($r !== false) {
unset($_rh['CONTENT_ENCODING']);
$rsps = $r;
}
}
$ret->code = $rcode;
$ret->msg = $rmsg;
$ret->headers = isset($_rh) ? $_rh : NULL;
$ret->body = $rsps;
$ret->method = $meth;
// $ret->host = $host;
$ret->port = $port;
$ret->path = $path;
$ret->request = $rqst;
return $ret;
// Old return:
// contents headers status-code status-message
// return array( $rsps, @$_rh, $rcode, $rmsg, $host, $port, $path, $rqst );
}
}
fclose($fs);
return false;
// no response
}
public static function file_exists($fn)
{
$ffn = PHPUNIT_DIR . $fn;
if (!file_exists($ffn)) {
$zfn = $ffn . '.gz';
if (!file_exists($zfn)) {
return false;
}
$gz = file_get_contents($zfn);
$data = hQuery::gzdecode($gz);
if (!file_put_contents($ffn, $data)) {
return false;
}
}
return $ffn;
}
