public function parsePageType(XPathHelper $xph)
{
if (null !== $xph->xpQuery("//*[@id='noResultsTitle']")) {
return "searchResult";
}
if (null !== $xph->xpQuery("//*[@id='s-result-count']")) {
return "searchResult";
}
if (null !== $xph->xpQuery("//*[@id='atfResults']")) {
return "searchResult";
}
if (null !== $xph->xpQuery("//div[@id='product-title_feature_div']/div[@class='buying'] | //form[@id='handleBuy']/div[@class='buying']")) {
return "product";
}
// TODO: fix this
//if ('Robot Check' == $xph->queryValue("//html/body/title/text()", 1))
return "captcha";
// default, save for analysis
$ci =& get_instance();
$html_contents = "\n\n----------------------------------------------------------\n" . date('Y-m-d H:i:s') . "\n" . $xph->dump();
file_put_contents($ci->config->item('file_root') . "output.amazon.failed.parse.txt", $html_contents, FILE_APPEND);
throw new Exception(__FUNCTION__ . ": unexpected page type at " . $xph->__toString());
}
function main()
{
$opt = getopt('x:t:i:k:hv:p:');
if (isset($opt['h'])) {
fatal_error($help);
}
if (isset($opt['v'])) {
XPathHelper::$_debug = intval($opt['v']);
}
if (isset($opt['p'])) {
XPathHelper::$_proxy = $opt['p'];
}
if (isset($opt['t'])) {
$target = $opt['t'];
} else {
$target = null;
}
if (!isset($opt['x'])) {
fatal_error("-x argument is mandatory");
}
if (isset($opt['s'])) {
XPathHelper::$_sleep = (int) $opt['s'];
}
if (isset($opt['i'])) {
$input = explode("\n", file_get_contents($opt['i']));
if (($tail = array_pop($input)) !== '') {
array_push($input, $tail);
}
} else {
$input = array($target);
}
if (isset($opt['k'])) {
$keyword = $opt['k'];
} else {
$keyword = '';
}
switch ($opt['x']) {
case 'product-search':
productSearch($keyword, $target);
break;
case 'product-details':
foreach ($input as $url) {
productDetails($keyword, $url);
}
break;
case 'product-details2':
foreach ($input as $url) {
productDetails2($url);
}
break;
case 'product-parse':
foreach ($input as $url) {
productParse($url);
}
break;
case 'product-listall':
productfinder_listAll($target);
break;
case 'product-search-upc':
if (isset($opt['k'])) {
priceCheck(array($keyword), $target);
} else {
priceCheck($input, $target);
}
break;
case 'product-search-upc2':
if (isset($opt['k'])) {
priceCheck2(array($keyword), $target);
} else {
priceCheck2($input, $target);
}
break;
case 'product-search-mpn':
if (isset($opt['k'])) {
fatal_error("unexpected -k option for action 'product-search-mpn'");
} else {
priceCheckMPN($input, $target);
}
break;
case 'fetch':
XPathHelper::$_curlopts[CURLOPT_FOLLOWLOCATION] = true;
$xph = new XPathHelper($target);
echo $xph->dump();
break;
default:
fatal_error("unexpected -x action: " . $opt['x']);
}
}