function debug_tables($schema)
{
global $DEBUG_TABLES;
if (!$DEBUG_TABLES) {
return;
}
$tables = scraperwiki::show_tables($schema);
print "Schema '" . $schema . "' contains " . count($tables) . " tables\n";
foreach (array_keys($tables) as $tableName) {
debug_table($schema, $tableName, TRUE);
}
}
foreach ($nextPage->find('a') as $element) {
$MyString = $element->href;
$MyString = htmlspecialchars_decode($MyString);
// print "Próxima página: " . $MyString . "\n";
}
}
}
//************************programa principal************************//
require 'scraperwiki/simple_html_dom.php';
$dom = new simple_html_dom();
//*Pegar os memberId dos avaliadores pelo sql e passar para a função*//
avaliaVendedor("blowitoutahere");
print "MyString = " . $MyString . "\n";
//limitando em percorrer 3 páginas
for ($pag = 0; $pag < 3; $pag++) {
$html = scraperWiki::scrape($MyString);
$dom->load($html);
foreach ($dom->find("table.FbOuterYukon") as $data) {
$tds = $data->find("td");
for ($i = 4; $i <= 197; $i += 8) {
$record = array('data_autocount' => $i, 'data_Feedback' => $tds[$i + 1]->plaintext, 'data_MemberID_AND_FeedbackScore' => $tds[$i + 2]->plaintext, 'data_Date-Time' => $tds[$i + 3]->plaintext, 'data_Item_Weight_Price_ItemNumber' => $tds[$i + 5]->plaintext, 'data_Price' => $tds[$i + 6]->plaintext);
// Salva o record na tabela // Salvar o $MyStringVendedor também.
saveData(array("Data_autocount", "data_Feedback", "data_MemberID_AND_FeedbackScore"), $record);
}
getLinks($data);
}
proxPaginaVendedor($dom);
}
print_r(scraperwiki::show_tables());
print_r(scraperwiki::sqliteexecute("select * from membersLinks"));
//print_r(scraperwiki::sqliteexecute("select * from allFeedBacks"));
$a = substr($aa->innertext, 0, strpos($aa->innertext, ' '));
$b = substr($bb->innertext, 0, strpos($bb->innertext, ' '));
if ($a == $b) {
return 0;
}
return $a < $b ? -1 : 1;
}
$tablename = "miway20121004";
require 'scraperwiki/simple_html_dom.php';
$html = scraperWiki::scrape("http://m.miway.ca/routes.jsp");
$dom = new simple_html_dom();
$dom->load($html);
$count = 0;
$toplinks = $dom->find('table tr td a[href^=routeStops.jsp]');
usort($toplinks, "toplinks_sort");
$tables = scraperwiki::show_tables();
if (count($tables) > 0 && $tables[$tablename]) {
$lastruninfo = scraperwiki::select("* FROM " . $tablename . " ORDER BY routeid DESC LIMIT 1");
$lastrouteid = $lastruninfo[0]["routeid"];
} else {
//this is a fresh run
$lastrouteid = -1;
}
foreach ($toplinks as $data) {
list($routeid, $nothing, $routename) = explode(" ", $data->innertext, 3);
// if ($routeid >= $lastrouteid) {
if ($routeid > 91) {
$cleanlink = preg_replace('/;jsessionid=[A-Z0-9]*/i', '', $data->href);
print "VISITING http://m.miway.ca/" . $cleanlink . "\n";
$route_html_content = scraperwiki::scrape("http://m.miway.ca/" . $cleanlink);
$route_html = str_get_html($route_html_content);
//current club ids
$url = "http://www.nrsr.sk/web/default.aspx?sid=poslanci/kluby/zoznam";
$html = scraperwiki::scrape($url);
//get dom
$dom = new simple_html_dom();
$dom->load($html);
$uls = $dom->find('ul[class=longlist]');
preg_match_all('/ID=([0-9]{1,})/', $uls[0], $matches);
$ids = $matches[1];
//current term
$divs = $dom->find('select[id=_sectionLayoutContainer_ctl01__currentTerm]');
$options = $divs[0]->find('option[selected=selected]');
$term = $options[0]->value;
//compare saved term
$saved_term = scraperwiki::get_var('current_term');
$info = scraperwiki::show_tables();
if ($term != $saved_term) {
if (isset($info['club'])) {
scraperwiki::sqliteexecute("delete from club");
scraperwiki::sqlitecommit();
}
if (isset($info['membership'])) {
scraperwiki::sqliteexecute("delete from membership");
scraperwiki::sqlitecommit();
}
}
scraperwiki::save_var('current_term', $term);
//current clubs
foreach ($ids as $i) {
$url = "http://www.nrsr.sk/web/Default.aspx?sid=poslanci/kluby/klub&ID=" . $i;
$html = scraperwiki::scrape($url);
$header = $detail->prev_sibling()->plaintext;
$header = strtolower(preg_replace('/ /', '_', $header));
$header = preg_replace('/:/', '', $header);
$detail_record[$header] = preg_replace('/\\s*$/', '', preg_replace('/ˆ\\s*/', '', $detail->plaintext));
}
if ($DEBUG) {
print_r($datail_record);
} else {
if ($detail_record['camera_type']) {
scraperwiki::save_sqlite(array('url'), $detail_record, $table_name = "cameras");
}
}
}
require 'scraperwiki/simple_html_dom.php';
$DEBUG = false;
$init = count(scraperwiki::show_tables()) == 0;
$lastbrandmodelcrawl = scraperwiki::get_var('lastcrawl');
if (time() > $lastbrandmodelcrawl + 3600 * 24 * 7) {
//Find the brands/Makers
$allbrands = scraperWiki::scrape("http://www.flickr.com/cameras/brands/");
$allbrands_dom = new simple_html_dom();
$allbrands_dom->load($allbrands);
$brands_node = $allbrands_dom->find("td.clTxt h4 a");
if (!$init) {
$known_brands = scraperwiki::select("count(url) as cnt from brands");
}
if ($init || $known_brands[0]['cnt'] != count($brands_node)) {
foreach ($brands_node as $data) {
$record = array('name' => $data->plaintext, 'url' => $data->href);
if ($DEBUG) {
print_r($record);
