function getAllRubrics($prevRubrics = array(), $parserNum = 0, $usePrev = false) { require_once "ProxiesService.php"; $proxyService = new ProxiesService(); $resProxy = $proxyService->getRandomProxyData(); $this->setProxyData($resProxy); if ($usePrev) { if (sizeof($prevRubrics) == 0) { print "Use main\n"; list($res, $parserId) = $this->getMainSiteRubrics(); if (PEAR::isError($res)) { return array(); } } else { print "Use previous\n"; $res = $prevRubrics; $parserId = $parserNum; } } else { list($res, $parserId) = $this->getMainSiteRubrics(); if (PEAR::isError($res)) { return; } } foreach ($res as $rubric) { print $rubric["id"] . ", " . $rubric["url"] . "\n"; $resProxy = $proxyService->getRandomProxyData(); $this->setProxyData($resProxy); $r1 = $this->getSiteRubrics1($rubric["url"], $rubric["id"], $parserId); if (PEAR::isError($r1)) { continue; } foreach ($r1 as $rubric1) { print $rubric1["id"] . ", "; $r2 = $this->getSiteRubrics2($rubric1["url"], $rubric1["id"], $parserId); } if ($usePrev) { break; } print "\n"; } if ($usePrev) { array_shift($res); return $res; } }
// ƒл¤ YellowPages1ContentGrabber запускать скрипт c параметрами 1 1 $parserId = StateUtils::getProcessNum() + 1; if ($parserId == 1) { $Parser =& new YellowPagesContentGrabber(); } else { $Parser =& new YellowPages1ContentGrabber(); } $Parser->showInfo = false; $Parser->waitBeforeRequest = 3; $rubrics =& new RubricsData(); $Parser->rubricsData =& $rubrics; $ProxyService = new ProxiesService(); $start = (int) FileUtils::loadContents(CALC_START_ID); $subRubrics = $rubrics->getSubRubricsForParser($parserId, $start, 1); if (PEAR::isError($subRubrics)) { print $subRubrics->getMessage(); exit; } if ($subRubrics->numRows() == 0) { // —брос счетчика последнего обработаной записи FileUtils::saveContents(CALC_START_ID, "0"); exit; } while ($row =& $subRubrics->fetchRow(DB_FETCHMODE_OBJECT)) { $res = $ProxyService->getRandomProxyData(); $Parser->setProxyData($res); $res = $Parser->getRealNum($row->rurl); $rubrics->setRealNum($res, $row->rid); print $row->rid . ", " . $row->rurl . ": " . $res . "\n"; FileUtils::saveContents(CALC_START_ID, $row->rid); }
function HandleHighlighting($useProxy = true) { $params = array(); if ($useProxy) { $proxyService = new ProxiesService(); $res = $proxyService->getRandomProxyData(); if (PEAR::isError($res)) { return $res; } $params["proxyData"] = $res; } $content = UrlUtils::loadPage($_REQUEST["url"], $params); print <<<HTML <html> <title>Подсветка {$_REQUEST["url"]}</title> <head> <style> em { color: red; font-size: 12pt; font-weight: bold; } pre { font-size: 9pt; } b { color: blue; } b b { color: red; } </style> </head> <body> <pre> HTML; print getHighlitedHtml($content); print <<<HTML </pre> </body> </html> HTML; }