} } $para = break_string($finalhtml, smart($word_count)); $st = new Summarizer(); $sums = $st->get_summary($para, false); $response["eng"]["summarized__smart"] = $sums; $response["hin"]["summarized__smart"] = TranslateArray($sums); if (isset($_GET["lcount"])) { $lcount = $_GET["lcount"]; $lines = explode(',', $lcount); for ($i = 0; $i < count($lines); $i++) { if ($i > 2) { $response["exception"] = "Max line count exceeded"; break; } $para = break_string($finalhtml, $lines[$i]); $st = new Summarizer(); $sums = $st->get_summary($para, false); $response["eng"]["summarized__" . $lines[$i] . "_lines"] = $sums; $response["hin"]["summarized__" . $lines[$i] . "_lines"] = TranslateArray($sums); } } $response["vid"] = $urlYoutube; $response["meta"]["originalLength"] = strlen($finalhtml); $response["meta"]["summaryLength"] = strlen(implode("", $sums)); $response["meta"]["summaryRatio"] = 100 - 100 * $response["meta"]["summaryLength"] / $response["meta"]["originalLength"]; $response["meta"]["source_text"] = "wikihow"; $response["meta"]["source_video"] = "youtube"; $response["meta"]["queryTime"] = microtime(true) - $start; $response["meta"]["graphRef"] = "http://nikunj.freakengineers.com/agro-api/showdata.php?total=" . ($response["meta"]["originalLength"] - $response["meta"]["summaryLength"]) . "&summr=" . $response["meta"]["summaryLength"] . "&st=" . rand(8000, 9000) / 1000 . "&myst=" . $response["meta"]["queryTime"] . "&nost=" . rand(1000, 2000) / 1000; $response["meta"]["rawRef"] = "http://nikunj.freakengineers.com/agro-api/showraw.php?query=" . urlencode($query);
function buildRSS($content) { /* $content is an array containing: [0] Name of Wiki template [1] Title of page [2] Content of page [3] Author of page [4] Date of modification */ global $_CONFIG; global $_EXTRAS; $content[2] = process($content[2], $content[0]); $content[4] = strtotime($content[4]); if ($_CONFIG['oneWiki']) { $base = $_CONFIG['base']; $url = $_CONFIG['host'] . $_CONFIG['base'] . "/" . $content[1]; } else { $base = $_CONFIG['base'] . "/" . $content[0]; $url = $_CONFIG['host'] . $_CONFIG['base'] . "/" . $content[0] . "/" . $content[1]; } $out = "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n" . "<rss version=\"2.0\" \n" . " xmlns:dc=\"http://purl.org/dc/elements/1.1/\"\n" . " xmlns:admin=\"http://webns.net/mvcb/\"\n" . " xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\"\n" . " xmlns:content=\"http://purl.org/rss/1.0/modules/content/\"\n" . ">\n" . "<channel>\n" . "<title>" . $content[0] . " - " . $content[1] . "</title>\n" . "<link>" . $url . "</link>\n" . "<description>A Wiki Page</description>\n" . "<dc:language>en-gb</dc:language>\n" . "<dc:creator>" . $content[3] . "</dc:creator>\n" . "<dc:rights>Copyright " . date("Y", $content[4]) . " " . $content[3] . "</dc:rights>\n" . "<dc:date>" . date("Y-m-d\\TH:i:00O", $content[4]) . "</dc:date>\n" . "<admin:generatorAgent rdf:resource=\"" . $_EXTRAS['versionURL'] . "\" />\n"; # ."<admin:errorReportsTo rdf:resource=\"mailto:".$_EP['admin']."\"/>\n" if ($_EXTRAS['data']) { // Data is of form array(url, title, description, date); foreach ($_EXTRAS['data'] as $data) { $desc = htmlspecialchars(break_string(strtr($data[2], "’", "'"), 400), ENT_NOQUOTES); $out .= "<item>\n" . "\t<title>" . htmlspecialchars($data[1], ENT_NOQUOTES) . "</title>\n" . "\t<link>" . $_CONFIG['host'] . $_CONFIG['base'] . $data[0] . "</link>\n" . "\t<description>" . $desc . "</description>\n" . "\t<guid isPermaLink=\"true\">" . $_CONFIG['host'] . $_CONFIG['base'] . $data[0] . "</guid>\n"; $out .= "\t<content:encoded><![CDATA[" . strtr($data[2], "’", "'") . "]]></content:encoded>\n" . "\t<dc:date>" . date("Y-m-d\\TH:i:00O", $data[3]) . "</dc:date>\n"; $out .= "</item>\n"; } } else { $desc = htmlspecialchars(break_string(strtr($content[2], "’", "'"), 400), ENT_NOQUOTES); $out .= "<item>\n" . "\t<title>" . htmlspecialchars($content[1], ENT_NOQUOTES) . "</title>\n" . "\t<link>" . $url . "</link>\n" . "\t<description>" . $desc . "</description>\n" . "\t<guid isPermaLink=\"true\">" . $url . "</guid>\n"; $out .= "\t<content:encoded><![CDATA[" . strtr($content[2], "’", "'") . "]]></content:encoded>\n" . "\t<dc:date>" . date("Y-m-d\\TH:i:00O", $content[4]) . "</dc:date>\n"; $out .= "</item>\n"; } $out .= "</channel>\n" . "</rss>"; #echo $out; return $out; }
$response["status"] = "missing query"; echo json_encode($response); die(""); } $response["status"] = "OK"; $response["author"] = "Nikunj Sharma"; $query = $_GET["query"]; $url = getUrl($query); $urlYoutube = getUrlYoutube($query); $html = file_get_html($url); $vartemp; foreach ($html->find('ol.steps_list_2 li') as $element) { $element->find(".step_num", 0)->innertext = ''; $vartemp .= $element->outertext . " "; } $finalhtml = html2text($vartemp); $finalhtml = preg_replace('/\\s+/', ' ', $finalhtml); $finalhtml = preg_replace('/\\[.*?\\]/', '', $finalhtml); $response["original"] = $finalhtml; $word_count = str_word_count($finalhtml); function smart($word_count) { if ($word_count <= 1200) { return 4; } else { return 6; } } $para = break_string($finalhtml, smart($word_count)); $st = new Summarizer(); $sums = $st->get_summary($para, true);