public static function newFromUrl($url, $langCode, $dbType)
{
$url = str_replace('http://www.wikihow.com/', '', $url);
$pageTitle = Misc::getUrlDecodedData($url);
$row = self::getDBRowFromPageTitle($pageTitle, $langCode, $dbType);
return is_null($row) ? null : BabelfishArticle::newFromDBRow($row, $dbType);
}
public function processUrlListByLang(&$urlList, $langCode)
{
$excludedKey = $this->getWAPConfig()->getExludedArticlesKeyName();
$excludeList = explode("\n", ConfigStorage::dbGetConfig($excludedKey));
$urlList = Misc::getUrlDecodedData($urlList);
$processedUrls = array(WAPArticle::STATE_INVALID => array(), WAPArticle::STATE_EXCLUDED => array(), WAPArticle::STATE_UNASSIGNED => array(), WAPArticle::STATE_ASSIGNED => array(), WAPArticle::STATE_COMPLETED => array(), WAPArticle::STATE_NEW => array());
$urls = $this->parseURLlist($urlList, $langCode);
foreach ($urls as $url) {
// If it's in the system, put it in the right bucket
$wa = $url['a'];
if ($wa->exists()) {
$url['aid'] = $aid = $wa->getArticleId();
if (in_array($aid, $excludeList)) {
$processedUrls[WAPArticle::STATE_EXCLUDED][] = $url;
} elseif ($wa->isCompleted()) {
$processedUrls[WAPArticle::STATE_COMPLETED][] = $url;
} elseif ($wa->isAssigned()) {
$processedUrls[WAPArticle::STATE_ASSIGNED][] = $url;
} elseif (!$wa->isAssigned()) {
$processedUrls[WAPArticle::STATE_UNASSIGNED][] = $url;
} else {
throw new Exception("Bad url: {$url['url']} {$url['lang']}. Report to your trusty wikiHow engineer.");
}
} else {
// If it's not in the system, see if it's even a valid title (exists and isn't a redirect)
// If it is then check to see if it's excluded.
// If it isn't excluded then it's considered a new article
$page = Misc::getPagesFromURLs(array($url['url']), array('page_id', 'page_is_redirect'), true);
$page = reset($page);
if (!empty($page) && $page['page_is_redirect'] != 1) {
$aid = $page['page_id'];
$url['aid'] = $aid;
if (in_array($aid, $excludeList)) {
$processedUrls[WAPArticle::STATE_EXCLUDED][] = $url;
} else {
$processedUrls[WAPArticle::STATE_NEW][] = $url;
}
} else {
$processedUrls[WAPArticle::STATE_INVALID][] = $url;
}
}
}
return $processedUrls;
}
function buildSQL(&$ids)
{
global $wgRequest;
$sql = Misc::getUrlDecodedData($wgRequest->getVal('sql'), false);
if (stripos($sql, "FROM titus") > 0) {
// Always get the language_code
if (stripos($sql, "ti_language_code") == FALSE) {
$sql = str_replace("FROM titus", ', ti_language_code as "ti_language_code" FROM titus', $sql);
}
$sql = str_replace("FROM titus", "FROM " . TitusDB::TITUS_TABLE_NAME, $sql);
//Hack to include ti_page_id and ti_language_code
if (!preg_match('@SELECT\\ +\\*@i', $sql) && !preg_match('@ti_page_id as "ti_page_id"@', $sql)) {
$sql = preg_replace('@SELECT @', 'SELECT ti_page_id as "ti_page_id",', $sql);
}
} else {
$sql = "SELECT * FROM " . TitusDB::TITUS_TABLE_NAME;
}
$pageCondition = "";
if ($this->language != "") {
$pageCondition = " ti_language_code='" . mysql_escape_string($this->language) . "'";
}
$langConditions = array();
if (is_array($ids) && sizeof($ids) > 0) {
$sz = sizeof($ids);
foreach ($ids as $id) {
if (!empty($id['page_id'])) {
if (!isset($langConditions[$id['language']])) {
$langConditions[$id['language']] = array();
}
$langConditions[$id['language']][] = $id['page_id'];
}
#$pageCondition .= "(ti_page_id ='" . $id['page_id'] . "' AND ti_language_code='" . $id['language'] ."')";
}
foreach ($langConditions as $lang => $langIds) {
if ($pageCondition != "") {
$pageCondition .= " OR ";
}
$pageCondition .= "(ti_language_code='" . $lang . "' AND ti_page_id in (" . implode(",", $langIds) . "))";
}
}
if (stripos($sql, "WHERE ")) {
if ($pageCondition) {
$pageCondition = " AND (" . $pageCondition . ")";
}
$sql = preg_replace("@WHERE (.+)\$@", "WHERE (\\1) {$pageCondition}", $sql);
} elseif ($pageCondition != "") {
$sql .= " WHERE {$pageCondition} {$orderBy}";
} else {
$sql .= " {$orderBy}";
}
return $sql;
}
/**
* Add a bunch of URLs to the database for adding their images to various languages
* @param urls New-line seperated list of URLs or Article ids from POST
* @param langs Comma-seperated list of languages to add the images on
*/
private function addImages($urls, $langs)
{
$urls = preg_split("@[\r\n]+@", Misc::getUrlDecodedData($urls));
$langs = preg_split("@,@", urldecode($langs));
$realUrls = array();
$langIds = array();
foreach ($urls as $url) {
if (is_numeric($url)) {
$langIds[] = array("lang" => "en", "id" => $url);
} else {
$realUrls[] = $url;
}
}
$pagesA = Misc::getPagesFromURLs($realUrls, array('page_id', 'page_title'));
$badURLs = array();
$results = array();
foreach ($realUrls as $url) {
if (!isset($pagesA[$url])) {
foreach ($langs as $lang) {
ImageTransfer::addBadURL($url, $lang, "URL not found");
}
$results[] = array('fromURL' => $url, 'toURL' => '', 'status' => 'Bad URL');
}
}
$pagesB = Misc::getPagesFromLangIds($langIds, array('page_id', 'page_title'));
$pages = array_merge($pagesA, $pagesB);
$fromIDs = array();
$urlLookup = array();
foreach ($pages as $page) {
$fromIDs[] = $page['page_id'];
$urlLookup[$page['page_id']] = Misc::getLangBaseURL('en') . '/' . $page['page_title'];
}
if (sizeof($fromIDs) == 0) {
return array();
}
// Fetch to cache the pages for later use in addImage
$this->batchFetchPages("en", $fromIDs);
$langTLs = array();
foreach ($langs as $lang) {
$langTLs[$lang] = TranslationLink::getLinks("en", $lang, array("tl_from_aid in (" . implode(',', $fromIDs) . ")"));
$newFromIds = array_map(function ($m) {
return $m->fromAID;
}, $langTLs[$lang]);
foreach ($fromIDs as $id) {
//Add error links
if (!in_array($id, $newFromIds)) {
$this->addImage($id, $lang, 0);
if (isset($urlLookup[$id])) {
$results[] = array('fromURL' => $urlLookup[$id], 'toURL' => '', 'status' => 'Could not find any translation link to ' . $lang);
}
}
}
TranslationLink::batchPopulateURLs($langTLs[$lang]);
// Cache the other language pages too in a batch
$langIds = array();
foreach ($langTLs[$lang] as $tl) {
$langIds[] = $tls->toAID;
}
$this->batchFetchPages($lang, $langIds);
}
foreach ($langs as $lang) {
foreach ($langTLs[$lang] as $tl) {
if (!$this->addImage($tl->fromAID, $tl->toLang, $tl->toAID)) {
$results[] = array('fromURL' => $tl->fromURL, 'toURL' => $tl->toURL, 'status' => 'Queued');
} else {
$results[] = array('fromURL' => $tl->fromURL, 'toURL' => $tl->toURL, 'status' => 'Failed to queue');
}
}
}
return $results;
}
function articleDetails()
{
global $wgRequest, $wgOut;
$url = Misc::getUrlDecodedData($wgRequest->getVal('url'));
$msg = "";
try {
$langCode = $this->cu->getLanguageTag();
// throws exception if no language tag set for user
$processedUrls = $this->wapDB->processUrlListByLang($url, $langCode);
$articleState = 'invalid';
$articleClass = $this->config->getArticleClassName();
$ca = $articleClass::newFromUrl($url, $langCode, $this->dbType);
foreach ($processedUrls as $state => $urls) {
if (!empty($urls)) {
$articleState = $state;
break;
}
}
$msg = "Not a valid wikiHow Article.";
$supportEmail = $this->config->getSupportEmailAddress();
switch ($articleState) {
case 'completed':
$msg = $this->cu->getId() == $ca->getUserId() ? " You have already completed this article" : "This article is already completed.";
break;
case 'assigned':
$msg = $this->cu->getId() == $ca->getUserId() ? "You have already reserved this article." : "This article is reserved by another user.";
break;
case 'excluded':
$msg = "This article is on the exclude list. It is permanently unavailable.";
break;
case 'unassigned':
$t = Title::newFromId($urls[0]['aid']);
if (!$t || !$t->exists()) {
// Make sure it's still a valid wikiHow title. Edge case where articles are deleted
// before nightly maintenance script cleans those up
// If it isn't, stick with the default message
} else {
if (!$this->cu->canView($ca)) {
$msg = "This article is on a special list, please email {$supportEmail} for permission to reserve.";
} else {
// Show the article
$vars = $this->getDefaultVars();
$vars['a'] = $ca;
$vars['cu'] = $this->cu;
$tmpl = new WAPTemplate($this->dbType);
$msg = $tmpl->getHtml('article_details.tmpl.php', $vars);
}
}
break;
case 'new':
$msg = $this->config->getNewArticleMessage($supportEmail);
break;
case 'invalid':
// Stick with the default message
break;
}
} catch (Exception $e) {
$msg = "No language tag set. Please contact " . $this->config->getSupportEmailAddress();
}
$msg = "<h4>{$msg}</h4>";
$wgOut->setArticleBodyOnly(true);
$wgOut->addHtml($msg);
}
function articleDetails()
{
global $wgRequest, $wgOut;
$wgOut->setArticleBodyOnly(true);
$vars = $this->getDefaultVars($this->dbType);
$articleClass = $this->config->getArticleClassName();
$url = Misc::getUrlDecodedData($wgRequest->getVal('url'));
$langs = $this->config->getSupportedLanguages();
foreach ($langs as $lang) {
$ca = $articleClass::newFromUrl($url, $lang, $this->dbType);
$vars['lang'] = $lang;
$vars['article'] = $ca;
$vars['user'] = $ca->getUser();
$vars['tags'] = $ca->getTags();
$tmpl = new WAPTemplate($this->dbType);
$wgOut->addHtml($tmpl->getHtml('article_details_admin.tmpl.php', $vars));
$wgOut->addHtml($html);
}
}
/**
* Execute special page. Only available to wikihow staff.
*/
public function execute()
{
global $wgRequest, $wgOut, $wgUser, $wgLang;
$user = $wgUser->getName();
$userGroups = $wgUser->getGroups();
$extraUsers = array('Gersh');
if ($wgUser->isBlocked() || !in_array('staff', $userGroups) && !in_array($user, $extraUsers)) {
$wgOut->setRobotpolicy('noindex,nofollow');
$wgOut->errorpage('nosuchspecialpage', 'nospecialpagetext');
return;
}
if ($wgRequest->wasPosted()) {
// handle more URLs at once
ini_set('memory_limit', '512M');
set_time_limit(0);
ignore_user_abort(true);
$wgOut->setArticleBodyOnly(true);
$dbr = wfGetDB(DB_SLAVE);
$action = $wgRequest->getVal('action', '');
self::$discardThreshold = $wgRequest->getInt('discard-threshold', 0);
$dataType = $wgRequest->getVal('data-type');
$pageList = $wgRequest->getVal('pages-list', '');
$domain = $wgRequest->getVal('domain', 'bt');
$pageList = Misc::getUrlDecodedData($pageList);
$urls = self::parseURLlist($pageList);
if (empty($urls)) {
$result = array('result' => '<i>ERROR: no URLs given</i><br/>');
print json_encode($result);
return;
}
if ('reset' == substr($action, 0, 5)) {
if ('reset' == $action) {
// both www and mobile
foreach (self::$domains as $domain => $foo) {
$html .= self::resetStats($urls, $domain);
}
} else {
// just one domain
$domain = substr($action, -2);
$html = self::resetStats($urls, $domain);
}
self::resetVarnishes($urls);
self::mailUserResetDone($wgUser, $urls);
} elseif ('fetch' == $action) {
$html = '';
if ($dataType == 'csv') {
self::outputCSV($urls);
return;
} else {
foreach (self::$domains as $domain => $foo) {
$html .= self::fetchStats($urls, $domain);
}
}
} else {
$html = 'ERROR: unknown action';
}
$result = array('result' => $html);
print json_encode($result);
return;
}
$wgOut->setHTMLTitle('Admin - Bounce Tests (Stu) - wikiHow');
$defaultDiscardThreshold = self::$discardThreshold;
$domain_opts = '';
foreach (self::$domains as $domain => $label) {
$domain_opts .= "<option value=\"{$domain}\">{$label}</option>\n";
}
$specialPage = $this->specialPage;
$tmpl = <<<EOHTML
<script src="/extensions/wikihow/common/download.jQuery.js"></script>
<form id="admin-form" method="post" action="/Special:{$specialPage}">
<div style="font-size: 16px; letter-spacing: 2px; margin-bottom: 15px;">
\tFetch or Reset Bounce Timer (Stu) Stats
</div>
<div style="font-size: 13px; margin-bottom: 10px; border: 1px solid #dddddd; padding: 10px;">
\t<div>
\t\tViews discard threshold: <input id="discard-threshold" type="text" size="4" name="discard-threshold" value="{$defaultDiscardThreshold}" />
\t\t<!--
\t\t Domain
\t\t<select name="domain" id="pages-domains">
\t\t<option value="all">All</option>
\t\t{$domain_opts}
\t\t</select>
\t\t//-->
\t</div>
\t<div style="margin-top: 5px;">
\t\t<input type="radio" name="data-type" value="summary" checked> Summary</input>
\t\t<input type="radio" name="data-type" value="csv"> CSV</input>
\t</div>
</div>
<div style="font-size: 13px; margin: 20px 0 7px 0;">
\tEnter a list of URL(s) such as <code style="font-weight: bold;">http://www.wikihow.com/Lose-Weight-Fast</code> to which this tool will apply. One per line.
</div>
<textarea id="pages-list" name="pages-list" type="text" rows="10" cols="70"></textarea><br/>
<button id="pages-fetch" disabled="disabled" style="padding: 5px;">fetch stats</button>
<button id="pages-resetbt" disabled="disabled" style="padding: 5px;">reset www</button>
<button id="pages-resetmb" disabled="disabled" style="padding: 5px;">reset mobile</button>
<button id="pages-reset" disabled="disabled" style="padding: 5px;">reset all</button>
<br/>
<br/>
<div id="pages-result">
</div>
</form>
<script>
(function(\$) {
\tfunction doServerAction(action) {
\t\tvar dataType = \$('input:radio[name=data-type]:checked').val();
\t\tvar url = '/Special:{$specialPage}/views.csv?action=' + action + '&data-type=' + dataType;
\t\tif ('summary' == dataType) {
\t\t\tvar form = \$('#admin-form').serializeArray();
\t\t\t\$('#pages-result').html('loading ...');
\t\t\tvar finished = false;
\t\t\t\$.post(url,
\t\t\t\tform,
\t\t\t\tfunction(data) {
\t\t\t\t\tfinished = true;
\t\t\t\t\t\$('#pages-result').html(data['result']);
\t\t\t\t\t\$('#pages-list').focus();
\t\t\t\t},
\t\t\t\t'json')
\t\t\t\t.complete(function () {
\t\t\t\t\tif (!finished) {
\t\t\t\t\t\t\$('#pages-result').html('Server call is taking too long. Wait for an email.');
\t\t\t\t\t}
\t\t\t\t});
\t\t} else { // csv
\t\t\tvar form = 'pages-list=' + encodeURIComponent(\$('#pages-list').val());
\t\t\t\$.download(url, form);
\t\t}
\t}
\t\$(document).ready(function() {
\t\t\$('#pages-resetbt, #pages-resetmb, #pages-reset, #pages-fetch')
\t\t\t.prop('disabled', false)
\t\t\t.click(function () {
\t\t\t\tvar action = \$(this).attr('id').replace(/^pages-/, '');
\t\t\t\tvar answer = true;
\t\t\t\tif ('reset' == action.substring(0,5)) {
\t\t\t\t\tvar count = \$('#pages-list').val().split(/\\n/).length;
\t\t\t\t\tvar domain = 'www';
\t\t\t\t\tif ('resetmb'==action) domain='mobile';
\t\t\t\t\telse if ('reset'==action) domain='all domains';
\t\t\t\t\tanswer = confirm('Are you sure you want to reset data for approx. ' + count + ' URL(s) on ' + domain + '?');
\t\t\t\t}
\t\t\t\tif (answer) {
\t\t\t\t\tdoServerAction(action);
\t\t\t\t}
\t\t\t\treturn false;
\t\t\t});
\t\t/*
\t\t\$('#pages-allcheck')
\t\t\t.click(function() {
\t\t\t\tif (\$(this).prop('checked')){
\t\t\t\t\t\$('#pages-reset').prop('disabled',false);
\t\t\t\t}else{
\t\t\t\t\t\$('#pages-reset').prop('disabled',true);
\t\t\t\t}
\t\t\t});
\t\t\$('#pages-domains')
\t\t\t.change(function(){
\t\t\t\tif (\$(this).attr('value')=='all'){
\t\t\t\t\t\$('#pages-reset').prop('disabled',true);
\t\t\t\t\t\$('#pages-allcheck').prop('disabled',false);
\t\t\t\t\t\$('#pages-check').css('color','');
\t\t\t\t}else{
\t\t\t\t\t\$('#pages-reset').prop('disabled',false);
\t\t\t\t\t\$('#pages-allcheck').prop('disabled',true).prop('checked',false);
\t\t\t\t\t\$('#pages-check').css('color','#ccc');
\t\t\t\t}
\t\t\t\t\$('#pages-result').html('');
\t\t\t});
\t\t*/
\t\t\$('#pages-list')
\t\t\t.focus();
\t});
})(jQuery);
</script>
EOHTML;
$wgOut->addHTML($tmpl);
}
function outputArticleIdReport()
{
global $wgRequest;
$ids = split("\n", trim(Misc::getUrlDecodedData($wgRequest->getVal('data'))));
$dbr = wfGetDB(DB_SLAVE);
$articles = array();
foreach ($ids as $id) {
$id = trim($id);
$t = Title::newFromId($id);
if ($t && $t->exists()) {
$articles[] = array("id" => $id, "url" => $t->getFullUrl());
}
}
if ($wgRequest->getVal('format') == 'csv') {
$output = $this->getArticleIdReportCSV($articles);
$this->sendFile('article_stats', $output);
} else {
$output = $this->getArticleIdReportHtml($articles);
$this->sendHtml($output);
}
}
function getQuery()
{
global $wgRequest;
$days = $wgRequest->getVal('days');
$userList = $wgRequest->getVal('users');
$usersType = $wgRequest->getVal('usersType');
$sql = Misc::getUrlDecodedData($wgRequest->getVal('sql'), false);
if ($sql == "") {
$sql = "select * from flavius_summary";
}
// We only get active users in the last 90 days
if ($usersType == 'active') {
$t = wfTimestamp(TS_MW, time() - 60 * 60 * 24 * 90);
$ltSQL = "fe_last_touched > '" . $t . "'";
if (preg_match("@where@i", $sql)) {
$sql .= ' AND ' . $ltSQL;
} else {
$sql .= ' WHERE ' . $ltSQL;
}
} elseif ($usersType != 'all') {
$users = preg_split("@[\r\n]+@", urldecode($userList));
$ids = array();
foreach ($users as $user) {
if (preg_match("@http://www\\.wikihow\\.com/User:(.+)@i", $user, $matches)) {
$u = User::newFromName($matches[1]);
if ($u) {
$ids[] = $u->getId();
}
}
}
if (sizeof($ids) > 0) {
if (preg_match("@where@i", $sql)) {
$sql .= ' AND fe_user in ( ' . implode(',', $ids) . ')';
} else {
$sql .= ' WHERE fe_user in (' . implode(',', $ids) . ')';
}
}
}
$rows = $this->getData($sql, $days);
foreach ($rows as $row) {
foreach ($row as $k => $v) {
print "{$v} ";
}
print "\n";
}
}
