public static function parse($contents)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$rows = $crawler->filter('table[class="mt8 episode_list js-watch-episode-list ascend"] tr[class="episode-list-data"]');
$result = array();
foreach ($rows as $episodeItem) {
$crawler = new Crawler($episodeItem);
$episode = new Episode();
$episode->setNumber($crawler->filter('td[class="episode-number nowrap"]')->text());
$episode->setTitle($crawler->filter('td[class="episode-title"] a')->text());
// MAL does not always provide the air date!
$date = $crawler->filter('td[class="episode-aired"]')->text();
if ($date !== 'N/A') {
$dateTime = new DateTime();
$episode->setAirDate($dateTime->createFromFormat('M j, Y', $date));
}
$extracted = $crawler->filter('td[class="episode-title"] span[class="di-ib"]');
if ($extracted->text() !== '' && $extracted->count() > 0) {
# English:
$extracted = explode('(', $extracted->text());
if (count($extracted) > 0) {
$other_titles['english'] = array(trim($extracted[0], chr(0xc2) . chr(0xa0)));
}
# Japanese:
if (count($extracted) > 1) {
$other_titles['japanese'] = array(trim(str_replace(')', '', $extracted[1])));
}
$episode->setOtherTitles($other_titles);
}
$result[] = $episode;
}
return $result;
}
public static function parse($contents, $id, $type)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$rows = $crawler->filter('div[class="spaceit_pad"]');
$title = preg_replace('/ (\\w+?) Details/', '$2', $crawler->filter('div[class="normal_header"]')->text());
$result = array();
if ($type === 'anime') {
foreach ($rows as $historyItem) {
$crawler = new Crawler($historyItem);
$date = explode(' ', $crawler->text());
$historyinfo['item'] = new Anime();
$historyinfo['item']->setId((int) $id);
$historyinfo['item']->setTitle($title);
$historyinfo['item']->setWatchedEpisodes((int) $date[1]);
$historyinfo['type'] = $type;
$historyinfo['time_updated'] = Date::formatTime($date[4] . ' ' . $date[6]);
$result[] = $historyinfo;
}
} else {
foreach ($rows as $historyItem) {
$crawler = new Crawler($historyItem);
$date = explode(' ', $crawler->text());
$historyinfo['item'] = new Manga();
$historyinfo['item']->setId((int) $id);
$historyinfo['item']->setTitle($title);
$historyinfo['item']->setChaptersRead((int) $date[1]);
$historyinfo['type'] = $type;
$historyinfo['time_updated'] = Date::formatTime($date[4] . ' ' . $date[6]);
$result[] = $historyinfo;
}
}
return $result;
}
public function show()
{
$options = Config::get('onepager.options');
$progressBar = Cache::get('progressBar', function () {
$c = curl_init('https://www.startnext.com/sanktionsfrei/widget/?w=200&h=300&l=de');
curl_setopt($c, CURLOPT_RETURNTRANSFER, true);
$html = curl_exec($c);
if (curl_error($c)) {
die(curl_error($c));
}
$status = curl_getinfo($c, CURLINFO_HTTP_CODE);
curl_close($c);
$percent = 0;
if ($status == 200) {
$crawler = new Crawler();
$crawler->addHTMLContent($html, 'UTF-8');
// get the percentage for the progressbar
$styleString = $crawler->filter('.bar.bar-1')->attr('style');
$stringArray = explode(':', $styleString);
$percent = substr($stringArray[1], 0, -2);
// get the text for the progressbar
$textArray = $crawler->filter('.status-text span')->extract(['_text']);
}
return ['percent' => $percent, 'progressText' => $textArray[0]];
}, 5);
return view('home', ['options' => $options, 'percent' => $progressBar['percent'], 'progressText' => $progressBar['progressText']]);
}
public static function parse($contents, $type)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$result = array();
$items = $crawler->filter('div [class="borderDark pt4 pb8 pl4 pr4 mb8"]');
foreach ($items as $item) {
$result[] = self::parseReviews($item, $type);
}
return $result;
}
public static function parse($contents, $type)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
//Filter into a set of tds from the source HTML table
$mediaitems = $crawler->filter('tr[class="ranking-list"]');
foreach ($mediaitems as $item) {
$resultset[] = self::parseRecord($item, $type);
}
return $resultset;
}
/**
* Get the CSRF Token.
*
* @return string A string representing the CSRF token required for login
*/
private function getCsrfToken()
{
$token = null;
//Get the csrf_token for login
$loginPageContent = $this->fetch('/login.php');
$crawler = new Crawler();
$crawler->addHTMLContent($loginPageContent, 'UTF-8');
$metaTags = $crawler->filter('meta[name="csrf_token"]');
foreach ($metaTags as $tag) {
$name = $tag->attributes->getNamedItem('name');
if ($name !== null && $name->value == 'csrf_token') {
$token = $tag->attributes->getNamedItem('content')->value;
}
}
return $token;
}
public static function parse($contents, $type)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$result = array();
$items = $crawler->filter('entry');
if ($type === 'anime') {
foreach ($items as $item) {
$result[] = self::parseAnime($item);
}
} else {
foreach ($items as $item) {
$result[] = self::parseManga($item);
}
}
return $result;
}
public static function parse($contents)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$rows = $crawler->filter('div[class="borderClass"]');
$result = array();
foreach ($rows as $historyItem) {
$crawler = new Crawler($historyItem);
$anime = new Anime();
$anime->setId(str_replace('#raArea1', '', $crawler->filter('a')->attr('id')));
$anime->setImageUrl(preg_replace('/r(.+?)\\/(.+?)\\?(.+?)$/', '$2', $crawler->filter('img')->attr('data-src')));
$anime->setTitle($crawler->filter('strong')->text());
$resultItem['item'] = $anime;
$resultItem['recommendations'] = self::parseInformation($crawler);
$result[] = $resultItem;
}
return $result;
}
public static function parse($contents)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$classDay = 'seasonal-anime-list js-seasonal-anime-list js-seasonal-anime-list-key-';
$recordContainer = 'div[class="seasonal-anime js-seasonal-anime"]';
$result = array();
$result['monday'] = ScheduleParser::parseDay($crawler->filter('div[class="' . $classDay . 'monday clearfix"] ' . $recordContainer));
$result['tuesday'] = ScheduleParser::parseDay($crawler->filter('div[class="' . $classDay . 'tuesday clearfix"] ' . $recordContainer));
$result['wednesday'] = ScheduleParser::parseDay($crawler->filter('div[class="' . $classDay . 'wednesday clearfix"] ' . $recordContainer));
$result['thursday'] = ScheduleParser::parseDay($crawler->filter('div[class="' . $classDay . 'thursday clearfix"] ' . $recordContainer));
$result['friday'] = ScheduleParser::parseDay($crawler->filter('div[class="' . $classDay . 'friday clearfix"] ' . $recordContainer));
$result['saturday'] = ScheduleParser::parseDay($crawler->filter('div[class="' . $classDay . 'saturday clearfix"] ' . $recordContainer));
$result['sunday'] = ScheduleParser::parseDay($crawler->filter('div[class="' . $classDay . 'sunday clearfix"] ' . $recordContainer));
$result['other'] = ScheduleParser::parseDay($crawler->filter('div[class="' . $classDay . 'other clearfix"] ' . $recordContainer));
$result['unknown'] = ScheduleParser::parseDay($crawler->filter('div[class="' . $classDay . 'unknown clearfix"] ' . $recordContainer));
return $result;
}
public static function parse($contents, $type)
{
$resultset = array();
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$menubar = true;
//Filter into a set of tds from the source HTML table
$mediaitems = $crawler->filterXPath('//div[@id="content"]/div/table/tr');
foreach ($mediaitems as $item) {
//tricky method to skip the menu bar which is also a <tr></tr>
if ($menubar === true) {
$menubar = false;
} else {
$resultset[] = self::parseRecord($item, $type);
}
}
return $resultset;
}
public static function parseMessage($contents, $id)
{
$crawler = new Crawler();
$crawler->addHTMLContent(str_replace('<br />', '', $contents), 'UTF-8');
$message = new Messages();
$message->setId((int) $id);
# Action id of the message.
# Example:
# <input type="button" onclick="document.location='/mymessages.php?go=send&replyid=21193061&threadid=16092543&toname=Ratan12'" class="inputButton" value=" Reply ">
$actionId = $crawler->filterXPath('//form[contains(@action,"delete")]/input[@name="id"]');
$message->setActionId((int) $actionId->attr('value'));
# Thread id of the message.
# Example:
# <a href="?go=read&id=0000000&threadid=00000000">
$threadId = $crawler->filterXPath('//div/input[contains(@value,"Reply")]');
$threadId = $threadId->attr('onclick');
if (preg_match('/threadid=([\\d]*)/', $threadId, $threadMatches)) {
$message->setThreadId((int) $threadMatches[1]);
}
# Username of the sender.
# Example:
# <a href="http://myanimelist.net/profile/ratan12">ratan12</a>
$message->setUsername($crawler->filterXPath('//td[@class="dialog-text"]/h2/a')->text());
# Time of the received message.
# Example:
# <small>50 minutes ago</small>
$time = $crawler->filterXPath('//td[@class="dialog-text"]/div[contains(@class,"lightLink")]');
if (count($time) > 0) {
$message->setTime($time->text());
}
# Subject.
# Example:
# <div style="margin-bottom: 4px; font-weight: bold;">re: coolmessage</div>
$messageSubject = $crawler->filterXPath('//td[@class="dialog-text"]/div[contains(@class,"fw-b")]')->text();
$message->setSubject($messageSubject);
# Message.
$messageText = $crawler->filterXPath('//td[@class="dialog-text"]');
if (preg_match('/Test Test<\\/div>(.*?)<div/s', $messageText->html(), $messageBody)) {
$message->setMessage($messageBody[1]);
}
return $message;
}
public static function parse($contents)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$items = $crawler->filter('td[style="padding-left: 5px;"] table[width="100%"]');
//Bypass Undefined variable error.
$staff = null;
$characters = null;
foreach ($items as $item) {
//Bypass to determine if the last table contains the staff members
$crawler = new Crawler($item);
if ($crawler->filter('td[width="27"]')->count() != 1) {
$staffitems = $crawler->children();
foreach ($staffitems as $staffitem) {
$staff[] = self::parseStaff($staffitem);
}
} else {
$characters[] = self::parseCharacters($crawler);
}
}
return array('Characters' => $characters, 'Staff' => $staff);
}
public static function parseTopic($contents)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$topicitems = $crawler->filter('div[class="forum_border_around "]');
foreach ($topicitems as $item) {
$set[] = self::parseTopicDetails($item);
}
$pages = $crawler->filter('div[class="fl-r pb4"]')->text();
if ($pages != '') {
$result['pages'] = (int) substr($pages, strpos($pages, ' (') + 2, strpos($pages, ')'));
} else {
$result['pages'] = 1;
}
$result['list'] = $set;
return $result;
}
/**
* @param $content string
*
* @return string
*/
private function removeLastItem($content)
{
$document = new \DOMDocument('1.0', \Yii::$app->charset);
$crawler = new Crawler();
$crawler->addHTMLContent($content, \Yii::$app->charset);
$root = $document->appendChild($document->createElement('_root'));
$crawler->rewind();
$root->appendChild($document->importNode($crawler->current(), true));
$domxpath = new \DOMXPath($document);
$crawlerInverse = $domxpath->query(CssSelector::toXPath($this->widgetItem . ':last-child'));
foreach ($crawlerInverse as $key => $elementToRemove) {
$parent = $elementToRemove->parentNode;
$parent->removeChild($elementToRemove);
}
$crawler->clear();
$crawler->add($document);
return $crawler->filter('body')->eq(0)->html();
}
/**
* Clear HTML widgetBody. Required to work with zero or more items.
*
* @param string $content
*/
private function removeItems($content)
{
$crawler = new Crawler();
$crawler->addHTMLContent($content, \Yii::$app->charset);
$crawler->filter($this->widgetItem)->each(function ($nodes) {
foreach ($nodes as $node) {
$node->parentNode->removeChild($node);
}
});
return $crawler->html();
}
public static function parseExtendedPersonal($contents, Anime $anime)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
#Personal tags
#<td class="borderClass"><textarea name="tags" rows="2" id="tagtext" cols="45" class="textarea">action, sci-fi</textarea></td>
$personalTags = $crawler->filter('textarea[id="add_anime_tags"]')->text();
if (strlen($personalTags) > 0) {
$personalTags = explode(',', $personalTags);
foreach ($personalTags as $tag) {
$tagArray[] = trim($tag);
}
$anime->setPersonalTags($tagArray);
}
#Start and Finish Dates
#<tr>
# <td class="borderClass">Start Date</td>
# <td class="borderClass">
# Month:
# <select name="startMonth" class="inputtext">
# <option value="00">
# <option value="1" >Jan<option value="2" selected>Feb<option value="3" >Mar<option value="4" >Apr<option value="5" >May<option value="6" >Jun<option value="7" >Jul<option value="8" >Aug<option value="9" >Sep<option value="10" >Oct<option value="11" >Nov<option value="12" >Dec </select>
# Day:
# <select name="startDay" class="inputtext">
# <option value="00">
# <option value="1" >1<option value="2" selected>2<option value="3" >3<option value="4" >4<option value="5" >5<option value="6" >6<option value="7" >7<option value="8" >8<option value="9" >9<option value="10" >10<option value="11" >11<option value="12" >12<option value="13" >13<option value="14" >14<option value="15" >15<option value="16" >16<option value="17" >17<option value="18" >18<option value="19" >19<option value="20" >20<option value="21" >21<option value="22" >22<option value="23" >23<option value="24" >24<option value="25" >25<option value="26" >26<option value="27" >27<option value="28" >28<option value="29" >29<option value="30" >30<option value="31" >31 </select>
# Year:
# <select name="startYear" class="inputtext">
# <option value="0000">
# <option value="2014" selected>2014<option value="2013" >2013<option value="2012" >2012<option value="2011" >2011<option value="2010" >2010<option value="2009" >2009<option value="2008" >2008<option value="2007" >2007<option value="2006" >2006<option value="2005" >2005<option value="2004" >2004<option value="2003" >2003<option value="2002" >2002<option value="2001" >2001<option value="2000" >2000<option value="1999" >1999<option value="1998" >1998<option value="1997" >1997<option value="1996" >1996<option value="1995" >1995<option value="1994" >1994<option value="1993" >1993<option value="1992" >1992<option value="1991" >1991<option value="1990" >1990<option value="1989" >1989<option value="1988" >1988<option value="1987" >1987<option value="1986" >1986<option value="1985" >1985<option value="1984" >1984 </select>
#
# <label><input type="checkbox" onchange="ChangeStartDate();" name="unknownStart" value="1"> <small>Unknown Date</label><br>Start Date represents the date you started watching the Anime <a href="javascript:setToday(1);">Insert Today</a></small>
# </td>
#</tr>
#<tr>
# <td class="borderClass">Finish Date</td>
# <td class="borderClass">
# Month:
# <select name="endMonth" class="inputtext" disabled>
# <option value="00">
# <option value="1" >Jan<option value="2" >Feb<option value="3" >Mar<option value="4" >Apr<option value="5" >May<option value="6" >Jun<option value="7" >Jul<option value="8" >Aug<option value="9" >Sep<option value="10" >Oct<option value="11" >Nov<option value="12" >Dec </select>
# Day:
# <select name="endDay" class="inputtext" disabled>
# <option value="00">
# <option value="1" >1<option value="2" >2<option value="3" >3<option value="4" >4<option value="5" >5<option value="6" >6<option value="7" >7<option value="8" >8<option value="9" >9<option value="10" >10<option value="11" >11<option value="12" >12<option value="13" >13<option value="14" >14<option value="15" >15<option value="16" >16<option value="17" >17<option value="18" >18<option value="19" >19<option value="20" >20<option value="21" >21<option value="22" >22<option value="23" >23<option value="24" >24<option value="25" >25<option value="26" >26<option value="27" >27<option value="28" >28<option value="29" >29<option value="30" >30<option value="31" >31 </select>
# Year:
# <select name="endYear" class="inputtext" disabled>
# <option value="0000">
# <option value="2014" >2014<option value="2013" >2013<option value="2012" >2012<option value="2011" >2011<option value="2010" >2010<option value="2009" >2009<option value="2008" >2008<option value="2007" >2007<option value="2006" >2006<option value="2005" >2005<option value="2004" >2004<option value="2003" >2003<option value="2002" >2002<option value="2001" >2001<option value="2000" >2000<option value="1999" >1999<option value="1998" >1998<option value="1997" >1997<option value="1996" >1996<option value="1995" >1995<option value="1994" >1994<option value="1993" >1993<option value="1992" >1992<option value="1991" >1991<option value="1990" >1990<option value="1989" >1989<option value="1988" >1988<option value="1987" >1987<option value="1986" >1986<option value="1985" >1985<option value="1984" >1984 </select>
#
# <small><label><input type="checkbox" onchange="ChangeEndDate();" checked name="unknownEnd" value="1"> Unknown Date</label><br>Do <u>not</u> fill out the Finish Date unless status is <em>Completed</em> <a href="javascript:setToday(2);">Insert Today</a></small>
# </td>
#</tr>
$isStarted = $crawler->filter('input[id="unknown_start"]')->attr('checked');
$isEnded = $crawler->filter('input[id="unknown_end"]')->attr('checked');
if ($isStarted != 'checked') {
//So, MAL allows users to put in just years, just years and months, or all three values.
//This mess here is to try and avoid things breaking.
if ($crawler->filter('select[id="add_anime_start_date_year"] option:selected')->count() > 0) {
$startYear = $crawler->filter('select[id="add_anime_start_date_year"] option:selected')->attr('value');
$startMonth = 6;
$startDay = 15;
if ($startYear !== '') {
if ($crawler->filter('select[id="add_anime_start_date_month"] option:selected')->count() > 0) {
$startMonth = $crawler->filter('select[id="add_anime_start_date_month"] option:selected')->attr('value');
if ($startMonth === '') {
$startMonth = 6;
}
if ($crawler->filter('select[id="add_anime_start_date_day"] option:selected')->count() > 0) {
$startDay = $crawler->filter('select[id="add_anime_start_date_day"] option:selected')->attr('value');
if ($startDay === '') {
$startDay = 15;
}
}
}
$anime->setWatchingStart(DateTime::createFromFormat('Y-n-j', "{$startYear}-{$startMonth}-{$startDay}"));
}
}
}
if ($isEnded != 'checked') {
//Same here, avoid breaking MAL's allowing of partial dates.
if ($crawler->filter('select[id="add_anime_finish_date_year"] option:selected')->count() > 0) {
$endYear = $crawler->filter('select[id="add_anime_finish_date_year"] option:selected')->attr('value');
$endMonth = 6;
$endDay = 15;
if ($endYear !== '') {
if ($crawler->filter('select[id="add_anime_finish_date_month"] option:selected')->count() > 0) {
$endMonth = $crawler->filter('select[id="add_anime_finish_date_month"] option:selected')->attr('value');
if ($endMonth === '') {
$endMonth = 6;
}
if ($crawler->filter('select[id="add_anime_finish_date_day"] option:selected')->count() > 0) {
$endDay = $crawler->filter('select[id="add_anime_finish_date_day"] option:selected')->attr('value');
if ($endDay === '') {
$endDay = 15;
}
}
}
$anime->setWatchingEnd(DateTime::createFromFormat('Y-n-j', "{$endYear}-{$endMonth}-{$endDay}"));
}
}
}
#Priority
#<td class="borderClass"><select name="priority" class="inputtext">
#<option value="0" selected>Low<option value="1" >Medium<option value="2" >High </select>
$priority = $crawler->filter('select[id="add_anime_priority"] option:selected')->attr('value');
$anime->setPriority($priority);
#Storage
#
#<td class="borderClass" align="left"><select name="storage" id="storage" onchange="StorageBooleanCheck(2);" class="inputtext">
# <option value="0">Select storage type
# <option value="1" >Hard Drive<option value="6" >External HD<option value="7" >NAS<option value="2" >DVD / CD<option value="4" >Retail DVD<option value="5" >VHS<option value="3" >None </select>
#<div style="margin: 3px 0px; display: none;" id="StorageDiv">Total <span id="storageDescription">DvD's</span> <input type="text" name="storageVal" id="storageValue" value="0.00" size="4" class="inputtext"></div>
#</td>
//Note that if storage isn't defined, nothing will be marked as selected. We thus have to get the value in two stages to avoid raising an exception.
$storage = $crawler->filter('select[id="add_anime_storage_type"] option:selected');
if (count($storage)) {
$anime->setStorage($storage->attr('value'));
}
#Storage Value - Either number of discs or size in GB
#<div style="margin: 3px 0px; display: none;" id="StorageDiv">Total <span id="storageDescription">DvD's</span> <input type="text" name="storageVal" id="storageValue" value="1.00" size="4" class="inputtext"></div>
$storageval = (double) $crawler->filter('input[id="add_anime_storage_value"]')->attr('value');
if ($storageval > 0) {
$anime->setStorageValue($storageval);
}
#Rewatched
#<label><input type="checkbox" id="add_anime_is_rewatching" name="add_anime[is_rewatching]" value="1" checked="checked">
$rewatch = $crawler->filter('input[id="add_anime_is_rewatching"]')->attr('checked');
if ($rewatch == null) {
$anime->setRewatching(false);
} else {
$anime->setRewatching(true);
}
#Times Rewatched
#<td class="borderClass"><input type="text" name="list_times_watched" value="0" size="4" class="inputtext">
$rewatchCount = $crawler->filter('input[id="add_anime_num_watched_times"]')->attr('value');
if ($rewatchCount > 0) {
$anime->setRewatchCount($rewatchCount);
}
#Rewatch Value
#<td class="borderClass"><select name="list_rewatch_value" class="inputtext">
# <option value="0">Select rewatch value<option value="1">Very Low<option value="2">Low<option value="3">Medium<option value="4">High<option selected value="5">Very High </select>
$rewatchValue = $crawler->filter('select[id="add_anime_rewatch_value"] option:selected');
if (count($rewatchValue)) {
$anime->setRewatchValue($rewatchValue->attr('value'));
}
#Comments
#<td class="borderClass"><textarea name="list_comments" rows="5" cols="45" class="textarea"></textarea></td>
$comments = trim($crawler->filter('textarea[id="add_anime_comments"]')->text());
if (strlen($comments)) {
$anime->setPersonalComments($comments);
}
return $anime;
}
public static function parseExtendedPersonal($contents, Manga $manga)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
#Personal tags
#<td align="left" class="borderClass"><textarea name="tags" rows="2" id="tagtext" cols="45" class="textarea"></textarea><div class="spaceit_pad"><small>Popular tags: <a href="javascript:void(0);" onclick="detailedadd_addTag('cooking');">cooking</a>, <a href="javascript:void(0);" onclick="detailedadd_addTag('seinen');">seinen</a>, <a href="javascript:void(0);" onclick="detailedadd_addTag('drama');">drama</a>, <a href="javascript:void(0);" onclick="detailedadd_addTag('slice of life');">slice of life</a></small></div></td>
$personalTags = $crawler->filter('textarea[id="add_manga_tags"]')->text();
if (strlen($personalTags) > 0) {
$personalTags = explode(',', $personalTags);
foreach ($personalTags as $tag) {
$tagArray[] = trim($tag);
}
$manga->setPersonalTags($tagArray);
}
#Start and Finish Dates
#<tr>
# <td align="left" class="borderClass">Start Date</td>
# <td align="left" class="borderClass">
# Month:
# <select name="startMonth" id="smonth" class="inputtext">
# <option value="00">
# <option value="01" >Jan<option value="02" >Feb<option value="03" >Mar<option value="04" >Apr<option value="05" >May<option value="06" >Jun<option value="07" >Jul<option value="08" >Aug<option value="09" selected>Sep<option value="10" >Oct<option value="11" >Nov<option value="12" >Dec </select>
# Day:
# <select name="startDay" class="inputtext">
# <option value="00">
# <option value="01" >1<option value="02" >2<option value="03" >3<option value="04" >4<option value="05" >5<option value="06" >6<option value="07" >7<option value="08" >8<option value="09" >9<option value="10" >10<option value="11" >11<option value="12" >12<option value="13" >13<option value="14" >14<option value="15" >15<option value="16" >16<option value="17" >17<option value="18" >18<option value="19" >19<option value="20" >20<option value="21" >21<option value="22" >22<option value="23" >23<option value="24" >24<option value="25" selected>25<option value="26" >26<option value="27" >27<option value="28" >28<option value="29" >29<option value="30" >30<option value="31" >31 </select>
# Year:
# <select name="startYear" class="inputtext">
# <option value="0000">
# <option value="2014" >2014<option value="2013" selected>2013<option value="2012" >2012<option value="2011" >2011<option value="2010" >2010<option value="2009" >2009<option value="2008" >2008<option value="2007" >2007<option value="2006" >2006<option value="2005" >2005<option value="2004" >2004<option value="2003" >2003<option value="2002" >2002<option value="2001" >2001<option value="2000" >2000<option value="1999" >1999<option value="1998" >1998<option value="1997" >1997<option value="1996" >1996<option value="1995" >1995<option value="1994" >1994<option value="1993" >1993<option value="1992" >1992<option value="1991" >1991<option value="1990" >1990<option value="1989" >1989<option value="1988" >1988<option value="1987" >1987<option value="1986" >1986<option value="1985" >1985<option value="1984" >1984 </select>
#
# <label><input type="checkbox" onchange="ChangeStartDate();" name="unknownStart" value="1"> <small>Unknown Date</label><br>Start Date represents the date you started watching the Anime <a href="javascript:setToday(1);">Insert Today</a></small>
# </td>
#</tr>
#<tr>
# <td align="left" class="borderClass">Finish Date</td>
# <td align="left" class="borderClass">
# Month:
# <select name="endMonth" id="emonth" class="inputtext" >
# <option value="00">
# <option value="01" >Jan<option value="02" >Feb<option value="03" >Mar<option value="04" >Apr<option value="05" >May<option value="06" >Jun<option value="07" >Jul<option value="08" >Aug<option value="09" >Sep<option value="10" selected>Oct<option value="11" >Nov<option value="12" >Dec </select>
# Day:
# <select name="endDay" class="inputtext" >
# <option value="00">
# <option value="01" >1<option value="02" >2<option value="03" >3<option value="04" >4<option value="05" >5<option value="06" >6<option value="07" >7<option value="08" >8<option value="09" >9<option value="10" >10<option value="11" selected>11<option value="12" >12<option value="13" >13<option value="14" >14<option value="15" >15<option value="16" >16<option value="17" >17<option value="18" >18<option value="19" >19<option value="20" >20<option value="21" >21<option value="22" >22<option value="23" >23<option value="24" >24<option value="25" >25<option value="26" >26<option value="27" >27<option value="28" >28<option value="29" >29<option value="30" >30<option value="31" >31 </select>
# Year:
# <select name="endYear" class="inputtext" >
# <option value="0000">
# <option value="2014" >2014<option value="2013" selected>2013<option value="2012" >2012<option value="2011" >2011<option value="2010" >2010<option value="2009" >2009<option value="2008" >2008<option value="2007" >2007<option value="2006" >2006<option value="2005" >2005<option value="2004" >2004<option value="2003" >2003<option value="2002" >2002<option value="2001" >2001<option value="2000" >2000<option value="1999" >1999<option value="1998" >1998<option value="1997" >1997<option value="1996" >1996<option value="1995" >1995<option value="1994" >1994<option value="1993" >1993<option value="1992" >1992<option value="1991" >1991<option value="1990" >1990<option value="1989" >1989<option value="1988" >1988<option value="1987" >1987<option value="1986" >1986<option value="1985" >1985<option value="1984" >1984 </select>
#
# <small><label><input type="checkbox" onchange="ChangeEndDate();" name="unknownEnd" value="1"> Unknown Date</label><br>Do <u>not</u> fill out the Finish Date unless status is <em>Completed</em> <a href="javascript:setToday(2);">Insert Today</a></small>
# </td>
#</tr>
$isStarted = $crawler->filter('input[id="unknown_start"]')->attr('checked');
$isEnded = $crawler->filter('input[id="unknown_end"]')->attr('checked');
if ($isStarted != 'checked') {
//So, MAL allows users to put in just years, just years and months, or all three values.
//This mess here is to try and avoid things breaking.
if ($crawler->filter('select[id="add_manga_start_date_year"] option:selected')->count() > 0) {
$startYear = $crawler->filter('select[id="add_manga_start_date_year"] option:selected')->attr('value');
$startMonth = 6;
$startDay = 15;
if ($startYear !== '') {
if ($crawler->filter('select[id="add_manga_start_date_month"] option:selected')->count() > 0) {
$startMonth = $crawler->filter('select[id="add_manga_start_date_month"] option:selected')->attr('value');
if ($crawler->filter('select[id="add_manga_start_date_day"] option:selected')->count() > 0) {
$startDay = $crawler->filter('select[id="add_manga_start_date_day"] option:selected')->attr('value');
}
}
$manga->setReadingStart(DateTime::createFromFormat('Y-n-j', "{$startYear}-{$startMonth}-{$startDay}"));
}
}
}
if ($isEnded != 'checked') {
//Same here, avoid breaking MAL's allowing of partial dates.
if ($crawler->filter('select[id="add_manga_finish_date_year"] option:selected')->count() > 0) {
$endYear = $crawler->filter('select[id="add_manga_finish_date_year"] option:selected')->attr('value');
$endMonth = 6;
$endDay = 15;
if ($endYear !== '') {
if ($crawler->filter('select[id="add_manga_finish_date_month"] option:selected')->count() > 0) {
$endMonth = $crawler->filter('select[id="add_manga_finish_date_month"] option:selected')->attr('value');
if ($crawler->filter('select[id="add_manga_finish_date_day"] option:selected')->count() > 0) {
$endDay = $crawler->filter('select[id="add_manga_finish_date_day"] option:selected')->attr('value');
}
}
$manga->setReadingEnd(DateTime::createFromFormat('Y-n-j', "{$endYear}-{$endMonth}-{$endDay}"));
}
}
}
#Priority
#<td align="left" class="borderClass">Priority</td>
#<td align="left" class="borderClass"><select name="priority" class="inputtext">
#<option value="0">Select</option>
#<option value="0" selected>Low<option value="1" >Medium<option value="2" >High </select>
#<div style="margin-top 3px;"><small>What is your priority level to read this manga?</small></div></td>
$priorityList = $crawler->filter('select[id="add_manga_priority"] option:selected');
if (count($priorityList)) {
$priority = $priorityList->attr('value');
$manga->setPriority($priority);
}
#Rewatched
#<label><input type="checkbox" id="add_manga_is_rereading" name="add_manga[is_rereading]" value="1" checked="checked">
$reread = $crawler->filter('input[id="add_manga_is_rereading"]')->attr('checked');
if ($reread == null) {
$manga->setRereading(false);
} else {
$manga->setRereading(true);
}
#Times Reread
#<td align="left" class="borderClass"><input type="text" class="inputtext" size="4" value="0" name="times_read">
$rereadCount = $crawler->filter('input[id="add_manga_num_read_times"]')->attr('value');
if ($rereadCount > 0) {
$manga->setRereadCount($rereadCount);
}
#Reread Value
#<td align="left" class="borderClass"><select class="inputtext" name="reread_value">
# <option value="0">Select reread value</option><option value="1">Very Low</option><option value="2">Low</option><option value="3">Medium</option><option value="4">High</option><option value="5">Very High </option></select>
$rereadValue = $crawler->filter('select[id="add_manga_reread_value"] option:selected');
if (count($rereadValue)) {
$manga->setRereadValue($rereadValue->attr('value'));
}
#Comments
#<td align="left" class="borderClass"><textarea class="textarea" cols="45" rows="5" name="comments"></textarea></td>
$comments = trim($crawler->filter('textarea[id="add_manga_comments"]')->text());
if (strlen($comments)) {
$manga->setPersonalComments($comments);
}
return $manga;
}
public static function parse($contents)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$personrecord = new Person();
# Person ID.
# Example:
# <input type="hidden" name="vaid" value="185">
$personrecord->setId((int) $crawler->filter('input[name="vaid"]')->attr('value'));
# Name
# Example:
# <div><h1 class="h1">Hanazawa, Kana</h1></div>
$personrecord->setName(trim($crawler->filter('div h1')->text()));
# Image
# Example:
# <a href="/people/185/Kana_Hanazawa/pictures"><img src="http://cdn.myanimelist.net/images/voiceactors/3/43500.jpg" alt="Hanazawa, Kana"></a>
$personrecord->setImageUrl($crawler->filter('div#content tr td div img')->attr('src'));
$leftcolumn = $crawler->filterXPath('//div[@id="content"]/table/tr/td[@class="borderClass"]');
# Given name
$extracted = $leftcolumn->filterXPath('//span[text()="Given name:"]');
if ($extracted->count() > 0) {
$personrecord->setGivenName(trim(str_replace($extracted->text(), '', $extracted->parents()->text())));
}
# Family name
# MAL messed this field up. It's not wrapped in a div, so the text is floating out in the td.
$extracted = $leftcolumn->filterXPath('//span[text()="Family name:"]');
if ($extracted->count() > 0) {
$matches = array();
# This regex matches "Family name:..." until it hits Birthday/Website/Alternate [name], one of which should
# be the field following the Family name field
preg_match('/Family name:.*?(?:(?!Birthday|Website|Alternate).)*/', $leftcolumn->text(), $matches);
if (count($matches) > 0) {
$personrecord->setFamilyName(trim(str_replace($extracted->text(), '', $matches[0])));
}
}
# Alternate names
$extracted = $leftcolumn->filterXPath('//span[text()="Alternate names:"]');
if ($extracted->count() > 0) {
$text = trim(str_replace($extracted->text(), '', $extracted->parents()->text()));
$alternateNames = explode(', ', $text);
$personrecord->setAlternateNames($alternateNames);
}
# Birthday
$extracted = $leftcolumn->filterXPath('//span[text()="Birthday:"]');
if ($extracted->count() > 0) {
$dateStr = trim(str_replace($extracted->text(), '', $extracted->parents()->text()));
$dateStr = str_replace(' ', ' ', $dateStr);
// Replace 2 spaces with 1, MAL seems to add 2 spaces in some dates
if (strpos($dateStr, ',') === false) {
if (strlen($dateStr) === 4) {
// Only a year, Example ID 11746
$personrecord->setBirthday(DateTime::createFromFormat('Y m d', $dateStr . ' 01 01'), 'year');
} elseif (count(explode(' ', $dateStr)) == 2) {
// Month + Year, i.e. "Feb 1999", Example ID 7277
$dateComponents = explode(' ', $dateStr);
$month = $dateComponents[0];
$year = $dateComponents[1];
$personrecord->setBirthday(DateTime::createFromFormat('M Y d', $month . ' ' . $year . ' 01'), 'month');
}
} else {
$dateComponents = explode(' ', $dateStr);
if (count($dateComponents) == 2) {
// Month + Day, i.e. "Jun 15,", Example ID 2608
$month = $dateComponents[0];
$day = substr($dateComponents[1], 0, -1);
$personrecord->setBirthday(DateTime::createFromFormat('M d Y', $month . ' ' . $day . ' 1970'), 'dayMonth');
} elseif (count($dateComponents) == 3) {
// Full date, i.e. "Feb 25, 1989", Example ID 185
$personrecord->setBirthday(DateTime::createFromFormat('M j, Y', $dateStr), 'day');
}
}
}
# Website
# This isn't in a div, but the anchor element is the next sibling of the span
$extracted = $leftcolumn->filterXPath('//span[text()="Website:"]');
if ($extracted->count() > 0) {
$personrecord->setWebsiteUrl(trim(str_replace($extracted->text(), '', $extracted->nextAll()->attr("href"))));
}
# Favorites count
$extracted = $leftcolumn->filterXPath('//span[text()="Member Favorites:"]');
if ($extracted->count() > 0) {
$personrecord->setFavoritedCount(trim(str_replace($extracted->text(), '', $extracted->parents()->text())));
}
# More Details
# Note: CSS classes are misspelled, need to keep an eye on this
$extracted = $leftcolumn->filter('div[class="people-informantion-more js-people-informantion-more"]');
if ($extracted->count() > 0) {
$personrecord->setMoreDetails($extracted->html());
}
# Extract from sections on the right column: Voice acting roles, anime staff positions, published manga
$rightcolumn = $crawler->filter('div[id="content"] td[style="padding-left: 5px;"]');
// Voice acting roles
$vaRoles = $rightcolumn->filterXPath('//div[text()="Voice Acting Roles"]');
if ($vaRoles->count() > 0) {
$rolesArray = array();
// Iterate over each row in the table
foreach ($vaRoles->nextAll()->children() as $item) {
$node = new Crawler($item);
// Fill in the character details
$characterAnchor = $node->filterXPath('//td[3]/a');
$characterName = $characterAnchor->text();
$characterUrl = $characterAnchor->attr('href');
$isMainCharacer = strpos($node->filterXPath('//td[3]/div')->text(), "Main") !== false;
$characterImage = preg_replace('/r(.+?)\\/(.+?)\\?(.+?)$/', '$2', $node->filterXPath('//td[4]/div/a/img')->attr('data-src'));
$match = preg_match('/\\/(character)\\/(\\d+)\\/.*?/', $characterUrl, $urlParts);
if ($match !== false && $match !== 0) {
$characterId = (int) $urlParts[2];
}
$itemArray = array('id' => $characterId, 'name' => $characterName, 'image_url' => $characterImage, 'main_role' => $isMainCharacer);
$itemArray['anime'] = new Anime();
$animeDetails = $node->filterXPath('//td[2]/a');
// Fill in the anime details
$itemArray['anime']->setImageUrl(preg_replace('/r(.+?)\\/(.+?)\\?(.+?)$/', '$2', $node->filterXPath('//td[1]/div/a/img')->attr('data-src')));
$itemArray['anime']->setTitle($animeDetails->text());
$match = preg_match('/\\/(anime)\\/(\\d+)\\/.*?/', $animeDetails->attr('href'), $urlParts);
if ($match !== false && $match !== 0) {
$itemArray['anime']->setId($urlParts[2]);
}
$rolesArray[] = $itemArray;
}
$personrecord->setVoiceActingRoles($rolesArray);
}
// Anime staff positions
$staffPositions = $rightcolumn->filterXPath('//div[text()="Anime Staff Positions"]');
if ($staffPositions->count() > 0) {
$positionsArray = array();
// Iterate over each row in the table
foreach ($staffPositions->nextAll()->children() as $item) {
$node = new Crawler($item);
// Fill in the position details
$position = $node->filterXPath('//td[2]/div/small')->text();
$positionDetails = $node->filterXPath('//td[2]/div')->text();
// Details are wrapped in parenthesis, so we'll find those parens and grab what's inside
// Sometimes what comes back is malformed - i.e. "ep. 1-15)," - it appears malformed on the site too
$match = preg_match('/\\((.*)\\)/', $positionDetails, $positionDetailsParts);
if ($match !== false && $match !== 0) {
$positionDetails = $positionDetailsParts[1];
} else {
$positionDetails = null;
}
$itemArray = array('position' => $position, 'details' => $positionDetails);
$itemArray['anime'] = new Anime();
$animeDetails = $node->filterXPath('//td[2]/a');
// Fill in the anime details
$itemArray['anime']->setImageUrl(preg_replace('/r(.+?)\\/(.+?)\\?(.+?)$/', '$2', $node->filterXPath('//td[1]/div/a/img')->attr('data-src')));
$itemArray['anime']->setTitle($animeDetails->text());
$match = preg_match('/\\/(anime)\\/(\\d+)\\/.*?/', $animeDetails->attr('href'), $urlParts);
if ($match !== false && $match !== 0) {
$itemArray['anime']->setId($urlParts[2]);
}
$positionsArray[] = $itemArray;
}
$personrecord->setAnimeStaffPositions($positionsArray);
}
// Published manga
$mangaPositions = $rightcolumn->filterXPath('//div[text()="Published Manga"]');
if ($mangaPositions->count() > 0) {
$positionsArray = array();
// Iterate over each row in the table
foreach ($mangaPositions->nextAll()->children() as $item) {
$node = new Crawler($item);
// Fill in the position
$position = $node->filterXPath('//td[2]/div/small')->text();
$itemArray = array('position' => $position);
$itemArray['manga'] = new Manga();
$mangaDetails = $node->filterXPath('//td[2]/a');
// Fill in the manga details
$itemArray['manga']->setImageUrl(preg_replace('/r(.+?)\\/(.+?)\\?(.+?)$/', '$2', $node->filterXPath('//td[1]/div/a/img')->attr('data-src')));
$itemArray['manga']->setTitle($mangaDetails->text());
$match = preg_match('/\\/(manga)\\/(\\d+)\\/.*?/', $mangaDetails->attr('href'), $urlParts);
if ($match !== false && $match !== 0) {
$itemArray['manga']->setId($urlParts[2]);
}
$positionsArray[] = $itemArray;
}
$personrecord->setPublishedManga($positionsArray);
}
return $personrecord;
}
/**
* Set timezone setting by parsing the MAL settings.
*
* Parse the timezone setting used by MAL when an user logged in.
* This depends on the location of the country when a user joined.
* After parsing it stores the timezone in a static variable.
*
* @param string $settings The HTML source of the settings page that contains the timezone.
*
* @return string
*/
public static function setTimeZone($settings)
{
$crawler = new Crawler();
$crawler->addHTMLContent($settings, 'UTF-8');
self::$timeZone = $crawler->filter('option[selected]')->text();
}
public static function parseHistory($contents)
{
$crawler = new Crawler();
$crawler->addHTMLContent($contents, 'UTF-8');
$maincontent = $crawler->filter('table')->filter('tr');
//Empty array so we return something non-null if the list is empty.
$historylist = array();
foreach ($maincontent as $historyentry) {
$crawler = new Crawler($historyentry);
// bypass for the MAL generated strings
if ($crawler->filter('a')->count() > 0) {
if (strpos($crawler->filter('a')->attr('href'), 'anime') !== false) {
$historyinfo['item'] = new Anime();
$historyinfo['item']->setWatchedEpisodes((int) $crawler->filter('strong')->text());
$historyinfo['type'] = 'anime';
} else {
$historyinfo['item'] = new Manga();
$historyinfo['item']->setChaptersRead((int) $crawler->filter('strong')->text());
$historyinfo['type'] = 'manga';
}
$historyinfo['item']->setTitle($crawler->filter('a')->text());
$historyinfo['item']->setId((int) str_replace('/' . $historyinfo['type'] . '.php?id=', '', $crawler->filter('a')->attr('href')));
$historyinfo['time_updated'] = Date::formatTime(substr($crawler->filter('td')->eq(1)->text(), 1));
$historylist[] = $historyinfo;
}
}
return $historylist;
}
