1 <?php

     2 /*

     3  * This file is part of the WikiTagBundle package.

     4  *

     5  * (c) IRI <http://www.iri.centrepompidou.fr/>

     6  *

     7  * For the full copyright and license information, please view the LICENSE

     8  * file that was distributed with this source code.

     9  */

    10 namespace IRI\Bundle\WikiTagBundle\Command;

    11 

    12 use IRI\Bundle\WikiTagBundle\Utils\WikiTagUtils;

    13 

    14 use Doctrine\ORM\QueryBuilder;

    15 use Symfony\Bundle\FrameworkBundle\Command\ContainerAwareCommand;

    16 use Symfony\Component\Console\Input\InputArgument;

    17 use Symfony\Component\Console\Input\InputInterface;

    18 use Symfony\Component\Console\Input\InputOption;

    19 use Symfony\Component\Console\Output\OutputInterface;

    20 

    21 class QueryWikipediaCommand extends ContainerAwareCommand

    22 {

    23 

    24     private function showProgress(OutputInterface $output, $current, $total, $label, $width)

    25     {

    26         $percent = (floatval($current)/floatval($total)) * 100.0;

    27         $marks = intval(floor(floatval($width) * ($percent / 100.0) ));

    28         $spaces = $width - $marks;

    29         

    30         $status_bar="\r[";

    31         $status_bar.=str_repeat("=", $marks);

    32         if($marks<$width){

    33             $status_bar.=">";

    34             $status_bar.=str_repeat(" ", $spaces);

    35         } else {

    36             $status_bar.="=";

    37         }

    38         

    39         $disp=str_pad(number_format($percent, 0),3, " ", STR_PAD_LEFT);

    40         

    41         $label = str_pad(substr($label,0,50), 50, " ");

    42         $current_str = str_pad($current, strlen("$total"), " ", STR_PAD_LEFT);

    43         

    44         $status_bar.="] $disp%  $current_str/$total : $label";

    45         

    46         $output->write("$status_bar  ");

    47         

    48         if($current == $total) {

    49             $output->writeln("");

    50         }

    51         

    52     }

    53     

    54     private function processTag($tag, $em)

    55     {

    56         $tag_label_normalized = WikiTagUtils::normalizeTag($tag->getLabel());

    57         $wp_response = WikiTagUtils::getWikipediaInfo($tag_label_normalized);

    58         

    59         $new_label = $wp_response['new_label'];

    60         $status = $wp_response['status'];

    61         $url = $wp_response['wikipedia_url'];

    62         $pageid = $wp_response['pageid'];

    63         $dbpedia_uri = $wp_response["dbpedia_uri"];

    64         $wikipedia_revision_id = $wp_response['revision_id'];

    65         

    66         # We save the datas

    67         if($new_label!=null){

    68             $tag->setLabel($new_label);

    69         }

    70         if($status!=null){

    71             $tag->setUrlStatus($status);

    72         }

    73         $tag->setWikipediaUrl($url);

    74         $tag->setWikipediaPageId($pageid);

    75         $tag->setDbpediaUri($dbpedia_uri);

    76         

    77         // Save datas.

    78         $em->persist($tag);

    79         

    80     }

    81     

    82           

    83     protected function configure()

    84     {

    85         parent::configure();

    86 

    87         $this

    88             ->setName('wikitag:query-wikipedia')

    89             ->setDescription('Query wikipedia for tags.')

    90             ->addOption("force","f",InputOption::VALUE_NONE, "Force remove tags")

    91             ->addOption("all","a",InputOption::VALUE_NONE, "Force remove tags")

    92             ->addOption("random","r",InputOption::VALUE_NONE, "randomize query on tags")

    93             ->addOption("site","S",InputOption::VALUE_OPTIONAL, "the url for the wikipedia site", "http://fr.wikipedia.org/w/api.php")

    94             ->addOption("limit","l",InputOption::VALUE_OPTIONAL, "number of tag to process", -1)

    95             ->addOption("start",null,InputOption::VALUE_OPTIONAL, "number of tag to ignore", 0);

    96     }

    97 

    98     protected function execute(InputInterface $input, OutputInterface $output)

    99     {

   100         

   101         $force = $input->getOption('force');

   102         $all = $input->getOption('all');

   103         $random = $input->getOption('random');

   104         $site = $input->getOption('site');

   105         $limit = intval($input->getOption('limit'));

   106         $start = intval($input->getOption('start'));

   107         

   108         $doctrine = $this->getContainer()->get('doctrine');

   109         $qb = $doctrine->getEntityManager()->createQueryBuilder();

   110         

   111         

   112         $qb->from('WikiTagBundle:Tag','t');

   113         

   114         if(!$all)

   115         {

   116             $qb->where($qb->expr()->isNull("t.urlStatus"));

   117         }

   118         

   119         if($start > 0)

   120         {

   121             $qb->setFirstResult($start);

   122         }

   123         

   124         if($limit>=0)

   125         {

   126             $qb->setMaxResults($limit);

   127         }

   128 

   129         $qb_count = clone $qb;

   130         

   131         $qb_count->select("t.id");

   132         

   133         $count = count($qb_count->getQuery()->getScalarResult());

   134         $doctrine->getEntityManager()->clear();

   135         

   136         if(! $force && $input->isInteractive())

   137         {

   138             $dialog = $this->getHelper('dialog');

   139             if (!$dialog->askConfirmation($output, "<question>This command will process $count tag(s). Continue ? (y/N) : </question>", false)) {

   140                 return;

   141             }

   142         }

   143         

   144         $qb->select("t");

   145         

   146         $done = 0;

   147         $iterable = $qb->getQuery()->iterate();

   148         $doctrine->getEntityManager()->beginTransaction();

   149         while (($row = $iterable->next()) !== false)

   150         {

   151             $done++;

   152             $tag = $row[0];

   153             

   154             $this->showProgress($output, $done, $count, $tag->getLabel(), 50);

   155             

   156             // process tag

   157             $this->processTag($tag, $doctrine->getEntityManager());

   158                         

   159             if($done%100 == 0)

   160             {

   161                 $doctrine->getEntityManager()->flush();

   162                 $doctrine->getEntityManager()->commit();

   163                 $doctrine->getEntityManager()->clear();

   164                 $doctrine->getEntityManager()->beginTransaction();

   165             }

   166         }

   167         $doctrine->getEntityManager()->flush();

   168         $doctrine->getEntityManager()->commit();

   169         

   170     }

   171 }