|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Comrade42\PhpBBParser\Command; |
|
4
|
|
|
|
|
5
|
|
|
use Goutte\Client; |
|
6
|
|
|
use Symfony\Component\BrowserKit\Cookie; |
|
7
|
|
|
use Symfony\Component\Console\Input\InputInterface; |
|
8
|
|
|
use Symfony\Component\Console\Output\OutputInterface; |
|
9
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
10
|
|
|
|
|
11
|
|
|
/** |
|
12
|
|
|
* Class ParseForumsCommand |
|
13
|
|
|
* @package Comrade42\PhpBBParser\Command |
|
14
|
|
|
*/ |
|
15
|
|
|
class ParseForumsCommand extends ContainerAwareCommand |
|
16
|
|
|
{ |
|
17
|
|
|
protected function configure() |
|
18
|
|
|
{ |
|
19
|
|
|
$this->setName('parse:forums')->setDescription('Parse forums'); |
|
20
|
|
|
} |
|
21
|
|
|
|
|
22
|
|
|
protected function execute(InputInterface $input, OutputInterface $output) |
|
23
|
|
|
{ |
|
24
|
|
|
/** @var \Doctrine\ORM\EntityManager $entityManager */ |
|
25
|
|
|
$entityManager = $this->container->get('doctrine'); |
|
26
|
|
|
/** @var \Comrade42\PhpBBParser\Bridge\BridgeInterface $entityBridge */ |
|
27
|
|
|
$entityBridge = $this->container->get('bridge'); |
|
28
|
|
|
/** @var \Symfony\Component\Console\Helper\DialogHelper $dialogHelper */ |
|
29
|
|
|
$dialogHelper = $this->getHelperSet()->get('dialog'); |
|
30
|
|
|
|
|
31
|
|
|
$client = new Client(); |
|
32
|
|
|
$client->getCookieJar()->set( |
|
33
|
|
|
new Cookie( |
|
34
|
|
|
$this->container->getParameter('fa_sid_cookie_name'), |
|
35
|
|
|
$this->container->getParameter('fa_sid_cookie_value') |
|
36
|
|
|
) |
|
37
|
|
|
); |
|
38
|
|
|
|
|
39
|
|
|
$url = rtrim($this->container->getParameter('forum_url'), '/') . '/forum'; |
|
40
|
|
|
$output->write("<info>⇒ HTTP GET: {$url}</info>\t"); |
|
41
|
|
|
|
|
42
|
|
|
$crawler = $client->request('GET', $url); |
|
43
|
|
|
|
|
44
|
|
|
$status = $client->getInternalResponse()->getStatus(); |
|
45
|
|
|
$output->writeln("<info>[{$status}]</info>"); |
|
46
|
|
|
|
|
47
|
|
|
if ($status != 200) return; |
|
48
|
|
|
|
|
49
|
|
|
$crawler->filter('#main-content div.forabg')->each( |
|
50
|
|
|
function (Crawler $node, $index) use ($output, $entityManager, $entityBridge, $dialogHelper) |
|
51
|
|
|
{ |
|
52
|
|
|
$categoryId = $index + 1; |
|
53
|
|
|
$parsedName = $node->filter('ul.topiclist li.header dd.dterm h2')->text(); |
|
54
|
|
|
|
|
55
|
|
|
$entity = $entityBridge->getCategoryEntity($entityManager, $categoryId); |
|
56
|
|
|
$categoryTitle = $entity->getTitle(); |
|
57
|
|
|
|
|
58
|
|
|
if (!empty($categoryTitle) && $categoryTitle != $parsedName && !$dialogHelper->askConfirmation( |
|
59
|
|
|
$output, |
|
60
|
|
|
"<question>Category name doesn't match for #{$categoryId} ({$categoryTitle} → {$parsedName}). Update category? [Y/n]:</question> " |
|
61
|
|
|
) |
|
62
|
|
|
) { |
|
63
|
|
|
return; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
$entity->fill($parsedName, $index); |
|
67
|
|
|
$entityManager->persist($entity); |
|
68
|
|
|
|
|
69
|
|
|
$node->filter('ul.forums li.row dd.dterm')->each( |
|
70
|
|
|
function (Crawler $node, $index) use ($output, $entityManager, $entityBridge, $dialogHelper, $categoryId) |
|
71
|
|
|
{ |
|
72
|
|
|
$link = $node->filter('a.forumtitle'); |
|
73
|
|
|
$forumId = substr($link->attr('href'), 2, -6); |
|
74
|
|
|
$parsedTitle = $link->text(); |
|
75
|
|
|
$description = trim(substr($node->text(), strlen($parsedTitle))); |
|
76
|
|
|
|
|
77
|
|
|
$entity = $entityBridge->getForumEntity($entityManager, $forumId); |
|
78
|
|
|
$forumTitle = $entity->getTitle(); |
|
79
|
|
|
|
|
80
|
|
|
if (!empty($forumTitle) && $forumTitle != $parsedTitle && !$dialogHelper->askConfirmation( |
|
81
|
|
|
$output, |
|
82
|
|
|
"<question>Forum name doesn't match for #{$forumId} ({$forumTitle} → {$parsedTitle}). Update forum? [Y/n]:</question> " |
|
83
|
|
|
) |
|
84
|
|
|
) { |
|
85
|
|
|
return; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
$entity->fill($categoryId, $parsedTitle, $description, $index); |
|
89
|
|
|
$entityManager->persist($entity); |
|
90
|
|
|
} |
|
91
|
|
|
); |
|
92
|
|
|
} |
|
93
|
|
|
); |
|
94
|
|
|
|
|
95
|
|
|
$entityManager->flush(); |
|
96
|
|
|
} |
|
97
|
|
|
} |
|
98
|
|
|
|