1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace App\Controller; |
6
|
|
|
|
7
|
|
|
use App\Model\Query; |
8
|
|
|
use App\Repository\CloudElasticRepository; |
9
|
|
|
use App\Repository\WikiDomainRepository; |
10
|
|
|
use GuzzleHttp\Client; |
11
|
|
|
use Psr\Cache\CacheItemPoolInterface; |
12
|
|
|
use Symfony\Bundle\FrameworkBundle\Controller\AbstractController; |
13
|
|
|
use Symfony\Component\HttpFoundation\JsonResponse; |
14
|
|
|
use Symfony\Component\HttpFoundation\Request; |
15
|
|
|
use Symfony\Component\HttpFoundation\Response; |
16
|
|
|
use Symfony\Component\Routing\Annotation\Route; |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* A DefaultController serves the main routes for the application and processing the submitted form. |
20
|
|
|
*/ |
21
|
|
|
class DefaultController extends AbstractController |
22
|
|
|
{ |
23
|
|
|
/** @var Client */ |
24
|
|
|
private $client; |
25
|
|
|
|
26
|
|
|
/** @var CacheItemPoolInterface */ |
27
|
|
|
private $cache; |
28
|
|
|
|
29
|
|
|
/** @var bool Whether the results were pulled from cache. */ |
30
|
|
|
private $fromCache = false; |
31
|
|
|
|
32
|
|
|
/** @var string Duration of cache for main results set, as accepted by DateInterval::createFromDateString() */ |
33
|
|
|
private const CACHE_TIME = '10 minutes'; |
34
|
|
|
|
35
|
|
|
/** @var string[]|null Map from wiki dbname to domain name */ |
36
|
|
|
private $domainLookup; |
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* DefaultController constructor. |
40
|
|
|
* @param CacheItemPoolInterface $cache |
41
|
|
|
*/ |
42
|
|
|
public function __construct(CacheItemPoolInterface $cache) |
43
|
|
|
{ |
44
|
|
|
$this->client = new Client([ |
45
|
|
|
'verify' => $_ENV['ELASTIC_INSECURE'] ? false : true, |
46
|
|
|
]); |
47
|
|
|
$this->cache = $cache; |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* Splash page, shown when user is logged out. |
52
|
|
|
* @Route("/splash") |
53
|
|
|
*/ |
54
|
|
|
public function splashAction(): Response |
55
|
|
|
{ |
56
|
|
|
return $this->render('jumbotron.html.twig'); |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* The main route. |
61
|
|
|
* @Route("/", name="home") |
62
|
|
|
* @param Request $request |
63
|
|
|
* @return Response |
64
|
|
|
*/ |
65
|
|
|
public function indexAction(Request $request): Response |
66
|
|
|
{ |
67
|
|
|
if (!$this->get('session')->get('logged_in_user')) { |
68
|
|
|
return $this->render('jumbotron.html.twig'); |
69
|
|
|
} |
70
|
|
|
$query = $request->query->get('q'); |
71
|
|
|
$regex = (bool)$request->query->get('regex'); |
72
|
|
|
$ignoreCase = (bool)$request->query->get('ignorecase'); |
73
|
|
|
[$namespaces, $namespaceIds] = $this->parseNamespaces($request); |
74
|
|
|
$purgeCache = (bool)$request->query->get('purge'); |
75
|
|
|
|
76
|
|
|
$ret = [ |
77
|
|
|
'q' => $query, |
78
|
|
|
'regex' => $regex, |
79
|
|
|
'max_results' => Query::MAX_RESULTS, |
80
|
|
|
'namespaces' => $namespaceIds, |
81
|
|
|
'ignore_case' => $ignoreCase, |
82
|
|
|
]; |
83
|
|
|
|
84
|
|
|
if ($query) { |
85
|
|
|
$ret = array_merge($ret, $this->getResults($query, $regex, $ignoreCase, $namespaceIds, $purgeCache)); |
86
|
|
|
$ret['from_cache'] = $this->fromCache; |
87
|
|
|
return $this->formatResponse($request, $query, $ret); |
88
|
|
|
} |
89
|
|
|
|
90
|
|
|
return $this->render('default/index.html.twig', $ret); |
91
|
|
|
} |
92
|
|
|
|
93
|
|
|
/** |
94
|
|
|
* Get the rendered template for the requested format. |
95
|
|
|
* @param Request $request |
96
|
|
|
* @param string $query Query string, used for filenames. |
97
|
|
|
* @param array $data Data that should be passed to the view. |
98
|
|
|
* @return Response |
99
|
|
|
*/ |
100
|
|
|
private function formatResponse(Request $request, string $query, array $data): Response |
101
|
|
|
{ |
102
|
|
|
$format = $request->query->get('format', 'html'); |
103
|
|
|
if ('' == $format) { |
104
|
|
|
// The default above doesn't work when the 'format' parameter is blank. |
105
|
|
|
$format = 'html'; |
106
|
|
|
} |
107
|
|
|
|
108
|
|
|
$formatMap = [ |
109
|
|
|
'html' => 'text/html', |
110
|
|
|
'wikitext' => 'text/plain', |
111
|
|
|
'csv' => 'text/csv', |
112
|
|
|
'tsv' => 'text/tab-separated-values', |
113
|
|
|
'json' => 'application/json', |
114
|
|
|
]; |
115
|
|
|
|
116
|
|
|
// Use HTML if unknown format requested. |
117
|
|
|
$format = isset($formatMap[$format]) ? $format : 'html'; |
118
|
|
|
|
119
|
|
|
// Nothing more needed if requesting JSON. |
120
|
|
|
if ('json' === $format) { |
121
|
|
|
return new JsonResponse($data); |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
$response = new Response(); |
125
|
|
|
|
126
|
|
|
$response->headers->set('Content-Type', $formatMap[$format]); |
127
|
|
|
if (in_array($format, ['csv', 'tsv'])) { |
128
|
|
|
$filename = $this->getFilenameForRequest($query); |
129
|
|
|
$response->headers->set( |
130
|
|
|
'Content-Disposition', |
131
|
|
|
"attachment; filename=\"{$filename}.$format\"" |
132
|
|
|
); |
133
|
|
|
} |
134
|
|
|
|
135
|
|
|
return $this->render("default/result.$format.twig", $data, $response); |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
/** |
139
|
|
|
* Returns pretty filename from the given query, with problematic characters filtered out. |
140
|
|
|
* @param string $query |
141
|
|
|
* @return string |
142
|
|
|
*/ |
143
|
|
|
private function getFilenameForRequest(string $query): string |
144
|
|
|
{ |
145
|
|
|
$filename = trim($query, '/'); |
146
|
|
|
return trim(preg_replace('/[-\/\\:;*?|<>%#"]+/', '-', $filename)); |
147
|
|
|
} |
148
|
|
|
|
149
|
|
|
/** |
150
|
|
|
* Get results based on given Request. |
151
|
|
|
* @param string $query |
152
|
|
|
* @param bool $regex |
153
|
|
|
* @param bool $ignoreCase |
154
|
|
|
* @param int[] $namespaceIds |
155
|
|
|
* @param bool $purgeCache |
156
|
|
|
* @return mixed[] |
157
|
|
|
*/ |
158
|
|
|
public function getResults( |
159
|
|
|
string $query, |
160
|
|
|
bool $regex, |
161
|
|
|
bool $ignoreCase, |
162
|
|
|
array $namespaceIds, |
163
|
|
|
bool $purgeCache = false |
164
|
|
|
): array { |
165
|
|
|
$cacheItem = md5($query.$regex.$ignoreCase.implode('|', $namespaceIds)); |
166
|
|
|
|
167
|
|
|
if (!$purgeCache && $this->cache->hasItem($cacheItem)) { |
168
|
|
|
$this->fromCache = true; |
169
|
|
|
return $this->cache->getItem($cacheItem)->get(); |
170
|
|
|
} |
171
|
|
|
|
172
|
|
|
$query = new Query($query, $namespaceIds, $regex, $ignoreCase); |
173
|
|
|
$params = $query->getParams(); |
174
|
|
|
$res = (new CloudElasticRepository($this->client, $params))->makeRequest(); |
175
|
|
|
$data = [ |
176
|
|
|
'regex' => $regex, |
177
|
|
|
'ignore_case' => $ignoreCase, |
178
|
|
|
'total' => $res['hits']['total'], |
179
|
|
|
'hits' => $this->formatHits($res), |
180
|
|
|
]; |
181
|
|
|
|
182
|
|
|
$cacheItem = $this->cache->getItem($cacheItem) |
183
|
|
|
->set($data) |
184
|
|
|
->expiresAfter(\DateInterval::createFromDateString(self::CACHE_TIME)); |
185
|
|
|
$this->cache->save($cacheItem); |
186
|
|
|
return $data; |
187
|
|
|
} |
188
|
|
|
|
189
|
|
|
/** |
190
|
|
|
* Build the data structure for each hit, giving the view what it needs. |
191
|
|
|
* @param mixed[] $data |
192
|
|
|
* @return mixed[] |
193
|
|
|
*/ |
194
|
|
|
private function formatHits(array $data): array |
195
|
|
|
{ |
196
|
|
|
$hits = $data['hits']['hits']; |
197
|
|
|
$newData = []; |
198
|
|
|
|
199
|
|
|
foreach ($hits as $hit) { |
200
|
|
|
$result = $hit['_source']; |
201
|
|
|
$title = ($result['namespace_text'] ? $result['namespace_text'].':' : '').$result['title']; |
202
|
|
|
$domain = $this->getWikiDomainFromDbName($result['wiki']); |
203
|
|
|
$newData[] = [ |
204
|
|
|
'wiki' => rtrim($domain, '.org'), |
205
|
|
|
'title' => $title, |
206
|
|
|
'url' => $this->getUrlForTitle($domain, $title), |
207
|
|
|
'source_text' => $this->highlightQuery( |
208
|
|
|
$hit['highlight']['source_text.plain'][0] ?? '' |
209
|
|
|
), |
210
|
|
|
]; |
211
|
|
|
} |
212
|
|
|
|
213
|
|
|
return $newData; |
214
|
|
|
} |
215
|
|
|
|
216
|
|
|
/** |
217
|
|
|
* Get the URL to the page with the given title on the given wiki. |
218
|
|
|
* @param string $domain |
219
|
|
|
* @param string $title |
220
|
|
|
* @return string |
221
|
|
|
*/ |
222
|
|
|
private function getUrlForTitle(string $domain, string $title): string |
223
|
|
|
{ |
224
|
|
|
return 'https://'.$domain.'/wiki/'.$title; |
225
|
|
|
} |
226
|
|
|
|
227
|
|
|
/** |
228
|
|
|
* Query Siteinfo API to get the domain of the wiki with the given database name. |
229
|
|
|
* @param string $wiki |
230
|
|
|
* @return string |
231
|
|
|
*/ |
232
|
|
|
private function getWikiDomainFromDbName(string $wiki): string |
233
|
|
|
{ |
234
|
|
|
if (null === $this->domainLookup) { |
235
|
|
|
$this->domainLookup = (new WikiDomainRepository($this->client, $this->cache))->load(); |
236
|
|
|
} |
237
|
|
|
return $this->domainLookup[$wiki] ?? 'WIKINOTFOUND'; |
238
|
|
|
} |
239
|
|
|
|
240
|
|
|
/** |
241
|
|
|
* Make the highlight text safe and wrap the search term in a span so that we can style it. |
242
|
|
|
* @param string $text |
243
|
|
|
* @return string |
244
|
|
|
*/ |
245
|
|
|
private function highlightQuery(string $text): string |
246
|
|
|
{ |
247
|
|
|
$text = htmlspecialchars($text); |
248
|
|
|
return strtr($text, [ |
249
|
|
|
Query::PRE_TAG => "<span class='highlight'>", |
250
|
|
|
Query::POST_TAG => "</span>", |
251
|
|
|
]); |
252
|
|
|
} |
253
|
|
|
|
254
|
|
|
/** |
255
|
|
|
* Parse the namespaces parameter of the query string. |
256
|
|
|
* @param Request $request |
257
|
|
|
* @return mixed[] [normalized comma-separated list as a string, array of ids as ints] |
258
|
|
|
*/ |
259
|
|
|
private function parseNamespaces(Request $request): array |
260
|
|
|
{ |
261
|
|
|
$param = $request->query->get('namespaces', ''); |
262
|
|
|
|
263
|
|
|
if ('' === $param) { |
264
|
|
|
$ids = []; |
265
|
|
|
} else { |
266
|
|
|
$ids = array_map( |
267
|
|
|
'intval', |
268
|
|
|
explode(',', $param) |
269
|
|
|
); |
270
|
|
|
} |
271
|
|
|
|
272
|
|
|
return [ |
273
|
|
|
implode(',', $ids), |
274
|
|
|
$ids, |
275
|
|
|
]; |
276
|
|
|
} |
277
|
|
|
} |
278
|
|
|
|