MetaGer.php 57.9 KB
Newer Older
1
2
3
4
<?php
namespace App;

use App;
5
use Cache;
6
use Carbon;
7
use Illuminate\Http\Request;
Phil Höfer's avatar
Phil Höfer committed
8
use Illuminate\Support\Facades\Redis;
9
use Jenssegers\Agent\Agent;
Dominik Hebeler's avatar
Bugfix    
Dominik Hebeler committed
10
use LaravelLocalization;
11
use Log;
12
use Predis\Connection\ConnectionException;
13
14
15

class MetaGer
{
16
17
18
19
20
21
22
23
24
25
    # Einstellungen für die Suche
    protected $fokus;
    protected $eingabe;
    protected $q;
    protected $category;
    protected $time;
    protected $page;
    protected $lang;
    protected $cache = "";
    protected $site;
26
    protected $hostBlacklist = [];
27
    protected $domainBlacklist = [];
28
29
30
31
32
33
34
35
36
    private $urlBlacklist = [];
    protected $stopWords = [];
    protected $phrases = [];
    protected $engines = [];
    protected $results = [];
    protected $ads = [];
    protected $warnings = [];
    protected $errors = [];
    protected $addedHosts = [];
37
    protected $availableFoki = [];
38
39
    protected $startCount = 0;
    protected $canCache = false;
40
    # Daten über die Abfrage$
41
    protected $ip;
Dominik Hebeler's avatar
Dominik Hebeler committed
42
    protected $useragent;
43
44
    protected $language;
    protected $agent;
45
    protected $apiKey = "";
Phil Höfer's avatar
Phil Höfer committed
46
    protected $apiAuthorized = false;
47
48
49
50
51
    # Konfigurationseinstellungen:
    protected $sumaFile;
    protected $mobile;
    protected $resultCount;
    protected $sprueche;
52
    protected $maps;
53
    protected $newtab;
54
    protected $domainsBlacklisted = [];
55
    protected $urlsBlacklisted = [];
56
    protected $url;
57
    protected $fullUrl;
58
    protected $languageDetect;
Dominik Hebeler's avatar
Dominik Hebeler committed
59
60
    protected $verificationId;
    protected $verificationCount;
61

62
63
    public function __construct()
    {
64
        # Timer starten
65
        $this->starttime = microtime(true);
66
        # Versuchen Blacklists einzulesen
67
        if (file_exists(config_path() . "/blacklistDomains.txt") && file_exists(config_path() . "/blacklistUrl.txt")) {
68
            $tmp = file_get_contents(config_path() . "/blacklistDomains.txt");
69
            $this->domainsBlacklisted = explode("\n", $tmp);
70
71
            $tmp = file_get_contents(config_path() . "/blacklistUrl.txt");
            $this->urlsBlacklisted = explode("\n", $tmp);
72
        } else {
73
            Log::warning("Achtung: Eine, oder mehrere Blacklist Dateien, konnten nicht geöffnet werden");
74
75
        }

76
        # Parser Skripte einhängen
77
78
79
80
        $dir = app_path() . "/Models/parserSkripte/";
        foreach (scandir($dir) as $filename) {
            $path = $dir . $filename;
            if (is_file($path)) {
81
                require_once $path;
82
83
84
            }
        }

85
        # Cachebarkeit testen
86
87
88
89
90
91
        try {
            Cache::has('test');
            $this->canCache = true;
        } catch (ConnectionException $e) {
            $this->canCache = false;
        }
92
    }
93

94
    # Erstellt aus den gesammelten Ergebnissen den View
95
    public function createView($quicktipResults = [])
96
    {
97
98
        # Hiermit werden die evtl. ausgewählten SuMas extrahiert, damit die Input-Boxen richtig gesetzt werden können
        $focusPages = [];
Aria Givi's avatar
Aria Givi committed
99

100
101
        foreach ($this->request->all() as $key => $value) {
            if ($value === 'on') {
Aria Givi's avatar
Aria Givi committed
102
                $focusPages[] = str_replace('param_', '', str_replace('engine_', '', $key));
103
104
105
            }
        }

106
        $viewResults = [];
107
        # Wir extrahieren alle notwendigen Variablen und geben Sie an unseren View:
108
        foreach ($this->results as $result) {
109
110
111
112
            $viewResults[] = get_object_vars($result);
        }
        # Wir müssen natürlich noch den Log für die durchgeführte Suche schreiben:
        $this->createLogs();
113
114
        if ($this->fokus === "bilder") {
            switch ($this->out) {
115
                case 'results':
116
                    return view('resultpages.results_images')
117
118
119
120
121
                        ->with('results', $viewResults)
                        ->with('eingabe', $this->eingabe)
                        ->with('mobile', $this->mobile)
                        ->with('warnings', $this->warnings)
                        ->with('errors', $this->errors)
Phil Höfer's avatar
Phil Höfer committed
122
                        ->with('apiAuthorized', $this->apiAuthorized)
123
                        ->with('metager', $this)
124
                        ->with('browser', (new Agent())->browser());
125
                default:
126
                    return view('resultpages.resultpage_images')
127
128
129
130
131
                        ->with('results', $viewResults)
                        ->with('eingabe', $this->eingabe)
                        ->with('mobile', $this->mobile)
                        ->with('warnings', $this->warnings)
                        ->with('errors', $this->errors)
Phil Höfer's avatar
Phil Höfer committed
132
                        ->with('apiAuthorized', $this->apiAuthorized)
133
                        ->with('metager', $this)
134
                        ->with('browser', (new Agent())->browser());
135
            }
136
137
138
        } else {
            switch ($this->out) {
                case 'results':
139
                    return view('resultpages.results')
140
141
142
143
144
                        ->with('results', $viewResults)
                        ->with('eingabe', $this->eingabe)
                        ->with('mobile', $this->mobile)
                        ->with('warnings', $this->warnings)
                        ->with('errors', $this->errors)
Phil Höfer's avatar
Phil Höfer committed
145
                        ->with('apiAuthorized', $this->apiAuthorized)
146
                        ->with('metager', $this)
Dominik Hebeler's avatar
Dominik Hebeler committed
147
148
                        ->with('browser', (new Agent())->browser())
                        ->with('fokus', $this->fokus);
149
150
                    break;
                case 'results-with-style':
151
                    return view('resultpages.resultpage')
152
153
154
155
156
                        ->with('results', $viewResults)
                        ->with('eingabe', $this->eingabe)
                        ->with('mobile', $this->mobile)
                        ->with('warnings', $this->warnings)
                        ->with('errors', $this->errors)
Phil Höfer's avatar
Phil Höfer committed
157
                        ->with('apiAuthorized', $this->apiAuthorized)
158
159
                        ->with('metager', $this)
                        ->with('suspendheader', "yes")
Dominik Hebeler's avatar
Dominik Hebeler committed
160
161
                        ->with('browser', (new Agent())->browser())
                        ->with('fokus', $this->fokus);
162
                    break;
Phil Höfer's avatar
Phil Höfer committed
163
                case 'rich':
164
                    return view('resultpages.metager3rich')
165
                        ->with('results', $viewResults)
Phil Höfer's avatar
Phil Höfer committed
166
167
168
169
                        ->with('eingabe', $this->eingabe)
                        ->with('mobile', $this->mobile)
                        ->with('warnings', $this->warnings)
                        ->with('errors', $this->errors)
Phil Höfer's avatar
Phil Höfer committed
170
                        ->with('apiAuthorized', $this->apiAuthorized)
Phil Höfer's avatar
Phil Höfer committed
171
                        ->with('metager', $this)
Dominik Hebeler's avatar
Dominik Hebeler committed
172
173
                        ->with('browser', (new Agent())->browser())
                        ->with('fokus', $this->fokus);
Phil Höfer's avatar
Phil Höfer committed
174
                    break;
175
                case 'rss20':
176
                    return view('resultpages.metager3resultsrss20')
177
178
                        ->with('results', $viewResults)
                        ->with('eingabe', $this->eingabe)
Phil Höfer's avatar
Phil Höfer committed
179
                        ->with('apiAuthorized', $this->apiAuthorized)
180
                        ->with('metager', $this)
Dominik Hebeler's avatar
Dominik Hebeler committed
181
182
                        ->with('resultcount', sizeof($viewResults))
                        ->with('fokus', $this->fokus);
183
                    break;
Aria Givi's avatar
Aria Givi committed
184
                case 'atom10':
185
186
                    return response()->view('resultpages.metager3resultsatom10', ['results' => $viewResults, 'eingabe' => $this->eingabe, 'metager' => $this, 'resultcount' => sizeof($viewResults)])
                        ->header('Content-Type', 'application/xml');
Aria Givi's avatar
Aria Givi committed
187
                    break;
188
                case 'result-count':
189
190
                    # Wir geben die Ergebniszahl und die benötigte Zeit zurück:
                    return sizeof($viewResults) . ";" . round((microtime(true) - $this->starttime), 2);
191
                    break;
192
                default:
193
                    return view('resultpages.resultpage')
194
                        ->with('eingabe', $this->eingabe)
195
                        ->with('focusPages', $focusPages)
196
197
198
                        ->with('mobile', $this->mobile)
                        ->with('warnings', $this->warnings)
                        ->with('errors', $this->errors)
Phil Höfer's avatar
Phil Höfer committed
199
                        ->with('apiAuthorized', $this->apiAuthorized)
200
                        ->with('metager', $this)
201
                        ->with('browser', (new Agent())->browser())
202
                        ->with('quicktips', $quicktipResults)
203
204
                        ->with('resultcount', count($this->results))
                        ->with('focus', $this->fokus);
205
206
                    break;
            }
207
208
209
        }
    }

Phil Höfer's avatar
Phil Höfer committed
210
    public function prepareResults()
211
    {
Phil Höfer's avatar
Phil Höfer committed
212
        $engines = $this->engines;
213

Phil Höfer's avatar
Phil Höfer committed
214
215
        // combine
        $combinedResults = $this->combineResults($engines);
216
217
218
219

        # Wir bestimmen die Sprache eines jeden Suchergebnisses
        $this->results = $this->addLangCodes($this->results);

Phil Höfer's avatar
Phil Höfer committed
220
221
222
223
224
        // sort
        //$sortedResults = $this->sortResults($engines);
        // filter
        // augment (boost&adgoal)
        // authorize
Phil Höfer's avatar
Phil Höfer committed
225
226
227
        if ($this->apiKey) {
            $this->apiAuthorized = $this->authorize($this->apiKey);
        }
Phil Höfer's avatar
Phil Höfer committed
228
        // misc (WiP)
229
        if ($this->fokus == "nachrichten") {
230
231
232
            $this->results = array_filter($this->results, function ($v, $k) {
                return !is_null($v->getRank());
            }, ARRAY_FILTER_USE_BOTH);
233
234
235
236
237
238
239
240
241
242
            uasort($this->results, function ($a, $b) {
                $datea = $a->getDate();
                $dateb = $b->getDate();
                return $dateb - $datea;
            });
        } else {
            uasort($this->results, function ($a, $b) {
                if ($a->getRank() == $b->getRank()) {
                    return 0;
                }
243

244
245
246
                return ($a->getRank() < $b->getRank()) ? 1 : -1;
            });
        }
247

248
249
        # Validate Results
        $newResults = [];
250
251
        foreach ($this->results as $result) {
            if ($result->isValid($this)) {
252
                $newResults[] = $result;
253
254
            }

255
256
257
        }
        $this->results = $newResults;

258
259
260
        #Adgoal Implementation
        $this->results = $this->parseAdgoal($this->results);

Dominik Hebeler's avatar
Dominik Hebeler committed
261
262
        # Human Verification
        $this->results = $this->humanVerification($this->results);
263
        $this->ads = $this->humanVerification($this->ads);
Dominik Hebeler's avatar
Dominik Hebeler committed
264

265
        $counter = 0;
266
        $firstRank = 0;
267

268
        if (isset($this->startForwards)) {
269
            $this->startCount = $this->startForwards;
270
        } elseif (isset($this->startBackwards)) {
271
            $this->startCount = $this->startBackwards - count($this->results) - 1;
272
        } else {
273
274
275
            $this->startCount = 0;
        }

276
277
        foreach ($this->results as $result) {
            if ($counter === 0) {
278
                $firstRank = $result->rank;
279
280
            }

281
            $counter++;
282
            $result->number = $counter + $this->startCount;
283
            $confidence = 0;
284
285
286
            if ($firstRank > 0) {
                $confidence = $result->rank / $firstRank;
            } else {
287
                $confidence = 0;
288
289
290
            }

            if ($confidence > 0.65) {
291
                $result->color = "#FF4000";
292
            } elseif ($confidence > 0.4) {
293
                $result->color = "#FF0080";
294
            } elseif ($confidence > 0.2) {
295
                $result->color = "#C000C0";
296
            } else {
297
                $result->color = "#000000";
298
299
            }

300
301
        }

302
        if ($this->validated) {
303
304
            $this->ads = [];
            $this->maps = false;
305
        }
306
307

        if (count($this->results) <= 0) {
308
309
310
311
312
313
            if (strlen($this->site) > 0) {
                $no_sitesearch_query = str_replace(urlencode("site:" . $this->site), "", $this->fullUrl);
                $this->errors[] = trans('metaGer.results.failedSitesearch', ['altSearch' => $no_sitesearch_query]);
            } else {
                $this->errors[] = trans('metaGer.results.failed');
            }
314
        }
315

316
        if ($this->canCache() && isset($this->next) && count($this->next) > 0 && count($this->results) > 0) {
317
            $page = $this->page + 1;
318
            $this->next = [
319
                'page' => $page,
320
                'startForwards' => $this->results[count($this->results) - 1]->number,
321
                'engines' => $this->next,
322
323
            ];
            Cache::put(md5(serialize($this->next)), serialize($this->next), 60);
324
325
        } else {
            $this->next = [];
326
327
        }

328
    }
329

330
331
    private function addLangCodes($results)
    {
332
333
334
335
336
        # Wenn es keine Ergebnisse gibt, brauchen wir uns gar nicht erst zu bemühen
        if (sizeof($results) === 0) {
            return $results;
        }

337
338
339
340
341
        # Bei der Spracheinstellung "all" wird nicht gefiltert
        if ($this->getLang() === "all") {
            return $results;
        } else {
            # Ansonsten müssen wir jedem Result einen Sprachcode hinzufügen
342
            $id = 0;
343
344
345
346
347
            $langStrings = [];
            foreach ($results as $result) {
                # Wir geben jedem Ergebnis eine ID um später die Sprachcodes zuordnen zu können
                $result->id = $id;

Dominik Hebeler's avatar
Dominik Hebeler committed
348
                $langStrings["result_" . $id] = utf8_encode($result->getLangString());
349
350
351
352
353
354
355
356

                $id++;
            }
            # Wir schreiben die Strings in eine temporäre JSON-Datei,
            # Da das Array unter umständen zu groß ist für eine direkte Übergabe an das Skript
            $filename = "/tmp/" . getmypid();
            file_put_contents($filename, json_encode($langStrings));
            $langDetectorPath = app_path() . "/Models/lang.pl";
357
358
            $lang = exec("echo '$filename' | $langDetectorPath");
            $lang = json_decode($lang, true);
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375

            # Wir haben nun die Sprachcodes der einzelnen Ergebnisse.
            # Diese müssen wir nur noch korrekt zuordnen, dann sind wir fertig.
            foreach ($lang as $key => $langCode) {
                # Prefix vom Key entfernen:
                $id = intval(str_replace("result_", "", $key));
                foreach ($this->results as $result) {
                    if ($result->id === $id) {
                        $result->langCode = $langCode;
                        break;
                    }
                }
            }
            return $results;
        }
    }

Phil Höfer's avatar
Phil Höfer committed
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
    public function combineResults($engines)
    {
        foreach ($engines as $engine) {
            if (isset($engine->next)) {
                $this->next[] = $engine->next;
            }
            if (isset($engine->last)) {
                $this->last[] = $engine->last;
            }
            foreach ($engine->results as $result) {
                if ($result->valid) {
                    $this->results[] = $result;
                }
            }
            foreach ($engine->ads as $ad) {
                $this->ads[] = $ad;
            }
        }
    }

396
397
    public function parseAdgoal($results)
    {
398
        $publicKey = getenv('adgoal_public');
399
        $privateKey = getenv('adgoal_private');
400
        if ($publicKey === false) {
401
402
403
            return $results;
        }
        $tldList = "";
404
405
        try {
            foreach ($results as $result) {
406
                $link = $result->anzeigeLink;
407
                if (strpos($link, "http") !== 0) {
408
409
410
411
412
413
414
415
416
417
                    $link = "http://" . $link;
                }
                $tldList .= parse_url($link, PHP_URL_HOST) . ",";
                $result->tld = parse_url($link, PHP_URL_HOST);
            }
            $tldList = rtrim($tldList, ",");

            # Hashwert
            $hash = md5("meta" . $publicKey . $tldList . "GER");

418
            # Query
419
420
            $query = urlencode($this->q);

421
            $link = "https://api.smartredirect.de/api_v2/CheckForAffiliateUniversalsearchMetager.php?p=" . $publicKey . "&k=" . $hash . "&tld=" . $tldList . "&q=" . $query;
422
423
424
            $answer = json_decode(file_get_contents($link));

            # Nun müssen wir nur noch die Links für die Advertiser ändern:
425
            foreach ($answer as $el) {
426
                $hoster = $el[0];
427
                $hash = $el[1];
428

429
                foreach ($results as $result) {
Dominik Hebeler's avatar
Dominik Hebeler committed
430
                    if ($hoster === $result->tld && !$result->partnershop) {
431
432
                        # Hier ist ein Advertiser:
                        # Das Logo hinzufügen:
433
                        if ($result->image !== "") {
434
                            $result->logo = "https://img.smartredirect.de/logos_v2/60x30/" . $hash . ".gif";
435
                        } else {
436
                            $result->image = "https://img.smartredirect.de/logos_v2/120x60/" . $hash . ".gif";
437
438
                        }

439
440
441
                        # Den Link hinzufügen:
                        $publicKey = $publicKey;
                        $targetUrl = $result->anzeigeLink;
442
                        if (strpos($targetUrl, "http") !== 0) {
443
                            $targetUrl = "http://" . $targetUrl;
444
445
                        }

446
447
448
                        $gateHash = md5($targetUrl . $privateKey);
                        $newLink = "https://api.smartredirect.de/api_v2/ClickGate.php?p=" . $publicKey . "&k=" . $gateHash . "&url=" . urlencode($targetUrl) . "&q=" . $query;
                        $result->link = $newLink;
449
450
451
452
                        $result->partnershop = true;
                    }
                }
            }
453
        } catch (\ErrorException $e) {
454
455
456
457
458
            return $results;
        }

        return $results;
    }
459

460
461
    public function humanVerification($results)
    {
Dominik Hebeler's avatar
Dominik Hebeler committed
462
        # Let's check if we need to implement a redirect for human verification
463
464
        if ($this->verificationCount > 10) {
            foreach ($results as $result) {
Dominik Hebeler's avatar
Dominik Hebeler committed
465
466
467
                $link = $result->link;
                $day = Carbon::now()->day;
                $pw = md5($this->verificationId . $day . $link . env("PROXY_PASSWORD"));
468
469
470
                $url = route('humanverification', ['mm' => $this->verificationId, 'pw' => $pw, "url" => urlencode(str_replace("/", "<<SLASH>>", base64_encode($link)))]);
                $proxyPw = md5($this->verificationId . $day . $result->proxyLink . env("PROXY_PASSWORD"));
                $proxyUrl = route('humanverification', ['mm' => $this->verificationId, 'pw' => $proxyPw, "url" => urlencode(str_replace("/", "<<SLASH>>", base64_encode($result->proxyLink)))]);
Dominik Hebeler's avatar
Dominik Hebeler committed
471
                $result->link = $url;
472
                $result->proxyLink = $proxyUrl;
Dominik Hebeler's avatar
Dominik Hebeler committed
473
474
            }
            return $results;
475
        } else {
Dominik Hebeler's avatar
Dominik Hebeler committed
476
477
478
479
            return $results;
        }
    }

Phil Höfer's avatar
Phil Höfer committed
480
481
482
483
484
485
    public function authorize($key)
    {
        $postdata = http_build_query(array(
            'dummy' => rand(),
        ));
        $opts = array('http' => array(
486
487
            'method' => 'POST',
            'header' => 'Content-type: application/x-www-form-urlencoded',
Phil Höfer's avatar
Phil Höfer committed
488
489
490
491
492
493
494
            'content' => $postdata,
        ),
        );

        $context = stream_context_create($opts);

        try {
495
            $link = "https://key.metager3.de/" . urlencode($key) . "/request-permission/api-access";
Phil Höfer's avatar
Phil Höfer committed
496
497
498
499
500
501
502
503
504
505
506
507
            $result = json_decode(file_get_contents($link, false, $context));
            if ($result->{'api-access'} == true) {
                return true;
            } else {
                return false;
            }

        } catch (\ErrorException $e) {
            return false;
        }
    }

508
509
    public function createQuicktips()
    {
510
511
512
513
514
        # Die quicktips werden als job erstellt und zur Abarbeitung freigegeben
        $quicktips = new \App\Models\Quicktips\Quicktips($this->q, $this->lang, $this->getTime(), $this->getHashCode());
        return $quicktips;
    }

Karl's avatar
Karl committed
515
516
517
518
    /*
     * Die Erstellung der Suchmaschinen bis die Ergebnisse da sind mit Unterfunktionen
     */

519
520
    public function createSearchEngines(Request $request)
    {
521
        # Wenn es kein Suchwort gibt
Dominik Hebeler's avatar
Dominik Hebeler committed
522
        if (!$request->filled("eingabe") || $this->q === "") {
523
            return;
524
        }
525

526
527
        $xml = simplexml_load_file($this->sumaFile);
        $sumas = $xml->xpath("suma");
528
        $enabledSearchengines = [];
529
530
        $overtureEnabled = false;
        $sumaCount = 0;
531

532
533
534
535
536
537
538
539
540
541
542
543
544
545
        /*
        * Erstellt eine Liste mit Foki, die verfügbar sind
        */
        $this->availableFoki = [];
        foreach ($sumas as $suma) {
            $foki = explode(",", trim($suma["type"]));
            foreach ($foki as $fokus) {
                if (!empty($fokus)) {
                    $this->availableFoki[$fokus] = "available";
                }

            }
        }

546
        $isCustomSearch = $this->startsWith($this->fokus, 'focus_');
547

548
        # Im Falle einer Custom-Suche ohne mindestens einer selektierter Suchmaschine wird eine Web-Suche durchgeführt
549
        if ($isCustomSearch && !$this->atLeastOneSearchengineSelected($request)) {
550
551
552
            $isCustomSearch = false;
            $this->fokus = 'web';
        }
553

Karl's avatar
Karl committed
554
        /* Erstellt die Liste der eingestellten Sumas
555
556
557
         * Der einzige Unterschied bei angepasstem Suchfokus ist,
         * dass nicht nach den Typen einer Suma,
         * sondern den im Request mitgegebenen Typen entschieden wird.
Karl's avatar
Karl committed
558
559
560
561
562
563
         * Ansonsten wird genau das selbe geprüft und gemacht:
         * Handelt es sich um spezielle Suchmaschinen die immer an sein müssen
         * Wenn es Overture ist vermerken dass Overture an ist
         * Suma Zähler erhöhen
         * Zu Liste hinzufügen
         */
564
        foreach ($sumas as $suma) {
565
            if (($this->sumaIsSelected($suma, $request, $isCustomSearch)
566
                || (!$this->isBildersuche()
567
                    && $this->sumaIsAdsuche($suma, $overtureEnabled)))
568
569
570
                && (!$this->sumaIsDisabled($suma))) {
                if ($this->sumaIsOverture($suma)) {
                    $overtureEnabled = true;
571
                }
572
573
                if ($this->sumaIsNotAdsuche($suma)) {
                    $sumaCount += 1;
574
                }
575
                $enabledSearchengines[] = $suma;
576
577
578
579
            }
        }

        # Sonderregelung für alle Suchmaschinen, die zu den Minisuchern gehören. Diese können alle gemeinsam über einen Link abgefragt werden
580
        $subcollections = [];
581
582

        $tmp = [];
583
584
585
586
        // Es gibt den Schalter "minism=on" Dieser soll bewirken, dass alle Minisucher angeschaltet werden.
        // Wenn also "minism=on" ist, dann durchsuchen wir statt den tatsächlich angeschalteten Suchmaschinen,
        // alle Suchmaschinen nach "minismCollection"
        if ($request->input("minism", "off") === "on") {
Dominik Hebeler's avatar
Dominik Hebeler committed
587
588
589
            // Wir laden alle Minisucher
            foreach ($sumas as $engine) {
                if (isset($engine["minismCollection"])) {
590
                    $subcollections[] = $engine["minismCollection"]->__toString();
Dominik Hebeler's avatar
Dominik Hebeler committed
591
592
                }
            }
Dominik Hebeler's avatar
Dominik Hebeler committed
593
594
595
596
597
598
            # Nur noch alle eventuell angeschalteten Minisucher deaktivieren
            foreach ($enabledSearchengines as $index => $engine) {
                if (!isset($engine["minismCollection"])) {
                    $tmp[] = $engine;
                }
            }
Dominik Hebeler's avatar
Dominik Hebeler committed
599
600
601
602
        } else {
            // Wir schalten eine Teilmenge, oder aber gar keine an
            foreach ($enabledSearchengines as $engine) {
                if (isset($engine['minismCollection'])) {
603
                    $subcollections[] = $engine['minismCollection']->__toString();
Dominik Hebeler's avatar
Dominik Hebeler committed
604
605
                } else {
                    $tmp[] = $engine;
Dominik Hebeler's avatar
Dominik Hebeler committed
606
                }
607
            }
608
609
        }
        $enabledSearchengines = $tmp;
610
611
        if (sizeof($subcollections) > 0) {
            $enabledSearchengines[] = $this->loadMiniSucher($xml, $subcollections);
612
        }
613
        if ($sumaCount <= 0) {
614
            $this->errors[] = trans('metaGer.settings.noneSelected');
615
        }
616
        $engines = [];
617
        # Wenn eine Sitesearch durchgeführt werden soll, überprüfen wir ob überhaupt eine der Suchmaschinen eine Sitesearch unterstützt
618
        $siteSearchFailed = $this->checkCanNotSitesearch($enabledSearchengines);
619
620

        $typeslist = [];
621
        $counter = 0;
622

623
624
        if ($this->requestIsCached($request)) {
            $engines = $this->getCachedEngines($request);
Dominik Hebeler's avatar
Dominik Hebeler committed
625
            # We need to edit some Options of the Cached Search Engines
Phil Höfer's avatar
Phil Höfer committed
626
            foreach ($engines as $engine) {
Dominik Hebeler's avatar
Dominik Hebeler committed
627
628
                $engine->setResultHash($this->getHashCode());
            }
629
        } else {
630
            $engines = $this->actuallyCreateSearchEngines($enabledSearchengines, $siteSearchFailed);
631
        }
632

633
        # Wir starten alle Suchen
634
635
        foreach ($engines as $engine) {
            $engine->startSearch($this);
636
        }
637

638
639
640
641
642
        /* Wir warten auf die Antwort der Suchmaschinen
         * Die Verbindung steht zu diesem Zeitpunkt und auch unsere Requests wurden schon gesendet.
         * Wir zählen die Suchmaschinen, die durch den Cache beantwortet wurden:
         * $enginesToLoad zählt einerseits die Suchmaschinen auf die wir warten und andererseits
         * welche Suchmaschinen nicht rechtzeitig geantwortet haben.
643
644
         */

645
        $enginesToLoad = [];
646
        $canBreak = false;
647
648
649
650
651
        foreach ($engines as $engine) {
            if ($engine->cached) {
                if ($overtureEnabled && ($engine->name === "overture" || $engine->name === "overtureAds")) {
                    $canBreak = true;
                }
652
653
            } else {
                $enginesToLoad[$engine->name] = false;
654
655
656
657
658
659
660
661
            }
        }

        $this->waitForResults($enginesToLoad, $overtureEnabled, $canBreak);

        $this->retrieveResults($engines);
    }

662
663
    # Spezielle Suchen und Sumas

664
    public function sumaIsSelected($suma, $request, $custom)
665
    {
666
        if ($custom) {
667
            if ($request->filled("engine_" . strtolower($suma["name"]))) {
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
                return true;
            }
        } else {
            $types = explode(",", $suma["type"]);
            if (in_array($this->fokus, $types)) {
                return true;
            }
        }
        return false;
    }

    public function actuallyCreateSearchEngines($enabledSearchengines, $siteSearchFailed)
    {
        $engines = [];
        foreach ($enabledSearchengines as $engine) {

            # Wenn diese Suchmaschine gar nicht eingeschaltet sein soll
            if (!$siteSearchFailed
                && strlen($this->site) > 0
                && (!isset($engine['hasSiteSearch'])
                    || $engine['hasSiteSearch']->__toString() === "0")) {
                continue;
            }

Dominik Hebeler's avatar
Dominik Hebeler committed
692
693
694
            if (!isset($engine["package"])) {
                die(var_dump($engine));
            }
695
696
697
698
699
            # Setze Pfad zu Parser
            $path = "App\Models\parserSkripte\\" . ucfirst($engine["package"]->__toString());

            # Prüfe ob Parser vorhanden
            if (!file_exists(app_path() . "/Models/parserSkripte/" . ucfirst($engine["package"]->__toString()) . ".php")) {
700
                Log::error("Konnte " . $engine["name"] . " nicht abfragen, da kein Parser existiert");
701
                $this->errors[] = trans('metaGer.engines.noParser', ['engine' => $engine["name"]]);
702
703
704
705
706
707
708
709
                continue;
            }

            # Es wird versucht die Suchengine zu erstellen
            $time = microtime();
            try {
                $tmp = new $path($engine, $this);
            } catch (\ErrorException $e) {
710
                Log::error("Konnte " . $engine["name"] . " nicht abfragen. " . var_dump($e));
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
                continue;
            }

            # Ausgabe bei Debug-Modus
            if ($tmp->enabled && isset($this->debug)) {
                $this->warnings[] = $tmp->service . "   Connection_Time: " . $tmp->connection_time . "    Write_Time: " . $tmp->write_time . " Insgesamt:" . ((microtime() - $time) / 1000);
            }

            # Wenn die neu erstellte Engine eingeschaltet ist, wird sie der Liste hinzugefügt
            if ($tmp->isEnabled()) {
                $engines[] = $tmp;
            }
        }
        return $engines;
    }

    public function isBildersuche()
    {
729
        return $this->fokus === "bilder";
730
731
732
733
    }

    public function sumaIsAdsuche($suma, $overtureEnabled)
    {
734
        $sumaName = $suma["name"]->__toString();
735
        return
736
737
            $sumaName === "qualigo"
            || $sumaName === "similar_product_ads"
Karl Hasselbring's avatar
Karl Hasselbring committed
738
            || (!$overtureEnabled && $sumaName === "overtureAds");
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
    }

    public function sumaIsDisabled($suma)
    {
        return
        isset($suma['disabled'])
        && $suma['disabled']->__toString() === "1";
    }

    public function sumaIsOverture($suma)
    {
        return
        $suma["name"]->__toString() === "overture"
        || $suma["name"]->__toString() === "overtureAds";
    }

    public function sumaIsNotAdsuche($suma)
    {
        return
        $suma["name"]->__toString() !== "qualigo"
        && $suma["name"]->__toString() !== "similar_product_ads"
        && $suma["name"]->__toString() !== "overtureAds";
    }

    public function requestIsCached($request)
    {
        return
Dominik Hebeler's avatar
Dominik Hebeler committed
766
        $request->filled('next')
767
768
769
770
771
772
        && Cache::has($request->input('next'))
        && unserialize(Cache::get($request->input('next')))['page'] > 1;
    }

    public function getCachedEngines($request)
    {
773
        $next = unserialize(Cache::get($request->input('next')));
774
        $this->page = $next['page'];
775
        $engines = $next['engines'];
776
777
778
779
780
781
782
783
784
785
786
        if (isset($next['startForwards'])) {
            $this->startForwards = $next['startForwards'];
        }
        if (isset($next['startBackwards'])) {
            $this->startBackwards = $next['startBackwards'];
        }
        return $engines;
    }

    public function loadMiniSucher($xml, $subcollections)
    {
787
        $minisucherEngine = $xml->xpath('suma[@name="minism"]')[0];
788
        $minisucherEngine["subcollections"] = implode(", ", $subcollections);
789
        $subcollectionsString = urlencode("(" . implode(" OR ", $subcollections) . ")");
790
        $minisucherEngine["formData"] = str_replace("<<SUBCOLLECTIONS>>", $subcollectionsString, $minisucherEngine["formData"]);
791
792
793
794
        $minisucherEngine["formData"] = str_replace("<<COUNT>>", sizeof($subcollections) * 10, $minisucherEngine["formData"]);
        return $minisucherEngine;
    }

795
    # Passt den Suchfokus an, falls für einen Fokus genau alle vorhandenen Sumas eingeschaltet sind
796
797
    public function adjustFocus($sumas, $enabledSearchengines)
    {
798
799
        # Findet für alle Foki die enthaltenen Sumas
        $foki = []; # [fokus][suma] => [suma]
800
        foreach ($sumas as $suma) {
801
            if ((!$this->sumaIsDisabled($suma)) && (!isset($suma['userSelectable']) || $suma['userSelectable']->__toString() === "1")) {
802
                if (isset($suma['type'])) {
803
804
805
                    # Wenn foki für diese Suchmaschine angegeben sind
                    $focuses = explode(",", $suma['type']->__toString());
                    foreach ($focuses as $foc) {
806
807
808
809
810
                        if (isset($suma['minismCollection'])) {
                            $foki[$foc][] = "minism";
                        } else {
                            $foki[$foc][] = $suma['name']->__toString();
                        }
811
                    }
812
                } else {
813
                    # Wenn keine foki für diese Suchmaschine angegeben sind
814
815
816
817
818
                    if (isset($suma['minismCollection'])) {
                        $foki["andere"][] = "minism";
                    } else {
                        $foki["andere"][] = $suma['name']->__toString();
                    }
819
820
821
822
                }
            }
        }

823
        # Findet die Namen der aktuell eingeschalteten Sumas
824
        $realEngNames = [];
825
        foreach ($enabledSearchengines as $realEng) {
826
            $nam = $realEng["name"]->__toString();
Karl Hasselbring's avatar
Karl Hasselbring committed
827
            if ($nam !== "qualigo" && $nam !== "overtureAds") {
828
829
830
                $realEngNames[] = $nam;
            }
        }
831

832
        # Anschließend werden diese beiden Listen verglichen (jeweils eine der Fokuslisten für jeden Fokus), um herauszufinden ob sie vielleicht identisch sind. Ist dies der Fall, so hat der Nutzer anscheinend Suchmaschinen eines kompletten Fokus eingestellt. Der Fokus wird dementsprechend angepasst.
833
        foreach ($foki as $fok => $engines) {
834
            $isFokus = true;
835
            $fokiEngNames = [];
836
            foreach ($engines as $eng) {
837
838
                $fokiEngNames[] = $eng;
            }
839
            # Jede eingeschaltete Engine ist für diesen Fokus geeignet
840
            foreach ($fokiEngNames as $fen) {
841
842
                # Bei Bildersuchen ist uns egal, ob alle Suchmaschinen aus dem Suchfokus eingeschaltet sind, da wir sie eh als Bildersuche anzeigen müssen
                if (!in_array($fen, $realEngNames) && $fok !== "bilder") {
843
844
845
                    $isFokus = false;
                }
            }
846
            # Jede im Fokus erwartete Engine ist auch eingeschaltet
847
848
            foreach ($realEngNames as $ren) {
                if (!in_array($ren, $fokiEngNames)) {
849
850
851
                    $isFokus = false;
                }
            }
852
            # Wenn die Listen identisch sind, setze den Fokus um
853
            if ($isFokus) {
854
855
856
                $this->fokus = $fok;
            }
        }
857
    }
858

859
860
861
862
863
864
865
    public function checkCanNotSitesearch($enabledSearchengines)
    {
        if (strlen($this->site) > 0) {
            $enginesWithSite = 0;
            foreach ($enabledSearchengines as $engine) {
                if (isset($engine['hasSiteSearch']) && $engine['hasSiteSearch']->__toString() === "1") {
                    $enginesWithSite++;
866
                }
867
868
            }
            if ($enginesWithSite === 0) {
869
                $this->errors[] = trans('metaGer.sitesearch.failed', ['site' => $this->site, 'searchLink' => $this->generateSearchLink("web", false)]);
870
871
                return true;
            } else {
872
                $this->warnings[] = trans('metaGer.sitesearch.success', ['site' => $this->site]);
873
                return false;
874
875
            }
        }
876
        return false;
877
878
879
880
    }

    public function waitForResults($enginesToLoad, $overtureEnabled, $canBreak)
    {
881

Phil Höfer's avatar
Phil Höfer committed
882
        $timeStart = microtime(true);
883
        $results = null;
884
        while (true) {
885
886
887
            $results = Redis::hgetall('search.' . $this->getHashCode());

            $ready = true;
Phil Höfer's avatar
Phil Höfer committed
888
            // When every
889
            $connected = true;
Phil Höfer's avatar
Phil Höfer committed
890
891
            foreach ($results as $key => $value) {
                if ($value === "waiting" || $value === "connected") {
892
                    $ready = false;
893
                }
Phil Höfer's avatar
Phil Höfer committed
894
                if ($value === "waiting") {
895
                    $connected = false;
896
                }
897
            }
898

899
900
            // If $ready is false at this point, we're waiting for more searchengines
            // But we have to check for the timeout, too
Phil Höfer's avatar
Phil Höfer committed
901
902
903
904
            if (!$connected) {
                $timeStart = microtime(true);
            }

905
906
907
908
            $time = (microtime(true) - $timeStart) * 1000;
            // We will apply the timeout only if it's not Yahoo we're waiting for since they are one the most
            // important search engines.
            $canTimeout = !((isset($results["overture"]) && $results["overture"] === "waiting") || (isset($results["overtureAds"]) && $results["overtureAds"] === "waiting"));
Phil Höfer's avatar
Phil Höfer committed
909
910
911
            if ($time > $this->time && $canTimeout) {
                $ready = true;
            }
912

Phil Höfer's avatar
Phil Höfer committed
913
            if ($ready) {
914
915
916
917
                break;
            }
            usleep(50000);
        }
918
919

        # Wir haben nun so lange wie möglich gewartet. Wir registrieren nun noch die Suchmaschinen, die geanwortet haben.
920
        foreach ($results as $key => $value) {
<