MetaGer.php 58 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
        #Adgoal Implementation
259
260
261
        if (!$this->apiAuthorized) {
            $this->results = $this->parseAdgoal($this->results);
        }
262

Dominik Hebeler's avatar
Dominik Hebeler committed
263
264
        # Human Verification
        $this->results = $this->humanVerification($this->results);
265
        $this->ads = $this->humanVerification($this->ads);
Dominik Hebeler's avatar
Dominik Hebeler committed
266

267
        $counter = 0;
268
        $firstRank = 0;
269

270
        if (isset($this->startForwards)) {
271
            $this->startCount = $this->startForwards;
272
        } elseif (isset($this->startBackwards)) {
273
            $this->startCount = $this->startBackwards - count($this->results) - 1;
274
        } else {
275
276
277
            $this->startCount = 0;
        }

278
279
        foreach ($this->results as $result) {
            if ($counter === 0) {
280
                $firstRank = $result->rank;
281
282
            }

283
            $counter++;
284
            $result->number = $counter + $this->startCount;
285
            $confidence = 0;
286
287
288
            if ($firstRank > 0) {
                $confidence = $result->rank / $firstRank;
            } else {
289
                $confidence = 0;
290
291
292
            }

            if ($confidence > 0.65) {
293
                $result->color = "#FF4000";
294
            } elseif ($confidence > 0.4) {
295
                $result->color = "#FF0080";
296
            } elseif ($confidence > 0.2) {
297
                $result->color = "#C000C0";
298
            } else {
299
                $result->color = "#000000";
300
301
            }

302
303
        }

304
        if ($this->validated) {
305
306
            $this->ads = [];
            $this->maps = false;
307
        }
308
309

        if (count($this->results) <= 0) {
310
311
312
313
314
315
            if (strlen($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');
            }
316
        }
317

318
        if ($this->canCache() && isset($this->next) && count($this->next) > 0 && count($this->results) > 0) {
319
            $page = $this->page + 1;
320
            $this->next = [
321
                'page' => $page,
322
                'startForwards' => $this->results[count($this->results) - 1]->number,
323
                'engines' => $this->next,
324
325
            ];
            Cache::put(md5(serialize($this->next)), serialize($this->next), 60);
326
327
        } else {
            $this->next = [];
328
329
        }

330
    }
331

332
333
    private function addLangCodes($results)
    {
334
335
336
337
338
        # Wenn es keine Ergebnisse gibt, brauchen wir uns gar nicht erst zu bemühen
        if (sizeof($results) === 0) {
            return $results;
        }

339
340
341
342
343
        # Bei der Spracheinstellung "all" wird nicht gefiltert
        if ($this->getLang() === "all") {
            return $results;
        } else {
            # Ansonsten müssen wir jedem Result einen Sprachcode hinzufügen
344
            $id = 0;
345
346
347
348
349
            $langStrings = [];
            foreach ($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
350
                $langStrings["result_" . $id] = utf8_encode($result->getLangString());
351
352
353
354
355
356
357
358

                $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";
359
360
            $lang = exec("echo '$filename' | $langDetectorPath");
            $lang = json_decode($lang, true);
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377

            # 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
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
    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;
            }
        }
    }

398
399
    public function parseAdgoal($results)
    {
400
        $publicKey = getenv('adgoal_public');
401
        $privateKey = getenv('adgoal_private');
402
        if ($publicKey === false) {
403
404
405
            return $results;
        }
        $tldList = "";
406
407
        try {
            foreach ($results as $result) {
408
                $link = $result->anzeigeLink;
409
                if (strpos($link, "http") !== 0) {
410
411
412
413
414
415
416
417
418
419
                    $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");

420
            # Query
421
422
            $query = urlencode($this->q);

423
            $link = "https://api.smartredirect.de/api_v2/CheckForAffiliateUniversalsearchMetager.php?p=" . $publicKey . "&k=" . $hash . "&tld=" . $tldList . "&q=" . $query;
424
425
426
            $answer = json_decode(file_get_contents($link));

            # Nun müssen wir nur noch die Links für die Advertiser ändern:
427
            foreach ($answer as $el) {
428
                $hoster = $el[0];
429
                $hash = $el[1];
430

431
                foreach ($results as $result) {
Dominik Hebeler's avatar
Dominik Hebeler committed
432
                    if ($hoster === $result->tld && !$result->partnershop) {
433
434
                        # Hier ist ein Advertiser:
                        # Das Logo hinzufügen:
435
                        if ($result->image !== "") {
436
                            $result->logo = "https://img.smartredirect.de/logos_v2/60x30/" . $hash . ".gif";
437
                        } else {
438
                            $result->image = "https://img.smartredirect.de/logos_v2/120x60/" . $hash . ".gif";
439
440
                        }

441
442
443
                        # Den Link hinzufügen:
                        $publicKey = $publicKey;
                        $targetUrl = $result->anzeigeLink;
444
                        if (strpos($targetUrl, "http") !== 0) {
445
                            $targetUrl = "http://" . $targetUrl;
446
447
                        }

448
449
450
                        $gateHash = md5($targetUrl . $privateKey);
                        $newLink = "https://api.smartredirect.de/api_v2/ClickGate.php?p=" . $publicKey . "&k=" . $gateHash . "&url=" . urlencode($targetUrl) . "&q=" . $query;
                        $result->link = $newLink;
451
452
453
454
                        $result->partnershop = true;
                    }
                }
            }
455
        } catch (\ErrorException $e) {
456
457
458
459
460
            return $results;
        }

        return $results;
    }
461

462
463
    public function humanVerification($results)
    {
Dominik Hebeler's avatar
Dominik Hebeler committed
464
        # Let's check if we need to implement a redirect for human verification
465
466
        if ($this->verificationCount > 10) {
            foreach ($results as $result) {
Dominik Hebeler's avatar
Dominik Hebeler committed
467
468
469
                $link = $result->link;
                $day = Carbon::now()->day;
                $pw = md5($this->verificationId . $day . $link . env("PROXY_PASSWORD"));
470
471
472
                $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
473
                $result->link = $url;
474
                $result->proxyLink = $proxyUrl;
Dominik Hebeler's avatar
Dominik Hebeler committed
475
476
            }
            return $results;
477
        } else {
Dominik Hebeler's avatar
Dominik Hebeler committed
478
479
480
481
            return $results;
        }
    }

Phil Höfer's avatar
Phil Höfer committed
482
483
484
485
486
487
    public function authorize($key)
    {
        $postdata = http_build_query(array(
            'dummy' => rand(),
        ));
        $opts = array('http' => array(
488
489
            'method' => 'POST',
            'header' => 'Content-type: application/x-www-form-urlencoded',
Phil Höfer's avatar
Phil Höfer committed
490
491
492
493
494
495
496
            'content' => $postdata,
        ),
        );

        $context = stream_context_create($opts);

        try {
497
            $link = "https://key.metager3.de/" . urlencode($key) . "/request-permission/api-access";
Phil Höfer's avatar
Phil Höfer committed
498
499
500
501
502
503
504
505
506
507
508
509
            $result = json_decode(file_get_contents($link, false, $context));
            if ($result->{'api-access'} == true) {
                return true;
            } else {
                return false;
            }

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

510
511
    public function createQuicktips()
    {
512
513
514
515
516
        # 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
517
518
519
520
    /*
     * Die Erstellung der Suchmaschinen bis die Ergebnisse da sind mit Unterfunktionen
     */

521
522
    public function createSearchEngines(Request $request)
    {
523
        # Wenn es kein Suchwort gibt
Dominik Hebeler's avatar
Dominik Hebeler committed
524
        if (!$request->filled("eingabe") || $this->q === "") {
525
            return;
526
        }
527

528
529
        $xml = simplexml_load_file($this->sumaFile);
        $sumas = $xml->xpath("suma");
530
        $enabledSearchengines = [];
531
532
        $overtureEnabled = false;
        $sumaCount = 0;
533

534
535
536
537
538
539
540
541
542
543
544
545
546
547
        /*
        * 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";
                }

            }
        }

548
        $isCustomSearch = $this->startsWith($this->fokus, 'focus_');
549

550
        # Im Falle einer Custom-Suche ohne mindestens einer selektierter Suchmaschine wird eine Web-Suche durchgeführt
551
        if ($isCustomSearch && !$this->atLeastOneSearchengineSelected($request)) {
552
553
554
            $isCustomSearch = false;
            $this->fokus = 'web';
        }
555

Karl's avatar
Karl committed
556
        /* Erstellt die Liste der eingestellten Sumas
557
558
559
         * 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
560
561
562
563
564
565
         * 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
         */
566
        foreach ($sumas as $suma) {
567
            if (($this->sumaIsSelected($suma, $request, $isCustomSearch)
568
                || (!$this->isBildersuche()
569
                    && $this->sumaIsAdsuche($suma, $overtureEnabled)))
570
571
572
                && (!$this->sumaIsDisabled($suma))) {
                if ($this->sumaIsOverture($suma)) {
                    $overtureEnabled = true;
573
                }
574
575
                if ($this->sumaIsNotAdsuche($suma)) {
                    $sumaCount += 1;
576
                }
577
                $enabledSearchengines[] = $suma;
578
579
580
581
            }
        }

        # Sonderregelung für alle Suchmaschinen, die zu den Minisuchern gehören. Diese können alle gemeinsam über einen Link abgefragt werden
582
        $subcollections = [];
583
584

        $tmp = [];
585
586
587
588
        // Es gibt den Schalter "minism=on" Dieser soll bewirken, dass alle Minisucher angeschaltet werden.
        // Wenn also "minism=on" ist, dann durchsuchen wir statt den tatsächlich angeschalteten Suchmaschinen,
        // alle Suchmaschinen nach "minismCollection"
        if ($request->input("minism", "off") === "on") {
Dominik Hebeler's avatar
Dominik Hebeler committed
589
590
591
            // Wir laden alle Minisucher
            foreach ($sumas as $engine) {
                if (isset($engine["minismCollection"])) {
592
                    $subcollections[] = $engine["minismCollection"]->__toString();
Dominik Hebeler's avatar
Dominik Hebeler committed
593
594
                }
            }
Dominik Hebeler's avatar
Dominik Hebeler committed
595
596
597
598
599
600
            # Nur noch alle eventuell angeschalteten Minisucher deaktivieren
            foreach ($enabledSearchengines as $index => $engine) {
                if (!isset($engine["minismCollection"])) {
                    $tmp[] = $engine;
                }
            }
Dominik Hebeler's avatar
Dominik Hebeler committed
601
602
603
604
        } else {
            // Wir schalten eine Teilmenge, oder aber gar keine an
            foreach ($enabledSearchengines as $engine) {
                if (isset($engine['minismCollection'])) {
605
                    $subcollections[] = $engine['minismCollection']->__toString();
Dominik Hebeler's avatar
Dominik Hebeler committed
606
607
                } else {
                    $tmp[] = $engine;
Dominik Hebeler's avatar
Dominik Hebeler committed
608
                }
609
            }
610
611
        }
        $enabledSearchengines = $tmp;
612
613
        if (sizeof($subcollections) > 0) {
            $enabledSearchengines[] = $this->loadMiniSucher($xml, $subcollections);
614
        }
615
        if ($sumaCount <= 0) {
616
            $this->errors[] = trans('metaGer.settings.noneSelected');
617
        }
618
        $engines = [];
619
        # Wenn eine Sitesearch durchgeführt werden soll, überprüfen wir ob überhaupt eine der Suchmaschinen eine Sitesearch unterstützt
620
        $siteSearchFailed = $this->checkCanNotSitesearch($enabledSearchengines);
621
622

        $typeslist = [];
623
        $counter = 0;
624

625
626
        if ($this->requestIsCached($request)) {
            $engines = $this->getCachedEngines($request);
Dominik Hebeler's avatar
Dominik Hebeler committed
627
            # We need to edit some Options of the Cached Search Engines
Phil Höfer's avatar
Phil Höfer committed
628
            foreach ($engines as $engine) {
Dominik Hebeler's avatar
Dominik Hebeler committed
629
630
                $engine->setResultHash($this->getHashCode());
            }
631
        } else {
632
            $engines = $this->actuallyCreateSearchEngines($enabledSearchengines, $siteSearchFailed);
633
        }
634

635
        # Wir starten alle Suchen
636
637
        foreach ($engines as $engine) {
            $engine->startSearch($this);
638
        }
639

640
641
642
643
644
        /* 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.
645
646
         */

647
        $enginesToLoad = [];
648
        $canBreak = false;
649
650
651
652
653
        foreach ($engines as $engine) {
            if ($engine->cached) {
                if ($overtureEnabled && ($engine->name === "overture" || $engine->name === "overtureAds")) {
                    $canBreak = true;
                }
654
655
            } else {
                $enginesToLoad[$engine->name] = false;
656
657
658
659
660
661
662
663
            }
        }

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

        $this->retrieveResults($engines);
    }

664
665
    # Spezielle Suchen und Sumas

666
    public function sumaIsSelected($suma, $request, $custom)
667
    {
668
        if ($custom) {
669
            if ($request->filled("engine_" . strtolower($suma["name"]))) {
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
                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
694
695
696
            if (!isset($engine["package"])) {
                die(var_dump($engine));
            }
697
698
699
700
701
            # 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")) {
702
                Log::error("Konnte " . $engine["name"] . " nicht abfragen, da kein Parser existiert");
703
                $this->errors[] = trans('metaGer.engines.noParser', ['engine' => $engine["name"]]);
704
705
706
707
708
709
710
711
                continue;
            }

            # Es wird versucht die Suchengine zu erstellen
            $time = microtime();
            try {
                $tmp = new $path($engine, $this);
            } catch (\ErrorException $e) {
712
                Log::error("Konnte " . $engine["name"] . " nicht abfragen. " . var_dump($e));
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
                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()
    {
731
        return $this->fokus === "bilder";
732
733
734
735
    }

    public function sumaIsAdsuche($suma, $overtureEnabled)
    {
736
        $sumaName = $suma["name"]->__toString();
737
        return
738
739
            $sumaName === "qualigo"
            || $sumaName === "similar_product_ads"
Karl Hasselbring's avatar
Karl Hasselbring committed
740
            || (!$overtureEnabled && $sumaName === "overtureAds");
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
766
767
    }

    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
768
        $request->filled('next')
769
770
771
772
773
774
        && Cache::has($request->input('next'))
        && unserialize(Cache::get($request->input('next')))['page'] > 1;
    }

    public function getCachedEngines($request)
    {
775
        $next = unserialize(Cache::get($request->input('next')));
776
        $this->page = $next['page'];
777
        $engines = $next['engines'];
778
779
780
781
782
783
784
785
786
787
788
        if (isset($next['startForwards'])) {
            $this->startForwards = $next['startForwards'];
        }
        if (isset($next['startBackwards'])) {
            $this->startBackwards = $next['startBackwards'];
        }
        return $engines;
    }

    public function loadMiniSucher($xml, $subcollections)
    {
789
        $minisucherEngine = $xml->xpath('suma[@name="minism"]')[0];
790
        $minisucherEngine["subcollections"] = implode(", ", $subcollections);
791
        $subcollectionsString = urlencode("(" . implode(" OR ", $subcollections) . ")");
792
        $minisucherEngine["formData"] = str_replace("<<SUBCOLLECTIONS>>", $subcollectionsString, $minisucherEngine["formData"]);
793
794
795
796
        $minisucherEngine["formData"] = str_replace("<<COUNT>>", sizeof($subcollections) * 10, $minisucherEngine["formData"]);
        return $minisucherEngine;
    }

797
    # Passt den Suchfokus an, falls für einen Fokus genau alle vorhandenen Sumas eingeschaltet sind
798
799
    public function adjustFocus($sumas, $enabledSearchengines)
    {
800
801
        # Findet für alle Foki die enthaltenen Sumas
        $foki = []; # [fokus][suma] => [suma]
802
        foreach ($sumas as $suma) {
803
            if ((!$this->sumaIsDisabled($suma)) && (!isset($suma['userSelectable']) || $suma['userSelectable']->__toString() === "1")) {
804
                if (isset($suma['type'])) {
805
806
807
                    # Wenn foki für diese Suchmaschine angegeben sind
                    $focuses = explode(",", $suma['type']->__toString());
                    foreach ($focuses as $foc) {
808
809
810
811
812
                        if (isset($suma['minismCollection'])) {
                            $foki[$foc][] = "minism";
                        } else {
                            $foki[$foc][] = $suma['name']->__toString();
                        }
813
                    }
814
                } else {
815
                    # Wenn keine foki für diese Suchmaschine angegeben sind
816
817
818
819
820
                    if (isset($suma['minismCollection'])) {
                        $foki["andere"][] = "minism";
                    } else {
                        $foki["andere"][] = $suma['name']->__toString();
                    }
821
822
823
824
                }
            }
        }

825
        # Findet die Namen der aktuell eingeschalteten Sumas
826
        $realEngNames = [];
827
        foreach ($enabledSearchengines as $realEng) {
828
            $nam = $realEng["name"]->__toString();
Karl Hasselbring's avatar
Karl Hasselbring committed
829
            if ($nam !== "qualigo" && $nam !== "overtureAds") {
830
831
832
                $realEngNames[] = $nam;
            }
        }
833

834
        # 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.
835
        foreach ($foki as $fok => $engines) {
836
            $isFokus = true;
837
            $fokiEngNames = [];
838
            foreach ($engines as $eng) {
839
840
                $fokiEngNames[] = $eng;
            }
841
            # Jede eingeschaltete Engine ist für diesen Fokus geeignet
842
            foreach ($fokiEngNames as $fen) {
843
844
                # Bei Bildersuchen ist uns egal, ob alle Suchmaschinen aus dem Suchfokus eingeschaltet sind, da wir sie eh als Bildersuche anzeigen müssen
                if (!in_array($fen, $realEngNames) && $fok !== "bilder") {
845
846
847
                    $isFokus = false;
                }
            }
848
            # Jede im Fokus erwartete Engine ist auch eingeschaltet
849
850
            foreach ($realEngNames as $ren) {
                if (!in_array($ren, $fokiEngNames)) {
851
852
853
                    $isFokus = false;
                }
            }
854
            # Wenn die Listen identisch sind, setze den Fokus um
855
            if ($isFokus) {
856
857
858
                $this->fokus = $fok;
            }
        }
859
    }
860

861
862
863
864
865
866
867
    public function checkCanNotSitesearch($enabledSearchengines)
    {
        if (strlen($this->site) > 0) {
            $enginesWithSite = 0;
            foreach ($enabledSearchengines as $engine) {
                if (isset($engine['hasSiteSearch']) && $engine['hasSiteSearch']->__toString() === "1") {
                    $enginesWithSite++;
868
                }
869
870
            }
            if ($enginesWithSite === 0) {
871
                $this->errors[] = trans('metaGer.sitesearch.failed', ['site' => $this->site, 'searchLink' => $this->generateSearchLink("web", false)]);
872
873
                return true;
            } else {
874
                $this->warnings[] = trans('metaGer.sitesearch.success', ['site' => $this->site]);
875
                return false;
876
877
            }
        }
878
        return false;
879
880
881
882
    }

    public function waitForResults($enginesToLoad, $overtureEnabled, $canBreak)
    {
883

Phil Höfer's avatar
Phil Höfer committed
884
        $timeStart = microtime(true);
885
        $results = null;
886
        while (true) {
887
888
889
            $results = Redis::hgetall('search.' . $this->getHashCode());

            $ready = true;
Phil Höfer's avatar
Phil Höfer committed
890
            // When every
891
            $connected = true;
Phil Höfer's avatar
Phil Höfer committed
892
893
            foreach ($results as $key => $value) {
                if ($value === "waiting" || $value === "connected") {
894
                    $ready = false;
895
                }
Phil Höfer's avatar
Phil Höfer committed
896
                if ($value === "waiting") {
897
                    $connected = false;
898
                }
899
            }
900

901
902
            // If $ready is false at this point, we're waiting for more searchengines
            // But we have to check for the timeout, too
Phil Höfer's avatar
Phil Höfer committed
903
904
905
906
            if (!$connected) {
                $timeStart = microtime(true);
            }

907
908
909
910
            $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"]) && $results["overtureAds"] === "waiting"));
Phil Höfer's avatar
Phil Höfer committed
911
912
913
            if ($time > $this->time && $canTimeout) {
                $ready = true;
            }
914

Phil Höfer's avatar
Phil Höfer committed
915
            if ($ready) {
916
917
918
919
                break;
            }
            usleep(50000);
        }
920
921

        # Wir haben nun so lange wie möglich gewartet. Wir registrieren nun noch die Suchmaschinen, die geanwortet haben.
922
        foreach ($results as $key => $value) {
923
924
925
            $enginesToLoad[$key] = true;
        }
        $this->enginesToLoad = $enginesToLoad;
926
    }
927

928
929
930
    public function retrieveResults($engines)
    {
        # Von geladenen Engines die Ergebnisse holen
931
932
933
        foreach ($engines as $engine) {
            if (!$engine->loaded) {
                try {
934
                    $engine->retrieveResults($this);
935
                } catch (\ErrorException $e) {
936
937
938
939
                    Log::error($e);
                }
            }
        }
940

941
        # Nicht fertige Engines verwefen
942
943
        foreach ($engines as $engine) {
            if (!$engine->loaded) {
944
                $engine->shutdown();
945
            }
946
947
948
        }

        $this->engines = $engines;
Dominik Hebeler's avatar