MetaGer.php 58.3 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
        foreach ($this->request->all() as $key => $value) {
101
102
            if (starts_with($key, 'engine_') && $value === 'on') {
                $focusPages[] = $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;
184
185
186
                case 'api':
                    return response()->view('resultpages.metager3resultsatom10', ['results' => $viewResults, 'eingabe' => $this->eingabe, 'metager' => $this, 'resultcount' => sizeof($viewResults), 'apiAuthorized' => $this->apiAuthorized])->header('Content-Type', 'application/xml');
                    break;
Aria Givi's avatar
Aria Givi committed
187
                case 'atom10':
188
                    return response()->view('resultpages.metager3resultsatom10', ['results' => $viewResults, 'eingabe' => $this->eingabe, 'metager' => $this, 'resultcount' => sizeof($viewResults), 'apiAuthorized' => true])
189
                        ->header('Content-Type', 'application/xml');
Aria Givi's avatar
Aria Givi committed
190
                    break;
191
                case 'result-count':
192
193
                    # Wir geben die Ergebniszahl und die benötigte Zeit zurück:
                    return sizeof($viewResults) . ";" . round((microtime(true) - $this->starttime), 2);
194
                    break;
195
                default:
196
                    return view('resultpages.resultpage')
197
                        ->with('eingabe', $this->eingabe)
198
                        ->with('focusPages', $focusPages)
199
200
201
                        ->with('mobile', $this->mobile)
                        ->with('warnings', $this->warnings)
                        ->with('errors', $this->errors)
Phil Höfer's avatar
Phil Höfer committed
202
                        ->with('apiAuthorized', $this->apiAuthorized)
203
                        ->with('metager', $this)
204
                        ->with('browser', (new Agent())->browser())
205
                        ->with('quicktips', $quicktipResults)
206
207
                        ->with('resultcount', count($this->results))
                        ->with('focus', $this->fokus);
208
209
                    break;
            }
210
211
212
        }
    }

Phil Höfer's avatar
Phil Höfer committed
213
    public function prepareResults()
214
    {
Phil Höfer's avatar
Phil Höfer committed
215
        $engines = $this->engines;
216

Phil Höfer's avatar
Phil Höfer committed
217
218
        // combine
        $combinedResults = $this->combineResults($engines);
219
220
221
222

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

Phil Höfer's avatar
Phil Höfer committed
223
224
225
226
227
        // sort
        //$sortedResults = $this->sortResults($engines);
        // filter
        // augment (boost&adgoal)
        // authorize
Phil Höfer's avatar
Phil Höfer committed
228
229
230
        if ($this->apiKey) {
            $this->apiAuthorized = $this->authorize($this->apiKey);
        }
Phil Höfer's avatar
Phil Höfer committed
231
        // misc (WiP)
232
        if ($this->fokus == "nachrichten") {
233
234
235
            $this->results = array_filter($this->results, function ($v, $k) {
                return !is_null($v->getRank());
            }, ARRAY_FILTER_USE_BOTH);
236
237
238
239
240
241
242
243
244
245
            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;
                }
246

247
248
249
                return ($a->getRank() < $b->getRank()) ? 1 : -1;
            });
        }
250

251
252
        # Validate Results
        $newResults = [];
253
254
        foreach ($this->results as $result) {
            if ($result->isValid($this)) {
255
                $newResults[] = $result;
256
257
            }

258
259
260
        }
        $this->results = $newResults;

261
        #Adgoal Implementation
262
263
264
        if (!$this->apiAuthorized) {
            $this->results = $this->parseAdgoal($this->results);
        }
265

Dominik Hebeler's avatar
Dominik Hebeler committed
266
267
        # Human Verification
        $this->results = $this->humanVerification($this->results);
268
        $this->ads = $this->humanVerification($this->ads);
Dominik Hebeler's avatar
Dominik Hebeler committed
269

270
        $counter = 0;
271
        $firstRank = 0;
272

273
        if (isset($this->startForwards)) {
274
            $this->startCount = $this->startForwards;
275
        } elseif (isset($this->startBackwards)) {
276
            $this->startCount = $this->startBackwards - count($this->results) - 1;
277
        } else {
278
279
280
            $this->startCount = 0;
        }

281
282
        foreach ($this->results as $result) {
            if ($counter === 0) {
283
                $firstRank = $result->rank;
284
285
            }

286
            $counter++;
287
            $result->number = $counter + $this->startCount;
288
            $confidence = 0;
289
290
291
            if ($firstRank > 0) {
                $confidence = $result->rank / $firstRank;
            } else {
292
                $confidence = 0;
293
294
295
            }

            if ($confidence > 0.65) {
296
                $result->color = "#FF4000";
297
            } elseif ($confidence > 0.4) {
298
                $result->color = "#FF0080";
299
            } elseif ($confidence > 0.2) {
300
                $result->color = "#C000C0";
301
            } else {
302
                $result->color = "#000000";
303
304
            }

305
306
        }

307
        if ($this->validated) {
308
309
            $this->ads = [];
            $this->maps = false;
310
        }
311
312

        if (count($this->results) <= 0) {
313
314
315
316
317
318
            if (strlen($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');
            }
319
        }
320

321
        if ($this->canCache() && isset($this->next) && count($this->next) > 0 && count($this->results) > 0) {
322
            $page = $this->page + 1;
323
            $this->next = [
324
                'page' => $page,
325
                'startForwards' => $this->results[count($this->results) - 1]->number,
326
                'engines' => $this->next,
327
328
            ];
            Cache::put(md5(serialize($this->next)), serialize($this->next), 60);
329
330
        } else {
            $this->next = [];
331
332
        }

333
    }
334

335
336
    private function addLangCodes($results)
    {
337
338
339
340
341
        # Wenn es keine Ergebnisse gibt, brauchen wir uns gar nicht erst zu bemühen
        if (sizeof($results) === 0) {
            return $results;
        }

342
343
344
345
346
        # Bei der Spracheinstellung "all" wird nicht gefiltert
        if ($this->getLang() === "all") {
            return $results;
        } else {
            # Ansonsten müssen wir jedem Result einen Sprachcode hinzufügen
347
            $id = 0;
348
349
350
351
352
            $langStrings = [];
            foreach ($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
353
                $langStrings["result_" . $id] = utf8_encode($result->getLangString());
354
355
356
357
358
359
360
361

                $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";
362
363
            $lang = exec("echo '$filename' | $langDetectorPath");
            $lang = json_decode($lang, true);
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380

            # 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
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
    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;
            }
        }
    }

401
402
    public function parseAdgoal($results)
    {
403
        $publicKey = getenv('adgoal_public');
404
        $privateKey = getenv('adgoal_private');
405
        if ($publicKey === false) {
406
407
408
            return $results;
        }
        $tldList = "";
409
410
        try {
            foreach ($results as $result) {
411
                $link = $result->anzeigeLink;
412
                if (strpos($link, "http") !== 0) {
413
414
415
416
417
418
419
420
421
422
                    $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");

423
            # Query
424
425
            $query = urlencode($this->q);

426
            $link = "https://api.smartredirect.de/api_v2/CheckForAffiliateUniversalsearchMetager.php?p=" . $publicKey . "&k=" . $hash . "&tld=" . $tldList . "&q=" . $query;
427
428
429
            $answer = json_decode(file_get_contents($link));

            # Nun müssen wir nur noch die Links für die Advertiser ändern:
430
            foreach ($answer as $el) {
431
                $hoster = $el[0];
432
                $hash = $el[1];
433

434
                foreach ($results as $result) {
Dominik Hebeler's avatar
Dominik Hebeler committed
435
                    if ($hoster === $result->tld && !$result->partnershop) {
436
437
                        # Hier ist ein Advertiser:
                        # Das Logo hinzufügen:
438
                        if ($result->image !== "") {
439
                            $result->logo = "https://img.smartredirect.de/logos_v2/60x30/" . $hash . ".gif";
440
                        } else {
441
                            $result->image = "https://img.smartredirect.de/logos_v2/120x60/" . $hash . ".gif";
442
443
                        }

444
445
446
                        # Den Link hinzufügen:
                        $publicKey = $publicKey;
                        $targetUrl = $result->anzeigeLink;
447
                        if (strpos($targetUrl, "http") !== 0) {
448
                            $targetUrl = "http://" . $targetUrl;
449
450
                        }

451
452
453
                        $gateHash = md5($targetUrl . $privateKey);
                        $newLink = "https://api.smartredirect.de/api_v2/ClickGate.php?p=" . $publicKey . "&k=" . $gateHash . "&url=" . urlencode($targetUrl) . "&q=" . $query;
                        $result->link = $newLink;
454
455
456
457
                        $result->partnershop = true;
                    }
                }
            }
458
        } catch (\ErrorException $e) {
459
460
461
462
463
            return $results;
        }

        return $results;
    }
464

465
466
    public function humanVerification($results)
    {
Dominik Hebeler's avatar
Dominik Hebeler committed
467
        # Let's check if we need to implement a redirect for human verification
468
469
        if ($this->verificationCount > 10) {
            foreach ($results as $result) {
Dominik Hebeler's avatar
Dominik Hebeler committed
470
471
472
                $link = $result->link;
                $day = Carbon::now()->day;
                $pw = md5($this->verificationId . $day . $link . env("PROXY_PASSWORD"));
473
474
475
                $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
476
                $result->link = $url;
477
                $result->proxyLink = $proxyUrl;
Dominik Hebeler's avatar
Dominik Hebeler committed
478
479
            }
            return $results;
480
        } else {
Dominik Hebeler's avatar
Dominik Hebeler committed
481
482
483
484
            return $results;
        }
    }

Phil Höfer's avatar
Phil Höfer committed
485
486
487
488
489
490
    public function authorize($key)
    {
        $postdata = http_build_query(array(
            'dummy' => rand(),
        ));
        $opts = array('http' => array(
491
492
            'method' => 'POST',
            'header' => 'Content-type: application/x-www-form-urlencoded',
Phil Höfer's avatar
Phil Höfer committed
493
494
495
496
497
498
499
            'content' => $postdata,
        ),
        );

        $context = stream_context_create($opts);

        try {
500
            $link = "https://key.metager3.de/" . urlencode($key) . "/request-permission/api-access";
Phil Höfer's avatar
Phil Höfer committed
501
502
503
504
505
506
507
508
509
510
511
512
            $result = json_decode(file_get_contents($link, false, $context));
            if ($result->{'api-access'} == true) {
                return true;
            } else {
                return false;
            }

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

513
514
    public function createQuicktips()
    {
515
516
517
518
519
        # 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
520
521
522
523
    /*
     * Die Erstellung der Suchmaschinen bis die Ergebnisse da sind mit Unterfunktionen
     */

524
525
    public function createSearchEngines(Request $request)
    {
526
        # Wenn es kein Suchwort gibt
Dominik Hebeler's avatar
Dominik Hebeler committed
527
        if (!$request->filled("eingabe") || $this->q === "") {
528
            return;
529
        }
530

531
532
        $xml = simplexml_load_file($this->sumaFile);
        $sumas = $xml->xpath("suma");
533
        $enabledSearchengines = [];
534
535
        $overtureEnabled = false;
        $sumaCount = 0;
536

537
        /*
538
539
         * Erstellt eine Liste mit Foki, die verfügbar sind
         */
540
541
542
543
544
545
546
547
548
549
550
        $this->availableFoki = [];
        foreach ($sumas as $suma) {
            $foki = explode(",", trim($suma["type"]));
            foreach ($foki as $fokus) {
                if (!empty($fokus)) {
                    $this->availableFoki[$fokus] = "available";
                }

            }
        }

551
        $isCustomSearch = $this->startsWith($this->fokus, 'focus_');
552

553
        # Im Falle einer Custom-Suche ohne mindestens einer selektierter Suchmaschine wird eine Web-Suche durchgeführt
554
        if ($isCustomSearch && !$this->atLeastOneSearchengineSelected($request)) {
555
556
557
            $isCustomSearch = false;
            $this->fokus = 'web';
        }
558

Karl's avatar
Karl committed
559
        /* Erstellt die Liste der eingestellten Sumas
560
561
562
         * 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
563
564
565
566
567
568
         * 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
         */
569
        foreach ($sumas as $suma) {
570
            if (($this->sumaIsSelected($suma, $request, $isCustomSearch)
571
                || (!$this->isBildersuche()
572
                    && $this->sumaIsAdsuche($suma, $overtureEnabled)))
573
574
575
                && (!$this->sumaIsDisabled($suma))) {
                if ($this->sumaIsOverture($suma)) {
                    $overtureEnabled = true;
576
                }
577
578
                if ($this->sumaIsNotAdsuche($suma)) {
                    $sumaCount += 1;
579
                }
580
                $enabledSearchengines[] = $suma;
581
582
583
584
            }
        }

        # Sonderregelung für alle Suchmaschinen, die zu den Minisuchern gehören. Diese können alle gemeinsam über einen Link abgefragt werden
585
        $subcollections = [];
586
587

        $tmp = [];
588
589
590
591
        // Es gibt den Schalter "minism=on" Dieser soll bewirken, dass alle Minisucher angeschaltet werden.
        // Wenn also "minism=on" ist, dann durchsuchen wir statt den tatsächlich angeschalteten Suchmaschinen,
        // alle Suchmaschinen nach "minismCollection"
        if ($request->input("minism", "off") === "on") {
Dominik Hebeler's avatar
Dominik Hebeler committed
592
593
594
            // Wir laden alle Minisucher
            foreach ($sumas as $engine) {
                if (isset($engine["minismCollection"])) {
595
                    $subcollections[] = $engine["minismCollection"]->__toString();
Dominik Hebeler's avatar
Dominik Hebeler committed
596
597
                }
            }
Dominik Hebeler's avatar
Dominik Hebeler committed
598
599
600
601
602
603
            # Nur noch alle eventuell angeschalteten Minisucher deaktivieren
            foreach ($enabledSearchengines as $index => $engine) {
                if (!isset($engine["minismCollection"])) {
                    $tmp[] = $engine;
                }
            }
Dominik Hebeler's avatar
Dominik Hebeler committed
604
605
606
607
        } else {
            // Wir schalten eine Teilmenge, oder aber gar keine an
            foreach ($enabledSearchengines as $engine) {
                if (isset($engine['minismCollection'])) {
608
                    $subcollections[] = $engine['minismCollection']->__toString();
Dominik Hebeler's avatar
Dominik Hebeler committed
609
610
                } else {
                    $tmp[] = $engine;
Dominik Hebeler's avatar
Dominik Hebeler committed
611
                }
612
            }
613
614
        }
        $enabledSearchengines = $tmp;
615
616
        if (sizeof($subcollections) > 0) {
            $enabledSearchengines[] = $this->loadMiniSucher($xml, $subcollections);
617
        }
618
        if ($sumaCount <= 0) {
619
            $this->errors[] = trans('metaGer.settings.noneSelected');
620
        }
621
        $engines = [];
622
        # Wenn eine Sitesearch durchgeführt werden soll, überprüfen wir ob überhaupt eine der Suchmaschinen eine Sitesearch unterstützt
623
        $siteSearchFailed = $this->checkCanNotSitesearch($enabledSearchengines);
624
625

        $typeslist = [];
626
        $counter = 0;
627

628
629
        if ($this->requestIsCached($request)) {
            $engines = $this->getCachedEngines($request);
Dominik Hebeler's avatar
Dominik Hebeler committed
630
            # We need to edit some Options of the Cached Search Engines
Phil Höfer's avatar
Phil Höfer committed
631
            foreach ($engines as $engine) {
Dominik Hebeler's avatar
Dominik Hebeler committed
632
633
                $engine->setResultHash($this->getHashCode());
            }
634
        } else {
635
            $engines = $this->actuallyCreateSearchEngines($enabledSearchengines, $siteSearchFailed);
636
        }
637

638
        # Wir starten alle Suchen
639
640
        foreach ($engines as $engine) {
            $engine->startSearch($this);
641
        }
642

643
644
645
646
647
        /* 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.
648
649
         */

650
        $enginesToLoad = [];
651
        $canBreak = false;
652
653
654
655
656
        foreach ($engines as $engine) {
            if ($engine->cached) {
                if ($overtureEnabled && ($engine->name === "overture" || $engine->name === "overtureAds")) {
                    $canBreak = true;
                }
657
658
            } else {
                $enginesToLoad[$engine->name] = false;
659
660
661
662
663
664
665
666
            }
        }

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

        $this->retrieveResults($engines);
    }

667
668
    # Spezielle Suchen und Sumas

669
    public function sumaIsSelected($suma, $request, $custom)
670
    {
671
        if ($custom) {
672
            if ($request->filled("engine_" . strtolower($suma["name"]))) {
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
                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
697
698
699
            if (!isset($engine["package"])) {
                die(var_dump($engine));
            }
700
701
702
703
704
            # 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")) {
705
                Log::error("Konnte " . $engine["name"] . " nicht abfragen, da kein Parser existiert");
706
                $this->errors[] = trans('metaGer.engines.noParser', ['engine' => $engine["name"]]);
707
708
709
710
711
712
713
714
                continue;
            }

            # Es wird versucht die Suchengine zu erstellen
            $time = microtime();
            try {
                $tmp = new $path($engine, $this);
            } catch (\ErrorException $e) {
715
                Log::error("Konnte " . $engine["name"] . " nicht abfragen. " . var_dump($e));
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
                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()
    {
734
        return $this->fokus === "bilder";
735
736
737
738
    }

    public function sumaIsAdsuche($suma, $overtureEnabled)
    {
739
        $sumaName = $suma["name"]->__toString();
740
        return
741
742
            $sumaName === "qualigo"
            || $sumaName === "similar_product_ads"
Karl Hasselbring's avatar
Karl Hasselbring committed
743
            || (!$overtureEnabled && $sumaName === "overtureAds");
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
    }

    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
771
        $request->filled('next')
772
773
774
775
776
777
        && Cache::has($request->input('next'))
        && unserialize(Cache::get($request->input('next')))['page'] > 1;
    }

    public function getCachedEngines($request)
    {
778
        $next = unserialize(Cache::get($request->input('next')));
779
        $this->page = $next['page'];
780
        $engines = $next['engines'];
781
782
783
784
785
786
787
788
789
790
791
        if (isset($next['startForwards'])) {
            $this->startForwards = $next['startForwards'];
        }
        if (isset($next['startBackwards'])) {
            $this->startBackwards = $next['startBackwards'];
        }
        return $engines;
    }

    public function loadMiniSucher($xml, $subcollections)
    {
792
        $minisucherEngine = $xml->xpath('suma[@name="minism"]')[0];
793
        $minisucherEngine["subcollections"] = implode(", ", $subcollections);
794
        $subcollectionsString = urlencode("(" . implode(" OR ", $subcollections) . ")");
795
        $minisucherEngine["formData"] = str_replace("<<SUBCOLLECTIONS>>", $subcollectionsString, $minisucherEngine["formData"]);
796
797
798
799
        $minisucherEngine["formData"] = str_replace("<<COUNT>>", sizeof($subcollections) * 10, $minisucherEngine["formData"]);
        return $minisucherEngine;
    }

800
    # Passt den Suchfokus an, falls für einen Fokus genau alle vorhandenen Sumas eingeschaltet sind
801
802
    public function adjustFocus($sumas, $enabledSearchengines)
    {
803
804
        # Findet für alle Foki die enthaltenen Sumas
        $foki = []; # [fokus][suma] => [suma]
805
        foreach ($sumas as $suma) {
806
            if ((!$this->sumaIsDisabled($suma)) && (!isset($suma['userSelectable']) || $suma['userSelectable']->__toString() === "1")) {
807
                if (isset($suma['type'])) {
808
809
810
                    # Wenn foki für diese Suchmaschine angegeben sind
                    $focuses = explode(",", $suma['type']->__toString());
                    foreach ($focuses as $foc) {
811
812
813
814
815
                        if (isset($suma['minismCollection'])) {
                            $foki[$foc][] = "minism";
                        } else {
                            $foki[$foc][] = $suma['name']->__toString();
                        }
816
                    }
817
                } else {
818
                    # Wenn keine foki für diese Suchmaschine angegeben sind
819
820
821
822
823
                    if (isset($suma['minismCollection'])) {
                        $foki["andere"][] = "minism";
                    } else {
                        $foki["andere"][] = $suma['name']->__toString();
                    }
824
825
826
827
                }
            }
        }

828
        # Findet die Namen der aktuell eingeschalteten Sumas
829
        $realEngNames = [];
830
        foreach ($enabledSearchengines as $realEng) {
831
            $nam = $realEng["name"]->__toString();
Karl Hasselbring's avatar
Karl Hasselbring committed
832
            if ($nam !== "qualigo" && $nam !== "overtureAds") {
833
834
835
                $realEngNames[] = $nam;
            }
        }
836

837
        # 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.
838
        foreach ($foki as $fok => $engines) {
839
            $isFokus = true;
840
            $fokiEngNames = [];
841
            foreach ($engines as $eng) {
842
843
                $fokiEngNames[] = $eng;
            }
844
            # Jede eingeschaltete Engine ist für diesen Fokus geeignet
845
            foreach ($fokiEngNames as $fen) {
846
847
                # Bei Bildersuchen ist uns egal, ob alle Suchmaschinen aus dem Suchfokus eingeschaltet sind, da wir sie eh als Bildersuche anzeigen müssen
                if (!in_array($fen, $realEngNames) && $fok !== "bilder") {
848
849
850
                    $isFokus = false;
                }
            }
851
            # Jede im Fokus erwartete Engine ist auch eingeschaltet
852
853
            foreach ($realEngNames as $ren) {
                if (!in_array($ren, $fokiEngNames)) {
854
855
856
                    $isFokus = false;
                }
            }
857
            # Wenn die Listen identisch sind, setze den Fokus um
858
            if ($isFokus) {
859
860
861
                $this->fokus = $fok;
            }
        }
862
    }
863

864
865
866
867
868
869
870
    public function checkCanNotSitesearch($enabledSearchengines)
    {
        if (strlen($this->site) > 0) {
            $enginesWithSite = 0;
            foreach ($enabledSearchengines as $engine) {
                if (isset($engine['hasSiteSearch']) && $engine['hasSiteSearch']->__toString() === "1") {
                    $enginesWithSite++;
871
                }
872
873
            }
            if ($enginesWithSite === 0) {
874
                $this->errors[] = trans('metaGer.sitesearch.failed', ['site' => $this->site, 'searchLink' => $this->generateSearchLink("web", false)]);
875
876
                return true;
            } else {
877
                $this->warnings[] = trans('metaGer.sitesearch.success', ['site' => $this->site]);
878
                return false;
879
880
            }
        }
881
        return false;
882
883
884
885
    }

    public function waitForResults($enginesToLoad, $overtureEnabled, $canBreak)
    {
886

Phil Höfer's avatar
Phil Höfer committed
887
        $timeStart = microtime(true);
888
        $results = null;
889
        while (true) {
890
891
892
            $results = Redis::hgetall('search.' . $this->getHashCode());

            $ready = true;
Phil Höfer's avatar
Phil Höfer committed
893
            // When every
894
            $connected = true;
Phil Höfer's avatar
Phil Höfer committed
895
896
            foreach ($results as $key => $value) {
                if ($value === "waiting" || $value === "connected") {
897
                    $ready = false;
898
                }
Phil Höfer's avatar
Phil Höfer committed
899
                if ($value === "waiting") {
900
                    $connected = false;
901
                }
902
            }
903

904
905
            // If $ready is false at this point, we're waiting for more searchengines
            // But we have to check for the timeout, too
Phil Höfer's avatar
Phil Höfer committed
906
907
908
909
            if (!$connected) {
                $timeStart = microtime(true);
            }

910
911
912
913
            $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"]) && $results["overtureAds"] === "waiting"));
Phil Höfer's avatar
Phil Höfer committed
914
915
916
            if ($time > $this->time && $canTimeout) {
                $ready = true;
            }