Searchengine.php 13.2 KB
Newer Older
1 2
<?php

3
namespace App\Models;
4

5
use App\Jobs\Searcher;
6
use App\MetaGer;
7 8
use Cache;
use Illuminate\Foundation\Bus\DispatchesJobs;
9
use Illuminate\Support\Facades\Redis;
10

11
abstract class Searchengine
12
{
13
    use DispatchesJobs;
14

Karl's avatar
Karl committed
15 16 17
    public $ch; # Curl Handle zum erhalten der Ergebnisse
    public $getString = ""; # Der String für die Get-Anfrage
    public $engine; # Die ursprüngliche Engine XML
Dominik Hebeler's avatar
Dominik Hebeler committed
18 19 20
    public $enabled = true; # true, wenn die Suchmaschine nicht explizit disabled ist
    public $results = []; # Die geladenen Ergebnisse
    public $ads = []; # Die geladenen Werbungen
Dominik Hebeler's avatar
Dominik Hebeler committed
21
    public $products = []; # Die geladenen Produkte
Dominik Hebeler's avatar
Dominik Hebeler committed
22 23
    public $loaded = false; # wahr, sobald die Ergebnisse geladen wurden
    public $cached = false;
Karl's avatar
Karl committed
24 25 26 27 28 29 30 31 32 33

    public $ip; # Die IP aus der metager
    public $uses; # Die Anzahl der Nutzungen dieser Suchmaschine
    public $homepage; # Die Homepage dieser Suchmaschine
    public $name; # Der Name dieser Suchmaschine
    public $disabled; # Ob diese Suchmaschine ausgeschaltet ist
    public $useragent; # Der HTTP Useragent
    public $startTime; # Die Zeit der Erstellung dieser Suchmaschine
    public $hash; # Der Hash-Wert dieser Suchmaschine

34 35 36
    private $user; # Username für HTTP-Auth (falls angegeben)
    private $password; # Passwort für HTTP-Auth (falls angegeben)

37 38
    private $headers; # Headers to add

Karl's avatar
Karl committed
39
    public $fp; # Wird für Artefakte benötigt
Dominik Hebeler's avatar
Dominik Hebeler committed
40 41 42
    public $socketNumber = null; # Wird für Artefakte benötigt
    public $counter = 0; # Wird eventuell für Artefakte benötigt
    public $write_time = 0; # Wird eventuell für Artefakte benötigt
Karl's avatar
Karl committed
43
    public $connection_time = 0; # Wird eventuell für Artefakte benötigt
44 45 46

    public function __construct(\SimpleXMLElement $engine, MetaGer $metager)
    {
Karl's avatar
Karl committed
47
        # Versucht möglichst viele attribute aus dem engine XML zu laden
48 49 50
        foreach ($engine->attributes() as $key => $value) {
            $this->$key = $value->__toString();
        }
Karl's avatar
Karl committed
51 52

        # Standardhomepage metager.de
53 54 55 56
        if (!isset($this->homepage)) {
            $this->homepage = "https://metager.de";
        }

Karl's avatar
Karl committed
57
        # Speichert die XML der Engine
58
        $this->engine = $engine->asXML();
59

Karl's avatar
Karl committed
60
        # Cache Standarddauer 60
61 62 63 64
        if (!isset($this->cacheDuration)) {
            $this->cacheDuration = 60;
        }

Karl's avatar
Karl committed
65 66
        $this->enabled = true;

67 68 69 70 71 72 73 74 75
        # Eine Suchmaschine kann automatisch temporär deaktiviert werden, wenn es Verbindungsprobleme gab:
        if (isset($this->disabled) && strtotime($this->disabled) <= time()) {
            # In diesem Fall ist der Timeout der Suchmaschine abgelaufen.
            $this->enable($metager->getSumaFile(), "Die Suchmaschine " . $this->name . " wurde wieder eingeschaltet.");
        } elseif (isset($this->disabled) && strtotime($this->disabled) > time()) {
            $this->enabled = false;
            return;
        }

Dominik Hebeler's avatar
Dominik Hebeler committed
76
        $this->useragent = $metager->getUserAgent();
Dominik Hebeler's avatar
Dominik Hebeler committed
77
        $this->ip = $metager->getIp();
78 79
        $this->startTime = microtime();

Karl's avatar
Karl committed
80
        # Suchstring generieren
81 82 83 84 85 86 87 88 89 90 91
        $q = "";
        if (isset($this->hasSiteSearch) && $this->hasSiteSearch === "1") {
            if (strlen($metager->getSite()) === 0) {
                $q = $metager->getQ();
            } else {
                $q = $metager->getQ() . " site:" . $metager->getSite();
            }

        } else {
            $q = $metager->getQ();
        }
Dominik Hebeler's avatar
Dominik Hebeler committed
92 93
        $this->getString = $this->generateGetString($q, $metager->getUrl(), $metager->getLanguage(), $metager->getCategory());
        $this->hash = md5($this->host . $this->getString . $this->port . $this->name);
94
        $this->resultHash = $metager->getHashCode();
Dominik Hebeler's avatar
Dominik Hebeler committed
95
        $this->canCache = $metager->canCache();
Phil Höfer's avatar
Phil Höfer committed
96
        if (!isset($this->additionalHeaders)) {$this->additionalHeaders = "";}
97 98 99 100
    }

    abstract public function loadResults($result);

101
    # Standardimplementierung der getNext Funktion, damit diese immer verwendet werden kann
102 103
    public function getNext(MetaGer $metager, $result)
    {
104 105 106

    }

Karl's avatar
Karl committed
107
    # Prüft, ob die Suche bereits gecached ist, ansonsted wird sie als Job dispatched
108 109
    public function startSearch(\App\MetaGer $metager)
    {
Dominik Hebeler's avatar
Dominik Hebeler committed
110
        if ($this->canCache && Cache::has($this->hash)) {
111
            $this->cached = true;
112
            $this->retrieveResults($metager);
113
        } else {
114 115 116 117 118 119 120 121
            // We will push the confirmation of the submission to the Result Hash
            Redis::hset('search.' . $this->resultHash, $this->name, "waiting");
            // We need to submit a action that one of our workers can understand
            // The missions are submitted to a redis queue in the following string format
            // <ResultHash>;<URL to fetch>
            // With <ResultHash> being the Hash Value where the fetcher will store the result.
            // and <URL to fetch> being the full URL to the searchengine
            $url = "";
Dominik Hebeler's avatar
Dominik Hebeler committed
122
            if ($this->port === "443") {
123
                $url = "https://";
Dominik Hebeler's avatar
Dominik Hebeler committed
124
            } else {
125 126
                $url = "http://";
            }
Dominik Hebeler's avatar
Dominik Hebeler committed
127
            $url .= $this->host;
Dominik Hebeler's avatar
Dominik Hebeler committed
128
            if ($this->port !== 80 && $this->port !== 443) {
Dominik Hebeler's avatar
Dominik Hebeler committed
129 130 131
                $url .= ":" . $this->port;
            }
            $url .= $this->getString;
Dominik Hebeler's avatar
Dominik Hebeler committed
132 133
            $url = base64_encode($url);
            $mission = $this->resultHash . ";" . $url . ";" . $metager->getTime();
134 135 136 137 138
            // Submit this mission to the corresponding Redis Queue
            // Since each Searcher is dedicated to one specific search engine
            // each Searcher has it's own queue lying under the redis key <name>.queue
            Redis::rpush($this->name . ".queue", $mission);

139
            /**
Dominik Hebeler's avatar
Dominik Hebeler committed
140 141 142 143 144 145 146 147 148 149 150
             * We have Searcher processes running for MetaGer
             * Each Searcher is dedicated to one specific Searchengine and fetches it's results.
             * We can have multiple Searchers for each engine, if needed.
             * At this point we need to decide, whether we need to start a new Searcher process or
             * if we have enough of them running.
             * The information for that is provided through the redis system. Each running searcher
             * gives information how long it has waited to be given the last fetcher job.
             * The longer this time value is, the less frequent the search engine is used and the less
             * searcher of that type we need.
             * But if it's too low, i.e. 100ms, then the searcher is near to it's full workload and needs assistence.
             **/
151 152 153 154 155
            $needSearcher = false;
            $searcherData = Redis::hgetall($this->name . ".stats");

            // We now have an array of statistical data from the searchers
            // Each searcher has one entry in it.
Dominik Hebeler's avatar
Dominik Hebeler committed
156
            // So if it's empty, then we have currently no searcher running and
157
            // of course need to spawn a new one.
Dominik Hebeler's avatar
Dominik Hebeler committed
158
            if (sizeof($searcherData) === 0) {
159
                $needSearcher = true;
Dominik Hebeler's avatar
Dominik Hebeler committed
160
            } else {
161 162 163 164 165
                // There we go:
                // There's at least one Fetcher running for this search engine.
                // Now we have to check if the current count is enough to fetch all the
                // searches or if it needs help.
                // Let's hardcode a minimum of 100ms between every search job.
Dominik Hebeler's avatar
Dominik Hebeler committed
166 167
                // First calculate the median of all Times
                $median = 0;
Dominik Hebeler's avatar
Dominik Hebeler committed
168
                foreach ($searcherData as $pid => $data) {
Dominik Hebeler's avatar
Dominik Hebeler committed
169 170 171 172
                    $data = explode(";", $data);
                    $median += floatval($data[1]);
                }
                $median /= sizeof($searcherData);
Dominik Hebeler's avatar
Dominik Hebeler committed
173
                if ($median < .1) {
Dominik Hebeler's avatar
Dominik Hebeler committed
174 175
                    $needSearcher = true;
                }
176
            }
Dominik Hebeler's avatar
Dominik Hebeler committed
177
            if ($needSearcher && Redis::get($this->name) !== "locked") {
Dominik Hebeler's avatar
Dominik Hebeler committed
178
                Redis::set($this->name, "locked");
179
                $this->dispatch(new Searcher($this->name, $this->user, $this->password, $this->headers));
180
            }
181 182 183
        }
    }

Karl's avatar
Karl committed
184
    # Ruft die Ranking-Funktion aller Ergebnisse auf.
185
    public function rank($eingabe)
186 187
    {
        foreach ($this->results as $result) {
188
            $result->rank($eingabe);
189 190 191
        }
    }

Karl's avatar
Karl committed
192
    # Magic ???
193 194 195 196 197 198 199 200 201
    private function setStatistic($key, $val)
    {

        $oldVal = floatval(Redis::hget($this->name, $key)) * $this->uses;
        $newVal = ($oldVal + max($val, 0)) / $this->uses;
        Redis::hset($this->name, $key, $newVal);
        $this->$key = $newVal;
    }

Karl's avatar
Karl committed
202
    # Entfernt wenn gesetzt das disabled="1" für diese Suchmaschine aus der sumas.xml
203 204 205 206 207
    public function enable($sumaFile, $message)
    {
        $xml = simplexml_load_file($sumaFile);
        unset($xml->xpath("//sumas/suma[@name='" . $this->name . "']")['0']['disabled']);
        $xml->saveXML($sumaFile);
Karl's avatar
Karl committed
208
        $this->enabled = true;
209 210
    }

Dominik Hebeler's avatar
Dominik Hebeler committed
211 212
    public function setResultHash($hash)
    {
213 214 215
        $this->resultHash = $hash;
    }

216 217 218 219 220
    public function closeFp()
    {
        fclose($this->fp);
    }

Karl's avatar
Karl committed
221
    # Öffnet einen neuen Socket für diese Engine
222 223 224 225 226 227 228 229 230 231
    public function getSocket()
    {
        $number = Redis::hget('search.' . $this->hash, $this->name);
        if ($number === null) {
            return null;
        } else {
            return pfsockopen($this->getHost() . ":" . $this->port . "/$number", $this->port, $errstr, $errno, 1);
        }
    }

Karl's avatar
Karl committed
232
    # Fragt die Ergebnisse von Redis ab und lädt Sie
233
    public function retrieveResults(MetaGer $metager)
234 235 236
    {
        if ($this->loaded) {
            return true;
237 238
        }

239
        $body = "";
Dominik Hebeler's avatar
Dominik Hebeler committed
240
        if ($this->canCache && $this->cacheDuration > 0 && Cache::has($this->hash)) {
241 242 243
            $body = Cache::get($this->hash);
        } elseif (Redis::hexists('search.' . $this->resultHash, $this->name)) {
            $body = Redis::hget('search.' . $this->resultHash, $this->name);
244
            Redis::hdel('search.' . $this->resultHash, $this->name);
Dominik Hebeler's avatar
Dominik Hebeler committed
245
            if ($this->canCache && $this->cacheDuration > 0) {
246 247 248
                Cache::put($this->hash, $body, $this->cacheDuration);
            }
        }
249
        if ($body !== "" && $body !== "connected" && $body !== "waiting") {
250
            $this->loadResults($body);
251
            $this->getNext($metager, $body);
252 253 254 255 256 257 258 259 260 261 262 263
            $this->loaded = true;
            return true;
        } else {
            return false;
        }
    }

    public function shutdown()
    {
        Redis::del($this->host . "." . $this->socketNumber);
    }

Karl's avatar
Karl committed
264
    # Erstellt den für die Get-Anfrage genutzten Host-Link
265 266 267 268 269 270 271 272 273 274 275 276
    protected function getHost()
    {
        $return = "";
        if ($this->port === "443") {
            $return .= "tls://";
        } else {
            $return .= "tcp://";
        }
        $return .= $this->host;
        return $return;
    }

Karl's avatar
Karl committed
277
    # Erstellt den für die Get-Anfrage genutzten String
278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294
    private function generateGetString($query, $url, $language, $category)
    {
        $getString = "";

        # Skript:
        if (strlen($this->skript) > 0) {
            $getString .= $this->skript;
        } else {
            $getString .= "/";
        }

        # FormData:
        if (strlen($this->formData) > 0) {
            $getString .= "?" . $this->formData;
        }

        # Wir müssen noch einige Platzhalter in dem GET-String ersetzen:
Karl's avatar
Karl committed
295
        # Useragent
296 297 298 299
        if (strpos($getString, "<<USERAGENT>>")) {
            $getString = str_replace("<<USERAGENT>>", $this->urlEncode($this->useragent), $getString);
        }

Karl's avatar
Karl committed
300
        # Query
301 302 303 304
        if (strpos($getString, "<<QUERY>>")) {
            $getString = str_replace("<<QUERY>>", $this->urlEncode($query), $getString);
        }

Karl's avatar
Karl committed
305
        # IP
306 307 308 309
        if (strpos($getString, "<<IP>>")) {
            $getString = str_replace("<<IP>>", $this->urlEncode($this->ip), $getString);
        }

Karl's avatar
Karl committed
310
        # Language
311 312 313 314
        if (strpos($getString, "<<LANGUAGE>>")) {
            $getString = str_replace("<<LANGUAGE>>", $this->urlEncode($language), $getString);
        }

Karl's avatar
Karl committed
315
        # Category
316 317 318 319
        if (strpos($getString, "<<CATEGORY>>")) {
            $getString = str_replace("<<CATEGORY>>", $this->urlEncode($category), $getString);
        }

Karl's avatar
Karl committed
320
        # Affildata
321 322 323 324 325 326
        if (strpos($getString, "<<AFFILDATA>>")) {
            $getString = str_replace("<<AFFILDATA>>", $this->getOvertureAffilData($url), $getString);
        }
        return $getString;
    }

Karl's avatar
Karl committed
327
    # Wandelt einen String nach aktuell gesetztem inputEncoding dieser Searchengine in URL-Format um
328 329 330 331 332 333 334 335 336
    protected function urlEncode($string)
    {
        if (isset($this->inputEncoding)) {
            return urlencode(mb_convert_encoding($string, $this->inputEncoding));
        } else {
            return urlencode($string);
        }
    }

Karl's avatar
Karl committed
337
    # Liefert Sonderdaten für Yahoo
338 339 340 341 342 343 344 345 346 347 348 349 350 351 352
    private function getOvertureAffilData($url)
    {
        $affil_data = 'ip=' . $this->ip;
        $affil_data .= '&ua=' . $this->useragent;
        $affilDataValue = $this->urlEncode($affil_data);
        # Wir benötigen die ServeUrl:
        $serveUrl = $this->urlEncode($url);

        return "&affilData=" . $affilDataValue . "&serveUrl=" . $serveUrl;
    }

    public function isEnabled()
    {
        return $this->enabled;
    }
Karl's avatar
Karl committed
353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374

    # Artefaktmethoden

    public function getCurlInfo()
    {
        return curl_getinfo($this->ch);
    }

    public function getCurlErrors()
    {
        return curl_errno($this->ch);
    }

    public function addCurlHandle($mh)
    {
        curl_multi_add_handle($mh, $this->ch);
    }

    public function removeCurlHandle($mh)
    {
        curl_multi_remove_handle($mh, $this->ch);
    }
375
}