Codetest

PHP Detected Guest 1 Views Size: 30.10 KB Posted on: Jul 13, 26 @ 2:54 AM

1scraper.php:
2
3<?php
4require_once 'utils.php';
5require_once __DIR__ . '/app/db.php';
6
7error_reporting(E_ALL);
8ini_set('display_errors', '1');
9
10function e($v){
11    return htmlspecialchars((string)$v, ENT_QUOTES, 'UTF-8');
12}
13
14/* ✅ LOG */
15function logAction($accion,$detalle){
16    try{
17        gm_db()->prepare("INSERT INTO scraping_logs (accion,detalle) VALUES(?,?)")
18              ->execute([$accion,$detalle]);
19    }catch(Exception $e){}
20}
21
22/* ✅ PARSER */
23function parseHistorico($html){
24
25    libxml_use_internal_errors(true);
26
27    $data = [
28        'evento'=>'',
29        'pista'=>'',
30        'distancia'=>'',
31        'hora'=>'',
32        'carrera'=>'',
33        'carrera_numero'=> null, // ✅ AÑADE ESTO
34        'estado'=>'Finalizada',
35        'runners'=>[]
36    ];
37
38    // ✅ PISTA
39    if(preg_match('/\b(Manawatu|Doncaster|Hove|Sunderland|Yarmouth)\b/i',$html,$m)){
40        $data['pista'] = $m[1];
41    }
42    
43    // ✅ EVENTO
44    if(preg_match('/\n([A-Za-z\s]+Heat)/',$html,$m)){
45        $data['evento'] = trim($m[1]);
46    }
47    
48    // ✅ DISTANCIA
49    if(preg_match('/(\d{2,4})m/',$html,$m)){
50        $data['distancia'] = $m[1];
51    }
52    
53    // ✅ HORA
54    if(preg_match('/Empezada\s*(\d{2}:\d{2}:\d{2})/',$html,$m)){
55        $data['hora'] = $m[1];
56    }
57    
58    // ✅ ESTADO
59    if(stripos($html,'Finalizada') !== false){
60        $data['estado'] = 'Finalizada';
61    }
62    
63    // ✅ CARRERA
64    if(preg_match('/Carrera\s*(\d+)/',$html,$m)){
65        $data['carrera_numero'] = (int)$m[1];
66    }
67
68    $data['runners'] = [];
69    
70    if(preg_match_all('/(\d+)º\s+(\d+)\s+([A-Za-z\s]+)/',$html,$matches,PREG_SET_ORDER)){
71    
72        foreach($matches as $m){
73    
74            $posicion = (int)$m[1];
75            $trap = (int)$m[2];
76            $nombre = trim($m[3]);
77    
78            if($posicion < 1 || $posicion > 8) continue;
79    
80            $data['runners'][] = [
81                'nombre' => $nombre,
82                'posicion' => $posicion,
83                'trap' => $trap
84            ];
85        }
86    }
87
88    // ✅ PARSER GLOBAL (robusto para Bet365 moderno)
89    
90    // 🏟 pista (ej: Doncaster)
91    if(empty($data['pista']) && preg_match('/\b(Doncaster|Manukau|Hove|Sunderland|Yarmouth)\b/i',$html,$m)){
92        $data['pista'] = $m[1];
93    }
94    
95    // 📏 distancia (ej: 275m)
96    if(empty($data['distancia']) && preg_match('/(\d{2,4})\s*m/i',$html,$m)){
97        $data['distancia'] = $m[1];
98    }
99    
100    // 🕒 hora (ej: 08:26:23)
101    if(empty($data['hora']) && preg_match('/(\d{2}:\d{2}:\d{2})/',$html,$m)){
102        $data['hora'] = $m[1];
103    }
104    
105    // 📊 estado
106    if(stripos($html,'finalizada') !== false){
107        $data['estado'] = 'Finalizada';
108    }elseif(stripos($html,'empezada') !== false){
109        $data['estado'] = 'En curso';
110    }
111    
112    // 🏁 número de carrera (DE LAS BURBUJAS)
113    if(empty($data['carrera_numero'])){
114    
115        if(preg_match('/>\s*(\d{1,2})\s*</',$html,$m)){
116            $data['carrera_numero'] = (int)$m[1];
117        }
118    }
119    
120    if(empty($data['pista'])){
121        if(preg_match('/\b([A-Z][a-z]+)\b\s*</',$html,$m)){
122            $data['pista'] = $m[1];
123        }
124    }
125    
126    
127    // ✅ detectar número desde burbujas
128    if(empty($data['carrera_numero']) && !empty($html)){
129        if(preg_match('/>\s*(\d{1,2})\s*</',$html,$m)){
130            $data['carrera_numero'] = (int)$m[1];
131        }
132    }
133
134
135        // ✅ FALLBACK GLOBAL (muy importante)
136    if(empty($data['carrera_numero'])){
137    
138        if(preg_match('/\b(\d{1,2})\b/', $html, $m)){
139            $data['carrera_numero'] = (int)$m[1];
140        }
141    }
142
143    return $data;
144}
145
146/* ✅ GUARDADO */
147function saveRaceDB($data){
148
149    $pdo = gm_db();
150    $pdo->beginTransaction();
151
152    try{
153
154        if(empty($data['runners'])){
155            throw new Exception("No hay runners");
156        }
157
158        $hash = md5(uniqid().serialize($data));
159
160       $stmt = $pdo->prepare("
161            INSERT INTO carreras
162            (pista, distancia, evento, hash, tipo, fuente, estado, fecha_carrera, pais, carrera_numero, codigo_unico)
163            VALUES (?, ?, ?, ?, 'Greyhound','html_manual','finalizada', ?, ?, ?, ?)
164        ");
165
166        
167        // ✅ mapa de pistas
168        $TRACK_MAP = [
169            'Manukau' => ['pais'=>'Nueva Zelanda'],
170            'Healesville' => ['pais'=>'Australia'],
171            'Capalaba' => ['pais'=>'Australia'],
172            'Mount Gambier' => ['pais'=>'Australia'],
173            'Sale' => ['pais'=>'Australia'],
174            'Rockhampton' => ['pais'=>'Australia'],
175            'Richmond' => ['pais'=>'Australia'],
176            'Nowra' => ['pais'=>'Australia'],
177            'Q2 Parklands' => ['pais'=>'Australia'],
178            'Gawler' => ['pais'=>'Australia'],
179            'Temora' => ['pais'=>'Australia'],
180            'Darwin' => ['pais'=>'Australia'],
181            'Valley' => ['pais'=>'UK'],
182            'Doncaster' => ['pais'=>'UK'],
183            'Yarmouth' => ['pais'=>'UK'],
184            'Sunderland' => ['pais'=>'UK'],
185            'Hove' => ['pais'=>'UK'],
186        ];
187        
188        // ✅ obtener país
189        $trackInfo = $TRACK_MAP[$data['pista']] ?? ['pais'=>'Desconocido'];
190        
191        // ✅ generar código único
192        $codigo = generarCodigoCarrera([
193            'pais' => $trackInfo['pais'],
194            'pista' => $data['pista'],
195            'carrera_numero' => $data['carrera_numero'] ?? 0,
196            'evento' => $data['evento'],
197            'fecha_carrera' => $data['hora'] 
198                ? date('Y-m-d').' '.$data['hora']
199                : date('Y-m-d') // ✅ fallback seguro
200        ]);
201
202        // ✅ asegurar unicidad del código
203        $stmtCheck = $pdo->prepare("
204            SELECT COUNT(*) FROM carreras WHERE codigo_unico = ?
205        ");
206        $stmtCheck->execute([$codigo]);
207        
208        if($stmtCheck->fetchColumn() > 0){
209        
210            // 🔥 generar sufijo incremental limpio
211            $suffix = 1;
212        
213            do {
214                $nuevoCodigo = $codigo . str_pad($suffix, 2, '0', STR_PAD_LEFT);
215        
216                $stmtCheck = $pdo->prepare("
217                    SELECT COUNT(*) FROM carreras WHERE codigo_unico = ?
218                ");
219                $stmtCheck->execute([$nuevoCodigo]);
220        
221                $existe = $stmtCheck->fetchColumn();
222                $suffix++;
223        
224            } while($existe > 0);
225        
226            $codigo = $nuevoCodigo;
227        }
228
229        // ✅ generar fecha carrera
230        $fechaCarrera = $data['hora'] 
231            ? date('Y-m-d').' '.$data['hora']
232            : date('Y-m-d H:i:s');
233        
234        // ✅ ejecutar
235        $stmt->execute([
236            $data['pista'],          // 1
237            $data['distancia'],      // 2
238            $data['evento'],         // 3
239            $hash,                  // 4
240            $fechaCarrera,          // 5
241            $trackInfo['pais'],     // 6
242            $data['carrera_numero'] ?? null, // 7
243            $codigo // ✅ NUEVO
244        ]);
245
246
247        $carreraId = $pdo->lastInsertId();
248
249        foreach($data['runners'] as $r){
250
251            $nombre = trim($r['nombre']);
252
253            $stmt=$pdo->prepare("SELECT id_galgo FROM galgos WHERE nombre=?");
254            $stmt->execute([$nombre]);
255            $gid=$stmt->fetchColumn();
256
257            if(!$gid){
258
259                $slug = strtolower(trim(preg_replace('/[^A-Za-z0-9]+/', '-', $nombre)));
260                $slug = trim($slug, '-');
261
262                $stmt=$pdo->prepare("SELECT COUNT(*) FROM galgos WHERE slug=?");
263                $stmt->execute([$slug]);
264
265                if($stmt->fetchColumn()>0){
266                    $slug .= "-".uniqid();
267                }
268
269                $stmt=$pdo->prepare("INSERT INTO galgos (nombre,slug) VALUES (?,?)");
270                $stmt->execute([$nombre,$slug]);
271
272                $gid=$pdo->lastInsertId();
273            }
274
275            $stmt=$pdo->prepare("
276                INSERT INTO participaciones (carrera_id,galgo_id,posicion,trap)
277                VALUES (?,?,?,?)
278            ");
279
280            $stmt->execute([
281                $carreraId,
282                $gid,
283                $r['posicion'],
284                $r['trap']
285            ]);
286        }
287
288        $pdo->commit();
289
290        return "✅ Guardado correctamente";
291
292    }catch(Exception $e){
293
294        $pdo->rollBack();
295
296        echo "<pre style='color:red'>";
297        echo "ERROR:\n".$e->getMessage();
298        echo "</pre>";
299        die();
300    }
301}
302
303/* ✅ CONTROL */
304$data=null;
305$msg=null;
306$html=$_POST['html'] ?? '';
307
308if($_SERVER['REQUEST_METHOD']=='POST'){
309
310    if($_POST['action']=='preview'){
311        $data=parseHistorico($html);
312    }
313
314    if($_POST['action']=='save'){
315        $payload=json_decode(base64_decode($_POST['payload']),true);
316        $msg=saveRaceDB($payload);
317        $data=$payload;
318    }
319    
320}
321
322$logs = gm_db()->query("SELECT * FROM scraping_logs ORDER BY id DESC LIMIT 10")->fetchAll();
323?>
324
325<!doctype html>
326<html>
327<head>
328<meta charset="utf-8">
329<title>GalgoMaster PRO</title>
330
331<style>
332body{
333    background:linear-gradient(135deg,#020617,#0f172a);
334    color:#e2e8f0;
335    font-family:system-ui;
336}
337
338/* nombre galgo */
339.dog-name{
340    display:block;
341    margin-top:6px;
342    font-size:15px;
343    font-weight:600;
344}
345
346/* cajón */
347.dog-trap{
348    margin-top:6px;
349    font-size:13px;
350    font-weight:700;
351    color:#ef4444;
352
353    background:rgba(239,68,68,0.1);
354    padding:4px 8px;
355    border-radius:6px;
356
357    display:inline-block;
358}
359
360/* hover más elegante */
361.box:hover .dog-trap{
362    background:#ef4444;
363    color:white;
364}
365
366.box{
367    background:#020617;
368    padding:20px;
369    border-radius:12px;
370    text-align:center;
371    transition:0.2s;
372    border:1px solid transparent;
373}
374
375.box:hover{
376    transform:translateY(-4px);
377    border:1px solid #22c55e;
378}
379
380/* layout */
381.container{max-width:1100px;margin:auto;}
382
383.card{
384    background:#111827;
385    padding:20px;
386    margin:20px 0;
387    border-radius:16px;
388    box-shadow:0 10px 30px rgba(0,0,0,.5);
389}
390
391/* textarea */
392textarea{
393    width:100%;
394    height:220px;
395    background:#020617;
396    color:#22c55e;
397    border-radius:12px;
398    padding:12px;
399}
400
401/* botones */
402button{
403    background:#22c55e;
404    border:0;
405    padding:12px 18px;
406    border-radius:10px;
407    font-weight:bold;
408    cursor:pointer;
409}
410
411/* header */
412.topbar{
413    display:flex;
414    justify-content:space-between;
415    align-items:center;
416}
417
418/* ✅ GRID */
419.grid{
420    display:grid;
421    grid-template-columns:repeat(4,1fr);
422    gap:15px;
423}
424
425/* ✅ BOX */
426.box{
427    background:#020617;
428    padding:20px;
429    border-radius:12px;
430    text-align:center;
431    transition:0.2s;
432}
433.box:hover{
434    transform:scale(1.05);
435}
436
437/* posiciones */
438.first{border-top:4px solid gold;}
439.second{border-top:4px solid silver;}
440.third{border-top:4px solid #f59e0b;}
441.normal{border-top:4px solid #334155;}
442
443.icon{font-size:32px;margin-bottom:10px;}
444
445/* logs */
446.logsBox{
447    max-height:200px;
448    overflow-y:auto;
449}
450.log{
451    font-size:12px;
452    padding:6px;
453    border-bottom:1px solid #333;
454}
455
456</style>
457</head>
458
459<body>
460
461<div class="container">
462
463<div class="topbar">
464<h1>🐕 GalgoMaster PRO</h1>
465<button onclick="location.href='historial.php'">📊 Historial</button>
466</div>
467
468<?php if($msg): ?>
469<div class="card"><?= e($msg) ?></div>
470<?php endif; ?>
471
472<div class="card">
473<form method="post">
474<textarea name="html"><?= e($html) ?></textarea>
475<br><br>
476<button name="action" value="preview">🔍 Analizar</button>
477</form>
478</div>
479
480<?php if($data): ?>
481
482<div class="card">
483
484<h2><?= e($data['evento']) ?></h2>
485<p>
486📍 <?= e($data['pista']) ?> — <?= e($data['distancia'] ?? '-') ?> m.  
487🏁 N° Carrera: <?= e($data['carrera'] ?? '-') ?>    
488🕒 Hora de Inicio: <?= e($data['hora'] ?? '-') ?>    
489📊 Estado: - <?= e($data['estado'] ?? '-') ?> -
490</p>
491
492<div class="grid">
493
494<?php foreach($data['runners'] as $r):
495
496$class="normal";
497$icon="🐕";
498
499if($r['posicion']==1){ $class="first"; $icon="🥇"; }
500elseif($r['posicion']==2){ $class="second"; $icon="🥈"; }
501elseif($r['posicion']==3){ $class="third"; $icon="🥉"; }
502
503?>
504
505<div class="box <?= $class ?>">
506<div class="icon"><?= $icon ?></div>
507<b class="dog-name"><?= e($r['nombre']) ?></b>
508
509<?php if(!empty($r['trap'])): ?>
510<div class="dog-trap">
511    Cajón N° <?= e($r['trap']) ?>
512</div>
513<?php endif; ?>
514</div>
515
516<?php endforeach; ?>
517
518</div>
519
520<br>
521
522<form method="post">
523<input type="hidden" name="action" value="save">
524<input type="hidden" name="payload" value="<?= base64_encode(json_encode($data)) ?>">
525<button>💾 Guardar</button>
526</form>
527
528</div>
529
530<?php endif; ?>
531
532<div class="card">
533<h3>📜 Logs</h3>
534
535<div class="logsBox">
536
537<?php foreach($logs as $l): ?>
538<div class="log">
539<?= e($l['fecha'] ?? '') ?> → <?= e($l['accion'] ?? '') ?>
540</div>
541<?php endforeach; ?>
542
543</div>
544
545</div>
546
547</div>
548
549</body>
550</html>
551
552
553
554
555app/Bet365Parser.php:
556
557<?php
558
559declare(strict_types=1);
560
561require_once __DIR__ . '/StyleDetector.php';
562require_once __DIR__ . '/RatingService.php';
563
564final class Bet365Parser
565{
566    public function parse(string $html): array
567    {
568        $html = html_entity_decode($html, ENT_QUOTES | ENT_HTML5, 'UTF-8');
569        $html = trim($html);
570
571        if ($html === '') {
572            throw new RuntimeException('El HTML está vacío.');
573        }
574
575        libxml_use_internal_errors(true);
576
577        $dom = new DOMDocument('1.0', 'UTF-8');
578
579        $loaded = $dom->loadHTML(
580            '<?xml encoding="UTF-8">' . $html,
581            LIBXML_NOWARNING | LIBXML_NOERROR | LIBXML_COMPACT
582        );
583
584        if (!$loaded) {
585            throw new RuntimeException('No se pudo interpretar el HTML.');
586        }
587
588        $xpath = new DOMXPath($dom);
589
590        $race = $this->parseRace($xpath);
591        $runners = $this->parseRunners($xpath, $race);
592
593        return [
594            'html_hash' => hash('sha256', $html),
595            'race' => $race,
596            'runners' => $runners,
597        ];
598    }
599
600    public function parseHistorico(string $html): array
601    {
602    libxml_use_internal_errors(true);
603
604    $dom = new DOMDocument();
605    $dom->loadHTML($html);
606
607    $xpath = new DOMXPath($dom);
608
609    /* ✅ INFO BÁSICA */
610    $pista = '';
611    $evento = '';
612    $hora = null;
613    $carrera = null;
614    $distancia = null;
615
616    // pista
617    $n = $xpath->query("//*[contains(@class,'rcr-20')]");
618    if ($n->length) $pista = trim($n->item(0)->textContent);
619
620    // nombre evento
621    $n = $xpath->query("//*[contains(@class,'rcr-67')]");
622    if ($n->length) $evento = trim($n->item(0)->textContent);
623
624    // info carrera
625    $info = $xpath->query("//*[contains(@class,'rcr-14')]");
626    foreach ($info as $i) {
627        $txt = trim($i->textContent);
628
629        if (strpos($txt, 'Carrera') !== false) {
630            preg_match('/(\d+)/', $txt, $m);
631            $carrera = $m[1] ?? null;
632        }
633
634        if (strpos($txt, 'Empezada') !== false) {
635            preg_match('/(\d{2}:\d{2}:\d{2})/', $txt, $m);
636            $hora = $m[1] ?? null;
637        }
638
639        if (preg_match('/(\d+)m/', $txt, $m)) {
640            $distancia = (int)$m[1];
641        }
642    }
643
644    /* ✅ RESULTADOS */
645    $runners = [];
646
647    $rows = $xpath->query("//*[contains(@class,'rrs-610')]");
648
649    foreach ($rows as $row) {
650
651        $posNode = $xpath->query(".//*[contains(@class,'rrs-1a')]", $row);
652        $nameNode = $xpath->query(".//*[contains(@class,'rrs-44')]", $row);
653        $trapNode = $xpath->query(".//img", $row);
654
655        if (!$nameNode->length) continue;
656
657        $name = trim($nameNode->item(0)->textContent);
658        if ($name === '') continue;
659
660        $pos = $posNode->length ? trim($posNode->item(0)->textContent) : null;
661
662        $posNum = null;
663        if ($pos && preg_match('/(\d+)/', $pos, $m)) {
664            $posNum = (int)$m[1];
665        }
666
667        $trap = $trapNode->length ? $trapNode->item(0)->getAttribute('alt') : null;
668
669        /* ✅ ODDS */
670        $oddNodes = $xpath->query(".//*[contains(@class,'rrs-763')]", $row);
671
672        $odds = [];
673        foreach ($oddNodes as $o) {
674            $val = trim($o->textContent);
675            if ($val !== '-' && $val !== '') {
676                $odds[] = (float)str_replace(',', '.', $val);
677            }
678        }
679
680        $avgOdds = count($odds) ? array_sum($odds)/count($odds) : null;
681
682        $runners[] = [
683            'trap' => $trap,
684            'nombre' => $name,
685            'posicion' => $pos,
686            'posicion_num' => $posNum,
687            'odds_promedio' => $avgOdds,
688            'odds' => $odds
689        ];
690    }
691
692    /* ✅ DISTANCIA ENTRE GALGOS */
693    $gap = null;
694    $gapNode = $xpath->query("//*[contains(@class,'rrs-035')]");
695    if ($gapNode->length) {
696        $gap = trim($gapNode->item(0)->textContent);
697    }
698
699    return [
700        'html_hash' => hash('sha256', $html),
701        'race' => [
702            'pista' => $pista,
703            'evento' => $evento,
704            'hora' => $hora,
705            'carrera' => $carrera,
706            'distancia' => $distancia,
707            'gap' => $gap
708        ],
709        'runners' => $runners
710    ];
711}
712
713    private function parseRace(DOMXPath $xpath): array
714    {
715        $pista = $this->firstText($xpath, "//*[contains(@class,'rcr-20')]");
716        $hora = $this->firstText($xpath, "//*[contains(@class,'rcr-31')]");
717
718        $carreraNumero = null;
719        $distancia = null;
720
721        $raceLabels = $this->texts($xpath, "//*[contains(@class,'rcr-4b')]");
722
723        foreach ($raceLabels as $label) {
724            if (stripos($label, 'Carrera') !== false) {
725                $carreraNumero = trim(str_replace('•', '', $label));
726            }
727
728            if (preg_match('/(\d{3,4})\s*m/i', $label, $m)) {
729                $distancia = (int) $m[1];
730            }
731        }
732
733        $preview = $this->firstText($xpath, "//*[contains(@class,'sr-0f27e2')]");
734
735        $fechaHora = null;
736
737        if ($hora && preg_match('/^\d{1,2}:\d{2}$/', $hora)) {
738            $fechaHora = date('Y-m-d') . ' ' . $hora . ':00';
739        }
740
741        return [
742            'pista' => $pista ?: 'Unknown',
743            'hora_texto' => $hora ?: null,
744            'fecha_texto' => date('Y-m-d'),
745            'fecha_hora' => $fechaHora,
746            'carrera_numero' => $carreraNumero,
747            'distancia' => $distancia,
748            'grado' => null,
749            'tipo' => 'Greyhound',
750            'going' => null,
751            'preview' => $preview ?: null,
752            'comentario_experto' => $preview ?: null,
753        ];
754    }
755
756    private function parseRunners(DOMXPath $xpath, array $race): array
757    {
758        $nodes = $xpath->query("//*[contains(concat(' ', normalize-space(@class), ' '), ' rg-cb ')]");
759
760        $runners = [];
761        $seen = [];
762
763        foreach ($nodes as $node) {
764            if (!$node instanceof DOMElement) {
765                continue;
766            }
767
768            $name = $this->firstTextRelative($xpath, $node, ".//*[contains(@class,'rg-0d')]");
769            $form = $this->firstAttributeRelative($xpath, $node, ".//*[contains(@class,'rg-0e')]", 'data-content');
770
771            if (!$form) {
772                $form = $this->firstTextRelative($xpath, $node, ".//*[contains(@class,'rg-0e')]");
773            }
774
775            if (!$name || !$form || !preg_match('/^?:-[1-6]{2,}$/', $form)) {
776                continue;
777            }
778
779            $trap = $this->extractTrap($xpath, $node);
780            if (!$trap) {
781                $trap = count($runners) + 1;
782            }
783
784            $oddsFinalText = $this->firstTextRelative($xpath, $node, ".//*[contains(@class,'rul-ce0412')]");
785            $oddsFinal = RatingService::decimalFromFraction($oddsFinalText);
786
787            $oddsHistory = $this->textsRelative($xpath, $node, ".//*[contains(@class,'sr-dc6edd')]//span");
788            $oddsInicial = null;
789
790            if (!empty($oddsHistory[0])) {
791                $oddsInicial = RatingService::decimalFromFraction($oddsHistory[0]);
792            }
793
794            if ($oddsInicial === null) {
795                $oddsInicial = $oddsFinal;
796            }
797
798            $history = $this->parseHistory($xpath, $node, $race['pista'] ?? null);
799
800            $remarksList = [];
801
802            foreach ($history as $row) {
803                if (!empty($row['remarks'])) {
804                    $remarksList[] = $row['remarks'];
805                }
806            }
807
808            $style = StyleDetector::detectFromMany($remarksList);
809
810            $rating = RatingService::calculate([
811                'forma_reciente' => $form,
812                'historial' => $history,
813                'estilo_detectado' => $style,
814            ], $race['distancia'] ?? null);
815
816            $value = RatingService::valueSignal($rating['rating'], $oddsFinal);
817
818            $latest = $history[0] ?? [];
819
820            $runner = [
821                'trap' => (int) $trap,
822                'nombre' => trim($name),
823                'forma_reciente' => trim($form),
824                'odds_inicial' => $oddsInicial,
825                'odds_final' => $oddsFinal,
826                'implied_probability' => RatingService::impliedProbability($oddsFinal),
827                'odds_movimiento' => $oddsHistory,
828                'cal_tm' => isset($latest['cal_tm']) ? $latest['cal_tm'] : null,
829                'win_time' => isset($latest['win_tm']) ? $latest['win_tm'] : null,
830                'splits' => isset($latest['splits']) ? $latest['splits'] : null,
831                'db' => isset($latest['db']) ? $latest['db'] : null,
832                'remarks' => implode(' | ', array_slice($remarksList, 0, 3)),
833                'estilo_detectado' => $style,
834                'rating_pre' => $rating['rating'],
835                'score_forma' => $rating['score_forma'],
836                'score_velocidad' => $rating['score_velocidad'],
837                'score_estilo' => $rating['score_estilo'],
838                'value_score' => $value['value_score'],
839                'value_flag' => $value['value_flag'],
840                'historial' => $history,
841                'raw_text' => $this->cleanText($node->textContent),
842            ];
843
844            $key = $runner['trap'] . '-' . mb_strtolower($runner['nombre'], 'UTF-8');
845
846            if (!isset($seen[$key])) {
847                $seen[$key] = true;
848                $runners[] = $runner;
849            }
850        }
851
852        usort($runners, fn ($a, $b) => $a['trap'] <=> $b['trap']);
853
854        return $runners;
855    }
856
857    private function parseHistory(DOMXPath $xpath, DOMElement $runnerNode, ?string $pista): array
858    {
859        $columns = [];
860
861        $columnNodes = $xpath->query(
862            ".//*[contains(concat(' ', normalize-space(@class), ' '), ' rg-a3 ')]",
863            $runnerNode
864        );
865
866        foreach ($columnNodes as $columnNode) {
867            if (!$columnNode instanceof DOMElement) {
868                continue;
869            }
870
871            $header = $this->firstTextRelative($xpath, $columnNode, ".//*[contains(@class,'rg-58')]");
872            $values = $this->textsRelative($xpath, $columnNode, ".//*[contains(@class,'rg-7f')]");
873
874            $header = trim($header);
875
876            if ($header !== '') {
877                $columns[$header] = $values;
878            }
879        }
880
881        if (!$columns) {
882            return [];
883        }
884
885        $maxRows = 0;
886
887        foreach ($columns as $values) {
888            $maxRows = max($maxRows, count($values));
889        }
890
891        $history = [];
892
893        for ($i = 0; $i < $maxRows; $i++) {
894            $dateText = $columns['Date'][$i] ?? null;
895            $disTrp = $columns['Dis / Trp'][$i] ?? null;
896            $fin = $columns['Fin'][$i] ?? null;
897            $sp = $columns['SP'][$i] ?? null;
898            $gr = $columns['Gr'][$i] ?? null;
899            $calTm = $columns['CalTm'][$i] ?? null;
900            $splitTm = $columns['SplitTm'][$i] ?? null;
901            $splits = $columns['Splits'][$i] ?? null;
902            $db = $columns['DB'][$i] ?? null;
903            $remarks = $columns['Remarks'][$i] ?? null;
904            $winTm = $columns['WinTm'][$i] ?? null;
905            $going = $columns['G'][$i] ?? null;
906
907            $distance = null;
908            $trap = null;
909
910            if ($disTrp && preg_match('/(\d{3,4})\s*m\s*\[(\d)\]/i', $disTrp, $m)) {
911                $distance = (int) $m[1];
912                $trap = (int) $m[2];
913            }
914
915            $history[] = [
916                'fecha_texto' => $dateText,
917                'fecha' => $this->normaliseShortDate($dateText),
918                'pista' => $pista,
919                'distancia' => $distance,
920                'trap' => $trap,
921                'posicion' => $fin,
922                'posicion_num' => $this->positionToInt($fin),
923                'sp' => $sp,
924                'sp_decimal' => RatingService::decimalFromFraction($sp),
925                'grade' => $gr ?: null,
926                'cal_tm' => is_numeric($calTm) ? (float) $calTm : null,
927                'split_tm' => $splitTm ?: null,
928                'splits' => $splits ?: null,
929                'db' => $db ?: null,
930                'remarks' => $remarks ?: null,
931                'win_tm' => is_numeric($winTm) ? (float) $winTm : null,
932                'going' => $going ?: null,
933            ];
934        }
935
936        return $history;
937    }
938
939    private function extractTrap(DOMXPath $xpath, DOMElement $node): ?int
940    {
941        $alt = $this->firstAttributeRelative(
942            $xpath,
943            $node,
944            ".//img[contains(@src,'dog') or contains(@src,'traps_uk')]",
945            'alt'
946        );
947
948        if ($alt && preg_match('/^[1-6]$/', trim($alt))) {
949            return (int) $alt;
950        }
951
952        $src = $this->firstAttributeRelative(
953            $xpath,
954            $node,
955            ".//img[contains(@src,'dog') or contains(@src,'traps_uk')]",
956            'src'
957        );
958
959        if ($src && preg_match('/dog([1-6])\.svg/i', $src, $m)) {
960            return (int) $m[1];
961        }
962
963        return null;
964    }
965
966    private function normaliseShortDate(?string $date): ?string
967    {
968        $date = trim((string) $date);
969
970        if (!preg_match('/^(\d{1,2})\/(\d{1,2})$/', $date, $m)) {
971            return null;
972        }
973
974        $day = (int) $m[1];
975        $month = (int) $m[2];
976        $year = (int) date('Y');
977
978        if (!checkdate($month, $day, $year)) {
979            return null;
980        }
981
982        return sprintf('%04d-%02d-%02d', $year, $month, $day);
983    }
984
985    private function positionToInt(?string $position): ?int
986    {
987        if (!$position) {
988            return null;
989        }
990
991        if (preg_match('/(\d+)/', $position, $m)) {
992            return (int) $m[1];
993        }
994
995        return null;
996    }
997
998    private function firstText(DOMXPath $xpath, string $query): ?string
999    {
1000        $nodes = $xpath->query($query);
1001
1002        if (!$nodes || $nodes->length === 0) {
1003            return null;
1004        }
1005
1006        return $this->cleanText($nodes->item(0)->textContent ?? '');
1007    }
1008
1009    private function firstTextRelative(DOMXPath $xpath, DOMNode $context, string $query): ?string
1010    {
1011        $nodes = $xpath->query($query, $context);
1012
1013        if (!$nodes || $nodes->length === 0) {
1014            return null;
1015        }
1016
1017        return $this->cleanText($nodes->item(0)->textContent ?? '');
1018    }
1019
1020    private function firstAttributeRelative(DOMXPath $xpath, DOMNode $context, string $query, string $attr): ?string
1021    {
1022        $nodes = $xpath->query($query, $context);
1023
1024        if (!$nodes || $nodes->length === 0) {
1025            return null;
1026        }
1027
1028        $node = $nodes->item(0);
1029
1030        if (!$node instanceof DOMElement) {
1031            return null;
1032        }
1033
1034        $value = $node->getAttribute($attr);
1035
1036        return $value !== '' ? trim($value) : null;
1037    }
1038
1039    private function texts(DOMXPath $xpath, string $query): array
1040    {
1041        $nodes = $xpath->query($query);
1042        $out = [];
1043
1044        if (!$nodes) {
1045            return $out;
1046        }
1047
1048        foreach ($nodes as $node) {
1049            $text = $this->cleanText($node->textContent ?? '');
1050
1051            if ($text !== '') {
1052                $out[] = $text;
1053            }
1054        }
1055
1056        return $out;
1057    }
1058
1059    private function textsRelative(DOMXPath $xpath, DOMNode $context, string $query): array
1060    {
1061        $nodes = $xpath->query($query, $context);
1062        $out = [];
1063
1064        if (!$nodes) {
1065            return $out;
1066        }
1067
1068        foreach ($nodes as $node) {
1069            $text = $this->cleanText($node->textContent ?? '');
1070            $out[] = $text;
1071        }
1072
1073        return $out;
1074    }
1075
1076    private function cleanText(string $text): string
1077    {
1078        $text = html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8');
1079        $text = preg_replace('/\s+/u', ' ', $text);
1080
1081        return trim((string) $text);
1082    }
1083}
1084
1085
1086
1087app/db.php: 
1088
1089<?php
1090
1091declare(strict_types=1);
1092
1093function gm_config(): array
1094{
1095    static $config = null;
1096
1097    if ($config === null) {
1098        $config = require __DIR__ . '/config.php';
1099
1100        if (!empty($config['timezone'])) {
1101            date_default_timezone_set($config['timezone']);
1102        }
1103    }
1104
1105    return $config;
1106}
1107
1108function gm_db(): PDO
1109{
1110    static $pdo = null;
1111
1112    if ($pdo instanceof PDO) {
1113        return $pdo;
1114    }
1115
1116    $config = gm_config();
1117
1118    $dsn = sprintf(
1119        'mysql:host=%s;dbname=%s;charset=utf8mb4',
1120        $config['db_host'],
1121        $config['db_name']
1122    );
1123
1124    $pdo = new PDO($dsn, $config['db_user'], $config['db_pass'], [
1125        PDO::ATTR_ERRMODE => PDO::ERRMODE_EXCEPTION,
1126        PDO::ATTR_DEFAULT_FETCH_MODE => PDO::FETCH_ASSOC,
1127        PDO::ATTR_EMULATE_PREPARES => false,
1128    ]);
1129
1130    return $pdo;
1131}
1132
1133function gm_slug(string $value): string
1134{
1135    $value = trim(mb_strtolower($value, 'UTF-8'));
1136    $ascii = @iconv('UTF-8', 'ASCII//TRANSLIT//IGNORE', $value);
1137
1138    if ($ascii !== false) {
1139        $value = $ascii;
1140    }
1141
1142    $value = preg_replace('/[^a-z0-9]+/i', '-', $value);
1143    $value = trim((string) $value, '-');
1144
1145    return $value !== '' ? $value : 'galgo-' . bin2hex(random_bytes(4));
1146}

Raw Paste

scraper.php:

<?php
require_once 'utils.php';
require_once __DIR__ . '/app/db.php';

error_reporting(E_ALL);
ini_set('display_errors', '1');

function e($v){
    return htmlspecialchars((string)$v, ENT_QUOTES, 'UTF-8');
}

/* ✅ LOG */
function logAction($accion,$detalle){
    try{
        gm_db()->prepare("INSERT INTO scraping_logs (accion,detalle) VALUES(?,?)")
              ->execute([$accion,$detalle]);
    }catch(Exception $e){}
}

/* ✅ PARSER */
function parseHistorico($html){

    libxml_use_internal_errors(true);

    $data = [
        'evento'=>'',
        'pista'=>'',
        'distancia'=>'',
        'hora'=>'',
        'carrera'=>'',
        'carrera_numero'=> null, // ✅ AÑADE ESTO
        'estado'=>'Finalizada',
        'runners'=>[]
    ];

    // ✅ PISTA
    if(preg_match('/\b(Manawatu|Doncaster|Hove|Sunderland|Yarmouth)\b/i',$html,$m)){
        $data['pista'] = $m[1];
    }
    
    // ✅ EVENTO
    if(preg_match('/\n([A-Za-z\s]+Heat)/',$html,$m)){
        $data['evento'] = trim($m[1]);
    }
    
    // ✅ DISTANCIA
    if(preg_match('/(\d{2,4})m/',$html,$m)){
        $data['distancia'] = $m[1];
    }
    
    // ✅ HORA
    if(preg_match('/Empezada\s*(\d{2}:\d{2}:\d{2})/',$html,$m)){
        $data['hora'] = $m[1];
    }
    
    // ✅ ESTADO
    if(stripos($html,'Finalizada') !== false){
        $data['estado'] = 'Finalizada';
    }
    
    // ✅ CARRERA
    if(preg_match('/Carrera\s*(\d+)/',$html,$m)){
        $data['carrera_numero'] = (int)$m[1];
    }

    $data['runners'] = [];
    
    if(preg_match_all('/(\d+)º\s+(\d+)\s+([A-Za-z\s]+)/',$html,$matches,PREG_SET_ORDER)){
    
        foreach($matches as $m){
    
            $posicion = (int)$m[1];
            $trap = (int)$m[2];
            $nombre = trim($m[3]);
    
            if($posicion < 1 || $posicion > 8) continue;
    
            $data['runners'][] = [
                'nombre' => $nombre,
                'posicion' => $posicion,
                'trap' => $trap
            ];
        }
    }

    // ✅ PARSER GLOBAL (robusto para Bet365 moderno)
    
    // 🏟 pista (ej: Doncaster)
    if(empty($data['pista']) && preg_match('/\b(Doncaster|Manukau|Hove|Sunderland|Yarmouth)\b/i',$html,$m)){
        $data['pista'] = $m[1];
    }
    
    // 📏 distancia (ej: 275m)
    if(empty($data['distancia']) && preg_match('/(\d{2,4})\s*m/i',$html,$m)){
        $data['distancia'] = $m[1];
    }
    
    // 🕒 hora (ej: 08:26:23)
    if(empty($data['hora']) && preg_match('/(\d{2}:\d{2}:\d{2})/',$html,$m)){
        $data['hora'] = $m[1];
    }
    
    // 📊 estado
    if(stripos($html,'finalizada') !== false){
        $data['estado'] = 'Finalizada';
    }elseif(stripos($html,'empezada') !== false){
        $data['estado'] = 'En curso';
    }
    
    // 🏁 número de carrera (DE LAS BURBUJAS)
    if(empty($data['carrera_numero'])){
    
        if(preg_match('/>\s*(\d{1,2})\s*</',$html,$m)){
            $data['carrera_numero'] = (int)$m[1];
        }
    }
    
    if(empty($data['pista'])){
        if(preg_match('/\b([A-Z][a-z]+)\b\s*</',$html,$m)){
            $data['pista'] = $m[1];
        }
    }
    
    
    // ✅ detectar número desde burbujas
    if(empty($data['carrera_numero']) && !empty($html)){
        if(preg_match('/>\s*(\d{1,2})\s*</',$html,$m)){
            $data['carrera_numero'] = (int)$m[1];
        }
    }


        // ✅ FALLBACK GLOBAL (muy importante)
    if(empty($data['carrera_numero'])){
    
        if(preg_match('/\b(\d{1,2})\b/', $html, $m)){
            $data['carrera_numero'] = (int)$m[1];
        }
    }

    return $data;
}

/* ✅ GUARDADO */
function saveRaceDB($data){

    $pdo = gm_db();
    $pdo->beginTransaction();

    try{

        if(empty($data['runners'])){
            throw new Exception("No hay runners");
        }

        $hash = md5(uniqid().serialize($data));

       $stmt = $pdo->prepare("
            INSERT INTO carreras
            (pista, distancia, evento, hash, tipo, fuente, estado, fecha_carrera, pais, carrera_numero, codigo_unico)
            VALUES (?, ?, ?, ?, 'Greyhound','html_manual','finalizada', ?, ?, ?, ?)
        ");

        
        // ✅ mapa de pistas
        $TRACK_MAP = [
            'Manukau' => ['pais'=>'Nueva Zelanda'],
            'Healesville' => ['pais'=>'Australia'],
            'Capalaba' => ['pais'=>'Australia'],
            'Mount Gambier' => ['pais'=>'Australia'],
            'Sale' => ['pais'=>'Australia'],
            'Rockhampton' => ['pais'=>'Australia'],
            'Richmond' => ['pais'=>'Australia'],
            'Nowra' => ['pais'=>'Australia'],
            'Q2 Parklands' => ['pais'=>'Australia'],
            'Gawler' => ['pais'=>'Australia'],
            'Temora' => ['pais'=>'Australia'],
            'Darwin' => ['pais'=>'Australia'],
            'Valley' => ['pais'=>'UK'],
            'Doncaster' => ['pais'=>'UK'],
            'Yarmouth' => ['pais'=>'UK'],
            'Sunderland' => ['pais'=>'UK'],
            'Hove' => ['pais'=>'UK'],
        ];
        
        // ✅ obtener país
        $trackInfo = $TRACK_MAP[$data['pista']] ?? ['pais'=>'Desconocido'];
        
        // ✅ generar código único
        $codigo = generarCodigoCarrera([
            'pais' => $trackInfo['pais'],
            'pista' => $data['pista'],
            'carrera_numero' => $data['carrera_numero'] ?? 0,
            'evento' => $data['evento'],
            'fecha_carrera' => $data['hora'] 
                ? date('Y-m-d').' '.$data['hora']
                : date('Y-m-d') // ✅ fallback seguro
        ]);

        // ✅ asegurar unicidad del código
        $stmtCheck = $pdo->prepare("
            SELECT COUNT(*) FROM carreras WHERE codigo_unico = ?
        ");
        $stmtCheck->execute([$codigo]);
        
        if($stmtCheck->fetchColumn() > 0){
        
            // 🔥 generar sufijo incremental limpio
            $suffix = 1;
        
            do {
                $nuevoCodigo = $codigo . str_pad($suffix, 2, '0', STR_PAD_LEFT);
        
                $stmtCheck = $pdo->prepare("
                    SELECT COUNT(*) FROM carreras WHERE codigo_unico = ?
                ");
                $stmtCheck->execute([$nuevoCodigo]);
        
                $existe = $stmtCheck->fetchColumn();
                $suffix++;
        
            } while($existe > 0);
        
            $codigo = $nuevoCodigo;
        }

        // ✅ generar fecha carrera
        $fechaCarrera = $data['hora'] 
            ? date('Y-m-d').' '.$data['hora']
            : date('Y-m-d H:i:s');
        
        // ✅ ejecutar
        $stmt->execute([
            $data['pista'],          // 1
            $data['distancia'],      // 2
            $data['evento'],         // 3
            $hash,                  // 4
            $fechaCarrera,          // 5
            $trackInfo['pais'],     // 6
            $data['carrera_numero'] ?? null, // 7
            $codigo // ✅ NUEVO
        ]);


        $carreraId = $pdo->lastInsertId();

        foreach($data['runners'] as $r){

            $nombre = trim($r['nombre']);

            $stmt=$pdo->prepare("SELECT id_galgo FROM galgos WHERE nombre=?");
            $stmt->execute([$nombre]);
            $gid=$stmt->fetchColumn();

            if(!$gid){

                $slug = strtolower(trim(preg_replace('/[^A-Za-z0-9]+/', '-', $nombre)));
                $slug = trim($slug, '-');

                $stmt=$pdo->prepare("SELECT COUNT(*) FROM galgos WHERE slug=?");
                $stmt->execute([$slug]);

                if($stmt->fetchColumn()>0){
                    $slug .= "-".uniqid();
                }

                $stmt=$pdo->prepare("INSERT INTO galgos (nombre,slug) VALUES (?,?)");
                $stmt->execute([$nombre,$slug]);

                $gid=$pdo->lastInsertId();
            }

            $stmt=$pdo->prepare("
                INSERT INTO participaciones (carrera_id,galgo_id,posicion,trap)
                VALUES (?,?,?,?)
            ");

            $stmt->execute([
                $carreraId,
                $gid,
                $r['posicion'],
                $r['trap']
            ]);
        }

        $pdo->commit();

        return "✅ Guardado correctamente";

    }catch(Exception $e){

        $pdo->rollBack();

        echo "<pre style='color:red'>";
        echo "ERROR:\n".$e->getMessage();
        echo "</pre>";
        die();
    }
}

/* ✅ CONTROL */
$data=null;
$msg=null;
$html=$_POST['html'] ?? '';

if($_SERVER['REQUEST_METHOD']=='POST'){

    if($_POST['action']=='preview'){
        $data=parseHistorico($html);
    }

    if($_POST['action']=='save'){
        $payload=json_decode(base64_decode($_POST['payload']),true);
        $msg=saveRaceDB($payload);
        $data=$payload;
    }
    
}

$logs = gm_db()->query("SELECT * FROM scraping_logs ORDER BY id DESC LIMIT 10")->fetchAll();
?>

<!doctype html>
<html>
<head>
<meta charset="utf-8">
<title>GalgoMaster PRO</title>

<style>
body{
    background:linear-gradient(135deg,#020617,#0f172a);
    color:#e2e8f0;
    font-family:system-ui;
}

/* nombre galgo */
.dog-name{
    display:block;
    margin-top:6px;
    font-size:15px;
    font-weight:600;
}

/* cajón */
.dog-trap{
    margin-top:6px;
    font-size:13px;
    font-weight:700;
    color:#ef4444;

    background:rgba(239,68,68,0.1);
    padding:4px 8px;
    border-radius:6px;

    display:inline-block;
}

/* hover más elegante */
.box:hover .dog-trap{
    background:#ef4444;
    color:white;
}

.box{
    background:#020617;
    padding:20px;
    border-radius:12px;
    text-align:center;
    transition:0.2s;
    border:1px solid transparent;
}

.box:hover{
    transform:translateY(-4px);
    border:1px solid #22c55e;
}

/* layout */
.container{max-width:1100px;margin:auto;}

.card{
    background:#111827;
    padding:20px;
    margin:20px 0;
    border-radius:16px;
    box-shadow:0 10px 30px rgba(0,0,0,.5);
}

/* textarea */
textarea{
    width:100%;
    height:220px;
    background:#020617;
    color:#22c55e;
    border-radius:12px;
    padding:12px;
}

/* botones */
button{
    background:#22c55e;
    border:0;
    padding:12px 18px;
    border-radius:10px;
    font-weight:bold;
    cursor:pointer;
}

/* header */
.topbar{
    display:flex;
    justify-content:space-between;
    align-items:center;
}

/* ✅ GRID */
.grid{
    display:grid;
    grid-template-columns:repeat(4,1fr);
    gap:15px;
}

/* ✅ BOX */
.box{
    background:#020617;
    padding:20px;
    border-radius:12px;
    text-align:center;
    transition:0.2s;
}
.box:hover{
    transform:scale(1.05);
}

/* posiciones */
.first{border-top:4px solid gold;}
.second{border-top:4px solid silver;}
.third{border-top:4px solid #f59e0b;}
.normal{border-top:4px solid #334155;}

.icon{font-size:32px;margin-bottom:10px;}

/* logs */
.logsBox{
    max-height:200px;
    overflow-y:auto;
}
.log{
    font-size:12px;
    padding:6px;
    border-bottom:1px solid #333;
}

</style>
</head>

<body>

<div class="container">

<div class="topbar">
<h1>🐕 GalgoMaster PRO</h1>
<button onclick="location.href='historial.php'">📊 Historial</button>
</div>

<?php if($msg): ?>
<div class="card"><?= e($msg) ?></div>
<?php endif; ?>

<div class="card">
<form method="post">
<textarea name="html"><?= e($html) ?></textarea>
<br><br>
<button name="action" value="preview">🔍 Analizar</button>
</form>
</div>

<?php if($data): ?>

<div class="card">

<h2><?= e($data['evento']) ?></h2>
<p>
📍 <?= e($data['pista']) ?> — <?= e($data['distancia'] ?? '-') ?> m.  
🏁 N° Carrera: <?= e($data['carrera'] ?? '-') ?>    
🕒 Hora de Inicio: <?= e($data['hora'] ?? '-') ?>    
📊 Estado: - <?= e($data['estado'] ?? '-') ?> -
</p>

<div class="grid">

<?php foreach($data['runners'] as $r):

$class="normal";
$icon="🐕";

if($r['posicion']==1){ $class="first"; $icon="🥇"; }
elseif($r['posicion']==2){ $class="second"; $icon="🥈"; }
elseif($r['posicion']==3){ $class="third"; $icon="🥉"; }

?>

<div class="box <?= $class ?>">
<div class="icon"><?= $icon ?></div>
<b class="dog-name"><?= e($r['nombre']) ?></b>

<?php if(!empty($r['trap'])): ?>
<div class="dog-trap">
    Cajón N° <?= e($r['trap']) ?>
</div>
<?php endif; ?>
</div>

<?php endforeach; ?>

</div>

<br>

<form method="post">
<input type="hidden" name="action" value="save">
<input type="hidden" name="payload" value="<?= base64_encode(json_encode($data)) ?>">
<button>💾 Guardar</button>
</form>

</div>

<?php endif; ?>

<div class="card">
<h3>📜 Logs</h3>

<div class="logsBox">

<?php foreach($logs as $l): ?>
<div class="log">
<?= e($l['fecha'] ?? '') ?> → <?= e($l['accion'] ?? '') ?>
</div>
<?php endforeach; ?>

</div>

</div>

</div>

</body>
</html>




app/Bet365Parser.php:

<?php

declare(strict_types=1);

require_once __DIR__ . '/StyleDetector.php';
require_once __DIR__ . '/RatingService.php';

final class Bet365Parser
{
    public function parse(string $html): array
    {
        $html = html_entity_decode($html, ENT_QUOTES | ENT_HTML5, 'UTF-8');
        $html = trim($html);

        if ($html === '') {
            throw new RuntimeException('El HTML está vacío.');
        }

        libxml_use_internal_errors(true);

        $dom = new DOMDocument('1.0', 'UTF-8');

        $loaded = $dom->loadHTML(
            '<?xml encoding="UTF-8">' . $html,
            LIBXML_NOWARNING | LIBXML_NOERROR | LIBXML_COMPACT
        );

        if (!$loaded) {
            throw new RuntimeException('No se pudo interpretar el HTML.');
        }

        $xpath = new DOMXPath($dom);

        $race = $this->parseRace($xpath);
        $runners = $this->parseRunners($xpath, $race);

        return [
            'html_hash' => hash('sha256', $html),
            'race' => $race,
            'runners' => $runners,
        ];
    }

    public function parseHistorico(string $html): array
    {
    libxml_use_internal_errors(true);

    $dom = new DOMDocument();
    $dom->loadHTML($html);

    $xpath = new DOMXPath($dom);

    /* ✅ INFO BÁSICA */
    $pista = '';
    $evento = '';
    $hora = null;
    $carrera = null;
    $distancia = null;

    // pista
    $n = $xpath->query("//*[contains(@class,'rcr-20')]");
    if ($n->length) $pista = trim($n->item(0)->textContent);

    // nombre evento
    $n = $xpath->query("//*[contains(@class,'rcr-67')]");
    if ($n->length) $evento = trim($n->item(0)->textContent);

    // info carrera
    $info = $xpath->query("//*[contains(@class,'rcr-14')]");
    foreach ($info as $i) {
        $txt = trim($i->textContent);

        if (strpos($txt, 'Carrera') !== false) {
            preg_match('/(\d+)/', $txt, $m);
            $carrera = $m[1] ?? null;
        }

        if (strpos($txt, 'Empezada') !== false) {
            preg_match('/(\d{2}:\d{2}:\d{2})/', $txt, $m);
            $hora = $m[1] ?? null;
        }

        if (preg_match('/(\d+)m/', $txt, $m)) {
            $distancia = (int)$m[1];
        }
    }

    /* ✅ RESULTADOS */
    $runners = [];

    $rows = $xpath->query("//*[contains(@class,'rrs-610')]");

    foreach ($rows as $row) {

        $posNode = $xpath->query(".//*[contains(@class,'rrs-1a')]", $row);
        $nameNode = $xpath->query(".//*[contains(@class,'rrs-44')]", $row);
        $trapNode = $xpath->query(".//img", $row);

        if (!$nameNode->length) continue;

        $name = trim($nameNode->item(0)->textContent);
        if ($name === '') continue;

        $pos = $posNode->length ? trim($posNode->item(0)->textContent) : null;

        $posNum = null;
        if ($pos && preg_match('/(\d+)/', $pos, $m)) {
            $posNum = (int)$m[1];
        }

        $trap = $trapNode->length ? $trapNode->item(0)->getAttribute('alt') : null;

        /* ✅ ODDS */
        $oddNodes = $xpath->query(".//*[contains(@class,'rrs-763')]", $row);

        $odds = [];
        foreach ($oddNodes as $o) {
            $val = trim($o->textContent);
            if ($val !== '-' && $val !== '') {
                $odds[] = (float)str_replace(',', '.', $val);
            }
        }

        $avgOdds = count($odds) ? array_sum($odds)/count($odds) : null;

        $runners[] = [
            'trap' => $trap,
            'nombre' => $name,
            'posicion' => $pos,
            'posicion_num' => $posNum,
            'odds_promedio' => $avgOdds,
            'odds' => $odds
        ];
    }

    /* ✅ DISTANCIA ENTRE GALGOS */
    $gap = null;
    $gapNode = $xpath->query("//*[contains(@class,'rrs-035')]");
    if ($gapNode->length) {
        $gap = trim($gapNode->item(0)->textContent);
    }

    return [
        'html_hash' => hash('sha256', $html),
        'race' => [
            'pista' => $pista,
            'evento' => $evento,
            'hora' => $hora,
            'carrera' => $carrera,
            'distancia' => $distancia,
            'gap' => $gap
        ],
        'runners' => $runners
    ];
}

    private function parseRace(DOMXPath $xpath): array
    {
        $pista = $this->firstText($xpath, "//*[contains(@class,'rcr-20')]");
        $hora = $this->firstText($xpath, "//*[contains(@class,'rcr-31')]");

        $carreraNumero = null;
        $distancia = null;

        $raceLabels = $this->texts($xpath, "//*[contains(@class,'rcr-4b')]");

        foreach ($raceLabels as $label) {
            if (stripos($label, 'Carrera') !== false) {
                $carreraNumero = trim(str_replace('•', '', $label));
            }

            if (preg_match('/(\d{3,4})\s*m/i', $label, $m)) {
                $distancia = (int) $m[1];
            }
        }

        $preview = $this->firstText($xpath, "//*[contains(@class,'sr-0f27e2')]");

        $fechaHora = null;

        if ($hora && preg_match('/^\d{1,2}:\d{2}$/', $hora)) {
            $fechaHora = date('Y-m-d') . ' ' . $hora . ':00';
        }

        return [
            'pista' => $pista ?: 'Unknown',
            'hora_texto' => $hora ?: null,
            'fecha_texto' => date('Y-m-d'),
            'fecha_hora' => $fechaHora,
            'carrera_numero' => $carreraNumero,
            'distancia' => $distancia,
            'grado' => null,
            'tipo' => 'Greyhound',
            'going' => null,
            'preview' => $preview ?: null,
            'comentario_experto' => $preview ?: null,
        ];
    }

    private function parseRunners(DOMXPath $xpath, array $race): array
    {
        $nodes = $xpath->query("//*[contains(concat(' ', normalize-space(@class), ' '), ' rg-cb ')]");

        $runners = [];
        $seen = [];

        foreach ($nodes as $node) {
            if (!$node instanceof DOMElement) {
                continue;
            }

            $name = $this->firstTextRelative($xpath, $node, ".//*[contains(@class,'rg-0d')]");
            $form = $this->firstAttributeRelative($xpath, $node, ".//*[contains(@class,'rg-0e')]", 'data-content');

            if (!$form) {
                $form = $this->firstTextRelative($xpath, $node, ".//*[contains(@class,'rg-0e')]");
            }

            if (!$name || !$form || !preg_match('/^?:-[1-6]{2,}$/', $form)) {
                continue;
            }

            $trap = $this->extractTrap($xpath, $node);
            if (!$trap) {
                $trap = count($runners) + 1;
            }

            $oddsFinalText = $this->firstTextRelative($xpath, $node, ".//*[contains(@class,'rul-ce0412')]");
            $oddsFinal = RatingService::decimalFromFraction($oddsFinalText);

            $oddsHistory = $this->textsRelative($xpath, $node, ".//*[contains(@class,'sr-dc6edd')]//span");
            $oddsInicial = null;

            if (!empty($oddsHistory[0])) {
                $oddsInicial = RatingService::decimalFromFraction($oddsHistory[0]);
            }

            if ($oddsInicial === null) {
                $oddsInicial = $oddsFinal;
            }

            $history = $this->parseHistory($xpath, $node, $race['pista'] ?? null);

            $remarksList = [];

            foreach ($history as $row) {
                if (!empty($row['remarks'])) {
                    $remarksList[] = $row['remarks'];
                }
            }

            $style = StyleDetector::detectFromMany($remarksList);

            $rating = RatingService::calculate([
                'forma_reciente' => $form,
                'historial' => $history,
                'estilo_detectado' => $style,
            ], $race['distancia'] ?? null);

            $value = RatingService::valueSignal($rating['rating'], $oddsFinal);

            $latest = $history[0] ?? [];

            $runner = [
                'trap' => (int) $trap,
                'nombre' => trim($name),
                'forma_reciente' => trim($form),
                'odds_inicial' => $oddsInicial,
                'odds_final' => $oddsFinal,
                'implied_probability' => RatingService::impliedProbability($oddsFinal),
                'odds_movimiento' => $oddsHistory,
                'cal_tm' => isset($latest['cal_tm']) ? $latest['cal_tm'] : null,
                'win_time' => isset($latest['win_tm']) ? $latest['win_tm'] : null,
                'splits' => isset($latest['splits']) ? $latest['splits'] : null,
                'db' => isset($latest['db']) ? $latest['db'] : null,
                'remarks' => implode(' | ', array_slice($remarksList, 0, 3)),
                'estilo_detectado' => $style,
                'rating_pre' => $rating['rating'],
                'score_forma' => $rating['score_forma'],
                'score_velocidad' => $rating['score_velocidad'],
                'score_estilo' => $rating['score_estilo'],
                'value_score' => $value['value_score'],
                'value_flag' => $value['value_flag'],
                'historial' => $history,
                'raw_text' => $this->cleanText($node->textContent),
            ];

            $key = $runner['trap'] . '-' . mb_strtolower($runner['nombre'], 'UTF-8');

            if (!isset($seen[$key])) {
                $seen[$key] = true;
                $runners[] = $runner;
            }
        }

        usort($runners, fn ($a, $b) => $a['trap'] <=> $b['trap']);

        return $runners;
    }

    private function parseHistory(DOMXPath $xpath, DOMElement $runnerNode, ?string $pista): array
    {
        $columns = [];

        $columnNodes = $xpath->query(
            ".//*[contains(concat(' ', normalize-space(@class), ' '), ' rg-a3 ')]",
            $runnerNode
        );

        foreach ($columnNodes as $columnNode) {
            if (!$columnNode instanceof DOMElement) {
                continue;
            }

            $header = $this->firstTextRelative($xpath, $columnNode, ".//*[contains(@class,'rg-58')]");
            $values = $this->textsRelative($xpath, $columnNode, ".//*[contains(@class,'rg-7f')]");

            $header = trim($header);

            if ($header !== '') {
                $columns[$header] = $values;
            }
        }

        if (!$columns) {
            return [];
        }

        $maxRows = 0;

        foreach ($columns as $values) {
            $maxRows = max($maxRows, count($values));
        }

        $history = [];

        for ($i = 0; $i < $maxRows; $i++) {
            $dateText = $columns['Date'][$i] ?? null;
            $disTrp = $columns['Dis / Trp'][$i] ?? null;
            $fin = $columns['Fin'][$i] ?? null;
            $sp = $columns['SP'][$i] ?? null;
            $gr = $columns['Gr'][$i] ?? null;
            $calTm = $columns['CalTm'][$i] ?? null;
            $splitTm = $columns['SplitTm'][$i] ?? null;
            $splits = $columns['Splits'][$i] ?? null;
            $db = $columns['DB'][$i] ?? null;
            $remarks = $columns['Remarks'][$i] ?? null;
            $winTm = $columns['WinTm'][$i] ?? null;
            $going = $columns['G'][$i] ?? null;

            $distance = null;
            $trap = null;

            if ($disTrp && preg_match('/(\d{3,4})\s*m\s*\[(\d)\]/i', $disTrp, $m)) {
                $distance = (int) $m[1];
                $trap = (int) $m[2];
            }

            $history[] = [
                'fecha_texto' => $dateText,
                'fecha' => $this->normaliseShortDate($dateText),
                'pista' => $pista,
                'distancia' => $distance,
                'trap' => $trap,
                'posicion' => $fin,
                'posicion_num' => $this->positionToInt($fin),
                'sp' => $sp,
                'sp_decimal' => RatingService::decimalFromFraction($sp),
                'grade' => $gr ?: null,
                'cal_tm' => is_numeric($calTm) ? (float) $calTm : null,
                'split_tm' => $splitTm ?: null,
                'splits' => $splits ?: null,
                'db' => $db ?: null,
                'remarks' => $remarks ?: null,
                'win_tm' => is_numeric($winTm) ? (float) $winTm : null,
                'going' => $going ?: null,
            ];
        }

        return $history;
    }

    private function extractTrap(DOMXPath $xpath, DOMElement $node): ?int
    {
        $alt = $this->firstAttributeRelative(
            $xpath,
            $node,
            ".//img[contains(@src,'dog') or contains(@src,'traps_uk')]",
            'alt'
        );

        if ($alt && preg_match('/^[1-6]$/', trim($alt))) {
            return (int) $alt;
        }

        $src = $this->firstAttributeRelative(
            $xpath,
            $node,
            ".//img[contains(@src,'dog') or contains(@src,'traps_uk')]",
            'src'
        );

        if ($src && preg_match('/dog([1-6])\.svg/i', $src, $m)) {
            return (int) $m[1];
        }

        return null;
    }

    private function normaliseShortDate(?string $date): ?string
    {
        $date = trim((string) $date);

        if (!preg_match('/^(\d{1,2})\/(\d{1,2})$/', $date, $m)) {
            return null;
        }

        $day = (int) $m[1];
        $month = (int) $m[2];
        $year = (int) date('Y');

        if (!checkdate($month, $day, $year)) {
            return null;
        }

        return sprintf('%04d-%02d-%02d', $year, $month, $day);
    }

    private function positionToInt(?string $position): ?int
    {
        if (!$position) {
            return null;
        }

        if (preg_match('/(\d+)/', $position, $m)) {
            return (int) $m[1];
        }

        return null;
    }

    private function firstText(DOMXPath $xpath, string $query): ?string
    {
        $nodes = $xpath->query($query);

        if (!$nodes || $nodes->length === 0) {
            return null;
        }

        return $this->cleanText($nodes->item(0)->textContent ?? '');
    }

    private function firstTextRelative(DOMXPath $xpath, DOMNode $context, string $query): ?string
    {
        $nodes = $xpath->query($query, $context);

        if (!$nodes || $nodes->length === 0) {
            return null;
        }

        return $this->cleanText($nodes->item(0)->textContent ?? '');
    }

    private function firstAttributeRelative(DOMXPath $xpath, DOMNode $context, string $query, string $attr): ?string
    {
        $nodes = $xpath->query($query, $context);

        if (!$nodes || $nodes->length === 0) {
            return null;
        }

        $node = $nodes->item(0);

        if (!$node instanceof DOMElement) {
            return null;
        }

        $value = $node->getAttribute($attr);

        return $value !== '' ? trim($value) : null;
    }

    private function texts(DOMXPath $xpath, string $query): array
    {
        $nodes = $xpath->query($query);
        $out = [];

        if (!$nodes) {
            return $out;
        }

        foreach ($nodes as $node) {
            $text = $this->cleanText($node->textContent ?? '');

            if ($text !== '') {
                $out[] = $text;
            }
        }

        return $out;
    }

    private function textsRelative(DOMXPath $xpath, DOMNode $context, string $query): array
    {
        $nodes = $xpath->query($query, $context);
        $out = [];

        if (!$nodes) {
            return $out;
        }

        foreach ($nodes as $node) {
            $text = $this->cleanText($node->textContent ?? '');
            $out[] = $text;
        }

        return $out;
    }

    private function cleanText(string $text): string
    {
        $text = html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8');
        $text = preg_replace('/\s+/u', ' ', $text);

        return trim((string) $text);
    }
}



app/db.php: 

<?php

declare(strict_types=1);

function gm_config(): array
{
    static $config = null;

    if ($config === null) {
        $config = require __DIR__ . '/config.php';

        if (!empty($config['timezone'])) {
            date_default_timezone_set($config['timezone']);
        }
    }

    return $config;
}

function gm_db(): PDO
{
    static $pdo = null;

    if ($pdo instanceof PDO) {
        return $pdo;
    }

    $config = gm_config();

    $dsn = sprintf(
        'mysql:host=%s;dbname=%s;charset=utf8mb4',
        $config['db_host'],
        $config['db_name']
    );

    $pdo = new PDO($dsn, $config['db_user'], $config['db_pass'], [
        PDO::ATTR_ERRMODE => PDO::ERRMODE_EXCEPTION,
        PDO::ATTR_DEFAULT_FETCH_MODE => PDO::FETCH_ASSOC,
        PDO::ATTR_EMULATE_PREPARES => false,
    ]);

    return $pdo;
}

function gm_slug(string $value): string
{
    $value = trim(mb_strtolower($value, 'UTF-8'));
    $ascii = @iconv('UTF-8', 'ASCII//TRANSLIT//IGNORE', $value);

    if ($ascii !== false) {
        $value = $ascii;
    }

    $value = preg_replace('/[^a-z0-9]+/i', '-', $value);
    $value = trim((string) $value, '-');

    return $value !== '' ? $value : 'galgo-' . bin2hex(random_bytes(4));
}

Fork Edit

Comments 0

No comments yet. Be the first.

Hello Guest! Paste is for source code and general debugging text.

Recent Pastes

Codetest

Codetest

Hello Guest! Paste is for source code and general debugging text.

We use cookies. To comply with GDPR in the EU and the UK we have to show you these.