feat(автоподбор): HtmlPhoneScanner — номера из кода (tel/schema/microdata/тело/email)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

app/app/Services/Autopodbor/Agent/Extract/HtmlPhoneScanner.php

@@ -0,0 +1,75 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Services\Autopodbor\Agent\Extract;
+
+final class HtmlPhoneScanner
+{
+    /** @return array{code: array<string,list<string>>, body: array<string,int>, emails: list<string>} */
+    public function scan(string $html): array
+    {
+        $code = [];
+        $add = function (string $raw, string $slot) use (&$code): void {
+            $n = $this->normalizeMaybe($raw);
+            if ($n === null) {
+                return;
+            }
+            $code[$n] ??= [];
+            if (! in_array($slot, $code[$n], true)) {
+                $code[$n][] = $slot;
+            }
+        };
+
+        if (preg_match_all('/tel:([+0-9()\s-]{7,})/i', $html, $m)) {
+            foreach ($m[1] as $x) {
+                $add($x, 'tel');
+            }
+        }
+        if (preg_match_all('/"telephone"\s*:\s*"([^"]+)"/i', $html, $m)) {
+            foreach ($m[1] as $x) {
+                $add($x, 'schema');
+            }
+        }
+        if (preg_match_all('/itemprop=["\']telephone["\'][^>]*content=["\']([^"\']+)/i', $html, $m)) {
+            foreach ($m[1] as $x) {
+                $add($x, 'microdata');
+            }
+        }
+
+        $body = [];
+        if (preg_match_all('/(?:\+7|8)[\s(\-]*\d{3}[\s)\-]*\d{3}[\s\-]*\d{2}[\s\-]*\d{2}/', $html, $m)) {
+            foreach ($m[0] as $x) {
+                $n = $this->normalizeMaybe($x);
+                if ($n !== null) {
+                    $body[$n] = ($body[$n] ?? 0) + 1;
+                }
+            }
+        }
+
+        $emails = [];
+        if (preg_match_all('/([a-z0-9._%+-]+)@[a-z0-9.-]+\.[a-z]{2,}/i', $html, $m)) {
+            foreach ($m[1] as $local) {
+                $d = preg_replace('/\D+/', '', $local) ?? '';
+                if (strlen($d) >= 7) {
+                    $emails[] = $d;
+                }
+            }
+        }
+
+        return ['code' => $code, 'body' => $body, 'emails' => $emails];
+    }
+
+    private function normalizeMaybe(string $raw): ?string
+    {
+        $digits = preg_replace('/\D+/', '', $raw) ?? '';
+        if (strlen($digits) === 11 && ($digits[0] === '8' || $digits[0] === '7')) {
+            return '7'.substr($digits, 1);
+        }
+        if (strlen($digits) === 10) {
+            return '7'.$digits;
+        }
+
+        return null;
+    }
+}

app/tests/Unit/Autopodbor/Extract/HtmlPhoneScannerTest.php

@@ -0,0 +1,26 @@
+<?php
+
+use App\Services\Autopodbor\Agent\Extract\HtmlPhoneScanner;
+
+it('берёт номера из tel:-ссылок', function () {
+    $html = '<a href="tel:+7 (843) 203-25-33">звонок</a><a href="tel:88432452533">2</a>';
+    $r = (new HtmlPhoneScanner)->scan($html);
+    expect($r['code'])->toHaveKey('78432032533')
+        ->and($r['code']['78432032533'])->toContain('tel')
+        ->and($r['code'])->toHaveKey('78432452533');
+});
+
+it('берёт номера из schema.org и microdata', function () {
+    $html = '<script type="application/ld+json">{"telephone":"+7(843)203-25-33"}</script>'
+          .'<span itemprop="telephone" content="+78432452533">x</span>';
+    $r = (new HtmlPhoneScanner)->scan($html);
+    expect($r['code']['78432032533'])->toContain('schema')
+        ->and($r['code']['78432452533'])->toContain('microdata');
+});
+
+it('считает вхождения в тело и берёт e-mail-цифры', function () {
+    $html = 'тел 8(843)203-25-33, ещё 8(843)203-25-33. почта 2032533@mail.ru';
+    $r = (new HtmlPhoneScanner)->scan($html);
+    expect($r['body']['78432032533'])->toBe(2)
+        ->and($r['emails'])->toContain('2032533');
+});