Fix tokenizer infinite loop and OOM on invalid bytes

src/Connection/ImapTokenizer.php

@@ -387,6 +387,20 @@ protected function readAtom(): Atom
             $this->advance();
         }
 
+        // If no value was read, we will throw an exception since
+        // an atom must contain at least one valid character.
+        if ($value === '') {
+            if ($char === null) {
+                throw new ImapStreamException('Unexpected end of stream while reading atom');
+            }
+
+            throw new ImapParserException(sprintf(
+                'Unexpected byte 0x%02X in response at buffer offset %d',
+                ord($char),
+                $this->position
+            ));
+        }
+
         if (strcasecmp($value, 'NIL') === 0) {
             return new Nil($value);
         }
@@ -437,7 +451,9 @@ protected function ensureBuffer(int $length): void
         while ((strlen($this->buffer) - $this->position) < $length) {
             $data = $this->stream->fgets();
 
-            if ($data === false) {
+            // If the stream did not return any data, we will stop
+            // filling the buffer until another read is attempted.
+            if ($data === false || $data === '') {
                 return;
             }
 

tests/Integration/MessagesTest.php

@@ -59,15 +59,21 @@ function folder(): Folder
 test('first or fail', function () {
     $folder = folder();
 
-    expect(fn () => $folder->messages()->firstOrFail())->toThrow(ItemNotFoundException::class);
-
     $uid = $folder->messages()->append(
         new DraftMessage(from: 'foo@example.com', text: 'hello world'),
     );
 
-    expect($folder->messages()->firstOrFail()->uid())->toBe($uid);
+    $message = $folder->messages()->firstOrFail();
+
+    expect($message->uid())->toBe($uid);
 });
 
+test('first or fail throws exception', function () {
+    $folder = folder();
+
+    $folder->messages()->firstOrFail();
+})->throws(ItemNotFoundException::class);
+
 test('find', function () {
     $folder = folder();
 
@@ -93,13 +99,17 @@ function folder(): Folder
         ),
     );
 
-    expect($folder->messages()->findOrFail($uid))->toBeInstanceOf(Message::class);
+    $message = $folder->messages()->findOrFail($uid);
 
-    expect(function () use ($folder) {
-        $folder->messages()->findOrFail(999);
-    })->toThrow(ItemNotFoundException::class);
+    expect($message)->toBeInstanceOf(Message::class);
 });
 
+test('find or fail throws exception', function () {
+    $folder = folder();
+
+    $folder->messages()->findOrFail(999);
+})->throws(ItemNotFoundException::class);
+
 test('get without fetches', function () {
     $folder = folder();
 

tests/Unit/Connection/ImapTokenizerTest.php

@@ -207,6 +207,74 @@
     $tokenizer->nextToken();
 })->throws(ImapParserException::class);
 
+test('tokenizer throws exception for an unexpected byte instead of returning an empty atom', function () {
+    $stream = new FakeStream;
+    $stream->open();
+
+    $stream->feedRaw("TAG1 OK L\x80GIN completed\r\n");
+
+    $tokenizer = new ImapTokenizer($stream);
+
+    expect($tokenizer->nextToken()->value)->toBe('TAG1');
+    expect($tokenizer->nextToken()->value)->toBe('OK');
+    expect($tokenizer->nextToken()->value)->toBe('L');
+
+    $tokenizer->nextToken();
+})->throws(ImapParserException::class, 'Unexpected byte 0x80 in response at buffer offset 9');
+
+test('tokenizer throws exception for an unexpected atom delimiter', function (string $delimiter, string $hex) {
+    $stream = new FakeStream;
+    $stream->open();
+
+    $stream->feedRaw($delimiter);
+
+    $tokenizer = new ImapTokenizer($stream);
+
+    expect(fn () => $tokenizer->nextToken())->toThrow(
+        ImapParserException::class,
+        "Unexpected byte 0x{$hex} in response at buffer offset 0"
+    );
+})->with([
+    'closing curly brace' => ['}', '7D'],
+    'closing angle bracket' => ['>', '3E'],
+]);
+
+test('tokenizer stops filling the buffer when the stream returns an empty string', function () {
+    $stream = new class extends FakeStream
+    {
+        public int $reads = 0;
+
+        public function fgets(): string|false
+        {
+            if ($this->reads++ > 1) {
+                throw new RuntimeException('The tokenizer continued reading after an empty stream read');
+            }
+
+            return '';
+        }
+    };
+
+    $stream->open();
+
+    $tokenizer = new ImapTokenizer($stream);
+
+    expect($tokenizer->nextToken())->toBeNull();
+    expect($stream->reads)->toBe(2);
+});
+
+test('tokenizer does not treat zero stream data as empty', function () {
+    $stream = new FakeStream;
+    $stream->open();
+
+    $stream->feedRaw(['0', "\r\n"]);
+
+    $tokenizer = new ImapTokenizer($stream);
+
+    $token = $tokenizer->nextToken();
+
+    expect($token->value)->toBe('0');
+});
+
 test('tokenizer throws exception for unterminated quoted string', function () {
     $stream = new FakeStream;
     $stream->open();