Another infinite loop in Tokenizer caused by Data

lib/Data.php

@@ -100,19 +100,21 @@ class Data {
     public function unconsume(int $length = 1) {
         assert($length > 0, new Exception(Exception::DATA_INVALID_DATA_CONSUMPTION_LENGTH, $length));
 
-        $this->data->seek(0 - $length);
+        if ($this->data->peekChar($length) !== '') {
+            $this->data->seek(0 - $length);
 
-        $string = $this->data->peekChar($length);
-        $numOfNewlines = substr_count($string, "\n");
+            $string = $this->data->peekChar($length);
+            $numOfNewlines = substr_count($string, "\n");
 
-        if ($numOfNewlines > 0) {
-            $this->_line -= $numOfNewlines;
+            if ($numOfNewlines > 0) {
+                $this->_line -= $numOfNewlines;
 
-            $count = $this->newlines;
-            $index = count($this->newlines) - ($numOfNewlines - 1);
-            $this->_column = 1 + (($count > 0 && isset($this->newlines[$index])) ? $this->data->posChar() - $this->newlines[$index] : $this->data->posChar());
-        } else {
-            $this->_column -= $length;
+                $count = $this->newlines;
+                $index = count($this->newlines) - ($numOfNewlines - 1);
+                $this->_column = 1 + (($count > 0 && isset($this->newlines[$index])) ? $this->data->posChar() - $this->newlines[$index] : $this->data->posChar());
+            } else {
+                $this->_column -= $length;
+            }
         }
 
         if (self::$debug) {

lib/ParseError.php

@@ -58,8 +58,8 @@ class ParseError {
         assert(count($arg) === $count, new Exception(Exception::INCORRECT_PARAMETERS_FOR_MESSAGE, $count));
 
         if ($count > 0) {
-            // Convert newlines and tabs in the arguments to words to better express what they
-            // are.
+            // Convert newlines and tabs in the arguments to words to better
+            // express what they are.
             $arg = array_map(function($value) {
                 if ($value === "\n") {
                     return 'Newline';
@@ -76,7 +76,8 @@ class ParseError {
             $message = sprintf($message, ...$arg);
         }
         // Wrap with preamble and location
-        // TODO: the file path should be middle-elided when necessary so that the message does not exceed 1024 bytes
+        // TODO: the file path should be middle-elided when necessary so that
+        // the message does not exceed 1024 bytes
         $message = sprintf("HTML5 Parse Error: \"%s\" in %s", $message, $file);
         if ($line) {
             $message .= sprintf(" on line %s, column %s", $line, $column);

lib/Tokenizer.php

@@ -366,14 +366,7 @@ class Tokenizer {
                 # U+003F QUESTION MARK (?)
                 elseif ($char === '?') {
                     # Parse error. Switch to the bogus comment state.
-
-                    // Making errors more expressive.
-                    if ($char !== '') {
-                        $this->error(ParseError::TAG_NAME_EXPECTED);
-                    } else {
-                        $this->error(ParseError::UNEXPECTED_EOF);
-                    }
-
+                    $this->error(ParseError::TAG_NAME_EXPECTED);
                     $this->state = self::BOGUS_COMMENT_STATE;
                 }
                 # Anything else