Fix remaining error positions

lib/Data.php

@@ -220,11 +220,15 @@ class Data {
 
     /** Returns an indexed array with the line and column positions of the requested offset from the current position */
     public function whereIs(int $relativePos): array {
+        if ($this->eof) {
+            $relativePos++;
+            if ($this->astrals[$this->data->posChar()] ?? false) {
+                $relativePos++;
+            }
+        }
         if ($relativePos === 0) {
             if (!$this->_column && $this->_line > 1) {
                 return [$this->_line - 1, $this->newlines[$this->data->posChar()] + 1];
-            } elseif ($this->astrals[$this->data->posChar()] ?? false) {
-                return [$this->_line, $this->_column + 1];
             } else {
                 return [$this->_line, $this->_column];
             }

lib/ParseError.php

@@ -146,12 +146,6 @@ class ParseError {
     ];
 
     const REPORT_OFFSETS = [
-        self::EOF_IN_TAG                                       => 1,
-        self::EOF_IN_COMMENT                                   => 1,
-        self::EOF_IN_DOCTYPE                                   => 1,
-        self::EOF_BEFORE_TAG_NAME                              => 1,
-        self::EOF_IN_SCRIPT_HTML_COMMENT_LIKE_TEXT             => 1,
-        self::EOF_IN_CDATA                                     => 1,
         self::INCORRECTLY_OPENED_COMMENT                       => 1,
         self::SURROGATE_CHARACTER_REFERENCE                    => 1,
         self::CHARACTER_REFERENCE_OUTSIDE_UNICODE_RANGE        => 1,
@@ -159,7 +153,8 @@ class ParseError {
         self::ABSENCE_OF_DIGITS_IN_NUMERIC_CHARACTER_REFERENCE => 1,
         self::NULL_CHARACTER_REFERENCE                         => 1,
         self::MISSING_SEMICOLON_AFTER_CHARACTER_REFERENCE      => 1,
-
+        self::CONTROL_CHARACTER_REFERENCE                      => 1,
+        self::UNKNOWN_NAMED_CHARACTER_REFERENCE                => 1,
     ];
 
     public function setHandler() {

tests/cases/TestTokenizer.php

@@ -189,6 +189,32 @@ class TestTokenizer extends \PHPUnit\Framework\TestCase {
             case ["<!\u{B}", ["Data state"]]:
                 $test['errors'] = array_reverse($test['errors']);
                 break;
+            // eof-in-<whatever> positions in some tests don't make sense
+            // https://github.com/html5lib/html5lib-tests/issues/125
+            case ["", ["CDATA section state"]]:
+                // there is no position 2
+                $test['errors'][0]['col']--;
+                break;
+            case ["\u{A}", ["CDATA section state"]]:
+                // the line break is, for some reason, not counted in the test
+                $test['errors'][0]['line']++;
+                $test['errors'][0]['col'] = 1;
+                break;
+            case ["<!----!\r\n>", ["Data state"]]:
+            case ["<!----!\n>", ["Data state"]]:
+            case ["<!----!\r>", ["Data state"]]:
+                // the line break is, for some reason, not counted in the test
+                $test['errors'][0]['line']++;
+                $test['errors'][0]['col'] = 2;
+                break;
+            case ["<!----! >", ["Data state"]]:
+                $test['errors'][0]['col']++;
+                break;
+            case [hex2bin("f4808080"), ["CDATA section state"]]:
+            case [hex2bin("3bf4808080"), ["CDATA section state"]]:
+                // malpaired surrogates count as two characters
+                $test['errors'][0]['col']++;
+                break;
         }
     }
 }

tests/cases/TestTreeConstructor.php

@@ -91,7 +91,7 @@ class TestTreeConstructor extends \PHPUnit\Framework\TestCase {
         $this->assertEquals($exp, $act, $treeBuilder->debugLog);
         if ($errors !== false) {
             // If $errors is false, the test does not include errors when there are in fact errors
-            //$this->assertCount(sizeof($errors), $actualErrors, var_export($errors, true).var_export($actualErrors, true));
+            $this->assertCount(sizeof($errors), $actualErrors, var_export($errors, true).var_export($actualErrors, true));
         }
     }