Some progress on multi-line comments in PHP

src/Row.php

@@ -22,12 +22,14 @@ class Row {
 
 	private array $phpTokenHighlightMap = [
 		// Delimiters
-		T_CLOSE_TAG => Highlight::DELIMITER,
 		T_ARRAY => Highlight::DELIMITER,
 		T_CURLY_OPEN => Highlight::DELIMITER,
 		T_DOLLAR_OPEN_CURLY_BRACES => Highlight::DELIMITER,
 		T_OPEN_TAG => Highlight::DELIMITER,
 		T_OPEN_TAG_WITH_ECHO => Highlight::DELIMITER,
+		T_CLOSE_TAG => Highlight::DELIMITER,
+		T_START_HEREDOC => Highlight::DELIMITER,
+		T_END_HEREDOC => Highlight::DELIMITER,
 
 		// Number literals
 		T_DNUMBER => Highlight::NUMBER,
@@ -432,15 +434,23 @@ class Row {
 
 	protected function updateSyntaxPHP():void
 	{
-		if ( ! isset($this->parent->syntax->tokens))
+		$rowNum = $this->idx + 1;
+
+		if (
+			( ! isset($this->parent->syntax->tokens))
+			|| ( ! array_key_exists($rowNum, $this->parent->syntax->tokens))
+			|| $this->idx > $this->parent->numRows)
 		{
 			return;
 		}
 
-		// The index for the tokens should exist
-		$tokens = $this->parent->syntax->tokens[$this->idx + 1];
+		$tokens = $this->parent->syntax->tokens[$rowNum];
+		if (empty($tokens))
+		{
+			return;
+		}
 
-		// $inComment = ($this->idx > 0 && $this->parent->rows[$this->idx - 1]->hlOpenComment);
+		$inComment = ($this->idx > 0 && $this->parent->rows[$this->idx - 1]->hlOpenComment);
 
 		// Keep track of where you are in the line, so that
 		// multiples of the same tokens can be effectively matched
@@ -455,15 +465,65 @@ class Row {
 
 			$char = $token['char'];
 			$charLen = strlen($char);
+			if ($charLen === 0)
+			{
+				continue;
+			}
 			$charStart = strpos($this->render, $char, $offset);
 			if ($charStart === FALSE)
 			{
-				$offset++;
 				continue;
 			}
 
 			$charEnd = $charStart + $charLen;
 
+			// Comments
+			if ($inComment)
+			{
+				if (substr($this->render, $offset, 2) === '*/')
+				{
+					$inComment = FALSE;
+					array_replace_range($this->hl, $offset, 2, Highlight::ML_COMMENT);
+					$offset += 2;
+					continue;
+				}
+
+				$this->hl[$offset] = Highlight::ML_COMMENT;
+				$offset++;
+				continue;
+			}
+			if (in_array($token['type'], [T_DOC_COMMENT, T_COMMENT], TRUE))
+			{
+				// Single line comments
+				if (strpos($token['char'], '//') !== FALSE)
+				{
+					array_replace_range($this->hl, $charStart, $charLen, Highlight::COMMENT);
+					break;
+				}
+
+				// Start of multi-line comment
+				$start = strpos($this->render, '/*', $offset);
+				$inComment = strpos($this->render, '*/', $offset) === FALSE;
+				array_replace_range($this->hl, $start, strlen($char) - $offset, Highlight::ML_COMMENT);
+				$offset = $start + strlen($char) - $offset;
+
+				if ($inComment)
+				{
+					break;
+				}
+
+				continue;
+			}
+
+			// Highlight specific tokens
+			if (($token['typeName'] !== 'RAW') && array_key_exists($token['type'], $this->phpTokenHighlightMap))
+			{
+				$hl = $this->phpTokenHighlightMap[$token['type']];
+				array_replace_range($this->hl, $charStart, $charLen, $hl);
+				$offset = $charEnd;
+				continue;
+			}
+
 			// Highlight raw characters
 			if (($token['typeName'] === 'RAW') && array_key_exists($token['char'], $this->phpCharacterHighlightMap))
 			{
@@ -473,14 +533,14 @@ class Row {
 				continue;
 			}
 
-			// Highlight specific tokens
-			if (array_key_exists($token['type'], $this->phpTokenHighlightMap))
-			{
-				$hl = $this->phpTokenHighlightMap[$token['type']];
-				array_replace_range($this->hl, $charStart, $charLen, $hl);
-				$offset = $charEnd;
-				continue;
+			$offset++;
 		}
+
+		$changed = $this->hlOpenComment !== $inComment;
+		$this->hlOpenComment = $inComment;
+		if ($changed && $this->idx + 1 < $this->parent->numRows)
+		{
+			$this->parent->rows[$this->idx + 1]->updateSyntax();
 		}
 	}
 }

src/functions.php

@@ -340,6 +340,12 @@ function get_php_tokens(string $code): array
 
 		[$type, $char, $currentLine] = $token;
 
+		// Only return the first line of a multi-line token
+		if ($char !== "\n" && strpos($char, "\n") !== FALSE)
+		{
+			$char = explode("\n", $char)[0];
+		}
+
 		$current = [
 			'type' => $type,
 			'typeName' => token_name($type),

test.php

@@ -3,6 +3,13 @@
 interface Ifoo {}
 
 abstract class Foo implements Ifoo {
+	/**
+	 * @param int $a
+	 * @param float $b
+	 * @param array $c
+	 * @param callable $d
+	 * @return string
+	 */
 	abstract public function bar(int $a, float $b, array $c, callable $d): string;
 
 	protected function doNothing(): void {}
@@ -46,7 +53,7 @@ $sql = <<<SQL
 	SELECT * FROM "foo" WHERE "bar"='baz' AND id={$x};
 SQL;
 
-// Nowdoc
+/* Nowdoc */
 $template = <<<'TEMPLATE'
 	<foo>{x}</foo>
 TEMPLATE;