Fix phpGH-17486: Incorrect error line numbers reported in Dom\HTMLDocument::createFromString

nielsdos · nielsdos · commit c7a8195deba8 · 2025-01-16T21:01:47.000+01:00
diff --git a/ext/dom/html_document.c b/ext/dom/html_document.c
@@ -880,6 +880,11 @@ PHP_METHOD(Dom_HTMLDocument, createFromString)
 		if (!result) {
 			goto fail_oom;
 		}
+
+		/* In the string case we have a single buffer that acts as a sliding window.
+		 * The `current_input_characters` field starts pointing at the start of the buffer, but needs to slide along the
+		 * sliding window as well. */
+		application_data.current_input_characters += chunk_size;
 	}
 
 	if (!dom_parse_decode_encode_finish(&ctx, document, parser, &decoding_encoding_ctx, &tokenizer_error_offset, &tree_error_offset)) {
diff --git a/ext/dom/tests/modern/html/parser/gh17486.phpt b/ext/dom/tests/modern/html/parser/gh17486.phpt
@@ -0,0 +1,37 @@
+--TEST--
+GH-17486 (Incorrect error line numbers reported in Dom\HTMLDocument::createFromString)
+--EXTENSIONS--
+dom
+--CREDITS--
+xPaw
+--FILE--
+<?php
+
+$repeated = str_repeat('a', 50000);
+
+$html = <<<HTML
+<!DOCTYPE html>
+<html lang="en">
+	<body>
+		<svg>
+			<path d="{$repeated}" />
+		</svg>
+		<div>&#xa;</div>
+	</body>
+</html>
+HTML;
+
+\Dom\HTMLDocument::createFromString($html);
+
+file_put_contents(__DIR__ . '/gh17486.tmp', $html);
+\Dom\HTMLDocument::createFromFile(__DIR__ . '/gh17486.tmp');
+
+?>
+--CLEAN--
+<?php
+@unlink(__DIR__ . '/gh17486.tmp');
+?>
+--EXPECTF--
+Warning: Dom\HTMLDocument::createFromString(): tokenizer error control-character-reference in Entity, line: 7, column: 9 in %s on line %d
+
+Warning: Dom\HTMLDocument::createFromFile(): tokenizer error control-character-reference in %sgh17486.tmp, line: 7, column: 9 in %s on line %d

Original file line number	Diff line number	Diff line change
`@@ -880,6 +880,11 @@ PHP_METHOD(Dom_HTMLDocument, createFromString)`
`880`	`880`	`if (!result) {`
`881`	`881`	`goto fail_oom;`
`882`	`882`	`}`
	`883`	`+`
	`884`	`+ /* In the string case we have a single buffer that acts as a sliding window.`
	`885`	+ * The `current_input_characters` field starts pointing at the start of the buffer, but needs to slide along the
	`886`	`+ * sliding window as well. */`
	`887`	`+ application_data.current_input_characters += chunk_size;`
`883`	`888`	`}`
`884`	`889`
`885`	`890`	`if (!dom_parse_decode_encode_finish(&ctx, document, parser, &decoding_encoding_ctx, &tokenizer_error_offset, &tree_error_offset)) {`