ext/standard: Check for UTF-8 first when resolving the charset of html functions

determine_charset() ran strlen() and walked the charset map with case-insensitive
compares on every call, while "UTF-8" (explicit or from default_charset) is what
nearly every call asks for. It now checks for "utf-8" in any case first and moves
the walk to a separate function, which skips strlen() and compares only the entries
that start with the same letter as the hint, so other charsets resolve faster too.

Author: nicolas-grekas

NEWS

@@ -21,5 +21,7 @@ PHP                                                                        NEWS
 - Standard:
   . Improved performance of array_splice() when inserting without removing
     elements. (mehmetcansahin)
+  . Improved performance of HTML functions when the charset is UTF-8.
+    (Nicolas Grekas)
 
 <<< NOTE: Insert NEWS from last stable release here prior to actual release! >>>

UPGRADING

@@ -74,3 +74,5 @@ PHP 8.7 UPGRADE NOTES
 - Standard:
   . Improved performance of array_splice() when inserting without removing
     elements.
+  . Improved performance of htmlspecialchars(), htmlentities(),
+    html_entity_decode() and get_html_translation_table() with UTF-8.

ext/standard/html.c

@@ -356,32 +356,56 @@ PHPAPI unsigned int php_next_utf8_char(
 }
 /* }}} */
 
+/* {{{ entity_charset lookup_charset
+ * Returns the charset identifier of a non-empty charset name, or UTF-8 if it is not supported. */
+static zend_never_inline enum entity_charset lookup_charset(const char *charset_hint, bool quiet)
+{
+	const unsigned char first = zend_tolower_ascii(charset_hint[0]);
+
+	/* now walk the charset map and look for the codeset */
+	for (size_t i = 0; i < sizeof(charset_map)/sizeof(charset_map[0]); i++) {
+		const char *codeset = charset_map[i].codeset;
+
+		if (zend_tolower_ascii(codeset[0]) != first) {
+			continue;
+		}
+		/* compare up to the NUL of codeset: a mismatch stops the loop at the NUL of charset_hint at the latest */
+		for (size_t j = 1; zend_tolower_ascii(charset_hint[j]) == zend_tolower_ascii(codeset[j]); j++) {
+			if (codeset[j] == '\0') {
+				return charset_map[i].charset;
+			}
+		}
+	}
+
+	if (!quiet) {
+		php_error_docref(NULL, E_WARNING, "Charset \"%s\" is not supported, assuming UTF-8",
+				charset_hint);
+	}
+
+	return cs_utf_8;
+}
+/* }}} */
+
 /* {{{ entity_charset determine_charset
  * Returns the charset identifier based on an explicitly provided charset,
  * the internal_encoding and default_charset ini settings, or UTF-8 by default. */
 static enum entity_charset determine_charset(const char *charset_hint, bool quiet)
 {
 	if (!charset_hint || !*charset_hint) {
 		charset_hint = get_default_charset();
-	}
-
-	if (charset_hint && *charset_hint) {
-		size_t len = strlen(charset_hint);
-		/* now walk the charset map and look for the codeset */
-		for (size_t i = 0; i < sizeof(charset_map)/sizeof(charset_map[0]); i++) {
-			if (len == charset_map[i].codeset_len &&
-			    zend_binary_strcasecmp(charset_hint, len, charset_map[i].codeset, len) == 0) {
-				return charset_map[i].charset;
-			}
+		if (!charset_hint) {
+			return cs_utf_8;
 		}
+	}
 
-		if (!quiet) {
-			php_error_docref(NULL, E_WARNING, "Charset \"%s\" is not supported, assuming UTF-8",
-					charset_hint);
-		}
+	/* Check the most used charset first, as lookup_charset() would: "utf-8" in any case */
+	if (zend_tolower_ascii(charset_hint[0]) == 'u' && zend_tolower_ascii(charset_hint[1]) == 't'
+			&& zend_tolower_ascii(charset_hint[2]) == 'f' && charset_hint[3] == '-' && charset_hint[4] == '8'
+			&& charset_hint[5] == '\0') {
+		return cs_utf_8;
 	}
 
-	return cs_utf_8;
+	return lookup_charset(charset_hint, quiet);
 }
 /* }}} */
 

ext/standard/tests/strings/html_charset_utf8.phpt

@@ -0,0 +1,32 @@
+--TEST--
+Charset names that are UTF-8 or close to it
+--FILE--
+<?php
+foreach (['UTF-8', 'utf-8', 'uTf-8', 'UTF8', 'UTF-', 'UTF-8 ', 'UTF-88', 'UTF_8'] as $charset) {
+    $escaped = htmlspecialchars("\xC3\xA9<", ENT_QUOTES, $charset);
+    echo var_export($charset, true), ": $escaped\n";
+}
+
+ini_set('default_charset', 'uTF-8');
+echo htmlspecialchars("\xC3\xA9<", ENT_QUOTES), "\n";
+?>
+--EXPECTF--
+'UTF-8': é&lt;
+'utf-8': é&lt;
+'uTf-8': é&lt;
+
+Warning: htmlspecialchars(): Charset "UTF8" is not supported, assuming UTF-8 in %s on line %d
+'UTF8': é&lt;
+
+Warning: htmlspecialchars(): Charset "UTF-" is not supported, assuming UTF-8 in %s on line %d
+'UTF-': é&lt;
+
+Warning: htmlspecialchars(): Charset "UTF-8 " is not supported, assuming UTF-8 in %s on line %d
+'UTF-8 ': é&lt;
+
+Warning: htmlspecialchars(): Charset "UTF-88" is not supported, assuming UTF-8 in %s on line %d
+'UTF-88': é&lt;
+
+Warning: htmlspecialchars(): Charset "UTF_8" is not supported, assuming UTF-8 in %s on line %d
+'UTF_8': é&lt;
+é&lt;