Ref #77388: Don't pass BAD_ESCAPE_IS_LITERAL

This option is considered dangerous and unwanted. To allow for more
graceful migration don't error on now ignored X modifier.

Closes GH-4430.

4 files changed, 13 insertions, 23 deletions

diff --git a/NEWS b/NEWS
index fd42944027..c69da31e4c 100644
--- a/NEWS
+++ b/NEWS
@@ -11,6 +11,7 @@ PHP                                                                        NEWS
   . Fixed bug #65547 (Default value for sunrise/sunset zenith still wrong).
     (cmb)
   . Fixed bug #69044 (discrepency between time and microtime). (krakjoe)
+
 - GD:
   . Removed deprecated image2wbmp(). (cmb)
   . Removed deprecated png2wbmp() and jpeg2wbmp(). (cmb)
@@ -27,6 +28,9 @@ PHP                                                                        NEWS
 - mysqlnd:
   . Fixed #60594 (mysqlnd exposes 160 lines of stats in phpinfo). (PeeHaa)
 
+- PCRE:
+  . Don't ignore invalid escape sequences. (sjon)
+
 - PDO:
   . Fixed bug #77849 (Disable cloning of PDO handle/connection objects).
     (camporter)
diff --git a/UPGRADING b/UPGRADING
index 4d0b05839c..6f6d8a57a5 100644
--- a/UPGRADING
+++ b/UPGRADING
@@ -145,6 +145,11 @@ PHP 8.0 UPGRADE NOTES
     as a string instead of an ASCII codepoint. The previous behavior may be
     restored with an explicit call to chr().
 
+- PCRE:
+  . When passing invalid escape sequences they are no longer intepreted as
+    literals. This behaviour previously required the X modifier - which is
+    now ignored.
+
 - PDO:
   . The method PDOStatement::setFetchMode() now accepts the following signature:
 
diff --git a/ext/pcre/php_pcre.c b/ext/pcre/php_pcre.c
index cae276134c..6c265df485 100644
--- a/ext/pcre/php_pcre.c
+++ b/ext/pcre/php_pcre.c
@@ -50,7 +50,6 @@ struct _pcre_cache_entry {
 	uint32_t capture_count;
 	uint32_t name_count;
 	uint32_t compile_options;
-	uint32_t extra_compile_options;
 	uint32_t refcount;
 };
 
@@ -167,7 +166,6 @@ static void php_pcre_free(void *block, void *data)
 	pefree(block, 1);
 }/*}}}*/
 
-#define PHP_PCRE_DEFAULT_EXTRA_COPTIONS PCRE2_EXTRA_BAD_ESCAPE_IS_LITERAL
 #define PHP_PCRE_PREALLOC_MDATA_SIZE 32
 
 static void php_pcre_init_pcre2(uint8_t jit)
@@ -188,12 +186,6 @@ static void php_pcre_init_pcre2(uint8_t jit)
 		}
 	}
 
-	/* XXX The 'X' modifier is the default behavior in PCRE2. This option is
-		called dangerous in the manual, as typos in patterns can cause
-		unexpected results. We might want to to switch to the default PCRE2
-		behavior, too, thus causing a certain BC break. */
-	pcre2_set_compile_extra_options(cctx, PHP_PCRE_DEFAULT_EXTRA_COPTIONS);
-
 	if (!mctx) {
 		mctx = pcre2_match_context_create(gctx);
 		if (!mctx) {
@@ -569,7 +561,6 @@ PHPAPI pcre_cache_entry* pcre_get_compiled_regex_cache(zend_string *regex)
 {
 	pcre2_code			*re = NULL;
 	uint32_t			 coptions = 0;
-	uint32_t			 extra_coptions = PHP_PCRE_DEFAULT_EXTRA_COPTIONS;
 	PCRE2_UCHAR	         error[128];
 	PCRE2_SIZE           erroffset;
 	int                  errnumber;
@@ -703,8 +694,8 @@ PHPAPI pcre_cache_entry* pcre_get_compiled_regex_cache(zend_string *regex)
 			case 'A':	coptions |= PCRE2_ANCHORED;		break;
 			case 'D':	coptions |= PCRE2_DOLLAR_ENDONLY;break;
 			case 'S':	/* Pass. */					break;
+			case 'X':	/* Pass. */					break;
 			case 'U':	coptions |= PCRE2_UNGREEDY;		break;
-			case 'X':	extra_coptions &= ~PCRE2_EXTRA_BAD_ESCAPE_IS_LITERAL;			break;
 			case 'u':	coptions |= PCRE2_UTF;
 	/* In  PCRE,  by  default, \d, \D, \s, \S, \w, and \W recognize only ASCII
        characters, even in UTF-8 mode. However, this can be changed by setting
@@ -767,19 +758,9 @@ PHPAPI pcre_cache_entry* pcre_get_compiled_regex_cache(zend_string *regex)
 		pcre2_set_character_tables(cctx, tables);
 	}
 
-	/* Set extra options for the compile context. */
-	if (PHP_PCRE_DEFAULT_EXTRA_COPTIONS != extra_coptions) {
-		pcre2_set_compile_extra_options(cctx, extra_coptions);
-	}
-
 	/* Compile pattern and display a warning if compilation failed. */
 	re = pcre2_compile((PCRE2_SPTR)pattern, pattern_len, coptions, &errnumber, &erroffset, cctx);
 
-	/* Reset the compile context extra options to default. */
-	if (PHP_PCRE_DEFAULT_EXTRA_COPTIONS != extra_coptions) {
-		pcre2_set_compile_extra_options(cctx, PHP_PCRE_DEFAULT_EXTRA_COPTIONS);
-	}
-
 	if (re == NULL) {
 		if (key != regex) {
 			zend_string_release_ex(key, 0);
@@ -823,7 +804,6 @@ PHPAPI pcre_cache_entry* pcre_get_compiled_regex_cache(zend_string *regex)
 	new_entry.re = re;
 	new_entry.preg_options = poptions;
 	new_entry.compile_options = coptions;
-	new_entry.extra_compile_options = extra_coptions;
 	new_entry.refcount = 0;
 
 	rc = pcre2_pattern_info(re, PCRE2_INFO_CAPTURECOUNT, &new_entry.capture_count);
diff --git a/ext/pcre/tests/pcre_extra.phpt b/ext/pcre/tests/pcre_extra.phpt
index 2bee408fbf..a1102424e8 100644
--- a/ext/pcre/tests/pcre_extra.phpt
+++ b/ext/pcre/tests/pcre_extra.phpt
@@ -1,5 +1,5 @@
 --TEST--
-X (PCRE_EXTRA) modififer
+X (PCRE_EXTRA) modififier is ignored (no error, no change)
 --FILE--
 <?php
 
@@ -8,7 +8,8 @@ var_dump(preg_match('/\y/X', '\y'));
 
 ?>
 --EXPECTF--
-int(1)
+Warning: preg_match(): Compilation failed: unrecognized character follows \ at offset 1 in %spcre_extra.php on line 3
+bool(false)
 
 Warning: preg_match(): Compilation failed: unrecognized character follows \ at offset 1 in %spcre_extra.php on line 4
 bool(false)