[DOC] Added a 4th parameter flag to htmlspecialchars() and htmlentities()

that makes the function not encode existing html entities. The feature is
disabled by default and can be activated by passing FALSE as the 4th param

diff --git a/NEWS b/NEWS
index 7fb9a4c21459ba9ce69c58d4916ae8e4a633c890..1d5c84ab2c932d194986c0ef93ee95d6a9fab5ef 100644 (file)
--- a/NEWS
+++ b/NEWS
@@ -5,6 +5,8 @@ PHP                                                                        NEWS
 - Optimized out a couple of per-request syscalls (Rasmus)
 - Optimized digest generation in md5() and sha1() functions. (Ilia)
 - Upgraded SQLite 3 to version 3.3.16 (Ilia)
+- Added a 4th parameter flag to htmlspecialchars() and htmlentities() that 
+  makes the function not encode existing html entities. (Ilia)
 - Added PDO::FETCH_KEY_PAIR mode that will fetch a 2 column result set into 
   an associated array. (Ilia)
 - Added function mysql_set_charset(). Allows connection encoding to be 

diff --git a/ext/standard/html.c b/ext/standard/html.c
index e2badee4748fa5097b266ff6a3acf52e56c5b2cb..8ac7b417c5fa96827b0317fce8cabf543676b241 100644 (file)
--- a/ext/standard/html.c
+++ b/ext/standard/html.c
@@ -1078,12 +1078,15 @@ empty_source:
 }
 /* }}} */
 
-
+PHPAPI char *php_escape_html_entities(unsigned char *old, int oldlen, int *newlen, int all, int quote_style, char *hint_charset TSRMLS_DC)
+{
+       return php_escape_html_entities_ex(old, oldlen, newlen, all, quote_style, hint_charset, 1 TSRMLS_CC);
+}
 
 
 /* {{{ php_escape_html_entities
  */
-PHPAPI char *php_escape_html_entities(unsigned char *old, int oldlen, int *newlen, int all, int quote_style, char *hint_charset TSRMLS_DC)
+PHPAPI char *php_escape_html_entities_ex(unsigned char *old, int oldlen, int *newlen, int all, int quote_style, char *hint_charset, zend_bool double_encode TSRMLS_DC)
 {
        int i, j, maxlen, len;
        char *replaced;
@@ -1145,8 +1148,34 @@ PHPAPI char *php_escape_html_entities(unsigned char *old, int oldlen, int *newle
                        int is_basic = 0;
 
                        if (this_char == '&') {
-                               memcpy(replaced + len, "&", sizeof("&") - 1);
-                               len += sizeof("&") - 1;
+                               if (double_encode) {
+encode_amp:
+                                       memcpy(replaced + len, "&", sizeof("&") - 1);
+                                       len += sizeof("&") - 1;
+                               } else {
+                                       char *e = memchr(old + i, ';', len - i);
+                                       char *s = old + i + 1;
+
+                                       if (!e || (e - s) > 10) { /* minor optimization to avoid "entities" over 10 chars in length */
+                                               goto encode_amp;
+                                       } else {
+                                               if (*s == '#') { /* numeric entities */
+                                                       s++;
+                                                       while (s < e) {
+                                                               if (!isdigit(*s++)) {
+                                                                       goto encode_amp;
+                                                               }
+                                                       }
+                                               } else { /* text entities */
+                                                       while (s < e) {
+                                                               if (!isalnum(*s++)) {
+                                                                       goto encode_amp;
+                                                               }
+                                                       }
+                                               }
+                                               replaced[len++] = '&';
+                                       }
+                               }
                                is_basic = 1;
                        } else {
                                for (j = 0; basic_entities[j].charcode != 0; j++) {
@@ -1193,12 +1222,13 @@ static void php_html_entities(INTERNAL_FUNCTION_PARAMETERS, int all)
        int len;
        long quote_style = ENT_COMPAT;
        char *replaced;
+       zend_bool double_encode = 1;
 
-       if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|ls", &str, &str_len, &quote_style, &hint_charset, &hint_charset_len) == FAILURE) {
+       if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|lsb", &str, &str_len, &quote_style, &hint_charset, &hint_charset_len, &double_encode) == FAILURE) {
                return;
        }
 
-       replaced = php_escape_html_entities(str, str_len, &len, all, quote_style, hint_charset TSRMLS_CC);
+       replaced = php_escape_html_entities_ex(str, str_len, &len, all, quote_style, hint_charset, double_encode TSRMLS_CC);
        RETVAL_STRINGL(replaced, len, 0);
 }
 /* }}} */

diff --git a/ext/standard/html.h b/ext/standard/html.h
index 3e896e18b5c8bc92d9ace5fe05b5e642a792c300..fec44bb0e4500dd193d4953bfc2c87bf10bee737 100644 (file)
--- a/ext/standard/html.h
+++ b/ext/standard/html.h
@@ -38,6 +38,7 @@ PHP_FUNCTION(html_entity_decode);
 PHP_FUNCTION(get_html_translation_table);
 
 PHPAPI char *php_escape_html_entities(unsigned char *old, int oldlen, int *newlen, int all, int quote_style, char *hint_charset TSRMLS_DC);
+PHPAPI char *php_escape_html_entities_ex(unsigned char *old, int oldlen, int *newlen, int all, int quote_style, char *hint_charset, zend_bool double_encode TSRMLS_DC);
 PHPAPI char *php_unescape_html_entities(unsigned char *old, int oldlen, int *newlen, int all, int quote_style, char *hint_charset TSRMLS_DC);
 
 #endif /* HTML_H */

diff --git a/ext/standard/tests/strings/htmlentities18.phpt b/ext/standard/tests/strings/htmlentities18.phpt
new file mode 100644 (file)
index 0000000..f171ada
--- /dev/null
+++ b/ext/standard/tests/strings/htmlentities18.phpt
@@ -0,0 +1,31 @@
+--TEST--
+htmlentities() / htmlspecialchars() "don't double encode" flag support
+--FILE--
+<?php
+$tests = array(
+       "abc",
+       "abc&amp;sfdsa",
+       "test&#043;s &amp; some more &#68;",
+       "&; &amp &#a; &9;",
+       "&kffjadfdhsjfhjasdhffasdfas;",
+       "&#8787978789",
+);
+
+foreach ($tests as $test) {
+       var_dump(htmlentities($test, ENT_QUOTES, NULL, FALSE));
+       var_dump(htmlspecialchars($test, ENT_QUOTES, NULL, FALSE));
+}
+?>
+--EXPECT--
+string(3) "abc"
+string(3) "abc"
+string(13) "abc&amp;sfdsa"
+string(13) "abc&amp;sfdsa"
+string(33) "test&#043;s &amp; some more &#68;"
+string(33) "test&#043;s &amp; some more &#68;"
+string(20) "&; &amp;amp &#a; &9;"
+string(20) "&; &amp;amp &#a; &9;"
+string(32) "&amp;kffjadfdhsjfhjasdhffasdfas;"
+string(32) "&amp;kffjadfdhsjfhjasdhffasdfas;"
+string(16) "&amp;#8787978789"
+string(16) "&amp;#8787978789"