MFH: fixed #40685: removed '&' in mb_decode_numericentity().

diff --git a/ext/mbstring/libmbfl/mbfl/mbfilter.c b/ext/mbstring/libmbfl/mbfl/mbfilter.c
index 34e33644b5d1906b17e7e32c2b1d258950bdc0ef..8f2c8bb14e5e28f2c0be5eb1696cbec2ad1a3612 100644 (file)
--- a/ext/mbstring/libmbfl/mbfl/mbfilter.c
+++ b/ext/mbstring/libmbfl/mbfl/mbfilter.c
@@ -2702,6 +2702,53 @@ collector_decode_htmlnumericentity(int c, void *data)
        return c;
 }
 
+int mbfl_filt_decode_htmlnumericentity_flush(mbfl_convert_filter *filter)
+{
+       struct collector_htmlnumericentity_data *pc = (struct collector_htmlnumericentity_data *)filter;
+       int n, s, r, d;
+
+       if (pc->status) {
+               switch (pc->status) {
+               case 1: /* '&' */
+                       (*pc->decoder->filter_function)(0x26, pc->decoder);             /* '&' */
+                       break;
+               case 2: /* '#' */
+                       (*pc->decoder->filter_function)(0x26, pc->decoder);             /* '&' */
+                       (*pc->decoder->filter_function)(0x23, pc->decoder);             /* '#' */
+                       break;
+               case 3: /* '0'-'9' */
+                       (*pc->decoder->filter_function)(0x26, pc->decoder);             /* '&' */
+                       (*pc->decoder->filter_function)(0x23, pc->decoder);             /* '#' */
+
+                       s = pc->cache;
+                       r = 1;
+                       n = pc->digit;
+                       while (n > 0) {
+                               r *= 10;
+                               n--;
+                       }
+                       s %= r;
+                       r /= 10;
+                       while (r > 0) {
+                               d = s/r;
+                               s %= r;
+                               r /= 10;
+                               (*pc->decoder->filter_function)(mbfl_hexchar_table[d], pc->decoder);
+                       }
+
+                       break;
+               default:
+                       break;
+               }
+       }
+
+       pc->status = 0;
+       pc->cache = 0;
+       pc->digit = 0;
+
+       return 0;
+}
+
 mbfl_string *
 mbfl_html_numeric_entity(
     mbfl_string *string,
@@ -2739,7 +2786,8 @@ mbfl_html_numeric_entity(
                encoder = mbfl_convert_filter_new(
                    string->no_encoding,
                    mbfl_no_encoding_wchar,
-                   collector_decode_htmlnumericentity, 0, &pc);
+                   collector_decode_htmlnumericentity, 
+                       (int (*)(void*))mbfl_filt_decode_htmlnumericentity_flush, &pc);
        }
        if (pc.decoder == NULL || encoder == NULL) {
                mbfl_convert_filter_delete(encoder);

diff --git a/ext/mbstring/tests/bug40685.phpt b/ext/mbstring/tests/bug40685.phpt
new file mode 100644 (file)
index 0000000..420b916
--- /dev/null
+++ b/ext/mbstring/tests/bug40685.phpt
@@ -0,0 +1,20 @@
+--TEST--
+Bug #40685 (mb_decode_numericentity() removes '&' in the string)
+--SKIPIF--
+<?php extension_loaded('mbstring') or die('skip mbstring not available'); ?>
+--FILE--
+<?php
+$map = array(0, 0x10FFFF, 0, 0xFFFFFF);
+var_dump(mb_decode_numericentity('&', $map, 'UTF-8'));
+var_dump(mb_decode_numericentity('&&&', $map, 'UTF-8'));
+var_dump(mb_decode_numericentity('&#', $map, 'UTF-8'));
+var_dump(mb_decode_numericentity('&#61', $map, 'UTF-8'));
+var_dump(mb_decode_numericentity('&#61;', $map, 'UTF-8'));
+?>
+--EXPECTF--
+string(1) "&"
+string(3) "&&&"
+string(2) "&#"
+string(4) "&#61"
+string(1) "="
+