test 4-byte sequences

diff --git a/php_http_utf8.h b/php_http_utf8.h
index d67910fb82aff5453541f01fe0b8d30e4a8e1db1..2503f0644e605db64aae577b6523e641b3e54098 100644 (file)
--- a/php_http_utf8.h
+++ b/php_http_utf8.h
@@ -555,7 +555,7 @@ static inline size_t utf8towc(unsigned *wc, const unsigned char *uc, size_t len)
 {
        unsigned char ub = utf8_mblen[*uc];
 
 {
        unsigned char ub = utf8_mblen[*uc];
 

-       if (!ub || ub > len || ub > 3) {
+       if (!ub || ub > len || ub > 4) {

                return 0;
        }
 
                return 0;
        }
 

diff --git a/tests/urlparser010.phpt b/tests/urlparser010.phpt
index 71f6943a396b0b42a18ef31f6ec93a878150ef44..ae63ae7b4050e4eba43b60049a565d9736e0342e 100644 (file)
--- a/tests/urlparser010.phpt
+++ b/tests/urlparser010.phpt
@@ -1,19 +1,24 @@
 --TEST--
 --TEST--

-url parser multibyte/utf-8/topct
+url parser multibyte/locale/topct

 --SKIPIF--
 <?php
 include "skipif.inc";
 --SKIPIF--
 <?php
 include "skipif.inc";

+if (!defined("http\\Url::PARSE_MBLOC") or
+       !stristr(setlocale(LC_CTYPE, NULL), ".utf")) {
+       die("skip need http\\Url::PARSE_MBLOC support and LC_CTYPE=*.UTF-8");
+}
+

 ?>
 --FILE--
 <?php
 echo "Test\n";
 
 $urls = array(
 ?>
 --FILE--
 <?php
 echo "Test\n";
 
 $urls = array(

-       "http://mike:paßwort@sörver.net/for/€/?by=¢#ø"
+       "http://mike:paßwort@𐌀𐌁𐌂.it/for/€/?by=¢#ø"

 );
 
 foreach ($urls as $url) {
 );
 
 foreach ($urls as $url) {

-       var_dump(new http\Url($url, null, http\Url::PARSE_MBUTF8|http\Url::PARSE_TOPCT));
+       var_dump(new http\Url($url, null, http\Url::PARSE_MBLOC|http\Url::PARSE_TOPCT));

 }
 ?>
 DONE
 }
 ?>
 DONE


@@ -27,7 +32,7 @@ object(http\Url)#%d (8) {
   ["pass"]=>
   string(12) "pa%C3%9Fwort"
   ["host"]=>
   ["pass"]=>
   string(12) "pa%C3%9Fwort"
   ["host"]=>

-  string(11) "sörver.net"
+  string(15) "𐌀𐌁𐌂.it"

   ["port"]=>
   NULL
   ["path"]=>
   ["port"]=>
   NULL
   ["path"]=>

diff --git a/tests/urlparser011.phpt b/tests/urlparser011.phpt
new file mode 100644 (file)
index 0000000..5dbd82a
--- /dev/null
+++ b/tests/urlparser011.phpt
@@ -0,0 +1,40 @@
+--TEST--
+url parser multibyte/utf-8/topct
+--SKIPIF--
+<?php
+include "skipif.inc";
+?>
+--FILE--
+<?php
+echo "Test\n";
+
+$urls = array(
+       "http://mike:paßwort@𐌀𐌁𐌂.it/for/€/?by=¢#ø"
+);
+
+foreach ($urls as $url) {
+       var_dump(new http\Url($url, null, http\Url::PARSE_MBUTF8|http\Url::PARSE_TOPCT));
+}
+?>
+DONE
+--EXPECTF--
+Test
+object(http\Url)#%d (8) {
+  ["scheme"]=>
+  string(4) "http"
+  ["user"]=>
+  string(4) "mike"
+  ["pass"]=>
+  string(12) "pa%C3%9Fwort"
+  ["host"]=>
+  string(15) "𐌀𐌁𐌂.it"
+  ["port"]=>
+  NULL
+  ["path"]=>
+  string(15) "/for/%E2%82%AC/"
+  ["query"]=>
+  string(9) "by=%C2%A2"
+  ["fragment"]=>
+  string(6) "%C3%B8"
+}
+DONE