Allow valid multi-byte utf-8 characters to be allowed as file names in phar archives.

diff --git a/NEWS b/NEWS
index 7b2b4048046e03c4e660c17ebf7ac06f031ab0df..6635a8e5522a91e12e47d8b32109eca6ed598d14 100644 (file)
--- a/NEWS
+++ b/NEWS
@@ -15,6 +15,10 @@ PHP                                                                        NEWS
   . Fixed bug #66908 (php-fpm reload leaks epoll_create() file descriptor). 
     (Julio Pintos)
 
+- Phar:
+  . Fix bug #64498 ($phar->buildFromDirectory can't compress file with an accent 
+    in its name). (PR #588) 
+
 ?? ??? 2014, PHP 5.4.28
 
 - Core:

diff --git a/ext/phar/phar_path_check.c b/ext/phar/phar_path_check.c
index c1aff7037db346bf989b80aa2f0e1794fda5502a..22c915dfac4f2b3d2739142934f35b837515b44c 100644 (file)
--- a/ext/phar/phar_path_check.c
+++ b/ext/phar/phar_path_check.c
@@ -1,4 +1,4 @@
-/* Generated by re2c 0.13.5 on Fri Feb 25 04:35:39 2011 */
+/* Generated by re2c 0.13.5 on Sun Apr 20 17:14:40 2014 */
 #line 1 "ext/phar/phar_path_check.re"
 /*
   +----------------------------------------------------------------------+
@@ -46,60 +46,69 @@ loop:
 #line 47 "ext/phar/phar_path_check.c"
 {
        YYCTYPE yych;
+       unsigned int yyaccept = 0;
 
        if ((YYLIMIT - YYCURSOR) < 4) YYFILL(4);
        yych = *YYCURSOR;
-       if (yych <= '.') {
-               if (yych <= '\n') {
-                       if (yych <= 0x00) goto yy13;
-                       if (yych <= '\t') goto yy10;
-                       goto yy12;
+       if (yych <= '>') {
+               if (yych <= 0x19) {
+                       if (yych <= 0x00) goto yy16;
+                       if (yych == '\n') goto yy15;
+                       goto yy14;
                } else {
-                       if (yych <= 0x19) goto yy10;
-                       if (yych == '*') goto yy6;
-                       goto yy15;
+                       if (yych <= '*') {
+                               if (yych <= ')') goto yy18;
+                               goto yy6;
+                       } else {
+                               if (yych != '/') goto yy18;
+                       }
                }
        } else {
-               if (yych <= '?') {
-                       if (yych <= '/') goto yy2;
-                       if (yych <= '>') goto yy15;
-                       goto yy8;
-               } else {
+               if (yych <= 0x7F) {
+                       if (yych <= '?') goto yy8;
                        if (yych == '\\') goto yy4;
-                       if (yych <= 0x7F) goto yy15;
-                       goto yy10;
+                       goto yy18;
+               } else {
+                       if (yych <= 0xDF) {
+                               if (yych <= 0xBF) goto yy14;
+                               goto yy10;
+                       } else {
+                               if (yych <= 0xEF) goto yy12;
+                               if (yych <= 0xF7) goto yy13;
+                               goto yy14;
+                       }
                }
        }
-yy2:
+       yyaccept = 0;
        yych = *(YYMARKER = ++YYCURSOR);
        if (yych <= '-') goto yy3;
-       if (yych <= '.') goto yy16;
-       if (yych <= '/') goto yy18;
+       if (yych <= '.') goto yy29;
+       if (yych <= '/') goto yy30;
 yy3:
-#line 93 "ext/phar/phar_path_check.re"
+#line 105 "ext/phar/phar_path_check.re"
        {
                        goto loop;
                }
-#line 84 "ext/phar/phar_path_check.c"
+#line 93 "ext/phar/phar_path_check.c"
 yy4:
        ++YYCURSOR;
-#line 60 "ext/phar/phar_path_check.re"
+#line 63 "ext/phar/phar_path_check.re"
        {
                        *error = "back-slash";
                        return pcr_err_back_slash;
                }
-#line 92 "ext/phar/phar_path_check.c"
+#line 101 "ext/phar/phar_path_check.c"
 yy6:
        ++YYCURSOR;
-#line 64 "ext/phar/phar_path_check.re"
+#line 67 "ext/phar/phar_path_check.re"
        {
                        *error = "star";
                        return pcr_err_star;
                }
-#line 100 "ext/phar/phar_path_check.c"
+#line 109 "ext/phar/phar_path_check.c"
 yy8:
        ++YYCURSOR;
-#line 68 "ext/phar/phar_path_check.re"
+#line 71 "ext/phar/phar_path_check.re"
        {
                        if (**s == '/') {
                                (*s)++;
@@ -108,22 +117,39 @@ yy8:
                        *error = NULL;
                        return pcr_use_query;
                }
-#line 112 "ext/phar/phar_path_check.c"
+#line 121 "ext/phar/phar_path_check.c"
 yy10:
        ++YYCURSOR;
+       if ((yych = *YYCURSOR) <= 0x7F) goto yy11;
+       if (yych <= 0xBF) goto yy27;
 yy11:
-#line 76 "ext/phar/phar_path_check.re"
+#line 88 "ext/phar/phar_path_check.re"
        {
                        *error ="illegal character";
                        return pcr_err_illegal_char;
                }
-#line 121 "ext/phar/phar_path_check.c"
+#line 132 "ext/phar/phar_path_check.c"
 yy12:
-       yych = *++YYCURSOR;
+       yyaccept = 1;
+       yych = *(YYMARKER = ++YYCURSOR);
+       if (yych <= 0x7F) goto yy11;
+       if (yych <= 0xBF) goto yy24;
        goto yy11;
 yy13:
+       yyaccept = 1;
+       yych = *(YYMARKER = ++YYCURSOR);
+       if (yych <= 0x7F) goto yy11;
+       if (yych <= 0xBF) goto yy19;
+       goto yy11;
+yy14:
+       yych = *++YYCURSOR;
+       goto yy11;
+yy15:
+       yych = *++YYCURSOR;
+       goto yy11;
+yy16:
        ++YYCURSOR;
-#line 80 "ext/phar/phar_path_check.re"
+#line 92 "ext/phar/phar_path_check.re"
        {
                        if (**s == '/') {
                                (*s)++;
@@ -137,49 +163,85 @@ yy13:
                        *error = NULL;
                        return pcr_is_ok;
                }
-#line 141 "ext/phar/phar_path_check.c"
-yy15:
+#line 167 "ext/phar/phar_path_check.c"
+yy18:
        yych = *++YYCURSOR;
        goto yy3;
-yy16:
+yy19:
        yych = *++YYCURSOR;
-       if (yych <= 0x00) goto yy21;
-       if (yych <= '-') goto yy17;
-       if (yych <= '.') goto yy20;
-       if (yych <= '/') goto yy21;
-yy17:
+       if (yych <= 0x7F) goto yy20;
+       if (yych <= 0xBF) goto yy21;
+yy20:
        YYCURSOR = YYMARKER;
-       goto yy3;
-yy18:
+       if (yyaccept <= 0) {
+               goto yy3;
+       } else {
+               goto yy11;
+       }
+yy21:
+       yych = *++YYCURSOR;
+       if (yych <= 0x7F) goto yy20;
+       if (yych >= 0xC0) goto yy20;
        ++YYCURSOR;
-#line 48 "ext/phar/phar_path_check.re"
+#line 85 "ext/phar/phar_path_check.re"
+       {
+                       goto loop;
+       }
+#line 191 "ext/phar/phar_path_check.c"
+yy24:
+       yych = *++YYCURSOR;
+       if (yych <= 0x7F) goto yy20;
+       if (yych >= 0xC0) goto yy20;
+       ++YYCURSOR;
+#line 82 "ext/phar/phar_path_check.re"
+       {
+                       goto loop;
+       }
+#line 201 "ext/phar/phar_path_check.c"
+yy27:
+       ++YYCURSOR;
+#line 79 "ext/phar/phar_path_check.re"
+       {
+                       goto loop;
+       }
+#line 208 "ext/phar/phar_path_check.c"
+yy29:
+       yych = *++YYCURSOR;
+       if (yych <= 0x00) goto yy33;
+       if (yych <= '-') goto yy20;
+       if (yych <= '.') goto yy32;
+       if (yych <= '/') goto yy33;
+       goto yy20;
+yy30:
+       ++YYCURSOR;
+#line 51 "ext/phar/phar_path_check.re"
        {
                        *error = "double slash";
                        return pcr_err_double_slash;
                }
-#line 161 "ext/phar/phar_path_check.c"
-yy20:
+#line 223 "ext/phar/phar_path_check.c"
+yy32:
        yych = *++YYCURSOR;
-       if (yych <= 0x00) goto yy23;
-       if (yych == '/') goto yy23;
-       goto yy17;
-yy21:
+       if (yych <= 0x00) goto yy35;
+       if (yych == '/') goto yy35;
+       goto yy20;
+yy33:
        ++YYCURSOR;
-#line 56 "ext/phar/phar_path_check.re"
+#line 59 "ext/phar/phar_path_check.re"
        {
                        *error = "current directory reference";
                        return pcr_err_curr_dir;
                }
-#line 174 "ext/phar/phar_path_check.c"
-yy23:
+#line 236 "ext/phar/phar_path_check.c"
+yy35:
        ++YYCURSOR;
-#line 52 "ext/phar/phar_path_check.re"
+#line 55 "ext/phar/phar_path_check.re"
        {
                        *error = "upper directory reference";
                        return pcr_err_up_dir;
                }
-#line 182 "ext/phar/phar_path_check.c"
+#line 244 "ext/phar/phar_path_check.c"
 }
-#line 96 "ext/phar/phar_path_check.re"
+#line 108 "ext/phar/phar_path_check.re"
 
 }

diff --git a/ext/phar/phar_path_check.re b/ext/phar/phar_path_check.re
index e43b3f846e5ec46746c4f35f8ce27a31926b909f..b182d784800ab8b23337d251eaaec4911b68cfb8 100644 (file)
--- a/ext/phar/phar_path_check.re
+++ b/ext/phar/phar_path_check.re
@@ -2,7 +2,7 @@
   +----------------------------------------------------------------------+
   | phar php single-file executable PHP extension                        |
   +----------------------------------------------------------------------+
-  | Copyright (c) 2007-2013 The PHP Group                                |
+  | Copyright (c) 2007-2014 The PHP Group                                |
   +----------------------------------------------------------------------+
   | This source file is subject to version 3.01 of the PHP license,      |
   | that is bundled with this package in the file LICENSE, and is        |
@@ -42,6 +42,9 @@ phar_path_check_result phar_path_check(char **s, int *len, const char **error)
 loop:
 /*!re2c
 END = "\x00";
+MB2 = ([\xC0-\xDF][\x80-\xBF]);
+MB3 = ([\xE0-\xEF][\x80-\xBF]{2});
+MB4 = ([\xF0-\xF7][\x80-\xBF]{3});
 ILL = [\x01-\x19\x80-\xFF];
 EOS = "/" | END;
 ANY = .;
@@ -73,6 +76,15 @@ ANY = .;
                        *error = NULL;
                        return pcr_use_query;
                }
+MB2 {
+                       goto loop;
+       }
+MB3 {
+                       goto loop;
+       }
+MB4 {
+                       goto loop;
+       }
 ILL {
                        *error ="illegal character";
                        return pcr_err_illegal_char;

diff --git a/ext/phar/tests/create_new_phar.phpt b/ext/phar/tests/create_new_phar.phpt
index ec57c272173f24437f4e951f0ae8c312c2659c4f..26794095e15f9aa269b3971213703c6ace799c85 100644 (file)
--- a/ext/phar/tests/create_new_phar.phpt
+++ b/ext/phar/tests/create_new_phar.phpt
@@ -9,8 +9,13 @@ phar.require_hash=1
 <?php
 
 file_put_contents('phar://' . dirname(__FILE__) . '/' . basename(__FILE__, '.php') . '.phar.php/a.php',
-       'brand new!');
+       "brand new!\n");
 include 'phar://' . dirname(__FILE__) . '/' . basename(__FILE__, '.php') . '.phar.php/a.php';
+
+$fileName = "ChineseFile\xE5\x84\xB7\xE9\xBB\x91.php";
+file_put_contents('phar://' . dirname(__FILE__) . '/' . basename(__FILE__, '.php') . '.phar.php/$fileName.php',
+       'Text in utf8 file.');
+include 'phar://' . dirname(__FILE__) . '/' . basename(__FILE__, '.php') . '.phar.php/$fileName.php';
 ?>
 
 ===DONE===
@@ -18,4 +23,5 @@ include 'phar://' . dirname(__FILE__) . '/' . basename(__FILE__, '.php') . '.pha
 <?php unlink(dirname(__FILE__) . '/' . basename(__FILE__, '.clean.php') . '.phar.php'); ?>
 --EXPECT--
 brand new!
+Text in utf8 file.
 ===DONE===

diff --git a/ext/phar/tests/create_path_error.phpt b/ext/phar/tests/create_path_error.phpt
index d3fc03586065fb1aebeffa6d7ac2f7618c6eea4e..7451d9cd3985e494c907c176865f67a6e458f496 100644 (file)
--- a/ext/phar/tests/create_path_error.phpt
+++ b/ext/phar/tests/create_path_error.phpt
@@ -22,15 +22,26 @@ var_dump(file_get_contents($pname . '/b.php'));
 
 function error_handler($errno, $errmsg)
 {
-       echo "Error: $errmsg\n";
+       echo "Error: $errmsg";
 }
 
 set_error_handler('error_handler');
 
-$checks = array('/', '.', '../', 'a/..', 'a/', 'b//a.php');
+$count = 0;
+$checks = array(
+    '/', '.', '../', 'a/..', 'a/', 'b//a.php', 
+    "Font\xE5\x84\xB7\xE9\xBB\x91pro.ttf", //two valid multi-byte characters
+    "\xF0\x9F\x98\x8D.ttf", // valid 4 byte char - smiling face with heart-shaped eyes
+    "Font\xE9\xBBpro.ttf", //Invalid multi-byte character - missing last byte
+    "Font\xBB\x91pro.ttf",   //Invalid multi-byte character - missing first byte
+    "\xFC\x81\x81\x81\x81pro.ttf", //RFC 3629 limited char points to 0000-10FFFF aka 5 byte utf-8 not valid
+);
 foreach($checks as $check)
 {
+       $count++;
+       echo "$count:";
        file_put_contents($pname . '/' . $check, "error");
+       echo "\n";
 }
 
 $phar = new Phar($fname);
@@ -54,9 +65,17 @@ foreach($checks as $check)
 --EXPECTF--
 string(5) "query"
 string(5) "query"
-Error: file_put_contents(phar://%s//): failed to open stream: phar error: file "" in phar "%s" cannot be empty
-Error: file_put_contents(phar://%s/.): failed to open stream: phar error: file "" in phar "%s" cannot be empty
-Error: file_put_contents(phar://%s/../): failed to open stream: phar error: file "" in phar "%s" cannot be empty
-Error: file_put_contents(phar://%s/a/..): failed to open stream: phar error: file "" in phar "%s" cannot be empty
+1:Error: file_put_contents(phar://%s//): failed to open stream: phar error: file "" in phar "%s" cannot be empty
+2:Error: file_put_contents(phar://%s/.): failed to open stream: phar error: file "" in phar "%s" cannot be empty
+3:Error: file_put_contents(phar://%s/../): failed to open stream: phar error: file "" in phar "%s" cannot be empty
+4:Error: file_put_contents(phar://%s/a/..): failed to open stream: phar error: file "" in phar "%s" cannot be empty
+5:
+6:
+7:
+8:
+9:Error: file_put_contents(phar:///%s): failed to open stream: phar error: invalid path "%s" contains illegal character
+10:Error: file_put_contents(phar:///%s): failed to open stream: phar error: invalid path "%s" contains illegal character
+11:Error: file_put_contents(phar:///%s): failed to open stream: phar error: invalid path "%s" contains illegal character
 Exception: Entry a does not exist and cannot be created: phar error: invalid path "a" contains illegal character
 ===DONE===
+