Unicode support.

32 files changed, 374 insertions, 0 deletions

diff --git a/tests/unicode/autodetect-SCSU.inc b/tests/unicode/autodetect-SCSU.inc
new file mode 100644
index 0000000000..d905c6c497
--- /dev/null
+++ b/tests/unicode/autodetect-SCSU.inc
@@ -0,0 +1,7 @@
+��<?php
+function µ��() {
+  echo "µ�� - ok\n";
+}
+
+µ��();
+?>
diff --git a/tests/unicode/autodetect-SCSU.phpt b/tests/unicode/autodetect-SCSU.phpt
new file mode 100755
index 0000000000..a3e3f45c0a
--- /dev/null
+++ b/tests/unicode/autodetect-SCSU.phpt
@@ -0,0 +1,11 @@
+--TEST--
+Script encoding autodetection (SCSU)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+include(dirname(__FILE__)."/autodetect-SCSU.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/autodetect-UTF16BE.inc b/tests/unicode/autodetect-UTF16BE.inc
new file mode 100644
index 0000000000..5e808dd60b
--- /dev/null
+++ b/tests/unicode/autodetect-UTF16BE.inc
diff --git a/tests/unicode/autodetect-UTF16BE.phpt b/tests/unicode/autodetect-UTF16BE.phpt
new file mode 100755
index 0000000000..439276961d
--- /dev/null
+++ b/tests/unicode/autodetect-UTF16BE.phpt
@@ -0,0 +1,11 @@
+--TEST--
+Script encoding autodetection (UTF-16BE)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+include(dirname(__FILE__)."/autodetect-UTF16BE.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/autodetect-UTF16LE.inc b/tests/unicode/autodetect-UTF16LE.inc
new file mode 100644
index 0000000000..666a4856ea
--- /dev/null
+++ b/tests/unicode/autodetect-UTF16LE.inc
diff --git a/tests/unicode/autodetect-UTF16LE.phpt b/tests/unicode/autodetect-UTF16LE.phpt
new file mode 100755
index 0000000000..4daaf7cad7
--- /dev/null
+++ b/tests/unicode/autodetect-UTF16LE.phpt
@@ -0,0 +1,11 @@
+--TEST--
+Script encoding autodetection (UTF-16LE)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+include(dirname(__FILE__)."/autodetect-UTF16LE.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/autodetect-UTF32BE.inc b/tests/unicode/autodetect-UTF32BE.inc
new file mode 100755
index 0000000000..1f26885771
--- /dev/null
+++ b/tests/unicode/autodetect-UTF32BE.inc
diff --git a/tests/unicode/autodetect-UTF32BE.phpt b/tests/unicode/autodetect-UTF32BE.phpt
new file mode 100755
index 0000000000..39e62b5445
--- /dev/null
+++ b/tests/unicode/autodetect-UTF32BE.phpt
@@ -0,0 +1,11 @@
+--TEST--
+Script encoding autodetection (UTF-32BE)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+include(dirname(__FILE__)."/autodetect-UTF32BE.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/autodetect-UTF32LE.inc b/tests/unicode/autodetect-UTF32LE.inc
new file mode 100755
index 0000000000..5bf6a1d286
--- /dev/null
+++ b/tests/unicode/autodetect-UTF32LE.inc
diff --git a/tests/unicode/autodetect-UTF32LE.phpt b/tests/unicode/autodetect-UTF32LE.phpt
new file mode 100755
index 0000000000..b966b04b2a
--- /dev/null
+++ b/tests/unicode/autodetect-UTF32LE.phpt
@@ -0,0 +1,11 @@
+--TEST--
+Script encoding autodetection (UTF-32LE)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+include(dirname(__FILE__)."/autodetect-UTF32LE.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/autodetect-UTF7.inc b/tests/unicode/autodetect-UTF7.inc
new file mode 100644
index 0000000000..dc2bda486c
--- /dev/null
+++ b/tests/unicode/autodetect-UTF7.inc
@@ -0,0 +1,7 @@
++/v8<?php
+function +BEIENQRBBEI() {
+  echo "+BEIENQRBBEI - ok+AFw-n";
+}
+
++BEIENQRBBEI();
+?>
diff --git a/tests/unicode/autodetect-UTF7.phpt b/tests/unicode/autodetect-UTF7.phpt
new file mode 100755
index 0000000000..89f8e6d305
--- /dev/null
+++ b/tests/unicode/autodetect-UTF7.phpt
@@ -0,0 +1,11 @@
+--TEST--
+Script encoding autodetection (UTF7)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+include(dirname(__FILE__)."/autodetect-UTF7.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/autodetect-UTF8.inc b/tests/unicode/autodetect-UTF8.inc
new file mode 100755
index 0000000000..e6af829d4c
--- /dev/null
+++ b/tests/unicode/autodetect-UTF8.inc
@@ -0,0 +1,7 @@
+<?php
+function тест() {
+  echo "тест - ok\n";
+}
+
+тест();
+?>
diff --git a/tests/unicode/autodetect-UTF8.phpt b/tests/unicode/autodetect-UTF8.phpt
new file mode 100755
index 0000000000..570f398b2b
--- /dev/null
+++ b/tests/unicode/autodetect-UTF8.phpt
@@ -0,0 +1,11 @@
+--TEST--
+Script encoding autodetection (UTF8)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+include(dirname(__FILE__)."/autodetect-UTF8.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/enc-HZ.inc b/tests/unicode/enc-HZ.inc
new file mode 100644
index 0000000000..ac38662dde
--- /dev/null
+++ b/tests/unicode/enc-HZ.inc
@@ -0,0 +1,7 @@
+~}<?php
+function ~{'d'V'c'd~}() {
+  echo "~{'d'V'c'd~} - ok\n";
+}
+
+~{'d'V'c'd~}();
+?>
diff --git a/tests/unicode/enc-HZ.phpt b/tests/unicode/enc-HZ.phpt
new file mode 100755
index 0000000000..9b327a1ee8
--- /dev/null
+++ b/tests/unicode/enc-HZ.phpt
@@ -0,0 +1,12 @@
+--TEST--
+Script encoding (HZ)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+ini_set("unicode.script_encoding", "HZ");
+include(dirname(__FILE__)."/enc-HZ.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/enc-ISO-2022-KR.inc b/tests/unicode/enc-ISO-2022-KR.inc
new file mode 100644
index 0000000000..fca3ab842c
--- /dev/null
+++ b/tests/unicode/enc-ISO-2022-KR.inc
@@ -0,0 +1,7 @@
+$)C<?php
+function ,d,V,c,d() {
+  echo ",d,V,c,d - ok�n";
+}
+
+,d,V,c,d();
+?>
diff --git a/tests/unicode/enc-ISO-2022-KR.phpt b/tests/unicode/enc-ISO-2022-KR.phpt
new file mode 100755
index 0000000000..94033f624e
--- /dev/null
+++ b/tests/unicode/enc-ISO-2022-KR.phpt
@@ -0,0 +1,12 @@
+--TEST--
+Script encoding (ISO-2022-KR)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+ini_set("unicode.script_encoding", "ISO-2022-KR");
+include(dirname(__FILE__)."/enc-ISO-2022-KR.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/enc-JIS.inc b/tests/unicode/enc-JIS.inc
new file mode 100644
index 0000000000..00d3ae690b
--- /dev/null
+++ b/tests/unicode/enc-JIS.inc
@@ -0,0 +1,7 @@
+<?php
+function $B'd'V'c'd(B() {
+  echo "$B'd'V'c'd(B - ok\n";
+}
+
+$B'd'V'c'd(B();
+?>
diff --git a/tests/unicode/enc-JIS.phpt b/tests/unicode/enc-JIS.phpt
new file mode 100755
index 0000000000..4bab19841a
--- /dev/null
+++ b/tests/unicode/enc-JIS.phpt
@@ -0,0 +1,12 @@
+--TEST--
+Script encoding (JIS)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+ini_set("unicode.script_encoding", "JIS");
+include(dirname(__FILE__)."/enc-JIS.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/enc-SJIS.inc b/tests/unicode/enc-SJIS.inc
new file mode 100644
index 0000000000..12e799306a
--- /dev/null
+++ b/tests/unicode/enc-SJIS.inc
@@ -0,0 +1,7 @@
+<?php
+function ���u����() {
+  echo "���u���� - ok\n";
+}
+
+���u����();
+?>
diff --git a/tests/unicode/enc-SJIS.phpt b/tests/unicode/enc-SJIS.phpt
new file mode 100755
index 0000000000..bd190ee86a
--- /dev/null
+++ b/tests/unicode/enc-SJIS.phpt
@@ -0,0 +1,12 @@
+--TEST--
+Script encoding (SJIS)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+ini_set("unicode.script_encoding", "SJIS");
+include(dirname(__FILE__)."/enc-SJIS.inc");
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/encrt-HZ.phpt b/tests/unicode/encrt-HZ.phpt
new file mode 100755
index 0000000000..c0f3812361
--- /dev/null
+++ b/tests/unicode/encrt-HZ.phpt
@@ -0,0 +1,17 @@
+--TEST--
+declare script encoding (HZ)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+declare(encoding="HZ");
+
+function ~{'d'V'c'd~}() {
+  echo "~{'d'V'c'd~} - ok\n";
+}
+
+~{'d'V'c'd~}();
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/encrt-ISO-2022-KR.phpt b/tests/unicode/encrt-ISO-2022-KR.phpt
new file mode 100755
index 0000000000..ebca20231a
--- /dev/null
+++ b/tests/unicode/encrt-ISO-2022-KR.phpt
@@ -0,0 +1,17 @@
+--TEST--
+declare script encoding (ISO-2022-KR)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+declare(encoding="ISO-2022-KR");
+
+function ,d,V,c,d() {
+  echo ",d,V,c,d - ok�n";
+}
+
+,d,V,c,d();
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/encrt-JIS.phpt b/tests/unicode/encrt-JIS.phpt
new file mode 100755
index 0000000000..082c1caf18
--- /dev/null
+++ b/tests/unicode/encrt-JIS.phpt
@@ -0,0 +1,17 @@
+--TEST--
+declare script encoding (JIS)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+declare(encoding="JIS");
+
+function $B'd'V'c'd(B() {
+  echo "$B'd'V'c'd(B - ok\n";
+}
+
+$B'd'V'c'd(B();
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/encrt-SJIS.phpt b/tests/unicode/encrt-SJIS.phpt
new file mode 100755
index 0000000000..918ddc910e
--- /dev/null
+++ b/tests/unicode/encrt-SJIS.phpt
@@ -0,0 +1,17 @@
+--TEST--
+declare script encoding (SJIS)
+--INI--
+unicode_semantics=on
+unicode.output_encoding=CP866
+--FILE--
+<?php
+declare(encoding="SJIS");
+
+function ���u����() {
+  echo "���u���� - ok\n";
+}
+
+���u����();
+?>
+--EXPECT--
+��� - ok
diff --git a/tests/unicode/func/strpos.phpt b/tests/unicode/func/strpos.phpt
new file mode 100644
index 0000000000..9bef742e97
--- /dev/null
+++ b/tests/unicode/func/strpos.phpt
@@ -0,0 +1,40 @@
+--TEST--
+Unicode: strpos() function test
+--FILE--
+<?php
+$a = "a™ᄒ\U020021z";
+var_dump(strpos($a, 'a'));
+var_dump(strpos($a, 'U'));
+var_dump(strpos($a, 'z'));
+var_dump(strpos($a, '\u2122'));
+var_dump(strpos($a, 0x1112));
+var_dump(strpos($a, 0x20021));
+
+$b = "\U020022z\U020021z";
+var_dump(strpos($b, 'z', 1));
+var_dump(strpos($b, 'z', 2));
+var_dump(strpos($b, 'z', 4));
+var_dump(strpos($b, 'z\U020021'));
+
+$c = "-A\u030a-Å-Å";
+var_dump(strpos($c, 'A'));
+var_dump(strpos($c, '\u030a'));
+var_dump(strpos($c, '\u00c5'));
+var_dump(strpos($c, '\u212b'));
+
+?>
+--EXPECT--
+int(0)
+bool(false)
+int(4)
+int(1)
+int(2)
+int(3)
+int(1)
+int(3)
+bool(false)
+int(1)
+int(1)
+int(2)
+int(4)
+int(6)
diff --git a/tests/unicode/func/strstr.phpt b/tests/unicode/func/strstr.phpt
new file mode 100644
index 0000000000..ad5727dada
--- /dev/null
+++ b/tests/unicode/func/strstr.phpt
@@ -0,0 +1,37 @@
+--TEST--
+Unicode: strpos() function test
+--FILE--
+<?php
+$a = "a™ᄒ\U020021z";
+var_dump(strstr($a, 'a'));
+var_dump(strstr($a, 'U'));
+var_dump(strstr($a, 'z'));
+var_dump(strstr($a, '\u2122'));
+var_dump(strstr($a, '\udc21'));
+var_dump(strstr($a, 0x1112));
+var_dump(strstr($a, 0x20021));
+
+$b = "\U020022z\U020021z";
+var_dump(strstr($b, '\U020021'));
+var_dump(strstr($b, 'z\U020021'));
+
+$c = "-A\u030a-Å-Å";
+var_dump(strstr($c, 'A'));
+var_dump(strstr($c, '\u030a'));
+var_dump(strstr($c, '\u00c5'));
+var_dump(strstr($c, '\u212b'));
+?>
+--EXPECT--
+unicode(5) "a™ᄒ𠀡z"
+bool(false)
+unicode(1) "z"
+unicode(4) "™ᄒ𠀡z"
+bool(false)
+unicode(3) "ᄒ𠀡z"
+unicode(2) "𠀡z"
+unicode(2) "𠀡z"
+unicode(3) "z𠀡z"
+unicode(6) "Å-Å-Å"
+unicode(5) "̊-Å-Å"
+unicode(3) "Å-Å"
+unicode(1) "Å"
diff --git a/tests/unicode/normalize_001.phpt b/tests/unicode/normalize_001.phpt
new file mode 100755
index 0000000000..46b456fc4e
--- /dev/null
+++ b/tests/unicode/normalize_001.phpt
@@ -0,0 +1,11 @@
+--TEST--
+Unicode identifiers normalization (${})
+--INI--
+unicode_semantics=on
+--FILE--
+<?php
+${"\u212B"} = "ok\n";
+echo ${"\u00C5"};
+?>
+--EXPECT--
+ok
diff --git a/tests/unicode/normalize_002.phpt b/tests/unicode/normalize_002.phpt
new file mode 100755
index 0000000000..ade41ec5fe
--- /dev/null
+++ b/tests/unicode/normalize_002.phpt
@@ -0,0 +1,13 @@
+--TEST--
+Unicode identifiers normalization ($$)
+--INI--
+unicode_semantics=on
+--FILE--
+<?php
+$a = "\u212B";
+$b = "\u00C5";
+$$a = "ok\n";
+echo $$b;
+?>
+--EXPECT--
+ok
diff --git a/tests/unicode/normalize_003.phpt b/tests/unicode/normalize_003.phpt
new file mode 100755
index 0000000000..21085acc24
--- /dev/null
+++ b/tests/unicode/normalize_003.phpt
@@ -0,0 +1,20 @@
+--TEST--
+Unicode identifiers normalization (indirect function call)
+--INI--
+unicode_semantics=on
+--FILE--
+<?php
+declare(encoding = "ISO-8859-1");
+
+function �() {
+  echo "ok\n";
+}
+
+$f1 = "\u212B";
+$f1();
+$f2 = "\u00C5";
+$f2();
+?>
+--EXPECT--
+ok
+ok
diff --git a/tests/unicode/normalize_004.phpt b/tests/unicode/normalize_004.phpt
new file mode 100755
index 0000000000..37f65ee940
--- /dev/null
+++ b/tests/unicode/normalize_004.phpt
@@ -0,0 +1,11 @@
+--TEST--
+Unicode identifiers normalization ($GLOBALS[])
+--INI--
+unicode_semantics=on
+--FILE--
+<?php
+$GLOBALS["\u212B"] = "ok\n";
+echo $GLOBALS["\u00C5"];
+?>
+--EXPECT--
+ok