php · php-pulls · Jan 4, 2015 · Nov 24, 2014 · Nov 24, 2014 · Nov 24, 2014
diff --git a/spec/09-lexical-structure.md b/spec/09-lexical-structure.md
@@ -542,6 +542,7 @@ A single-quoted string literal is always a constant expression.
     <i>dq-simple-escape-sequence</i>
     <i>dq-octal-escape-sequence</i>
     <i>dq-hexadecimal-escape-sequence</i>
+    <i>dq-unicode-escape-sequence</i>
 
   <i>dq-simple-escape-sequence:: one of</i>
     \"   \\   \$   \e   \f   \n   \r   \t   \v
@@ -554,6 +555,13 @@ A single-quoted string literal is always a constant expression.
   <i>dq-hexadecimal-escape-sequence::</i>
     \x  <i>hexadecimal-digit   hexadecimal-digit<sub>opt</sub></i>
     \X  <i>hexadecimal-digit   hexadecimal-digit<sub>opt</sub></i>
+
+  <i>dq-unicode-escape-sequence::</i>
+    \u{  codepoint-digits  }
+
+  <i>codepoint-digits::</i>
+     <i>hexadecimal-digit</i>
+     <i>hexadecimal-digit   codepoint-digits</i>
 </pre>
 
 *octal-digit* and *hexadecimal-digit* are defined in [§§](#integer-literals).
@@ -586,13 +594,28 @@ Escape sequence | Character name | Unicode character
 \v  | Vertical Tab | U+000B
 \ooo |  1–3-digit octal digit value ooo
 \xhh or \Xhh  | 1–2-digit hexadecimal digit value hh
+\u{xxxxxx} | UTF-8 encoding of Unicode codepoint U+xxxxxx | U+xxxxxx
 
 Within a double-quoted string literal, except when recognized as the
 start of an escape sequence, a backslash (\\) is retained verbatim.
 
 Within a double-quoted string literal a dollar ($) character not
 escaped by a backslash (\\) is handled using a variable substitution rules
-described below. 
+described below.
+
+The `\u{xxxxxx}` escape sequence produces the UTF-8 encoding of the Unicode 
+codepoint with the hexadecimal number specified within the curly braces.
+Implementations MUST NOT allow Unicode codepoints beyond U+10FFFF as this is
+outside the range UTF-8 can encode (see
+[RFC 3629](http://tools.ietf.org/html/rfc3629#section-3)). If a codepoint 
+larger than U+10FFFF is specified, implementations MUST error.
+Implementations MUST pass through `\u` verbatim and not interpret it as an
+escape sequence if it is not followed by an opening `{`, but if it is,
+implementations MUST produce an error if there is no terminating `}` or the
+contents are not a valid codepoint. Implementations MUST support leading zeroes,
+but MUST NOT support leading or trailing whitespace for the codepoint between
+the opening and terminating braces. Implementations MUST allow Unicode 
+codepoints that are not Unicode scalar values, such as high and low surrogates.
 
 **Variable substitution**
 
@@ -695,6 +718,7 @@ echo "\$myC->p1 = >$myC->p1<\n";  // → $myC->p1 = >2<
     <i>hd-simple-escape-sequence</i>
     <i>dq-octal-escape-sequence</i>
     <i>dq-hexadecimal-escape-sequence</i>
+    <i>dq-unicode-escape-sequence</i>
 
   <i>hd-simple-escape-sequence:: one of</i>
     \\   \$   \e   \f   \n   \r   \t   \v

diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape.php b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape.php
@@ -0,0 +1,8 @@
+<?php
+
+var_dump("\u{61}"); // ASCII "a" - characters below U+007F just encode as ASCII, as it's UTF-8
+var_dump("\u{FF}"); // y with diaeresis
+var_dump("\u{ff}"); // case-insensitive
+var_dump("\u{2603}"); // Unicode snowman
+var_dump("\u{1F602}"); // FACE WITH TEARS OF JOY emoji
+var_dump("\u{0000001F602}"); // Leading zeroes permitted
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape.php.expect b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape.php.expect
@@ -0,0 +1,6 @@
+string(1) "a"
+string(2) "ÿ"
+string(2) "ÿ"
+string(3) "☃"
+string(4) "😂"
+string(4) "😂"
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_empty.php b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_empty.php
@@ -0,0 +1,3 @@
+<?php
+
+var_dump("\u{}");
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_empty.php.expectf b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_empty.php.expectf
@@ -0,0 +1 @@
+Fatal error: Invalid UTF-8 codepoint escape sequence in %s on line %d
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_incomplete.php b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_incomplete.php
@@ -0,0 +1,3 @@
+<?php
+
+var_dunp("\u{blah");
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_incomplete.php.expectf b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_incomplete.php.expectf
@@ -0,0 +1 @@
+Fatal error: Invalid UTF-8 codepoint escape sequence in %s on line %d
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_large_codepoint.php b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_large_codepoint.php
@@ -0,0 +1,3 @@
+<?php
+
+var_dump("\u{110000}"); // U+10FFFF + 1
diff --git a/...xical_structure/unicode_string_escape_sequence/unicode_escape_large_codepoint.php.expectf b/...xical_structure/unicode_string_escape_sequence/unicode_escape_large_codepoint.php.expectf
@@ -0,0 +1 @@
+Fatal error: Invalid UTF-8 codepoint escape sequence: Codepoint too large in %s on line %d
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_legacy.php b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_legacy.php
@@ -0,0 +1,6 @@
+<?php
+
+// These are ignored to avoid breaking JSON string literals
+var_dump("\u");
+var_dump("\u202e");
+var_dump("\ufoobar");
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_legacy.php.expectf b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_legacy.php.expectf
@@ -0,0 +1,3 @@
+string(2) "\u"
+string(6) "\u202e"
+string(8) "\ufoobar"
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_sign.php b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_sign.php
@@ -0,0 +1,3 @@
+<?php
+
+var_dump("\u{+1F602}");
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_sign.php.expectf b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_sign.php.expectf
@@ -0,0 +1 @@
+Fatal error: Invalid UTF-8 codepoint escape sequence in %s on line %d
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_sign2.php b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_sign2.php
@@ -0,0 +1,3 @@
+<?php
+
+var_dump("\u{-1F602}");
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_sign2.php.expectf b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_sign2.php.expectf
@@ -0,0 +1 @@
+Fatal error: Invalid UTF-8 codepoint escape sequence in %s on line %d
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_surrogates.php b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_surrogates.php
@@ -0,0 +1,8 @@
+<?php
+
+// Surrogate pairs are non-well-formed UTF-8 - however, it is sometimes useful
+// to be able to produce these (e.g. CESU-8 handling)
+
+var_dump(bin2hex("\u{D801}"));
+var_dump(bin2hex("\u{DC00}"));
+var_dump(bin2hex("\u{D801}\u{DC00}")); // CESU-8 encoding of U+10400
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_surrogates.php.expect b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_surrogates.php.expect
@@ -0,0 +1,3 @@
+string(6) "eda081"
+string(6) "edb080"
+string(12) "eda081edb080"
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_whitespace.php b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_whitespace.php
@@ -0,0 +1,3 @@
+<?php
+
+var_dump("\u{1F602 }");
diff --git a/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_whitespace.php.expectf b/tests/lexical_structure/unicode_string_escape_sequence/unicode_escape_whitespace.php.expectf
@@ -0,0 +1 @@
+Fatal error: Invalid UTF-8 codepoint escape sequence in %s on line %d
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1 @@
		Fatal error: Invalid UTF-8 codepoint escape sequence in %s on line %d