Fix handling of upper/lowercase, and whitespace

diff --git a/src/etc/unicode.py b/src/etc/unicode.py
index 7c49b9b57726ae840c481800f2c8b93aaf506b9a..d35263c100f4b400dd4b1eea62fa5e33080e1685 100755 (executable)
--- a/src/etc/unicode.py
+++ b/src/etc/unicode.py
@@ -41,7 +41,7 @@ def load_unicode_data(f):
             continue
         [code, name, gencat, combine, bidi,
          decomp, deci, digit, num, mirror,
-         old, iso, upcase, lowcsae, titlecase ] = fields
+         old, iso, upcase, lowcase, titlecase ] = fields
 
         code = int(code, 16)
 
@@ -89,11 +89,9 @@ def load_unicode_data(f):
 
     return (canon_decomp, compat_decomp, gencats, combines)
 
-
-def load_derived_core_properties(f):
+def load_properties(f, interestingprops):
     fetch(f)
-    derivedprops = {}
-    interestingprops = ["XID_Start", "XID_Continue", "Alphabetic"]
+    props = {}
     re1 = re.compile("^([0-9A-F]+) +; (\w+)")
     re2 = re.compile("^([0-9A-F]+)\.\.([0-9A-F]+) +; (\w+)")
 
@@ -118,10 +116,10 @@ def load_derived_core_properties(f):
             continue
         d_lo = int(d_lo, 16)
         d_hi = int(d_hi, 16)
-        if prop not in derivedprops:
-            derivedprops[prop] = []
-        derivedprops[prop].append((d_lo, d_hi))
-    return derivedprops
+        if prop not in props:
+            props[prop] = []
+        props[prop].append((d_lo, d_hi))
+    return props
 
 def escape_char(c):
     if c <= 0xff:
@@ -376,5 +374,9 @@ emit_property_module(rf, "general_category", gencats)
 
 emit_decomp_module(rf, canon_decomp, compat_decomp, combines)
 
-derived = load_derived_core_properties("DerivedCoreProperties.txt")
+derived = load_properties("DerivedCoreProperties.txt",
+        ["XID_Start", "XID_Continue", "Alphabetic", "Lowercase", "Uppercase"])
 emit_property_module(rf, "derived_property", derived)
+
+props = load_properties("PropList.txt", ["White_Space"])
+emit_property_module(rf, "property", props)

diff --git a/src/libstd/char.rs b/src/libstd/char.rs
index b372993e3e50665d6040b63d66c964c2aae2e434..3019e47c9f93e7e7be7e1486c8c1c62c5ee8f133 100644 (file)
--- a/src/libstd/char.rs
+++ b/src/libstd/char.rs
@@ -14,7 +14,7 @@
 use option::{None, Option, Some};
 use iter::{Iterator, range_step};
 use str::StrSlice;
-use unicode::{derived_property, general_category, decompose};
+use unicode::{derived_property, property, general_category, decompose};
 use to_str::ToStr;
 use str;
 
@@ -89,30 +89,28 @@ pub fn is_XID_continue(c: char) -> bool { derived_property::XID_Continue(c) }
 
 ///
 /// Indicates whether a character is in lower case, defined
-/// in terms of the Unicode General Category 'Ll'
+/// in terms of the Unicode Derived Core Property 'Lowercase'.
 ///
 #[inline]
-pub fn is_lowercase(c: char) -> bool { general_category::Ll(c) }
+pub fn is_lowercase(c: char) -> bool { derived_property::Lowercase(c) }
 
 ///
 /// Indicates whether a character is in upper case, defined
-/// in terms of the Unicode General Category 'Lu'.
+/// in terms of the Unicode Derived Core Property 'Uppercase'.
 ///
 #[inline]
-pub fn is_uppercase(c: char) -> bool { general_category::Lu(c) }
+pub fn is_uppercase(c: char) -> bool { derived_property::Uppercase(c) }
 
 ///
 /// Indicates whether a character is whitespace. Whitespace is defined in
-/// terms of the Unicode General Categories 'Zs', 'Zl', 'Zp'
-/// additional 'Cc'-category control codes in the range [0x09, 0x0d]
+/// terms of the Unicode Property 'White_Space'.
 ///
 #[inline]
 pub fn is_whitespace(c: char) -> bool {
+    // As an optimization ASCII whitespace characters are checked separately
     c == ' '
         || ('\x09' <= c && c <= '\x0d')
-        || general_category::Zs(c)
-        || general_category::Zl(c)
-        || general_category::Zp(c)
+        || property::White_Space(c)
 }
 
 ///

diff --git a/src/libstd/unicode.rs b/src/libstd/unicode.rs
index 397bf8ae549dcbe536323b0bfbff4ef99aba61b7..f19ce98ff1a331e440064db208b7ba62e8df8dca 100644 (file)
--- a/src/libstd/unicode.rs
+++ b/src/libstd/unicode.rs
@@ -4087,6 +4087,640 @@ pub fn Alphabetic(c: char) -> bool {
         bsearch_range_table(c, Alphabetic_table)
     }
 
+    static Lowercase_table : &'static [(char,char)] = &[
+        ('\x61', '\x7a'), ('\xaa', '\xaa'),
+        ('\xb5', '\xb5'), ('\xba', '\xba'),
+        ('\xdf', '\xf6'), ('\xf8', '\xff'),
+        ('\u0101', '\u0101'), ('\u0103', '\u0103'),
+        ('\u0105', '\u0105'), ('\u0107', '\u0107'),
+        ('\u0109', '\u0109'), ('\u010b', '\u010b'),
+        ('\u010d', '\u010d'), ('\u010f', '\u010f'),
+        ('\u0111', '\u0111'), ('\u0113', '\u0113'),
+        ('\u0115', '\u0115'), ('\u0117', '\u0117'),
+        ('\u0119', '\u0119'), ('\u011b', '\u011b'),
+        ('\u011d', '\u011d'), ('\u011f', '\u011f'),
+        ('\u0121', '\u0121'), ('\u0123', '\u0123'),
+        ('\u0125', '\u0125'), ('\u0127', '\u0127'),
+        ('\u0129', '\u0129'), ('\u012b', '\u012b'),
+        ('\u012d', '\u012d'), ('\u012f', '\u012f'),
+        ('\u0131', '\u0131'), ('\u0133', '\u0133'),
+        ('\u0135', '\u0135'), ('\u0137', '\u0138'),
+        ('\u013a', '\u013a'), ('\u013c', '\u013c'),
+        ('\u013e', '\u013e'), ('\u0140', '\u0140'),
+        ('\u0142', '\u0142'), ('\u0144', '\u0144'),
+        ('\u0146', '\u0146'), ('\u0148', '\u0149'),
+        ('\u014b', '\u014b'), ('\u014d', '\u014d'),
+        ('\u014f', '\u014f'), ('\u0151', '\u0151'),
+        ('\u0153', '\u0153'), ('\u0155', '\u0155'),
+        ('\u0157', '\u0157'), ('\u0159', '\u0159'),
+        ('\u015b', '\u015b'), ('\u015d', '\u015d'),
+        ('\u015f', '\u015f'), ('\u0161', '\u0161'),
+        ('\u0163', '\u0163'), ('\u0165', '\u0165'),
+        ('\u0167', '\u0167'), ('\u0169', '\u0169'),
+        ('\u016b', '\u016b'), ('\u016d', '\u016d'),
+        ('\u016f', '\u016f'), ('\u0171', '\u0171'),
+        ('\u0173', '\u0173'), ('\u0175', '\u0175'),
+        ('\u0177', '\u0177'), ('\u017a', '\u017a'),
+        ('\u017c', '\u017c'), ('\u017e', '\u0180'),
+        ('\u0183', '\u0183'), ('\u0185', '\u0185'),
+        ('\u0188', '\u0188'), ('\u018c', '\u018d'),
+        ('\u0192', '\u0192'), ('\u0195', '\u0195'),
+        ('\u0199', '\u019b'), ('\u019e', '\u019e'),
+        ('\u01a1', '\u01a1'), ('\u01a3', '\u01a3'),
+        ('\u01a5', '\u01a5'), ('\u01a8', '\u01a8'),
+        ('\u01aa', '\u01ab'), ('\u01ad', '\u01ad'),
+        ('\u01b0', '\u01b0'), ('\u01b4', '\u01b4'),
+        ('\u01b6', '\u01b6'), ('\u01b9', '\u01ba'),
+        ('\u01bd', '\u01bf'), ('\u01c6', '\u01c6'),
+        ('\u01c9', '\u01c9'), ('\u01cc', '\u01cc'),
+        ('\u01ce', '\u01ce'), ('\u01d0', '\u01d0'),
+        ('\u01d2', '\u01d2'), ('\u01d4', '\u01d4'),
+        ('\u01d6', '\u01d6'), ('\u01d8', '\u01d8'),
+        ('\u01da', '\u01da'), ('\u01dc', '\u01dd'),
+        ('\u01df', '\u01df'), ('\u01e1', '\u01e1'),
+        ('\u01e3', '\u01e3'), ('\u01e5', '\u01e5'),
+        ('\u01e7', '\u01e7'), ('\u01e9', '\u01e9'),
+        ('\u01eb', '\u01eb'), ('\u01ed', '\u01ed'),
+        ('\u01ef', '\u01f0'), ('\u01f3', '\u01f3'),
+        ('\u01f5', '\u01f5'), ('\u01f9', '\u01f9'),
+        ('\u01fb', '\u01fb'), ('\u01fd', '\u01fd'),
+        ('\u01ff', '\u01ff'), ('\u0201', '\u0201'),
+        ('\u0203', '\u0203'), ('\u0205', '\u0205'),
+        ('\u0207', '\u0207'), ('\u0209', '\u0209'),
+        ('\u020b', '\u020b'), ('\u020d', '\u020d'),
+        ('\u020f', '\u020f'), ('\u0211', '\u0211'),
+        ('\u0213', '\u0213'), ('\u0215', '\u0215'),
+        ('\u0217', '\u0217'), ('\u0219', '\u0219'),
+        ('\u021b', '\u021b'), ('\u021d', '\u021d'),
+        ('\u021f', '\u021f'), ('\u0221', '\u0221'),
+        ('\u0223', '\u0223'), ('\u0225', '\u0225'),
+        ('\u0227', '\u0227'), ('\u0229', '\u0229'),
+        ('\u022b', '\u022b'), ('\u022d', '\u022d'),
+        ('\u022f', '\u022f'), ('\u0231', '\u0231'),
+        ('\u0233', '\u0239'), ('\u023c', '\u023c'),
+        ('\u023f', '\u0240'), ('\u0242', '\u0242'),
+        ('\u0247', '\u0247'), ('\u0249', '\u0249'),
+        ('\u024b', '\u024b'), ('\u024d', '\u024d'),
+        ('\u024f', '\u0293'), ('\u0295', '\u02af'),
+        ('\u02b0', '\u02b8'), ('\u02c0', '\u02c1'),
+        ('\u02e0', '\u02e4'), ('\u0345', '\u0345'),
+        ('\u0371', '\u0371'), ('\u0373', '\u0373'),
+        ('\u0377', '\u0377'), ('\u037a', '\u037a'),
+        ('\u037b', '\u037d'), ('\u0390', '\u0390'),
+        ('\u03ac', '\u03ce'), ('\u03d0', '\u03d1'),
+        ('\u03d5', '\u03d7'), ('\u03d9', '\u03d9'),
+        ('\u03db', '\u03db'), ('\u03dd', '\u03dd'),
+        ('\u03df', '\u03df'), ('\u03e1', '\u03e1'),
+        ('\u03e3', '\u03e3'), ('\u03e5', '\u03e5'),
+        ('\u03e7', '\u03e7'), ('\u03e9', '\u03e9'),
+        ('\u03eb', '\u03eb'), ('\u03ed', '\u03ed'),
+        ('\u03ef', '\u03f3'), ('\u03f5', '\u03f5'),
+        ('\u03f8', '\u03f8'), ('\u03fb', '\u03fc'),
+        ('\u0430', '\u045f'), ('\u0461', '\u0461'),
+        ('\u0463', '\u0463'), ('\u0465', '\u0465'),
+        ('\u0467', '\u0467'), ('\u0469', '\u0469'),
+        ('\u046b', '\u046b'), ('\u046d', '\u046d'),
+        ('\u046f', '\u046f'), ('\u0471', '\u0471'),
+        ('\u0473', '\u0473'), ('\u0475', '\u0475'),
+        ('\u0477', '\u0477'), ('\u0479', '\u0479'),
+        ('\u047b', '\u047b'), ('\u047d', '\u047d'),
+        ('\u047f', '\u047f'), ('\u0481', '\u0481'),
+        ('\u048b', '\u048b'), ('\u048d', '\u048d'),
+        ('\u048f', '\u048f'), ('\u0491', '\u0491'),
+        ('\u0493', '\u0493'), ('\u0495', '\u0495'),
+        ('\u0497', '\u0497'), ('\u0499', '\u0499'),
+        ('\u049b', '\u049b'), ('\u049d', '\u049d'),
+        ('\u049f', '\u049f'), ('\u04a1', '\u04a1'),
+        ('\u04a3', '\u04a3'), ('\u04a5', '\u04a5'),
+        ('\u04a7', '\u04a7'), ('\u04a9', '\u04a9'),
+        ('\u04ab', '\u04ab'), ('\u04ad', '\u04ad'),
+        ('\u04af', '\u04af'), ('\u04b1', '\u04b1'),
+        ('\u04b3', '\u04b3'), ('\u04b5', '\u04b5'),
+        ('\u04b7', '\u04b7'), ('\u04b9', '\u04b9'),
+        ('\u04bb', '\u04bb'), ('\u04bd', '\u04bd'),
+        ('\u04bf', '\u04bf'), ('\u04c2', '\u04c2'),
+        ('\u04c4', '\u04c4'), ('\u04c6', '\u04c6'),
+        ('\u04c8', '\u04c8'), ('\u04ca', '\u04ca'),
+        ('\u04cc', '\u04cc'), ('\u04ce', '\u04cf'),
+        ('\u04d1', '\u04d1'), ('\u04d3', '\u04d3'),
+        ('\u04d5', '\u04d5'), ('\u04d7', '\u04d7'),
+        ('\u04d9', '\u04d9'), ('\u04db', '\u04db'),
+        ('\u04dd', '\u04dd'), ('\u04df', '\u04df'),
+        ('\u04e1', '\u04e1'), ('\u04e3', '\u04e3'),
+        ('\u04e5', '\u04e5'), ('\u04e7', '\u04e7'),
+        ('\u04e9', '\u04e9'), ('\u04eb', '\u04eb'),
+        ('\u04ed', '\u04ed'), ('\u04ef', '\u04ef'),
+        ('\u04f1', '\u04f1'), ('\u04f3', '\u04f3'),
+        ('\u04f5', '\u04f5'), ('\u04f7', '\u04f7'),
+        ('\u04f9', '\u04f9'), ('\u04fb', '\u04fb'),
+        ('\u04fd', '\u04fd'), ('\u04ff', '\u04ff'),
+        ('\u0501', '\u0501'), ('\u0503', '\u0503'),
+        ('\u0505', '\u0505'), ('\u0507', '\u0507'),
+        ('\u0509', '\u0509'), ('\u050b', '\u050b'),
+        ('\u050d', '\u050d'), ('\u050f', '\u050f'),
+        ('\u0511', '\u0511'), ('\u0513', '\u0513'),
+        ('\u0515', '\u0515'), ('\u0517', '\u0517'),
+        ('\u0519', '\u0519'), ('\u051b', '\u051b'),
+        ('\u051d', '\u051d'), ('\u051f', '\u051f'),
+        ('\u0521', '\u0521'), ('\u0523', '\u0523'),
+        ('\u0525', '\u0525'), ('\u0527', '\u0527'),
+        ('\u0561', '\u0587'), ('\u1d00', '\u1d2b'),
+        ('\u1d2c', '\u1d6a'), ('\u1d6b', '\u1d77'),
+        ('\u1d78', '\u1d78'), ('\u1d79', '\u1d9a'),
+        ('\u1d9b', '\u1dbf'), ('\u1e01', '\u1e01'),
+        ('\u1e03', '\u1e03'), ('\u1e05', '\u1e05'),
+        ('\u1e07', '\u1e07'), ('\u1e09', '\u1e09'),
+        ('\u1e0b', '\u1e0b'), ('\u1e0d', '\u1e0d'),
+        ('\u1e0f', '\u1e0f'), ('\u1e11', '\u1e11'),
+        ('\u1e13', '\u1e13'), ('\u1e15', '\u1e15'),
+        ('\u1e17', '\u1e17'), ('\u1e19', '\u1e19'),
+        ('\u1e1b', '\u1e1b'), ('\u1e1d', '\u1e1d'),
+        ('\u1e1f', '\u1e1f'), ('\u1e21', '\u1e21'),
+        ('\u1e23', '\u1e23'), ('\u1e25', '\u1e25'),
+        ('\u1e27', '\u1e27'), ('\u1e29', '\u1e29'),
+        ('\u1e2b', '\u1e2b'), ('\u1e2d', '\u1e2d'),
+        ('\u1e2f', '\u1e2f'), ('\u1e31', '\u1e31'),
+        ('\u1e33', '\u1e33'), ('\u1e35', '\u1e35'),
+        ('\u1e37', '\u1e37'), ('\u1e39', '\u1e39'),
+        ('\u1e3b', '\u1e3b'), ('\u1e3d', '\u1e3d'),
+        ('\u1e3f', '\u1e3f'), ('\u1e41', '\u1e41'),
+        ('\u1e43', '\u1e43'), ('\u1e45', '\u1e45'),
+        ('\u1e47', '\u1e47'), ('\u1e49', '\u1e49'),
+        ('\u1e4b', '\u1e4b'), ('\u1e4d', '\u1e4d'),
+        ('\u1e4f', '\u1e4f'), ('\u1e51', '\u1e51'),
+        ('\u1e53', '\u1e53'), ('\u1e55', '\u1e55'),
+        ('\u1e57', '\u1e57'), ('\u1e59', '\u1e59'),
+        ('\u1e5b', '\u1e5b'), ('\u1e5d', '\u1e5d'),
+        ('\u1e5f', '\u1e5f'), ('\u1e61', '\u1e61'),
+        ('\u1e63', '\u1e63'), ('\u1e65', '\u1e65'),
+        ('\u1e67', '\u1e67'), ('\u1e69', '\u1e69'),
+        ('\u1e6b', '\u1e6b'), ('\u1e6d', '\u1e6d'),
+        ('\u1e6f', '\u1e6f'), ('\u1e71', '\u1e71'),
+        ('\u1e73', '\u1e73'), ('\u1e75', '\u1e75'),
+        ('\u1e77', '\u1e77'), ('\u1e79', '\u1e79'),
+        ('\u1e7b', '\u1e7b'), ('\u1e7d', '\u1e7d'),
+        ('\u1e7f', '\u1e7f'), ('\u1e81', '\u1e81'),
+        ('\u1e83', '\u1e83'), ('\u1e85', '\u1e85'),
+        ('\u1e87', '\u1e87'), ('\u1e89', '\u1e89'),
+        ('\u1e8b', '\u1e8b'), ('\u1e8d', '\u1e8d'),
+        ('\u1e8f', '\u1e8f'), ('\u1e91', '\u1e91'),
+        ('\u1e93', '\u1e93'), ('\u1e95', '\u1e9d'),
+        ('\u1e9f', '\u1e9f'), ('\u1ea1', '\u1ea1'),
+        ('\u1ea3', '\u1ea3'), ('\u1ea5', '\u1ea5'),
+        ('\u1ea7', '\u1ea7'), ('\u1ea9', '\u1ea9'),
+        ('\u1eab', '\u1eab'), ('\u1ead', '\u1ead'),
+        ('\u1eaf', '\u1eaf'), ('\u1eb1', '\u1eb1'),
+        ('\u1eb3', '\u1eb3'), ('\u1eb5', '\u1eb5'),
+        ('\u1eb7', '\u1eb7'), ('\u1eb9', '\u1eb9'),
+        ('\u1ebb', '\u1ebb'), ('\u1ebd', '\u1ebd'),
+        ('\u1ebf', '\u1ebf'), ('\u1ec1', '\u1ec1'),
+        ('\u1ec3', '\u1ec3'), ('\u1ec5', '\u1ec5'),
+        ('\u1ec7', '\u1ec7'), ('\u1ec9', '\u1ec9'),
+        ('\u1ecb', '\u1ecb'), ('\u1ecd', '\u1ecd'),
+        ('\u1ecf', '\u1ecf'), ('\u1ed1', '\u1ed1'),
+        ('\u1ed3', '\u1ed3'), ('\u1ed5', '\u1ed5'),
+        ('\u1ed7', '\u1ed7'), ('\u1ed9', '\u1ed9'),
+        ('\u1edb', '\u1edb'), ('\u1edd', '\u1edd'),
+        ('\u1edf', '\u1edf'), ('\u1ee1', '\u1ee1'),
+        ('\u1ee3', '\u1ee3'), ('\u1ee5', '\u1ee5'),
+        ('\u1ee7', '\u1ee7'), ('\u1ee9', '\u1ee9'),
+        ('\u1eeb', '\u1eeb'), ('\u1eed', '\u1eed'),
+        ('\u1eef', '\u1eef'), ('\u1ef1', '\u1ef1'),
+        ('\u1ef3', '\u1ef3'), ('\u1ef5', '\u1ef5'),
+        ('\u1ef7', '\u1ef7'), ('\u1ef9', '\u1ef9'),
+        ('\u1efb', '\u1efb'), ('\u1efd', '\u1efd'),
+        ('\u1eff', '\u1f07'), ('\u1f10', '\u1f15'),
+        ('\u1f20', '\u1f27'), ('\u1f30', '\u1f37'),
+        ('\u1f40', '\u1f45'), ('\u1f50', '\u1f57'),
+        ('\u1f60', '\u1f67'), ('\u1f70', '\u1f7d'),
+        ('\u1f80', '\u1f87'), ('\u1f90', '\u1f97'),
+        ('\u1fa0', '\u1fa7'), ('\u1fb0', '\u1fb4'),
+        ('\u1fb6', '\u1fb7'), ('\u1fbe', '\u1fbe'),
+        ('\u1fc2', '\u1fc4'), ('\u1fc6', '\u1fc7'),
+        ('\u1fd0', '\u1fd3'), ('\u1fd6', '\u1fd7'),
+        ('\u1fe0', '\u1fe7'), ('\u1ff2', '\u1ff4'),
+        ('\u1ff6', '\u1ff7'), ('\u2071', '\u2071'),
+        ('\u207f', '\u207f'), ('\u2090', '\u209c'),
+        ('\u210a', '\u210a'), ('\u210e', '\u210f'),
+        ('\u2113', '\u2113'), ('\u212f', '\u212f'),
+        ('\u2134', '\u2134'), ('\u2139', '\u2139'),
+        ('\u213c', '\u213d'), ('\u2146', '\u2149'),
+        ('\u214e', '\u214e'), ('\u2170', '\u217f'),
+        ('\u2184', '\u2184'), ('\u24d0', '\u24e9'),
+        ('\u2c30', '\u2c5e'), ('\u2c61', '\u2c61'),
+        ('\u2c65', '\u2c66'), ('\u2c68', '\u2c68'),
+        ('\u2c6a', '\u2c6a'), ('\u2c6c', '\u2c6c'),
+        ('\u2c71', '\u2c71'), ('\u2c73', '\u2c74'),
+        ('\u2c76', '\u2c7b'), ('\u2c7c', '\u2c7d'),
+        ('\u2c81', '\u2c81'), ('\u2c83', '\u2c83'),
+        ('\u2c85', '\u2c85'), ('\u2c87', '\u2c87'),
+        ('\u2c89', '\u2c89'), ('\u2c8b', '\u2c8b'),
+        ('\u2c8d', '\u2c8d'), ('\u2c8f', '\u2c8f'),
+        ('\u2c91', '\u2c91'), ('\u2c93', '\u2c93'),
+        ('\u2c95', '\u2c95'), ('\u2c97', '\u2c97'),
+        ('\u2c99', '\u2c99'), ('\u2c9b', '\u2c9b'),
+        ('\u2c9d', '\u2c9d'), ('\u2c9f', '\u2c9f'),
+        ('\u2ca1', '\u2ca1'), ('\u2ca3', '\u2ca3'),
+        ('\u2ca5', '\u2ca5'), ('\u2ca7', '\u2ca7'),
+        ('\u2ca9', '\u2ca9'), ('\u2cab', '\u2cab'),
+        ('\u2cad', '\u2cad'), ('\u2caf', '\u2caf'),
+        ('\u2cb1', '\u2cb1'), ('\u2cb3', '\u2cb3'),
+        ('\u2cb5', '\u2cb5'), ('\u2cb7', '\u2cb7'),
+        ('\u2cb9', '\u2cb9'), ('\u2cbb', '\u2cbb'),
+        ('\u2cbd', '\u2cbd'), ('\u2cbf', '\u2cbf'),
+        ('\u2cc1', '\u2cc1'), ('\u2cc3', '\u2cc3'),
+        ('\u2cc5', '\u2cc5'), ('\u2cc7', '\u2cc7'),
+        ('\u2cc9', '\u2cc9'), ('\u2ccb', '\u2ccb'),
+        ('\u2ccd', '\u2ccd'), ('\u2ccf', '\u2ccf'),
+        ('\u2cd1', '\u2cd1'), ('\u2cd3', '\u2cd3'),
+        ('\u2cd5', '\u2cd5'), ('\u2cd7', '\u2cd7'),
+        ('\u2cd9', '\u2cd9'), ('\u2cdb', '\u2cdb'),
+        ('\u2cdd', '\u2cdd'), ('\u2cdf', '\u2cdf'),
+        ('\u2ce1', '\u2ce1'), ('\u2ce3', '\u2ce4'),
+        ('\u2cec', '\u2cec'), ('\u2cee', '\u2cee'),
+        ('\u2cf3', '\u2cf3'), ('\u2d00', '\u2d25'),
+        ('\u2d27', '\u2d27'), ('\u2d2d', '\u2d2d'),
+        ('\ua641', '\ua641'), ('\ua643', '\ua643'),
+        ('\ua645', '\ua645'), ('\ua647', '\ua647'),
+        ('\ua649', '\ua649'), ('\ua64b', '\ua64b'),
+        ('\ua64d', '\ua64d'), ('\ua64f', '\ua64f'),
+        ('\ua651', '\ua651'), ('\ua653', '\ua653'),
+        ('\ua655', '\ua655'), ('\ua657', '\ua657'),
+        ('\ua659', '\ua659'), ('\ua65b', '\ua65b'),
+        ('\ua65d', '\ua65d'), ('\ua65f', '\ua65f'),
+        ('\ua661', '\ua661'), ('\ua663', '\ua663'),
+        ('\ua665', '\ua665'), ('\ua667', '\ua667'),
+        ('\ua669', '\ua669'), ('\ua66b', '\ua66b'),
+        ('\ua66d', '\ua66d'), ('\ua681', '\ua681'),
+        ('\ua683', '\ua683'), ('\ua685', '\ua685'),
+        ('\ua687', '\ua687'), ('\ua689', '\ua689'),
+        ('\ua68b', '\ua68b'), ('\ua68d', '\ua68d'),
+        ('\ua68f', '\ua68f'), ('\ua691', '\ua691'),
+        ('\ua693', '\ua693'), ('\ua695', '\ua695'),
+        ('\ua697', '\ua697'), ('\ua723', '\ua723'),
+        ('\ua725', '\ua725'), ('\ua727', '\ua727'),
+        ('\ua729', '\ua729'), ('\ua72b', '\ua72b'),
+        ('\ua72d', '\ua72d'), ('\ua72f', '\ua731'),
+        ('\ua733', '\ua733'), ('\ua735', '\ua735'),
+        ('\ua737', '\ua737'), ('\ua739', '\ua739'),
+        ('\ua73b', '\ua73b'), ('\ua73d', '\ua73d'),
+        ('\ua73f', '\ua73f'), ('\ua741', '\ua741'),
+        ('\ua743', '\ua743'), ('\ua745', '\ua745'),
+        ('\ua747', '\ua747'), ('\ua749', '\ua749'),
+        ('\ua74b', '\ua74b'), ('\ua74d', '\ua74d'),
+        ('\ua74f', '\ua74f'), ('\ua751', '\ua751'),
+        ('\ua753', '\ua753'), ('\ua755', '\ua755'),
+        ('\ua757', '\ua757'), ('\ua759', '\ua759'),
+        ('\ua75b', '\ua75b'), ('\ua75d', '\ua75d'),
+        ('\ua75f', '\ua75f'), ('\ua761', '\ua761'),
+        ('\ua763', '\ua763'), ('\ua765', '\ua765'),
+        ('\ua767', '\ua767'), ('\ua769', '\ua769'),
+        ('\ua76b', '\ua76b'), ('\ua76d', '\ua76d'),
+        ('\ua76f', '\ua76f'), ('\ua770', '\ua770'),
+        ('\ua771', '\ua778'), ('\ua77a', '\ua77a'),
+        ('\ua77c', '\ua77c'), ('\ua77f', '\ua77f'),
+        ('\ua781', '\ua781'), ('\ua783', '\ua783'),
+        ('\ua785', '\ua785'), ('\ua787', '\ua787'),
+        ('\ua78c', '\ua78c'), ('\ua78e', '\ua78e'),
+        ('\ua791', '\ua791'), ('\ua793', '\ua793'),
+        ('\ua7a1', '\ua7a1'), ('\ua7a3', '\ua7a3'),
+        ('\ua7a5', '\ua7a5'), ('\ua7a7', '\ua7a7'),
+        ('\ua7a9', '\ua7a9'), ('\ua7f8', '\ua7f9'),
+        ('\ua7fa', '\ua7fa'), ('\ufb00', '\ufb06'),
+        ('\ufb13', '\ufb17'), ('\uff41', '\uff5a'),
+        ('\U00010428', '\U0001044f'), ('\U0001d41a', '\U0001d433'),
+        ('\U0001d44e', '\U0001d454'), ('\U0001d456', '\U0001d467'),
+        ('\U0001d482', '\U0001d49b'), ('\U0001d4b6', '\U0001d4b9'),
+        ('\U0001d4bb', '\U0001d4bb'), ('\U0001d4bd', '\U0001d4c3'),
+        ('\U0001d4c5', '\U0001d4cf'), ('\U0001d4ea', '\U0001d503'),
+        ('\U0001d51e', '\U0001d537'), ('\U0001d552', '\U0001d56b'),
+        ('\U0001d586', '\U0001d59f'), ('\U0001d5ba', '\U0001d5d3'),
+        ('\U0001d5ee', '\U0001d607'), ('\U0001d622', '\U0001d63b'),
+        ('\U0001d656', '\U0001d66f'), ('\U0001d68a', '\U0001d6a5'),
+        ('\U0001d6c2', '\U0001d6da'), ('\U0001d6dc', '\U0001d6e1'),
+        ('\U0001d6fc', '\U0001d714'), ('\U0001d716', '\U0001d71b'),
+        ('\U0001d736', '\U0001d74e'), ('\U0001d750', '\U0001d755'),
+        ('\U0001d770', '\U0001d788'), ('\U0001d78a', '\U0001d78f'),
+        ('\U0001d7aa', '\U0001d7c2'), ('\U0001d7c4', '\U0001d7c9'),
+        ('\U0001d7cb', '\U0001d7cb')
+    ];
+
+    pub fn Lowercase(c: char) -> bool {
+        bsearch_range_table(c, Lowercase_table)
+    }
+
+    static Uppercase_table : &'static [(char,char)] = &[
+        ('\x41', '\x5a'), ('\xc0', '\xd6'),
+        ('\xd8', '\xde'), ('\u0100', '\u0100'),
+        ('\u0102', '\u0102'), ('\u0104', '\u0104'),
+        ('\u0106', '\u0106'), ('\u0108', '\u0108'),
+        ('\u010a', '\u010a'), ('\u010c', '\u010c'),
+        ('\u010e', '\u010e'), ('\u0110', '\u0110'),
+        ('\u0112', '\u0112'), ('\u0114', '\u0114'),
+        ('\u0116', '\u0116'), ('\u0118', '\u0118'),
+        ('\u011a', '\u011a'), ('\u011c', '\u011c'),
+        ('\u011e', '\u011e'), ('\u0120', '\u0120'),
+        ('\u0122', '\u0122'), ('\u0124', '\u0124'),
+        ('\u0126', '\u0126'), ('\u0128', '\u0128'),
+        ('\u012a', '\u012a'), ('\u012c', '\u012c'),
+        ('\u012e', '\u012e'), ('\u0130', '\u0130'),
+        ('\u0132', '\u0132'), ('\u0134', '\u0134'),
+        ('\u0136', '\u0136'), ('\u0139', '\u0139'),
+        ('\u013b', '\u013b'), ('\u013d', '\u013d'),
+        ('\u013f', '\u013f'), ('\u0141', '\u0141'),
+        ('\u0143', '\u0143'), ('\u0145', '\u0145'),
+        ('\u0147', '\u0147'), ('\u014a', '\u014a'),
+        ('\u014c', '\u014c'), ('\u014e', '\u014e'),
+        ('\u0150', '\u0150'), ('\u0152', '\u0152'),
+        ('\u0154', '\u0154'), ('\u0156', '\u0156'),
+        ('\u0158', '\u0158'), ('\u015a', '\u015a'),
+        ('\u015c', '\u015c'), ('\u015e', '\u015e'),
+        ('\u0160', '\u0160'), ('\u0162', '\u0162'),
+        ('\u0164', '\u0164'), ('\u0166', '\u0166'),
+        ('\u0168', '\u0168'), ('\u016a', '\u016a'),
+        ('\u016c', '\u016c'), ('\u016e', '\u016e'),
+        ('\u0170', '\u0170'), ('\u0172', '\u0172'),
+        ('\u0174', '\u0174'), ('\u0176', '\u0176'),
+        ('\u0178', '\u0179'), ('\u017b', '\u017b'),
+        ('\u017d', '\u017d'), ('\u0181', '\u0182'),
+        ('\u0184', '\u0184'), ('\u0186', '\u0187'),
+        ('\u0189', '\u018b'), ('\u018e', '\u0191'),
+        ('\u0193', '\u0194'), ('\u0196', '\u0198'),
+        ('\u019c', '\u019d'), ('\u019f', '\u01a0'),
+        ('\u01a2', '\u01a2'), ('\u01a4', '\u01a4'),
+        ('\u01a6', '\u01a7'), ('\u01a9', '\u01a9'),
+        ('\u01ac', '\u01ac'), ('\u01ae', '\u01af'),
+        ('\u01b1', '\u01b3'), ('\u01b5', '\u01b5'),
+        ('\u01b7', '\u01b8'), ('\u01bc', '\u01bc'),
+        ('\u01c4', '\u01c4'), ('\u01c7', '\u01c7'),
+        ('\u01ca', '\u01ca'), ('\u01cd', '\u01cd'),
+        ('\u01cf', '\u01cf'), ('\u01d1', '\u01d1'),
+        ('\u01d3', '\u01d3'), ('\u01d5', '\u01d5'),
+        ('\u01d7', '\u01d7'), ('\u01d9', '\u01d9'),
+        ('\u01db', '\u01db'), ('\u01de', '\u01de'),
+        ('\u01e0', '\u01e0'), ('\u01e2', '\u01e2'),
+        ('\u01e4', '\u01e4'), ('\u01e6', '\u01e6'),
+        ('\u01e8', '\u01e8'), ('\u01ea', '\u01ea'),
+        ('\u01ec', '\u01ec'), ('\u01ee', '\u01ee'),
+        ('\u01f1', '\u01f1'), ('\u01f4', '\u01f4'),
+        ('\u01f6', '\u01f8'), ('\u01fa', '\u01fa'),
+        ('\u01fc', '\u01fc'), ('\u01fe', '\u01fe'),
+        ('\u0200', '\u0200'), ('\u0202', '\u0202'),
+        ('\u0204', '\u0204'), ('\u0206', '\u0206'),
+        ('\u0208', '\u0208'), ('\u020a', '\u020a'),
+        ('\u020c', '\u020c'), ('\u020e', '\u020e'),
+        ('\u0210', '\u0210'), ('\u0212', '\u0212'),
+        ('\u0214', '\u0214'), ('\u0216', '\u0216'),
+        ('\u0218', '\u0218'), ('\u021a', '\u021a'),
+        ('\u021c', '\u021c'), ('\u021e', '\u021e'),
+        ('\u0220', '\u0220'), ('\u0222', '\u0222'),
+        ('\u0224', '\u0224'), ('\u0226', '\u0226'),
+        ('\u0228', '\u0228'), ('\u022a', '\u022a'),
+        ('\u022c', '\u022c'), ('\u022e', '\u022e'),
+        ('\u0230', '\u0230'), ('\u0232', '\u0232'),
+        ('\u023a', '\u023b'), ('\u023d', '\u023e'),
+        ('\u0241', '\u0241'), ('\u0243', '\u0246'),
+        ('\u0248', '\u0248'), ('\u024a', '\u024a'),
+        ('\u024c', '\u024c'), ('\u024e', '\u024e'),
+        ('\u0370', '\u0370'), ('\u0372', '\u0372'),
+        ('\u0376', '\u0376'), ('\u0386', '\u0386'),
+        ('\u0388', '\u038a'), ('\u038c', '\u038c'),
+        ('\u038e', '\u038f'), ('\u0391', '\u03a1'),
+        ('\u03a3', '\u03ab'), ('\u03cf', '\u03cf'),
+        ('\u03d2', '\u03d4'), ('\u03d8', '\u03d8'),
+        ('\u03da', '\u03da'), ('\u03dc', '\u03dc'),
+        ('\u03de', '\u03de'), ('\u03e0', '\u03e0'),
+        ('\u03e2', '\u03e2'), ('\u03e4', '\u03e4'),
+        ('\u03e6', '\u03e6'), ('\u03e8', '\u03e8'),
+        ('\u03ea', '\u03ea'), ('\u03ec', '\u03ec'),
+        ('\u03ee', '\u03ee'), ('\u03f4', '\u03f4'),
+        ('\u03f7', '\u03f7'), ('\u03f9', '\u03fa'),
+        ('\u03fd', '\u042f'), ('\u0460', '\u0460'),
+        ('\u0462', '\u0462'), ('\u0464', '\u0464'),
+        ('\u0466', '\u0466'), ('\u0468', '\u0468'),
+        ('\u046a', '\u046a'), ('\u046c', '\u046c'),
+        ('\u046e', '\u046e'), ('\u0470', '\u0470'),
+        ('\u0472', '\u0472'), ('\u0474', '\u0474'),
+        ('\u0476', '\u0476'), ('\u0478', '\u0478'),
+        ('\u047a', '\u047a'), ('\u047c', '\u047c'),
+        ('\u047e', '\u047e'), ('\u0480', '\u0480'),
+        ('\u048a', '\u048a'), ('\u048c', '\u048c'),
+        ('\u048e', '\u048e'), ('\u0490', '\u0490'),
+        ('\u0492', '\u0492'), ('\u0494', '\u0494'),
+        ('\u0496', '\u0496'), ('\u0498', '\u0498'),
+        ('\u049a', '\u049a'), ('\u049c', '\u049c'),
+        ('\u049e', '\u049e'), ('\u04a0', '\u04a0'),
+        ('\u04a2', '\u04a2'), ('\u04a4', '\u04a4'),
+        ('\u04a6', '\u04a6'), ('\u04a8', '\u04a8'),
+        ('\u04aa', '\u04aa'), ('\u04ac', '\u04ac'),
+        ('\u04ae', '\u04ae'), ('\u04b0', '\u04b0'),
+        ('\u04b2', '\u04b2'), ('\u04b4', '\u04b4'),
+        ('\u04b6', '\u04b6'), ('\u04b8', '\u04b8'),
+        ('\u04ba', '\u04ba'), ('\u04bc', '\u04bc'),
+        ('\u04be', '\u04be'), ('\u04c0', '\u04c1'),
+        ('\u04c3', '\u04c3'), ('\u04c5', '\u04c5'),
+        ('\u04c7', '\u04c7'), ('\u04c9', '\u04c9'),
+        ('\u04cb', '\u04cb'), ('\u04cd', '\u04cd'),
+        ('\u04d0', '\u04d0'), ('\u04d2', '\u04d2'),
+        ('\u04d4', '\u04d4'), ('\u04d6', '\u04d6'),
+        ('\u04d8', '\u04d8'), ('\u04da', '\u04da'),
+        ('\u04dc', '\u04dc'), ('\u04de', '\u04de'),
+        ('\u04e0', '\u04e0'), ('\u04e2', '\u04e2'),
+        ('\u04e4', '\u04e4'), ('\u04e6', '\u04e6'),
+        ('\u04e8', '\u04e8'), ('\u04ea', '\u04ea'),
+        ('\u04ec', '\u04ec'), ('\u04ee', '\u04ee'),
+        ('\u04f0', '\u04f0'), ('\u04f2', '\u04f2'),
+        ('\u04f4', '\u04f4'), ('\u04f6', '\u04f6'),
+        ('\u04f8', '\u04f8'), ('\u04fa', '\u04fa'),
+        ('\u04fc', '\u04fc'), ('\u04fe', '\u04fe'),
+        ('\u0500', '\u0500'), ('\u0502', '\u0502'),
+        ('\u0504', '\u0504'), ('\u0506', '\u0506'),
+        ('\u0508', '\u0508'), ('\u050a', '\u050a'),
+        ('\u050c', '\u050c'), ('\u050e', '\u050e'),
+        ('\u0510', '\u0510'), ('\u0512', '\u0512'),
+        ('\u0514', '\u0514'), ('\u0516', '\u0516'),
+        ('\u0518', '\u0518'), ('\u051a', '\u051a'),
+        ('\u051c', '\u051c'), ('\u051e', '\u051e'),
+        ('\u0520', '\u0520'), ('\u0522', '\u0522'),
+        ('\u0524', '\u0524'), ('\u0526', '\u0526'),
+        ('\u0531', '\u0556'), ('\u10a0', '\u10c5'),
+        ('\u10c7', '\u10c7'), ('\u10cd', '\u10cd'),
+        ('\u1e00', '\u1e00'), ('\u1e02', '\u1e02'),
+        ('\u1e04', '\u1e04'), ('\u1e06', '\u1e06'),
+        ('\u1e08', '\u1e08'), ('\u1e0a', '\u1e0a'),
+        ('\u1e0c', '\u1e0c'), ('\u1e0e', '\u1e0e'),
+        ('\u1e10', '\u1e10'), ('\u1e12', '\u1e12'),
+        ('\u1e14', '\u1e14'), ('\u1e16', '\u1e16'),
+        ('\u1e18', '\u1e18'), ('\u1e1a', '\u1e1a'),
+        ('\u1e1c', '\u1e1c'), ('\u1e1e', '\u1e1e'),
+        ('\u1e20', '\u1e20'), ('\u1e22', '\u1e22'),
+        ('\u1e24', '\u1e24'), ('\u1e26', '\u1e26'),
+        ('\u1e28', '\u1e28'), ('\u1e2a', '\u1e2a'),
+        ('\u1e2c', '\u1e2c'), ('\u1e2e', '\u1e2e'),
+        ('\u1e30', '\u1e30'), ('\u1e32', '\u1e32'),
+        ('\u1e34', '\u1e34'), ('\u1e36', '\u1e36'),
+        ('\u1e38', '\u1e38'), ('\u1e3a', '\u1e3a'),
+        ('\u1e3c', '\u1e3c'), ('\u1e3e', '\u1e3e'),
+        ('\u1e40', '\u1e40'), ('\u1e42', '\u1e42'),
+        ('\u1e44', '\u1e44'), ('\u1e46', '\u1e46'),
+        ('\u1e48', '\u1e48'), ('\u1e4a', '\u1e4a'),
+        ('\u1e4c', '\u1e4c'), ('\u1e4e', '\u1e4e'),
+        ('\u1e50', '\u1e50'), ('\u1e52', '\u1e52'),
+        ('\u1e54', '\u1e54'), ('\u1e56', '\u1e56'),
+        ('\u1e58', '\u1e58'), ('\u1e5a', '\u1e5a'),
+        ('\u1e5c', '\u1e5c'), ('\u1e5e', '\u1e5e'),
+        ('\u1e60', '\u1e60'), ('\u1e62', '\u1e62'),
+        ('\u1e64', '\u1e64'), ('\u1e66', '\u1e66'),
+        ('\u1e68', '\u1e68'), ('\u1e6a', '\u1e6a'),
+        ('\u1e6c', '\u1e6c'), ('\u1e6e', '\u1e6e'),
+        ('\u1e70', '\u1e70'), ('\u1e72', '\u1e72'),
+        ('\u1e74', '\u1e74'), ('\u1e76', '\u1e76'),
+        ('\u1e78', '\u1e78'), ('\u1e7a', '\u1e7a'),
+        ('\u1e7c', '\u1e7c'), ('\u1e7e', '\u1e7e'),
+        ('\u1e80', '\u1e80'), ('\u1e82', '\u1e82'),
+        ('\u1e84', '\u1e84'), ('\u1e86', '\u1e86'),
+        ('\u1e88', '\u1e88'), ('\u1e8a', '\u1e8a'),
+        ('\u1e8c', '\u1e8c'), ('\u1e8e', '\u1e8e'),
+        ('\u1e90', '\u1e90'), ('\u1e92', '\u1e92'),
+        ('\u1e94', '\u1e94'), ('\u1e9e', '\u1e9e'),
+        ('\u1ea0', '\u1ea0'), ('\u1ea2', '\u1ea2'),
+        ('\u1ea4', '\u1ea4'), ('\u1ea6', '\u1ea6'),
+        ('\u1ea8', '\u1ea8'), ('\u1eaa', '\u1eaa'),
+        ('\u1eac', '\u1eac'), ('\u1eae', '\u1eae'),
+        ('\u1eb0', '\u1eb0'), ('\u1eb2', '\u1eb2'),
+        ('\u1eb4', '\u1eb4'), ('\u1eb6', '\u1eb6'),
+        ('\u1eb8', '\u1eb8'), ('\u1eba', '\u1eba'),
+        ('\u1ebc', '\u1ebc'), ('\u1ebe', '\u1ebe'),
+        ('\u1ec0', '\u1ec0'), ('\u1ec2', '\u1ec2'),
+        ('\u1ec4', '\u1ec4'), ('\u1ec6', '\u1ec6'),
+        ('\u1ec8', '\u1ec8'), ('\u1eca', '\u1eca'),
+        ('\u1ecc', '\u1ecc'), ('\u1ece', '\u1ece'),
+        ('\u1ed0', '\u1ed0'), ('\u1ed2', '\u1ed2'),
+        ('\u1ed4', '\u1ed4'), ('\u1ed6', '\u1ed6'),
+        ('\u1ed8', '\u1ed8'), ('\u1eda', '\u1eda'),
+        ('\u1edc', '\u1edc'), ('\u1ede', '\u1ede'),
+        ('\u1ee0', '\u1ee0'), ('\u1ee2', '\u1ee2'),
+        ('\u1ee4', '\u1ee4'), ('\u1ee6', '\u1ee6'),
+        ('\u1ee8', '\u1ee8'), ('\u1eea', '\u1eea'),
+        ('\u1eec', '\u1eec'), ('\u1eee', '\u1eee'),
+        ('\u1ef0', '\u1ef0'), ('\u1ef2', '\u1ef2'),
+        ('\u1ef4', '\u1ef4'), ('\u1ef6', '\u1ef6'),
+        ('\u1ef8', '\u1ef8'), ('\u1efa', '\u1efa'),
+        ('\u1efc', '\u1efc'), ('\u1efe', '\u1efe'),
+        ('\u1f08', '\u1f0f'), ('\u1f18', '\u1f1d'),
+        ('\u1f28', '\u1f2f'), ('\u1f38', '\u1f3f'),
+        ('\u1f48', '\u1f4d'), ('\u1f59', '\u1f59'),
+        ('\u1f5b', '\u1f5b'), ('\u1f5d', '\u1f5d'),
+        ('\u1f5f', '\u1f5f'), ('\u1f68', '\u1f6f'),
+        ('\u1fb8', '\u1fbb'), ('\u1fc8', '\u1fcb'),
+        ('\u1fd8', '\u1fdb'), ('\u1fe8', '\u1fec'),
+        ('\u1ff8', '\u1ffb'), ('\u2102', '\u2102'),
+        ('\u2107', '\u2107'), ('\u210b', '\u210d'),
+        ('\u2110', '\u2112'), ('\u2115', '\u2115'),
+        ('\u2119', '\u211d'), ('\u2124', '\u2124'),
+        ('\u2126', '\u2126'), ('\u2128', '\u2128'),
+        ('\u212a', '\u212d'), ('\u2130', '\u2133'),
+        ('\u213e', '\u213f'), ('\u2145', '\u2145'),
+        ('\u2160', '\u216f'), ('\u2183', '\u2183'),
+        ('\u24b6', '\u24cf'), ('\u2c00', '\u2c2e'),
+        ('\u2c60', '\u2c60'), ('\u2c62', '\u2c64'),
+        ('\u2c67', '\u2c67'), ('\u2c69', '\u2c69'),
+        ('\u2c6b', '\u2c6b'), ('\u2c6d', '\u2c70'),
+        ('\u2c72', '\u2c72'), ('\u2c75', '\u2c75'),
+        ('\u2c7e', '\u2c80'), ('\u2c82', '\u2c82'),
+        ('\u2c84', '\u2c84'), ('\u2c86', '\u2c86'),
+        ('\u2c88', '\u2c88'), ('\u2c8a', '\u2c8a'),
+        ('\u2c8c', '\u2c8c'), ('\u2c8e', '\u2c8e'),
+        ('\u2c90', '\u2c90'), ('\u2c92', '\u2c92'),
+        ('\u2c94', '\u2c94'), ('\u2c96', '\u2c96'),
+        ('\u2c98', '\u2c98'), ('\u2c9a', '\u2c9a'),
+        ('\u2c9c', '\u2c9c'), ('\u2c9e', '\u2c9e'),
+        ('\u2ca0', '\u2ca0'), ('\u2ca2', '\u2ca2'),
+        ('\u2ca4', '\u2ca4'), ('\u2ca6', '\u2ca6'),
+        ('\u2ca8', '\u2ca8'), ('\u2caa', '\u2caa'),
+        ('\u2cac', '\u2cac'), ('\u2cae', '\u2cae'),
+        ('\u2cb0', '\u2cb0'), ('\u2cb2', '\u2cb2'),
+        ('\u2cb4', '\u2cb4'), ('\u2cb6', '\u2cb6'),
+        ('\u2cb8', '\u2cb8'), ('\u2cba', '\u2cba'),
+        ('\u2cbc', '\u2cbc'), ('\u2cbe', '\u2cbe'),
+        ('\u2cc0', '\u2cc0'), ('\u2cc2', '\u2cc2'),
+        ('\u2cc4', '\u2cc4'), ('\u2cc6', '\u2cc6'),
+        ('\u2cc8', '\u2cc8'), ('\u2cca', '\u2cca'),
+        ('\u2ccc', '\u2ccc'), ('\u2cce', '\u2cce'),
+        ('\u2cd0', '\u2cd0'), ('\u2cd2', '\u2cd2'),
+        ('\u2cd4', '\u2cd4'), ('\u2cd6', '\u2cd6'),
+        ('\u2cd8', '\u2cd8'), ('\u2cda', '\u2cda'),
+        ('\u2cdc', '\u2cdc'), ('\u2cde', '\u2cde'),
+        ('\u2ce0', '\u2ce0'), ('\u2ce2', '\u2ce2'),
+        ('\u2ceb', '\u2ceb'), ('\u2ced', '\u2ced'),
+        ('\u2cf2', '\u2cf2'), ('\ua640', '\ua640'),
+        ('\ua642', '\ua642'), ('\ua644', '\ua644'),
+        ('\ua646', '\ua646'), ('\ua648', '\ua648'),
+        ('\ua64a', '\ua64a'), ('\ua64c', '\ua64c'),
+        ('\ua64e', '\ua64e'), ('\ua650', '\ua650'),
+        ('\ua652', '\ua652'), ('\ua654', '\ua654'),
+        ('\ua656', '\ua656'), ('\ua658', '\ua658'),
+        ('\ua65a', '\ua65a'), ('\ua65c', '\ua65c'),
+        ('\ua65e', '\ua65e'), ('\ua660', '\ua660'),
+        ('\ua662', '\ua662'), ('\ua664', '\ua664'),
+        ('\ua666', '\ua666'), ('\ua668', '\ua668'),
+        ('\ua66a', '\ua66a'), ('\ua66c', '\ua66c'),
+        ('\ua680', '\ua680'), ('\ua682', '\ua682'),
+        ('\ua684', '\ua684'), ('\ua686', '\ua686'),
+        ('\ua688', '\ua688'), ('\ua68a', '\ua68a'),
+        ('\ua68c', '\ua68c'), ('\ua68e', '\ua68e'),
+        ('\ua690', '\ua690'), ('\ua692', '\ua692'),
+        ('\ua694', '\ua694'), ('\ua696', '\ua696'),
+        ('\ua722', '\ua722'), ('\ua724', '\ua724'),
+        ('\ua726', '\ua726'), ('\ua728', '\ua728'),
+        ('\ua72a', '\ua72a'), ('\ua72c', '\ua72c'),
+        ('\ua72e', '\ua72e'), ('\ua732', '\ua732'),
+        ('\ua734', '\ua734'), ('\ua736', '\ua736'),
+        ('\ua738', '\ua738'), ('\ua73a', '\ua73a'),
+        ('\ua73c', '\ua73c'), ('\ua73e', '\ua73e'),
+        ('\ua740', '\ua740'), ('\ua742', '\ua742'),
+        ('\ua744', '\ua744'), ('\ua746', '\ua746'),
+        ('\ua748', '\ua748'), ('\ua74a', '\ua74a'),
+        ('\ua74c', '\ua74c'), ('\ua74e', '\ua74e'),
+        ('\ua750', '\ua750'), ('\ua752', '\ua752'),
+        ('\ua754', '\ua754'), ('\ua756', '\ua756'),
+        ('\ua758', '\ua758'), ('\ua75a', '\ua75a'),
+        ('\ua75c', '\ua75c'), ('\ua75e', '\ua75e'),
+        ('\ua760', '\ua760'), ('\ua762', '\ua762'),
+        ('\ua764', '\ua764'), ('\ua766', '\ua766'),
+        ('\ua768', '\ua768'), ('\ua76a', '\ua76a'),
+        ('\ua76c', '\ua76c'), ('\ua76e', '\ua76e'),
+        ('\ua779', '\ua779'), ('\ua77b', '\ua77b'),
+        ('\ua77d', '\ua77e'), ('\ua780', '\ua780'),
+        ('\ua782', '\ua782'), ('\ua784', '\ua784'),
+        ('\ua786', '\ua786'), ('\ua78b', '\ua78b'),
+        ('\ua78d', '\ua78d'), ('\ua790', '\ua790'),
+        ('\ua792', '\ua792'), ('\ua7a0', '\ua7a0'),
+        ('\ua7a2', '\ua7a2'), ('\ua7a4', '\ua7a4'),
+        ('\ua7a6', '\ua7a6'), ('\ua7a8', '\ua7a8'),
+        ('\ua7aa', '\ua7aa'), ('\uff21', '\uff3a'),
+        ('\U00010400', '\U00010427'), ('\U0001d400', '\U0001d419'),
+        ('\U0001d434', '\U0001d44d'), ('\U0001d468', '\U0001d481'),
+        ('\U0001d49c', '\U0001d49c'), ('\U0001d49e', '\U0001d49f'),
+        ('\U0001d4a2', '\U0001d4a2'), ('\U0001d4a5', '\U0001d4a6'),
+        ('\U0001d4a9', '\U0001d4ac'), ('\U0001d4ae', '\U0001d4b5'),
+        ('\U0001d4d0', '\U0001d4e9'), ('\U0001d504', '\U0001d505'),
+        ('\U0001d507', '\U0001d50a'), ('\U0001d50d', '\U0001d514'),
+        ('\U0001d516', '\U0001d51c'), ('\U0001d538', '\U0001d539'),
+        ('\U0001d53b', '\U0001d53e'), ('\U0001d540', '\U0001d544'),
+        ('\U0001d546', '\U0001d546'), ('\U0001d54a', '\U0001d550'),
+        ('\U0001d56c', '\U0001d585'), ('\U0001d5a0', '\U0001d5b9'),
+        ('\U0001d5d4', '\U0001d5ed'), ('\U0001d608', '\U0001d621'),
+        ('\U0001d63c', '\U0001d655'), ('\U0001d670', '\U0001d689'),
+        ('\U0001d6a8', '\U0001d6c0'), ('\U0001d6e2', '\U0001d6fa'),
+        ('\U0001d71c', '\U0001d734'), ('\U0001d756', '\U0001d76e'),
+        ('\U0001d790', '\U0001d7a8'), ('\U0001d7ca', '\U0001d7ca')
+    ];
+
+    pub fn Uppercase(c: char) -> bool {
+        bsearch_range_table(c, Uppercase_table)
+    }
+
     static XID_Continue_table : &'static [(char,char)] = &[
         ('\x30', '\x39'), ('\x41', '\x5a'),
         ('\x5f', '\x5f'), ('\x61', '\x7a'),
@@ -4856,3 +5490,31 @@ pub fn XID_Start(c: char) -> bool {
     }
 
 }
+pub mod property {
+
+    fn bsearch_range_table(c: char, r: &'static [(char,char)]) -> bool {
+        use cmp::{Equal, Less, Greater};
+        use vec::ImmutableVector;
+        use option::None;
+        r.bsearch(|&(lo,hi)| {
+            if lo <= c && c <= hi { Equal }
+            else if hi < c { Less }
+            else { Greater }
+        }) != None
+    }
+
+
+    static White_Space_table : &'static [(char,char)] = &[
+        ('\x09', '\x0d'), ('\x20', '\x20'),
+        ('\x85', '\x85'), ('\xa0', '\xa0'),
+        ('\u1680', '\u1680'), ('\u2000', '\u200a'),
+        ('\u2028', '\u2028'), ('\u2029', '\u2029'),
+        ('\u202f', '\u202f'), ('\u205f', '\u205f'),
+        ('\u3000', '\u3000')
+    ];
+
+    pub fn White_Space(c: char) -> bool {
+        bsearch_range_table(c, White_Space_table)
+    }
+
+}

diff --git a/src/test/pretty/block-comment-wchar.pp b/src/test/pretty/block-comment-wchar.pp
index c60cc1a36cabae2e902cb4cf949ea17129f11033..15e6e52e3baf324dde31be4ff64a960b37bdde35 100644 (file)
--- a/src/test/pretty/block-comment-wchar.pp
+++ b/src/test/pretty/block-comment-wchar.pp
@@ -51,7 +51,7 @@ fn f() {
       CR4+2:                         (should align)
     */
     /*
-    // (NEL deliberately omitted)
+      NEL4+2:                        (should align)
     */
     /*
       Ogham Space Mark 4+2:          (should align)
@@ -103,11 +103,10 @@ fn f() {
 fn main() {
     // Taken from http://www.unicode.org/Public/UNIDATA/PropList.txt
     let chars =
-        ['\x0A', '\x0B', '\x0C', '\x0D', '\x20',
-         // '\x85', // for some reason Rust thinks NEL isn't whitespace
-         '\xA0', '\u1680', '\u2000', '\u2001', '\u2002', '\u2003', '\u2004',
-         '\u2005', '\u2006', '\u2007', '\u2008', '\u2009', '\u200A', '\u2028',
-         '\u2029', '\u202F', '\u205F', '\u3000'];
+        ['\x0A', '\x0B', '\x0C', '\x0D', '\x20', '\x85', '\xA0', '\u1680',
+         '\u2000', '\u2001', '\u2002', '\u2003', '\u2004', '\u2005', '\u2006',
+         '\u2007', '\u2008', '\u2009', '\u200A', '\u2028', '\u2029', '\u202F',
+         '\u205F', '\u3000'];
     for c in chars.iter() {
         let ws = c.is_whitespace();
         println!("{:?} {:?}" , c , ws);

diff --git a/src/test/pretty/block-comment-wchar.rs b/src/test/pretty/block-comment-wchar.rs
index 52936da41b29f88548efd0510bd68816c6cf6c80..98b9fd0507799fc4d0c0284fcf514aa12eb52b9d 100644 (file)
--- a/src/test/pretty/block-comment-wchar.rs
+++ b/src/test/pretty/block-comment-wchar.rs
@@ -51,7 +51,7 @@ fn nested() {
 \r\r\r\r  CR4+2:                         (should align)
     */
     /*
-    // (NEL deliberately omitted)
+\85\85\85\85  NEL4+2:                        (should align)
     */
     /*
       Ogham Space Mark 4+2:          (should align)
@@ -97,8 +97,7 @@ fn nested() {
 fn main() {
     // Taken from http://www.unicode.org/Public/UNIDATA/PropList.txt
     let chars =
-        ['\x0A', '\x0B', '\x0C', '\x0D', '\x20',
-         // '\x85', // for some reason Rust thinks NEL isn't whitespace
+        ['\x0A', '\x0B', '\x0C', '\x0D', '\x20', '\x85',
          '\xA0', '\u1680', '\u2000', '\u2001', '\u2002', '\u2003',
          '\u2004', '\u2005', '\u2006', '\u2007', '\u2008', '\u2009', '\u200A',
          '\u2028', '\u2029', '\u202F', '\u205F', '\u3000'];