regex: Refactor predefined character classes.

backend.native/tests/external/stdlib/text/RegexTest/harmony/AllCodePointsTest.kt

@@ -9,21 +9,15 @@ fun codePointToString(codePoint: Int): String {
     return fromCharArray(charArray, 0, charArray.size)
 }
 
-fun box() {}
-
 // TODO: Here is a performance problem: an execution of this test requires much more time than it in Kotlin/JVM.
-fun box1() {
+fun box() {
     // Regression for HARMONY-3145
     var p = Regex("(\\p{all})+")
     var res = true
     var cnt = 0
     var s: String
     for (i in 0..1114111) {
-        if (i % 200000 == 0) {
-            println(i)
-        }
         s = codePointToString(i)
-        // if (!s.matches(p.toString().toRegex())) { TODO: Uncomment when caching is done.
         if (!s.matches(p)) {
             cnt++
             res = false
@@ -37,11 +31,7 @@ fun box1() {
     cnt = 0
 
     for (i in 0..1114111) {
-        if (i % 200000 == 0) {
-            println(i)
-        }
         s = codePointToString(i)
-        // if (!s.matches(p.toString().toRegex())) { TODO: Uncomment when caching is done.
         if (!s.matches(p)) {
             cnt++
             res = false

runtime/src/main/kotlin/kotlin/text/regex/AbstractCharClass.kt

@@ -65,13 +65,12 @@ internal abstract class AbstractCharClass : SpecialToken() {
      * Returns BitSet representing this character class or `null`
      * if this character class does not have character representation;
      */
-    // TODO: @C++?. Or implement a BitSet
     open internal val bits: BitSet?
         get() = null
 
     fun hasLowHighSurrogates(): Boolean {
         return if (altSurrogates)
-            lowHighSurrogates.nextClearBit(0) != -1 // TODO: What if the bitset is empty?
+            lowHighSurrogates.nextClearBit(0) != -1
         else
             lowHighSurrogates.nextSetBit(0) != -1
     }
@@ -81,7 +80,6 @@ internal abstract class AbstractCharClass : SpecialToken() {
     open val instance: AbstractCharClass
         get() = this
 
-    // TODO: refactor
     val surrogates: AbstractCharClass by lazy {
         val result = object : AbstractCharClass() {
             override fun contains(ch: Int): Boolean {
@@ -98,7 +96,6 @@ internal abstract class AbstractCharClass : SpecialToken() {
     }
 
 
-    // TODO: refactor
     val withoutSurrogates: AbstractCharClass by lazy {
         val result = object : AbstractCharClass() {
             override fun contains(ch: Int): Boolean {
@@ -109,7 +106,6 @@ internal abstract class AbstractCharClass : SpecialToken() {
                 else
                     false
 
-
                 return this@AbstractCharClass.contains(ch) && !containslHS
             }
         }
@@ -125,7 +121,6 @@ internal abstract class AbstractCharClass : SpecialToken() {
      * Although this method will not alternate all the already set characters,
      * just overall meaning of the class.
      */
-    // TODO: replace with a property
     fun setNegative(value: Boolean): AbstractCharClass {
         if (alt xor value) {
             alt = !alt
@@ -141,8 +136,7 @@ internal abstract class AbstractCharClass : SpecialToken() {
         return alt
     }
 
-    // TODO: replace getValue with property access?
+    internal abstract class CachedCharClass {
-    internal abstract class LazyCharClass {
         private val posValue: AbstractCharClass by lazy { computeValue() }
         private val negValue: AbstractCharClass by lazy { computeValue().setNegative(true) }
 
@@ -150,147 +144,113 @@ internal abstract class AbstractCharClass : SpecialToken() {
         protected abstract fun computeValue(): AbstractCharClass
     }
 
-    internal class LazyDigit : LazyCharClass() {
+    internal class CachedDigit : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass = CharClass().add('0', '9')
-            return CharClass().add('0', '9')
-        }
     }
 
-    // TODO: DO we need it? But we have LazyCharCLass.negValue. May be it's connected with mayContainSupplCodepoints?
+    internal class CachedNonDigit : CachedCharClass() {
-    // TODO: Don't use bitmaps in preset classes?
+        override fun computeValue(): AbstractCharClass =
-    internal class LazyNonDigit : LazyCharClass() {
+                CharClass().add('0', '9').setNegative(true).apply { mayContainSupplCodepoints = true }
-        override fun computeValue(): AbstractCharClass {
-            val result = CharClass().add('0', '9').setNegative(true)
-
-            result.mayContainSupplCodepoints = true
-            return result
-        }
     }
 
-    internal class LazySpace : LazyCharClass() {
+    internal class CachedSpace : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        /* 9-13 - \t\n\x0B\f\r; 32 - ' ' */
-            /* 9-13 - \t\n\x0B\f\r; 32 - ' ' */
+        override fun computeValue(): AbstractCharClass = CharClass().add(9, 13).add(32)
-            return CharClass().add(9, 13).add(32)
-        }
     }
 
-    internal class LazyNonSpace : LazyCharClass() {
+    internal class CachedNonSpace : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass =
-            val result = LazySpace().getValue(negative = true)
+                CachedSpace().getValue(negative = true).apply { mayContainSupplCodepoints = true }
-
-            result.mayContainSupplCodepoints = true
-            return result
-        }
     }
 
-    internal class LazyWord : LazyCharClass() {
+    internal class CachedWord : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass = CharClass().add('a', 'z').add('A', 'Z').add('0', '9').add('_')
-            return CharClass().add('a', 'z').add('A', 'Z').add('0', '9')
-                    .add('_')
-        }
     }
 
-    internal class LazyNonWord : LazyCharClass() {
+    internal class CachedNonWord : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass =
-            val result = LazyWord().getValue(negative = true)
+                CachedWord().getValue(negative = true).apply { mayContainSupplCodepoints = true }
-
-            result.mayContainSupplCodepoints = true
-            return result
-        }
     }
 
-    // TODO: It is only for latin in Harmony.
+    internal class CachedLower : CachedCharClass() {
-    internal class LazyLower : LazyCharClass() {
+        override fun computeValue(): AbstractCharClass = CharClass().add('a', 'z')
-        override fun computeValue(): AbstractCharClass {
-            return CharClass().add('a', 'z')
-        }
     }
 
-    internal class LazyUpper : LazyCharClass() {
+    internal class CachedUpper : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass = CharClass().add('A', 'Z')
-            return CharClass().add('A', 'Z')
-        }
     }
 
-    internal class LazyASCII : LazyCharClass() {
+    internal class CachedASCII : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass = CharClass().add(0x00, 0x7F)
-            return CharClass().add(0x00, 0x7F)
-        }
     }
 
-    internal class LazyAlpha : LazyCharClass() {
+    internal class CachedAlpha : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass = CharClass().add('a', 'z').add('A', 'Z')
-            return CharClass().add('a', 'z').add('A', 'Z')
-        }
     }
 
-    internal class LazyAlnum : LazyCharClass() {
+    internal class CachedAlnum : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass =
-            // TODO: Get rid of the cast?
+                (CachedAlpha().getValue(negative = false) as CharClass).add('0', '9')
-            return (LazyAlpha().getValue(negative = false) as CharClass).add('0', '9')
-        }
     }
 
-    internal class LazyPunct : LazyCharClass() {
+    internal class CachedPunct : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        /* Punctuation !"#$%&'()*+,-./:;<=>?@ [\]^_` {|}~ */
-            /* Punctuation !"#$%&'()*+,-./:;<=>?@ [\]^_` {|}~ */
+        override fun computeValue(): AbstractCharClass = CharClass().add(0x21, 0x40).add(0x5B, 0x60).add(0x7B, 0x7E)
-            return CharClass().add(0x21, 0x40).add(0x5B, 0x60).add(0x7B,
-                    0x7E)
-        }
     }
 
-    internal class LazyGraph : LazyCharClass() {
+    internal class CachedGraph : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        /* plus punctuation */
-            /* plus punctuation */
+        override fun computeValue(): AbstractCharClass =
-            return (LazyAlnum().getValue(negative = false) as CharClass)
+                (CachedAlnum().getValue(negative = false) as CharClass)
-                    .add(0x21, 0x40)
+                        .add(0x21, 0x40)
-                    .add(0x5B, 0x60)
+                        .add(0x5B, 0x60)
-                    .add(0x7B, 0x7E)
+                        .add(0x7B, 0x7E)
-        }
     }
 
-    internal class LazyPrint : LazyCharClass() {
+    internal class CachedPrint : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass =
-            return (LazyGraph().getValue(negative = true) as CharClass).add(0x20)
+                (CachedGraph().getValue(negative = true) as CharClass).add(0x20)
-        }
     }
 
-    internal class LazyBlank : LazyCharClass() {
+    internal class CachedBlank : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass = CharClass().add(' ').add('\t')
-            return CharClass().add(' ').add('\t')
-        }
     }
 
-    internal class LazyCntrl : LazyCharClass() {
+    internal class CachedCntrl : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass = CharClass().add(0x00, 0x1F).add(0x7F)
-            return CharClass().add(0x00, 0x1F).add(0x7F)
-        }
     }
 
-    internal class LazyXDigit : LazyCharClass() {
+    internal class CachedXDigit : CachedCharClass() {
-        override fun computeValue(): AbstractCharClass {
+        override fun computeValue(): AbstractCharClass = CharClass().add('0', '9').add('a', 'f').add('A', 'F')
-            return CharClass().add('0', '9').add('a', 'f').add('A', 'F')
-        }
     }
 
-    internal class LazyRange(var start: Int, var end: Int) : LazyCharClass() {
+    internal class CachedRange(var start: Int, var end: Int) : CachedCharClass() {
-
+        override fun computeValue(): AbstractCharClass =
-        public override fun computeValue(): AbstractCharClass {
+            object: AbstractCharClass() {
-            val chCl = CharClass().add(start, end)
+                override fun contains(ch: Int): Boolean = alt xor (ch in start..end)
-            return chCl
+            }.apply {
-        }
+                if (end >= Char.MIN_SUPPLEMENTARY_CODE_POINT) {
+                    mayContainSupplCodepoints = true
+                }
+                val minSurrogate = Char.MIN_SURROGATE.toInt()
+                val maxSurrogate = Char.MAX_SURROGATE.toInt()
+                // There is an intersection with surrogate characters.
+                if (end >= minSurrogate && start <= maxSurrogate && start <= end) {
+                    val surrogatesStart = maxOf(start, minSurrogate) - minSurrogate
+                    val surrogatesEnd = minOf(end, maxSurrogate) - minSurrogate
+                    lowHighSurrogates.set(surrogatesStart..surrogatesEnd)
+                }
+            }
     }
 
-    internal class LazySpecialsBlock : LazyCharClass() {
+    internal class CachedSpecialsBlock : CachedCharClass() {
-        public override fun computeValue(): AbstractCharClass {
+        public override fun computeValue(): AbstractCharClass = CharClass().add(0xFEFF, 0xFEFF).add(0xFFF0, 0xFFFD)
-            return CharClass().add(0xFEFF, 0xFEFF).add(0xFFF0, 0xFFFD)
-        }
     }
 
-    internal class LazyCategoryScope(
+    internal class CachedCategoryScope(
             val category: Int,
             val mayContainSupplCodepoints: Boolean,
-            val containsAllSurrogates: Boolean = false) : LazyCharClass() {
+            val containsAllSurrogates: Boolean = false) : CachedCharClass() {
 
         override fun computeValue(): AbstractCharClass {
             val result = UnicodeCategoryScope(category)
@@ -303,10 +263,10 @@ internal abstract class AbstractCharClass : SpecialToken() {
         }
     }
 
-    internal class LazyCategory(
+    internal class CachedCategory(
             val category: Int,
             val mayContainSupplCodepoints: Boolean,
-            val containsAllSurrogates: Boolean = false) : LazyCharClass() {
+            val containsAllSurrogates: Boolean = false) : CachedCharClass() {
 
         override fun computeValue(): AbstractCharClass {
             val result = UnicodeCategory(category)
@@ -318,236 +278,229 @@ internal abstract class AbstractCharClass : SpecialToken() {
         }
     }
 
-    // -----------------------------------------------------------------
+
-    // Static methods and predefined classes
-    // -----------------------------------------------------------------
     companion object {
-
         //Char.MAX_SURROGATE - Char.MIN_SURROGATE + 1
-        var SURROGATE_CARDINALITY = 2048
+        const val SURROGATE_CARDINALITY = 2048
 
-        var space: LazyCharClass = LazySpace()
+        private var classCache: MutableMap<String, CachedCharClass>? = null
-        var digit: LazyCharClass = LazyDigit()
 
         /**
-         * character classes generated from
+         * Character classes.
-         * http://www.unicode.org/reports/tr18/
+         * See http://www.unicode.org/reports/tr18/, http://www.unicode.org/Public/4.1.0/ucd/Blocks.txt
-         * http://www.unicode.org/Public/4.1.0/ucd/Blocks.txt
          */
-        // TODO: @C++ Harmony uses ListResourceBundle class here.  TODO: perfrom a effector lookup on the C++ side.
+        // TODO: Make a faster implementation.
-        // Temporary solution: when
+        fun createClass(name: String) : CachedCharClass =
-        // TODO: @C++. Or make it in code. Or do something else.
+                when (name) {
-        // We can lazily create objects of char classes in C++ hashmap or Array.
+                    "Lower" -> CachedLower()
-        // The main point of it is laziness
+                    "Upper" -> CachedUpper()
-        fun getClass(name: String) : LazyCharClass {
+                    "ASCII" -> CachedASCII()
+                    "Alpha" -> CachedAlpha()
+                    "Digit" -> CachedDigit()
+                    "Alnum" -> CachedAlnum()
+                    "Punct" -> CachedPunct()
+                    "Graph" -> CachedGraph()
+                    "Print" -> CachedPrint()
+                    "Blank" -> CachedBlank()
+                    "Cntrl" -> CachedCntrl()
+                    "XDigit" -> CachedXDigit()
+                    "Space" -> CachedSpace()
+                    "w" -> CachedWord()
+                    "W" -> CachedNonWord()
+                    "s" -> CachedSpace()
+                    "S" -> CachedNonSpace()
+                    "d" -> CachedDigit()
+                    "D" -> CachedNonDigit()
+                    "BasicLatin" -> CachedRange(0x0000, 0x007F)
+                    "Latin-1Supplement" -> CachedRange(0x0080, 0x00FF)
+                    "LatinExtended-A" -> CachedRange(0x0100, 0x017F)
+                    "LatinExtended-B" -> CachedRange(0x0180, 0x024F)
+                    "IPAExtensions" -> CachedRange(0x0250, 0x02AF)
+                    "SpacingModifierLetters" -> CachedRange(0x02B0, 0x02FF)
+                    "CombiningDiacriticalMarks" -> CachedRange(0x0300, 0x036F)
+                    "Greek" -> CachedRange(0x0370, 0x03FF)
+                    "Cyrillic" -> CachedRange(0x0400, 0x04FF)
+                    "CyrillicSupplement" -> CachedRange(0x0500, 0x052F)
+                    "Armenian" -> CachedRange(0x0530, 0x058F)
+                    "Hebrew" -> CachedRange(0x0590, 0x05FF)
+                    "Arabic" -> CachedRange(0x0600, 0x06FF)
+                    "Syriac" -> CachedRange(0x0700, 0x074F)
+                    "ArabicSupplement" -> CachedRange(0x0750, 0x077F)
+                    "Thaana" -> CachedRange(0x0780, 0x07BF)
+                    "Devanagari" -> CachedRange(0x0900, 0x097F)
+                    "Bengali" -> CachedRange(0x0980, 0x09FF)
+                    "Gurmukhi" -> CachedRange(0x0A00, 0x0A7F)
+                    "Gujarati" -> CachedRange(0x0A80, 0x0AFF)
+                    "Oriya" -> CachedRange(0x0B00, 0x0B7F)
+                    "Tamil" -> CachedRange(0x0B80, 0x0BFF)
+                    "Telugu" -> CachedRange(0x0C00, 0x0C7F)
+                    "Kannada" -> CachedRange(0x0C80, 0x0CFF)
+                    "Malayalam" -> CachedRange(0x0D00, 0x0D7F)
+                    "Sinhala" -> CachedRange(0x0D80, 0x0DFF)
+                    "Thai" -> CachedRange(0x0E00, 0x0E7F)
+                    "Lao" -> CachedRange(0x0E80, 0x0EFF)
+                    "Tibetan" -> CachedRange(0x0F00, 0x0FFF)
+                    "Myanmar" -> CachedRange(0x1000, 0x109F)
+                    "Georgian" -> CachedRange(0x10A0, 0x10FF)
+                    "HangulJamo" -> CachedRange(0x1100, 0x11FF)
+                    "Ethiopic" -> CachedRange(0x1200, 0x137F)
+                    "EthiopicSupplement" -> CachedRange(0x1380, 0x139F)
+                    "Cherokee" -> CachedRange(0x13A0, 0x13FF)
+                    "UnifiedCanadianAboriginalSyllabics" -> CachedRange(0x1400, 0x167F)
+                    "Ogham" -> CachedRange(0x1680, 0x169F)
+                    "Runic" -> CachedRange(0x16A0, 0x16FF)
+                    "Tagalog" -> CachedRange(0x1700, 0x171F)
+                    "Hanunoo" -> CachedRange(0x1720, 0x173F)
+                    "Buhid" -> CachedRange(0x1740, 0x175F)
+                    "Tagbanwa" -> CachedRange(0x1760, 0x177F)
+                    "Khmer" -> CachedRange(0x1780, 0x17FF)
+                    "Mongolian" -> CachedRange(0x1800, 0x18AF)
+                    "Limbu" -> CachedRange(0x1900, 0x194F)
+                    "TaiLe" -> CachedRange(0x1950, 0x197F)
+                    "NewTaiLue" -> CachedRange(0x1980, 0x19DF)
+                    "KhmerSymbols" -> CachedRange(0x19E0, 0x19FF)
+                    "Buginese" -> CachedRange(0x1A00, 0x1A1F)
+                    "PhoneticExtensions" -> CachedRange(0x1D00, 0x1D7F)
+                    "PhoneticExtensionsSupplement" -> CachedRange(0x1D80, 0x1DBF)
+                    "CombiningDiacriticalMarksSupplement" -> CachedRange(0x1DC0, 0x1DFF)
+                    "LatinExtendedAdditional" -> CachedRange(0x1E00, 0x1EFF)
+                    "GreekExtended" -> CachedRange(0x1F00, 0x1FFF)
+                    "GeneralPunctuation" -> CachedRange(0x2000, 0x206F)
+                    "SuperscriptsandSubscripts" -> CachedRange(0x2070, 0x209F)
+                    "CurrencySymbols" -> CachedRange(0x20A0, 0x20CF)
+                    "CombiningMarksforSymbols" -> CachedRange(0x20D0, 0x20FF)
+                    "LetterlikeSymbols" -> CachedRange(0x2100, 0x214F)
+                    "NumberForms" -> CachedRange(0x2150, 0x218F)
+                    "Arrows" -> CachedRange(0x2190, 0x21FF)
+                    "MathematicalOperators" -> CachedRange(0x2200, 0x22FF)
+                    "MiscellaneousTechnical" -> CachedRange(0x2300, 0x23FF)
+                    "ControlPictures" -> CachedRange(0x2400, 0x243F)
+                    "OpticalCharacterRecognition" -> CachedRange(0x2440, 0x245F)
+                    "EnclosedAlphanumerics" -> CachedRange(0x2460, 0x24FF)
+                    "BoxDrawing" -> CachedRange(0x2500, 0x257F)
+                    "BlockElements" -> CachedRange(0x2580, 0x259F)
+                    "GeometricShapes" -> CachedRange(0x25A0, 0x25FF)
+                    "MiscellaneousSymbols" -> CachedRange(0x2600, 0x26FF)
+                    "Dingbats" -> CachedRange(0x2700, 0x27BF)
+                    "MiscellaneousMathematicalSymbols-A" -> CachedRange(0x27C0, 0x27EF)
+                    "SupplementalArrows-A" -> CachedRange(0x27F0, 0x27FF)
+                    "BraillePatterns" -> CachedRange(0x2800, 0x28FF)
+                    "SupplementalArrows-B" -> CachedRange(0x2900, 0x297F)
+                    "MiscellaneousMathematicalSymbols-B" -> CachedRange(0x2980, 0x29FF)
+                    "SupplementalMathematicalOperators" -> CachedRange(0x2A00, 0x2AFF)
+                    "MiscellaneousSymbolsandArrows" -> CachedRange(0x2B00, 0x2BFF)
+                    "Glagolitic" -> CachedRange(0x2C00, 0x2C5F)
+                    "Coptic" -> CachedRange(0x2C80, 0x2CFF)
+                    "GeorgianSupplement" -> CachedRange(0x2D00, 0x2D2F)
+                    "Tifinagh" -> CachedRange(0x2D30, 0x2D7F)
+                    "EthiopicExtended" -> CachedRange(0x2D80, 0x2DDF)
+                    "SupplementalPunctuation" -> CachedRange(0x2E00, 0x2E7F)
+                    "CJKRadicalsSupplement" -> CachedRange(0x2E80, 0x2EFF)
+                    "KangxiRadicals" -> CachedRange(0x2F00, 0x2FDF)
+                    "IdeographicDescriptionCharacters" -> CachedRange(0x2FF0, 0x2FFF)
+                    "CJKSymbolsandPunctuation" -> CachedRange(0x3000, 0x303F)
+                    "Hiragana" -> CachedRange(0x3040, 0x309F)
+                    "Katakana" -> CachedRange(0x30A0, 0x30FF)
+                    "Bopomofo" -> CachedRange(0x3100, 0x312F)
+                    "HangulCompatibilityJamo" -> CachedRange(0x3130, 0x318F)
+                    "Kanbun" -> CachedRange(0x3190, 0x319F)
+                    "BopomofoExtended" -> CachedRange(0x31A0, 0x31BF)
+                    "CJKStrokes" -> CachedRange(0x31C0, 0x31EF)
+                    "KatakanaPhoneticExtensions" -> CachedRange(0x31F0, 0x31FF)
+                    "EnclosedCJKLettersandMonths" -> CachedRange(0x3200, 0x32FF)
+                    "CJKCompatibility" -> CachedRange(0x3300, 0x33FF)
+                    "CJKUnifiedIdeographsExtensionA" -> CachedRange(0x3400, 0x4DB5)
+                    "YijingHexagramSymbols" -> CachedRange(0x4DC0, 0x4DFF)
+                    "CJKUnifiedIdeographs" -> CachedRange(0x4E00, 0x9FFF)
+                    "YiSyllables" -> CachedRange(0xA000, 0xA48F)
+                    "YiRadicals" -> CachedRange(0xA490, 0xA4CF)
+                    "ModifierToneLetters" -> CachedRange(0xA700, 0xA71F)
+                    "SylotiNagri" -> CachedRange(0xA800, 0xA82F)
+                    "HangulSyllables" -> CachedRange(0xAC00, 0xD7A3)
+                    "HighSurrogates" -> CachedRange(0xD800, 0xDB7F)
+                    "HighPrivateUseSurrogates" -> CachedRange(0xDB80, 0xDBFF)
+                    "LowSurrogates" -> CachedRange(0xDC00, 0xDFFF)
+                    "PrivateUseArea" -> CachedRange(0xE000, 0xF8FF)
+                    "CJKCompatibilityIdeographs" -> CachedRange(0xF900, 0xFAFF)
+                    "AlphabeticPresentationForms" -> CachedRange(0xFB00, 0xFB4F)
+                    "ArabicPresentationForms-A" -> CachedRange(0xFB50, 0xFDFF)
+                    "VariationSelectors" -> CachedRange(0xFE00, 0xFE0F)
+                    "VerticalForms" -> CachedRange(0xFE10, 0xFE1F)
+                    "CombiningHalfMarks" -> CachedRange(0xFE20, 0xFE2F)
+                    "CJKCompatibilityForms" -> CachedRange(0xFE30, 0xFE4F)
+                    "SmallFormVariants" -> CachedRange(0xFE50, 0xFE6F)
+                    "ArabicPresentationForms-B" -> CachedRange(0xFE70, 0xFEFF)
+                    "HalfwidthandFullwidthForms" -> CachedRange(0xFF00, 0xFFEF)
+                    "all" -> CachedRange(0x00, 0x10FFFF)
+                    "Specials" -> CachedSpecialsBlock()
+                    "Cn" -> CachedCategory(CharCategory.UNASSIGNED.value, true)
+                    "IsL" -> CachedCategoryScope(0x3E, true)
+                    "Lu" -> CachedCategory(CharCategory.UPPERCASE_LETTER.value, true)
+                    "Ll" -> CachedCategory(CharCategory.LOWERCASE_LETTER.value, true)
+                    "Lt" -> CachedCategory(CharCategory.TITLECASE_LETTER.value, false)
+                    "Lm" -> CachedCategory(CharCategory.MODIFIER_LETTER.value, false)
+                    "Lo" -> CachedCategory(CharCategory.OTHER_LETTER.value, true)
+                    "IsM" -> CachedCategoryScope(0x1C0, true)
+                    "Mn" -> CachedCategory(CharCategory.NON_SPACING_MARK.value, true)
+                    "Me" -> CachedCategory(CharCategory.ENCLOSING_MARK.value, false)
+                    "Mc" -> CachedCategory(CharCategory.COMBINING_SPACING_MARK.value, true)
+                    "N" -> CachedCategoryScope(0xE00, true)
+                    "Nd" -> CachedCategory(CharCategory.DECIMAL_DIGIT_NUMBER.value, true)
+                    "Nl" -> CachedCategory(CharCategory.LETTER_NUMBER.value, true)
+                    "No" -> CachedCategory(CharCategory.OTHER_NUMBER.value, true)
+                    "IsZ" -> CachedCategoryScope(0x7000, false)
+                    "Zs" -> CachedCategory(CharCategory.SPACE_SEPARATOR.value, false)
+                    "Zl" -> CachedCategory(CharCategory.LINE_SEPARATOR.value, false)
+                    "Zp" -> CachedCategory(CharCategory.PARAGRAPH_SEPARATOR.value, false)
+                    "IsC" -> CachedCategoryScope(0xF0000, true, true)
+                    "Cc" -> CachedCategory(CharCategory.CONTROL.value, false)
+                    "Cf" -> CachedCategory(CharCategory.FORMAT.value, true)
+                    "Co" -> CachedCategory(CharCategory.PRIVATE_USE.value, true)
+                    "Cs" -> CachedCategory(CharCategory.SURROGATE.value, false, true)
+                    "IsP" -> CachedCategoryScope(1 shl CharCategory.DASH_PUNCTUATION.value or
+                            (1 shl CharCategory.START_PUNCTUATION.value) or
+                            (1 shl CharCategory.END_PUNCTUATION.value) or
+                            (1 shl CharCategory.CONNECTOR_PUNCTUATION.value) or
+                            (1 shl CharCategory.OTHER_PUNCTUATION.value) or
+                            (1 shl CharCategory.INITIAL_QUOTE_PUNCTUATION.value) or
+                            (1 shl CharCategory.FINAL_QUOTE_PUNCTUATION.value), true)
+                    "Pd" -> CachedCategory(CharCategory.DASH_PUNCTUATION.value, false)
+                    "Ps" -> CachedCategory(CharCategory.START_PUNCTUATION.value, false)
+                    "Pe" -> CachedCategory(CharCategory.END_PUNCTUATION.value, false)
+                    "Pc" -> CachedCategory(CharCategory.CONNECTOR_PUNCTUATION.value, false)
+                    "Po" -> CachedCategory(CharCategory.OTHER_PUNCTUATION.value, true)
+                    "IsS" -> CachedCategoryScope(0x7E000000, true)
+                    "Sm" -> CachedCategory(CharCategory.MATH_SYMBOL.value, true)
+                    "Sc" -> CachedCategory(CharCategory.CURRENCY_SYMBOL.value, false)
+                    "Sk" -> CachedCategory(CharCategory.MODIFIER_SYMBOL.value, false)
+                    "So" -> CachedCategory(CharCategory.OTHER_SYMBOL.value, true)
+                    "Pi" -> CachedCategory(CharCategory.INITIAL_QUOTE_PUNCTUATION.value, false)
+                    "Pf" -> CachedCategory(CharCategory.FINAL_QUOTE_PUNCTUATION.value, false)
+                    else -> throw PatternSyntaxException("No such character class")
+                }
 
-            return when (name) {
+        fun intersects(ch1: Int, ch2: Int): Boolean = ch1 == ch2
+        fun intersects(cc: AbstractCharClass, ch: Int): Boolean = cc.contains(ch)
 
-                "Lower" -> LazyLower()
+        fun intersects(cc1: AbstractCharClass, cc2: AbstractCharClass): Boolean {
-                "Upper" -> LazyUpper()
+            if (cc1.bits == null || cc2.bits == null) {
-                "ASCII" -> LazyASCII()
-                "Alpha" -> LazyAlpha()
-                "Digit" -> digit
-                "Alnum" -> LazyAlnum()
-                "Punct" -> LazyPunct()
-                "Graph" -> LazyGraph()
-                "Print" -> LazyPrint()
-                "Blank" -> LazyBlank()
-                "Cntrl" -> LazyCntrl()
-                "XDigit" -> LazyXDigit()
-                "Space" -> space
-                "w" -> LazyWord()
-                "W" -> LazyNonWord()
-                "s" -> space
-                "S" -> LazyNonSpace()
-                "d" -> digit
-                "D" -> LazyNonDigit()
-                "BasicLatin" -> LazyRange(0x0000, 0x007F)
-                "Latin-1Supplement" -> LazyRange(0x0080, 0x00FF)
-                "LatinExtended-A" -> LazyRange(0x0100, 0x017F)
-                "LatinExtended-B" -> LazyRange(0x0180, 0x024F)
-                "IPAExtensions" -> LazyRange(0x0250, 0x02AF)
-                "SpacingModifierLetters" -> LazyRange(0x02B0, 0x02FF)
-                "CombiningDiacriticalMarks" -> LazyRange(0x0300, 0x036F)
-                "Greek" -> LazyRange(0x0370, 0x03FF)
-                "Cyrillic" -> LazyRange(0x0400, 0x04FF)
-                "CyrillicSupplement" -> LazyRange(0x0500, 0x052F)
-                "Armenian" -> LazyRange(0x0530, 0x058F)
-                "Hebrew" -> LazyRange(0x0590, 0x05FF)
-                "Arabic" -> LazyRange(0x0600, 0x06FF)
-                "Syriac" -> LazyRange(0x0700, 0x074F)
-                "ArabicSupplement" -> LazyRange(0x0750, 0x077F)
-                "Thaana" -> LazyRange(0x0780, 0x07BF)
-                "Devanagari" -> LazyRange(0x0900, 0x097F)
-                "Bengali" -> LazyRange(0x0980, 0x09FF)
-                "Gurmukhi" -> LazyRange(0x0A00, 0x0A7F)
-                "Gujarati" -> LazyRange(0x0A80, 0x0AFF)
-                "Oriya" -> LazyRange(0x0B00, 0x0B7F)
-                "Tamil" -> LazyRange(0x0B80, 0x0BFF)
-                "Telugu" -> LazyRange(0x0C00, 0x0C7F)
-                "Kannada" -> LazyRange(0x0C80, 0x0CFF)
-                "Malayalam" -> LazyRange(0x0D00, 0x0D7F)
-                "Sinhala" -> LazyRange(0x0D80, 0x0DFF)
-                "Thai" -> LazyRange(0x0E00, 0x0E7F)
-                "Lao" -> LazyRange(0x0E80, 0x0EFF)
-                "Tibetan" -> LazyRange(0x0F00, 0x0FFF)
-                "Myanmar" -> LazyRange(0x1000, 0x109F)
-                "Georgian" -> LazyRange(0x10A0, 0x10FF)
-                "HangulJamo" -> LazyRange(0x1100, 0x11FF)
-                "Ethiopic" -> LazyRange(0x1200, 0x137F)
-                "EthiopicSupplement" -> LazyRange(0x1380, 0x139F)
-                "Cherokee" -> LazyRange(0x13A0, 0x13FF)
-                "UnifiedCanadianAboriginalSyllabics" -> LazyRange(0x1400, 0x167F)
-                "Ogham" -> LazyRange(0x1680, 0x169F)
-                "Runic" -> LazyRange(0x16A0, 0x16FF)
-                "Tagalog" -> LazyRange(0x1700, 0x171F)
-                "Hanunoo" -> LazyRange(0x1720, 0x173F)
-                "Buhid" -> LazyRange(0x1740, 0x175F)
-                "Tagbanwa" -> LazyRange(0x1760, 0x177F)
-                "Khmer" -> LazyRange(0x1780, 0x17FF)
-                "Mongolian" -> LazyRange(0x1800, 0x18AF)
-                "Limbu" -> LazyRange(0x1900, 0x194F)
-                "TaiLe" -> LazyRange(0x1950, 0x197F)
-                "NewTaiLue" -> LazyRange(0x1980, 0x19DF)
-                "KhmerSymbols" -> LazyRange(0x19E0, 0x19FF)
-                "Buginese" -> LazyRange(0x1A00, 0x1A1F)
-                "PhoneticExtensions" -> LazyRange(0x1D00, 0x1D7F)
-                "PhoneticExtensionsSupplement" -> LazyRange(0x1D80, 0x1DBF)
-                "CombiningDiacriticalMarksSupplement" -> LazyRange(0x1DC0, 0x1DFF)
-                "LatinExtendedAdditional" -> LazyRange(0x1E00, 0x1EFF)
-                "GreekExtended" -> LazyRange(0x1F00, 0x1FFF)
-                "GeneralPunctuation" -> LazyRange(0x2000, 0x206F)
-                "SuperscriptsandSubscripts" -> LazyRange(0x2070, 0x209F)
-                "CurrencySymbols" -> LazyRange(0x20A0, 0x20CF)
-                "CombiningMarksforSymbols" -> LazyRange(0x20D0, 0x20FF)
-                "LetterlikeSymbols" -> LazyRange(0x2100, 0x214F)
-                "NumberForms" -> LazyRange(0x2150, 0x218F)
-                "Arrows" -> LazyRange(0x2190, 0x21FF)
-                "MathematicalOperators" -> LazyRange(0x2200, 0x22FF)
-                "MiscellaneousTechnical" -> LazyRange(0x2300, 0x23FF)
-                "ControlPictures" -> LazyRange(0x2400, 0x243F)
-                "OpticalCharacterRecognition" -> LazyRange(0x2440, 0x245F)
-                "EnclosedAlphanumerics" -> LazyRange(0x2460, 0x24FF)
-                "BoxDrawing" -> LazyRange(0x2500, 0x257F)
-                "BlockElements" -> LazyRange(0x2580, 0x259F)
-                "GeometricShapes" -> LazyRange(0x25A0, 0x25FF)
-                "MiscellaneousSymbols" -> LazyRange(0x2600, 0x26FF)
-                "Dingbats" -> LazyRange(0x2700, 0x27BF)
-                "MiscellaneousMathematicalSymbols-A" -> LazyRange(0x27C0, 0x27EF)
-                "SupplementalArrows-A" -> LazyRange(0x27F0, 0x27FF)
-                "BraillePatterns" -> LazyRange(0x2800, 0x28FF)
-                "SupplementalArrows-B" -> LazyRange(0x2900, 0x297F)
-                "MiscellaneousMathematicalSymbols-B" -> LazyRange(0x2980, 0x29FF)
-                "SupplementalMathematicalOperators" -> LazyRange(0x2A00, 0x2AFF)
-                "MiscellaneousSymbolsandArrows" -> LazyRange(0x2B00, 0x2BFF)
-                "Glagolitic" -> LazyRange(0x2C00, 0x2C5F)
-                "Coptic" -> LazyRange(0x2C80, 0x2CFF)
-                "GeorgianSupplement" -> LazyRange(0x2D00, 0x2D2F)
-                "Tifinagh" -> LazyRange(0x2D30, 0x2D7F)
-                "EthiopicExtended" -> LazyRange(0x2D80, 0x2DDF)
-                "SupplementalPunctuation" -> LazyRange(0x2E00, 0x2E7F)
-                "CJKRadicalsSupplement" -> LazyRange(0x2E80, 0x2EFF)
-                "KangxiRadicals" -> LazyRange(0x2F00, 0x2FDF)
-                "IdeographicDescriptionCharacters" -> LazyRange(0x2FF0, 0x2FFF)
-                "CJKSymbolsandPunctuation" -> LazyRange(0x3000, 0x303F)
-                "Hiragana" -> LazyRange(0x3040, 0x309F)
-                "Katakana" -> LazyRange(0x30A0, 0x30FF)
-                "Bopomofo" -> LazyRange(0x3100, 0x312F)
-                "HangulCompatibilityJamo" -> LazyRange(0x3130, 0x318F)
-                "Kanbun" -> LazyRange(0x3190, 0x319F)
-                "BopomofoExtended" -> LazyRange(0x31A0, 0x31BF)
-                "CJKStrokes" -> LazyRange(0x31C0, 0x31EF)
-                "KatakanaPhoneticExtensions" -> LazyRange(0x31F0, 0x31FF)
-                "EnclosedCJKLettersandMonths" -> LazyRange(0x3200, 0x32FF)
-                "CJKCompatibility" -> LazyRange(0x3300, 0x33FF)
-                "CJKUnifiedIdeographsExtensionA" -> LazyRange(0x3400, 0x4DB5)
-                "YijingHexagramSymbols" -> LazyRange(0x4DC0, 0x4DFF)
-                "CJKUnifiedIdeographs" -> LazyRange(0x4E00, 0x9FFF)
-                "YiSyllables" -> LazyRange(0xA000, 0xA48F)
-                "YiRadicals" -> LazyRange(0xA490, 0xA4CF)
-                "ModifierToneLetters" -> LazyRange(0xA700, 0xA71F)
-                "SylotiNagri" -> LazyRange(0xA800, 0xA82F)
-                "HangulSyllables" -> LazyRange(0xAC00, 0xD7A3)
-                "HighSurrogates" -> LazyRange(0xD800, 0xDB7F)
-                "HighPrivateUseSurrogates" -> LazyRange(0xDB80, 0xDBFF)
-                "LowSurrogates" -> LazyRange(0xDC00, 0xDFFF)
-                "PrivateUseArea" -> LazyRange(0xE000, 0xF8FF)
-                "CJKCompatibilityIdeographs" -> LazyRange(0xF900, 0xFAFF)
-                "AlphabeticPresentationForms" -> LazyRange(0xFB00, 0xFB4F)
-                "ArabicPresentationForms-A" -> LazyRange(0xFB50, 0xFDFF)
-                "VariationSelectors" -> LazyRange(0xFE00, 0xFE0F)
-                "VerticalForms" -> LazyRange(0xFE10, 0xFE1F)
-                "CombiningHalfMarks" -> LazyRange(0xFE20, 0xFE2F)
-                "CJKCompatibilityForms" -> LazyRange(0xFE30, 0xFE4F)
-                "SmallFormVariants" -> LazyRange(0xFE50, 0xFE6F)
-                "ArabicPresentationForms-B" -> LazyRange(0xFE70, 0xFEFF)
-                "HalfwidthandFullwidthForms" -> LazyRange(0xFF00, 0xFFEF)
-                "all" -> LazyRange(0x00, 0x10FFFF)
-                "Specials" -> LazySpecialsBlock()
-                "Cn" -> LazyCategory(CharCategory.UNASSIGNED.value, true)
-                "IsL" -> LazyCategoryScope(0x3E, true)
-                "Lu" -> LazyCategory(CharCategory.UPPERCASE_LETTER.value, true)
-                "Ll" -> LazyCategory(CharCategory.LOWERCASE_LETTER.value, true)
-                "Lt" -> LazyCategory(CharCategory.TITLECASE_LETTER.value, false)
-                "Lm" -> LazyCategory(CharCategory.MODIFIER_LETTER.value, false)
-                "Lo" -> LazyCategory(CharCategory.OTHER_LETTER.value, true)
-                "IsM" -> LazyCategoryScope(0x1C0, true)
-                "Mn" -> LazyCategory(CharCategory.NON_SPACING_MARK.value, true)
-                "Me" -> LazyCategory(CharCategory.ENCLOSING_MARK.value, false)
-                "Mc" -> LazyCategory(CharCategory.COMBINING_SPACING_MARK.value, true)
-                "N" -> LazyCategoryScope(0xE00, true)
-                "Nd" -> LazyCategory(CharCategory.DECIMAL_DIGIT_NUMBER.value, true)
-                "Nl" -> LazyCategory(CharCategory.LETTER_NUMBER.value, true)
-                "No" -> LazyCategory(CharCategory.OTHER_NUMBER.value, true)
-                "IsZ" -> LazyCategoryScope(0x7000, false)
-                "Zs" -> LazyCategory(CharCategory.SPACE_SEPARATOR.value, false)
-                "Zl" -> LazyCategory(CharCategory.LINE_SEPARATOR.value, false)
-                "Zp" -> LazyCategory(CharCategory.PARAGRAPH_SEPARATOR.value, false)
-                "IsC" -> LazyCategoryScope(0xF0000, true, true)
-                "Cc" -> LazyCategory(CharCategory.CONTROL.value, false)
-                "Cf" -> LazyCategory(CharCategory.FORMAT.value, true)
-                "Co" -> LazyCategory(CharCategory.PRIVATE_USE.value, true)
-                "Cs" -> LazyCategory(CharCategory.SURROGATE.value, false, true)
-                "IsP" -> LazyCategoryScope(1 shl CharCategory.DASH_PUNCTUATION.value or
-                        (1 shl CharCategory.START_PUNCTUATION.value) or
-                        (1 shl CharCategory.END_PUNCTUATION.value) or
-                        (1 shl CharCategory.CONNECTOR_PUNCTUATION.value) or
-                        (1 shl CharCategory.OTHER_PUNCTUATION.value) or
-                        (1 shl CharCategory.INITIAL_QUOTE_PUNCTUATION.value) or
-                        (1 shl CharCategory.FINAL_QUOTE_PUNCTUATION.value), true)
-                "Pd" -> LazyCategory(CharCategory.DASH_PUNCTUATION.value, false)
-                "Ps" -> LazyCategory(CharCategory.START_PUNCTUATION.value, false)
-                "Pe" -> LazyCategory(CharCategory.END_PUNCTUATION.value, false)
-                "Pc" -> LazyCategory(CharCategory.CONNECTOR_PUNCTUATION.value, false)
-                "Po" -> LazyCategory(CharCategory.OTHER_PUNCTUATION.value, true)
-                "IsS" -> LazyCategoryScope(0x7E000000, true)
-                "Sm" -> LazyCategory(CharCategory.MATH_SYMBOL.value, true)
-                "Sc" -> LazyCategory(CharCategory.CURRENCY_SYMBOL.value, false)
-                "Sk" -> LazyCategory(CharCategory.MODIFIER_SYMBOL.value, false)
-                "So" -> LazyCategory(CharCategory.OTHER_SYMBOL.value, true)
-                "Pi" -> LazyCategory(CharCategory.INITIAL_QUOTE_PUNCTUATION.value, false)
-                "Pf" -> LazyCategory(CharCategory.FINAL_QUOTE_PUNCTUATION.value, false)
-                else -> throw PatternSyntaxException("No such character class")
-            }
-        }
-
-        fun intersects(ch1: Int, ch2: Int): Boolean {
-            return ch1 == ch2
-        }
-
-        fun intersects(cc: AbstractCharClass, ch: Int): Boolean {
-            return cc.contains(ch)
-        }
-
-        fun intersects(cc1: AbstractCharClass,
-                       cc2: AbstractCharClass): Boolean {
-            if (cc1.bits == null || cc2.bits == null)
                 return true
+            }
             return cc1.bits!!.intersects(cc2.bits!!)
         }
 
         fun getPredefinedClass(name: String, negative: Boolean): AbstractCharClass {
-            return (getClass(name)).getValue(negative)
+            var cache = classCache
+            if (cache == null) {
+                cache = mutableMapOf()
+                classCache = cache
+            }
+            var cachedClass = cache[name]
+            if (cachedClass == null) {
+                cachedClass = createClass(name)
+                cache[name] = cachedClass
+            }
+            return cachedClass.getValue(negative)
         }
     }
 }

runtime/src/main/kotlin/kotlin/text/regex/CharClass.kt

@@ -22,9 +22,7 @@ package kotlin.text.regex
  *
  * @author Nikolay A. Kuznetsov
  */
-/*
+// TODO: replace the implementation with one using BitSet for first 256 symbols and a hash table / tree for the rest of UTF.
- * TODO: replace the implementation with one using BitSet for first 256 symbols and a hash table / tree for the rest of UTF.
- */
 internal class CharClass(val ignoreCase: Boolean = false, negative: Boolean = false)  : AbstractCharClass()  {
 
     var invertedSurrogates = false
@@ -38,7 +36,6 @@ internal class CharClass(val ignoreCase: Boolean = false, negative: Boolean = fa
      */
     var inverted = false
 
-    // TODO: May be we can get rid of it?
     var hideBits = false
 
     internal var bits_ = BitSet()
@@ -155,9 +152,6 @@ internal class CharClass(val ignoreCase: Boolean = false, negative: Boolean = fa
                 }
             }
         // Some of charclasses hides its bits
-        // Looks like hideBits and nonBitSet are used when we add another class which have no bitmask.
-        // TODO: We can potentially remove nonBitSet and hideBits. What is alt?
-        // TODO: The same for or operation.
         } else {
             val curAlt = alt
 
@@ -222,17 +216,28 @@ internal class CharClass(val ignoreCase: Boolean = false, negative: Boolean = fa
     fun add(start: Int, end: Int): CharClass {
         if (start > end)
             throw IllegalArgumentException()
-        //no intersection with surrogate characters
+        val minSurrogate = Char.MIN_SURROGATE.toInt()
-        if (!ignoreCase && (end < Char.MIN_SURROGATE.toInt() || start > Char.MAX_SURROGATE.toInt())) {
+        val maxSurrogate = Char.MAX_SURROGATE.toInt()
-            if (!inverted) {
+        if (ignoreCase) {
-                bits_.set(start, end + 1)
+            // TODO: Make a faster implementation.
-            } else {
-                bits_.clear(start, end + 1)
-            }
-        } else {
             for (i in start..end) {
                 add(i)
             }
+        } else {
+            // No intersection with surrogate characters.
+            if (end < minSurrogate || start > maxSurrogate) {
+                bits_.set(start, end + 1, !inverted)
+            } else {
+                val surrogatesStart = maxOf(start, minSurrogate)
+                val surrogatesEnd = minOf(end, maxSurrogate)
+                bits_.set(start, end + 1, !inverted)
+                lowHighSurrogates.set(surrogatesStart - minSurrogate,
+                        surrogatesEnd - minSurrogate + 1,
+                        !invertedSurrogates)
+                if (!mayContainSupplCodepoints && end >= Char.MIN_SUPPLEMENTARY_CODE_POINT) {
+                    mayContainSupplCodepoints = true
+                }
+            }
         }
         return this
     }
@@ -489,7 +494,7 @@ internal class CharClass(val ignoreCase: Boolean = false, negative: Boolean = fa
      */
     override operator fun contains(ch: Int): Boolean {
         if (nonBitSet == null) {
-            return this.alt xor bits_.get(ch) // TODO alt xor bits. It must make sense.
+            return alt xor bits_.get(ch)
         } else {
             return alt xor nonBitSet!!.contains(ch)
         }