integrate fusible support

evanchooly · Jun 23, 2023 · b06d15a · b06d15a
1 parent 00dc599
commit b06d15a
Show file tree

Hide file tree

Showing 16 changed files with 55 additions and 11 deletions.
diff --git a/src/main/kotlin/com/antwerkz/expression/types/AnyOf.kt b/src/main/kotlin/com/antwerkz/expression/types/AnyOf.kt
@@ -1,7 +1,5 @@
 package com.antwerkz.expression.types
 
-import java.util.function.Predicate
-
 internal class AnyOf : Type("anyOf") {
     companion object {
         private fun fuseElements(elements: List<Type>): Pair<String, List<Type>> {
@@ -10,6 +8,8 @@ internal class AnyOf : Type("anyOf") {
                 fusables.joinToString("") { el ->
                     if (el is CharType || el is AnyOfChars) {
                         el.value.toString()
+                    } else if (el !is RangeType) {
+                        el.evaluate()
                     } else {
                         val value = el.value as List<*>
                         "${value[0]}-${value[1]}"
@@ -19,15 +19,11 @@ internal class AnyOf : Type("anyOf") {
         }
 
         private fun partition(elements: List<Type>): Pair<List<Type>, List<Type>> {
-            val predicate =
-                Predicate<Type> { type ->
-                    type is RangeType || type is CharType || type is AnyOfChars
-                }
             val fused = mutableListOf<Type>()
             val rest = mutableListOf<Type>()
 
             elements.forEach {
-                if (predicate.test(it)) {
+                if (it.fusible) {
                     fused += it
                 } else {
                     rest += it

diff --git a/src/main/kotlin/com/antwerkz/expression/types/AnyOfChars.kt b/src/main/kotlin/com/antwerkz/expression/types/AnyOfChars.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal class AnyOfChars(chars: String) : Type(chars) {
+    init {
+        fusible = true
+    }
     override fun copy() = AnyOfChars(value as String).copy(this)
 
     override fun evaluate() = "[${value}]"

diff --git a/src/main/kotlin/com/antwerkz/expression/types/CarriageReturn.kt b/src/main/kotlin/com/antwerkz/expression/types/CarriageReturn.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal object CarriageReturn : Type("carriageReturn") {
+    init {
+        fusible = true
+    }
     override fun copy() = CarriageReturn
 
     override fun evaluate() = "\\r"

diff --git a/src/main/kotlin/com/antwerkz/expression/types/CharType.kt b/src/main/kotlin/com/antwerkz/expression/types/CharType.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal class CharType(value: String) : Type(value) {
+    init {
+        fusible = true
+    }
     override fun copy() = CharType(value as String).copy(this)
 
     override fun evaluate() = value as String

diff --git a/src/main/kotlin/com/antwerkz/expression/types/Digit.kt b/src/main/kotlin/com/antwerkz/expression/types/Digit.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal object Digit : Type("digit") {
+    init {
+        fusible = true
+    }
     override fun copy() = Digit
 
     override fun evaluate() = "\\d"

diff --git a/src/main/kotlin/com/antwerkz/expression/types/NewLine.kt b/src/main/kotlin/com/antwerkz/expression/types/NewLine.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal object NewLine : Type("newline") {
+    init {
+        fusible = true
+    }
     override fun copy() = NewLine
 
     override fun evaluate() = "\\n"

diff --git a/src/main/kotlin/com/antwerkz/expression/types/NonDigit.kt b/src/main/kotlin/com/antwerkz/expression/types/NonDigit.kt
@@ -1,5 +1,8 @@
 package com.antwerkz.expression.types
 internal object NonDigit : Type("nonDigit") {
+    init {
+        fusible = true
+    }
     override fun copy() = NonDigit
 
     override fun evaluate() = "\\D"

diff --git a/src/main/kotlin/com/antwerkz/expression/types/NonWhiteSpaceChar.kt b/src/main/kotlin/com/antwerkz/expression/types/NonWhiteSpaceChar.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal object NonWhiteSpaceChar : Type("nonWhitespaceChar") {
+    init {
+        fusible = true
+    }
     override fun copy() = NonWhiteSpaceChar
 
     override fun evaluate() = "\\S"

diff --git a/src/main/kotlin/com/antwerkz/expression/types/NonWord.kt b/src/main/kotlin/com/antwerkz/expression/types/NonWord.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal object NonWord : Type("nonWord") {
+    init {
+        fusible = true
+    }
     override fun copy() = NonWord
 
     override fun evaluate() = "\\W"

diff --git a/src/main/kotlin/com/antwerkz/expression/types/NullByte.kt b/src/main/kotlin/com/antwerkz/expression/types/NullByte.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal object NullByte : Type("nullByte") {
+    init {
+        fusible = true
+    }
     override fun copy() = NullByte
 
     override fun evaluate() = "\\0"

diff --git a/src/main/kotlin/com/antwerkz/expression/types/RangeType.kt b/src/main/kotlin/com/antwerkz/expression/types/RangeType.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal class RangeType(start: Char, end: Char) : Type(listOf(start, end)) {
+    init {
+        fusible = true
+    }
     @Suppress("UNCHECKED_CAST")
     override fun copy(): Type {
         val list = value as List<Char>

diff --git a/src/main/kotlin/com/antwerkz/expression/types/Tab.kt b/src/main/kotlin/com/antwerkz/expression/types/Tab.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal object Tab : Type("tab") {
+    init {
+        fusible = true
+    }
     override fun copy() = Tab
 
     override fun evaluate() = "\\t"

diff --git a/src/main/kotlin/com/antwerkz/expression/types/Type.kt b/src/main/kotlin/com/antwerkz/expression/types/Type.kt
@@ -1,11 +1,12 @@
 package com.antwerkz.expression.types
 
 internal abstract class Type(var value: Any? = null) {
-    var containsChildren: Boolean = false
-    var quantifierRequiresGroup: Boolean = false
+    var containsChildren = false
+    var quantifierRequiresGroup = false
+    var fusible = false
     var times: List<Int> = emptyList()
     var name: String? = null
-    var index: Int = 0
+    var index = 0
 
     constructor(original: Type) : this(original.value) {
         this.name = original.name

diff --git a/src/main/kotlin/com/antwerkz/expression/types/WhitespaceChar.kt b/src/main/kotlin/com/antwerkz/expression/types/WhitespaceChar.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal object WhitespaceChar : Type("whitespaceChar") {
+    init {
+        fusible = true
+    }
     override fun copy() = WhitespaceChar
 
     override fun evaluate() = "\\s"

diff --git a/src/main/kotlin/com/antwerkz/expression/types/Word.kt b/src/main/kotlin/com/antwerkz/expression/types/Word.kt
@@ -1,6 +1,9 @@
 package com.antwerkz.expression.types
 
 internal object Word : Type("word") {
+    init {
+        fusible = true
+    }
     override fun copy() = Word
 
     override fun evaluate() = "\\w"

diff --git a/src/test/kotlin/com/antwerkz/expressive/SuperExpressiveTest.kt b/src/test/kotlin/com/antwerkz/expressive/SuperExpressiveTest.kt
@@ -101,7 +101,7 @@ class SuperExpressiveTest {
         testRegexEquality("\\t", tab())
 
         testRegexEquality(
-            "(?:hello|\\d|\\w|[\\.#])",
+            """(?:hello|[\d\w\.#])""",
             anyOf { string("hello").digit().word().char('.').char('#') }
         )
 
@@ -187,6 +187,14 @@ class SuperExpressiveTest {
         testRegexEquality("h", char('h'))
     }
 
+    @Test
+    fun testFusing() {
+        testRegexEquality(
+            """(?:hello|[\d\w\.#])""",
+            anyOf { string("hello").digit().word().char('.').char('#') }
+        )
+    }
+
     @Test
     fun simpleSubexpression() {
         testRegexEquality(