refactor: optimize CharItem to handle special characters in toString method

lib/src/main/kotlin/org/example/Parser.kt

@@ -2,16 +2,45 @@ package org.example
 
 import com.github.h0tk3y.betterParse.combinators.*
 import com.github.h0tk3y.betterParse.grammar.Grammar
+import com.github.h0tk3y.betterParse.grammar.parseToEnd
 import com.github.h0tk3y.betterParse.grammar.parser
 import com.github.h0tk3y.betterParse.lexer.literalToken
 import com.github.h0tk3y.betterParse.lexer.regexToken
+import com.github.h0tk3y.betterParse.lexer.token
 import com.github.h0tk3y.betterParse.parser.Parser
 
 class RegexParser : Grammar<RegexItem>() {
+  private var groupCounter = 0
+
+  val escapedCharacter by regexToken("\\\\[+*?.$^()|\\[\\]]")
   val postfixOperator by regexToken("[+*?]")
+  val anchorOperator by regexToken("[$^]")
   val alternationSymbol by literalToken("|")
   val openParenSymbol by literalToken("(")
   val closeParenSymbol by literalToken(")")
+  val bracketContent by
+          token(
+                  name = "bracketContent",
+                  matcher = { seq, from ->
+                    if (seq[from] != '[') {
+                      0 // 대괄호로 시작하지 않으면 매칭 실패
+                    } else {
+                      // 대괄호의 시작 위치에서부터 ']'를 찾음
+                      var to = seq.indexOf(']', from)
+                      // 이스케이프 ']' 건너 뛰기
+                      while (to >= 0 && to > from && seq[to - 1] == '\\') {
+                        to = seq.indexOf(']', to + 1)
+                      }
+                      if (to < 0) {
+                        0
+                      } else if (to == from + 1) {
+                        0 // 빈 대괄호는 허용하지 않음
+                      } else {
+                        to - from + 1 // 대괄호의 시작 위치부터 ']'까지의 길이
+                      }
+                    }
+                  }
+          )
 
   val dot by literalToken(".")
   val charToken by regexToken("[a-zA-Z0-9]")
@@ -20,24 +49,27 @@ class RegexParser : Grammar<RegexItem>() {
 
   val item: Parser<RegexItem> by
           char or
+                  (anchorOperator map { AnchorItem(it.text) }) or
                   (dot asJust DotItem()) or
-                  (skip(openParenSymbol) and (parser(::rootParser)) and skip(closeParenSymbol))
+                  (escapedCharacter map { CharItem(it.text.substring(1)) }) or
+                  (bracketContent map { BracketItem(it.text.substring(1, it.text.length - 1)) }) or
+                  (skip(openParenSymbol) and
+                          (parser(::rootParser)) and
+                          skip(closeParenSymbol) map
+                          {
+                            val groupName = "${groupCounter++}"
+                            GroupItem(it, groupName)
+                          })
 
   val term: Parser<RegexItem> by
-          (item and optional(postfixOperator)) map
+          (item and optional(postfixOperator)) map { (item, op) ->
-                  { result ->
+            when (op?.text) {
-                    result.t1.let { first ->
+              "+" -> PlusItem(item)
-                      result.t2?.let {
+              "*" -> StarItem(item)
-                        when (it.text) {
+              "?" -> QuestionItem(item)
-                          "+" -> PlusItem(first)
+              else -> item
-                          "*" -> StarItem(first)
+            }
-                          "?" -> QuestionItem(first)
+          }
-                          else -> first
-                        }
-                      }
-                              ?: first
-                    }
-                  }
   val andThen: Parser<RegexItem> by
           oneOrMore(term) map { items -> items.reduce { left, right -> AndThenItem(left, right) } }
   val termWithAlternation: Parser<RegexItem> by
@@ -48,3 +80,7 @@ class RegexParser : Grammar<RegexItem>() {
 
   override val rootParser: Parser<RegexItem> by termWithAlternation
 }
+
+fun compileRegex(input: String): RegexItem {
+  return RegexParser().parseToEnd(input)
+}

lib/src/main/kotlin/org/example/RegexItem.kt

@@ -1,175 +1,208 @@
 package org.example
 
 data class State(
-        val matched: String,
+    val input: String,
-        val remaining: String,
+    val startIndex: Int,
-)
+    val endIndex: Int,
-
+    val captures: Map<String, String> = emptyMap()
-data class AvailableState(val seq: Sequence<State> = emptySequence()) : Sequence<State> by seq {
+) {
-  val isEmpty: Boolean
+  val matched: String
-    get() = seq.none()
+    get() = input.substring(startIndex, endIndex)
+  val remaining: String
+    get() = input.substring(endIndex)
 }
 
-class MatchResult(val available: AvailableState) {
+typealias AvailableState = Sequence<State>
 
-  val isSuccess: Boolean
+// 재귀 하향 분석기.
-    get() = !this.available.isEmpty
+// 백트랙킹 기반.
+interface RegexItem {
+  override fun toString(): String
+  fun findMatch(str: String, position: Int = 0): AvailableState
+}
 
-  val isFailure: Boolean
+fun RegexItem.findAll(item: String): Sequence<State> {
-    get() = this.available.isEmpty
+  return sequence {
+    var position = 0
+    // 문자열의 끝까지 반복합니다. 비어있어도 한번은 시도합니다.
+    while (position <= item.length) {
+      // findMatch 메서드를 호출하여 매칭을 시도합니다.
+      val matchResult = findMatch(item, position).firstOrNull()
+      if (matchResult == null) {
+        // 매칭이 실패하면 position을 증가시키고 다시 시도합니다.
+        position++
+        continue
+      }
+      // 매칭이 성공하면 MatchResult를 생성하여 반환합니다.
+      yield(matchResult)
 
-  override fun toString(): String {
+      // 다음 위치로 이동합니다.
-    return if (isSuccess) {
+      position =
-      "MatchResult(success)"
+          if (matchResult.startIndex == matchResult.endIndex) {
-    } else {
+            position + 1
-      "MatchResult(failure)"
+          } else {
+            matchResult.endIndex
+          }
     }
   }
 }
 
-// 재귀 하향 분석기.
+fun RegexItem.find(item: String): State? {
-interface RegexItem {
+  // findAll 에서 첫 번째 매칭 결과를 반환합니다.
-  override fun toString(): String
+  return this.findAll(item).firstOrNull()
-  fun findMatch(str: String): AvailableState
 }
 
-fun RegexItem.match(item: String): MatchResult {
+fun RegexItem.containsMatchIn(item: String): Boolean {
-  // 기본 매칭 함수. AvailableState를 MatchResult로 변환
+  // 매칭 결과가 성공인지 확인하는 헬퍼 함수
-  return MatchResult(this.findMatch(item))
+  return this.find(item) != null
 }
 
 class AndThenItem(val left: RegexItem, val right: RegexItem) : RegexItem {
   override fun toString(): String = "${left}${right}"
-  override fun findMatch(str: String): AvailableState {
+  override fun findMatch(str: String, position: Int): AvailableState {
-    val leftMatch = left.findMatch(str)
+    val leftMatch = left.findMatch(str, position)
-    if (leftMatch.isEmpty) {
+    return leftMatch.flatMap { leftState ->
-      return AvailableState() // If left match fails, return empty sequence
+      right.findMatch(str, leftState.endIndex).map { rightState ->
+        // If right match is successful, combine the matched parts
+        State(
+            str,
+            leftState.startIndex,
+            rightState.endIndex,
+            leftState.captures + rightState.captures // Combine captures
+        )
+      }
     }
-    // If left match is successful, try to match the right item with the remaining string
-    // from the left match.
-
-    return AvailableState(
-            leftMatch.flatMap { state ->
-              val rightMatch = right.findMatch(state.remaining)
-              if (!rightMatch.isEmpty) {
-                // If right match is successful, combine the matched parts
-                rightMatch.map { rightState ->
-                  State(state.matched + rightState.matched, rightState.remaining)
-                }
-              } else {
-                // If right match fails, return an empty sequence
-                emptySequence()
-              }
-            }
-    )
   }
 }
 
 class CharItem(val value: String) : RegexItem {
-  override fun toString(): String = value
+  companion object {
-  override fun findMatch(str: String): AvailableState {
+    private val META_CHARS = setOf("\\", "+", "*", "?", ".", "(", ")", "|", "[", "]", "^", "$")
+  }
+  override fun toString(): String =
+      if (value in META_CHARS) "\\$value" else value
+
+  override fun findMatch(str: String, position: Int): AvailableState {
     return when {
       // 첫번째 문자가 value와 일치하는지 확인
-      str.isNotEmpty() && str[0].toString() == value -> {
+      position < str.length && str[position].toString() == value -> {
-        AvailableState(sequenceOf(State(value, str.substring(1))))
+        sequenceOf(State(str, position, position + 1))
       }
-      else -> AvailableState()
+      else -> emptySequence() // 일치하지 않으면 빈 시퀀스 반환
     }
   }
 }
 
-fun matchMany(
+class BracketItem(val content: String) : RegexItem {
-        str: String,
+  override fun toString(): String = "[$content]"
-        item: RegexItem,
+
-): Sequence<State> {
+  // TODO: 범위 처리
+  override fun findMatch(str: String, position: Int): AvailableState {
+    // 대괄호 안의 내용과 일치하는 첫 문자를 찾음
+    return when {
+      position < str.length && content.contains(str[position]) -> {
+        sequenceOf(State(str, position, position + 1))
+      }
+      else -> emptySequence() // 일치하지 않으면 빈 시퀀스 반환
+    }
+  }
+}
+
+class GroupItem(val item: RegexItem, val name: String) : RegexItem {
+  override fun toString(): String = "(${item})"
+
+  override fun findMatch(str: String, position: Int): AvailableState {
+    // 그룹은 내부 아이템과 동일하게 매칭을 시도
+    val ret = item.findMatch(str, position)
+    // 매칭된 상태에 그룹 이름을 추가하여 반환
+    return ret.map { state -> state.copy(captures = state.captures + (name to state.matched)) }
+  }
+}
+
+class AnchorItem(val anchor: String) : RegexItem {
+  override fun toString(): String = anchor
+
+  override fun findMatch(str: String, position: Int): AvailableState {
+    // 앵커는 문자열의 시작(^) 또는 끝($)과 매칭됨
+    return when (anchor) {
+      "^" ->
+          if (position == 0) {
+            sequenceOf(State(str, 0, 0))
+          } else {
+            emptySequence() // 시작 앵커가 실패하면 빈 시퀀스 반환
+          }
+      "$" ->
+          if (position == str.length) {
+            sequenceOf(State(str, str.length, str.length))
+          } else {
+            emptySequence() // 끝 앵커가 실패하면 빈 시퀀스 반환
+          }
+      // 다른 앵커는 지원하지 않음
+      else -> throw IllegalArgumentException("Unknown anchor: $anchor")
+    }
+  }
+}
+
+fun matchMany(str: String, item: RegexItem, position: Int): Sequence<State> {
   // 욕심쟁이 매칭을 위한 헬퍼 함수
-  return item.findMatch(str).seq.flatMap { state ->
+  return item.findMatch(str, position).flatMap { state ->
-    if (state.remaining.isEmpty()) {
+    if (state.endIndex == str.length) {
       sequenceOf(state) // If remaining is empty, return the matched state
     } else {
       // Otherwise, continue matching with the remaining string
-      matchMany(state.remaining, item).map { nextState ->
+      matchMany(str, item, state.endIndex).map { nextState ->
-        State(state.matched + nextState.matched, nextState.remaining)
+        State(str, state.startIndex, nextState.endIndex)
       } + sequenceOf(state) // Include the current state as well
     }
   }
 }
 
-// fun matchMany(str: String, item: RegexItem): Sequence<State> = sequence {
-//   val stack = mutableListOf(item.findMatch(str).seq)
-//   while (stack.isNotEmpty()) {
-//     val current = stack.removeAt(stack.lastIndex)
-//     for (state in current) {
-//       yield(state) // Yield the current state
-//       if (state.remaining.isNotEmpty()) {
-//         // If there is remaining string, continue matching
-//         stack.add(item.findMatch(state.remaining).seq.map { nextState ->
-//           State(state.matched + nextState.matched, nextState.remaining)
-//         })
-//       }
-//     }
-//   }
-// }
-
 class PlusItem(val item: RegexItem) : RegexItem {
   override fun toString(): String = "${item}+"
-  override fun findMatch(str: String): AvailableState {
+  override fun findMatch(str: String, position: Int): AvailableState {
-    return AvailableState(matchMany(str, item))
+    return matchMany(str, item, position)
   }
 }
 
 class StarItem(val item: RegexItem) : RegexItem {
   override fun toString(): String = "${item}*"
-  override fun findMatch(str: String): AvailableState {
+  override fun findMatch(str: String, position: Int): AvailableState {
-    // *는 0개 이상의 매칭을 의미하므로, 먼저 시도해보고 실패하면 빈 시퀀스를 반환
+    // *는 0번 또는 1번 이상 일치합니다.
-    val matchResult = this.item.findMatch(str)
+    // 욕심쟁이(greedy) 방식으로 구현하기 위해, 가장 긴 매치(1번 이상)를 먼저 찾고, 그 다음에 0번 매치를 추가합니다.
-    if (matchResult.isEmpty) {
+    val oneOrMoreMatches = matchMany(str, item, position)
-      // If the item does not match, return an empty sequence
+    val zeroMatch = sequenceOf(State(str, position, position))
-      return AvailableState(sequenceOf(State("", str)))
+    return oneOrMoreMatches + zeroMatch
-    }
-    // If it matches, return the successful match and continue matching with the remaining string
-    return AvailableState(
-            matchResult.flatMap { state ->
-              sequenceOf(state) + matchMany(state.remaining, this.item)
-            }
-    )
   }
 }
 
 class QuestionItem(val item: RegexItem) : RegexItem {
   override fun toString(): String = "${item}?"
-  override fun findMatch(str: String): AvailableState {
+  override fun findMatch(str: String, position: Int): AvailableState {
-    // ?는 0개 또는 1개 매칭을 의미하므로, 먼저 시도해보고 실패하면 빈 시퀀스를 반환
+    // ?는 0번 또는 1번 일치합니다.
-    val matchResult = this.item.findMatch(str)
+    val oneMatch = item.findMatch(str, position)
-    if (matchResult.isEmpty) {
+    val zeroMatch = sequenceOf(State(str, position, position))
-      // If the item does not match, return an empty sequence
+    // 1번 매치를 0번 매치보다 우선합니다.
-      return AvailableState(sequenceOf(State("", str)))
+    return oneMatch + zeroMatch
-    }
-    // If it matches, return the successful match
-    return AvailableState(matchResult.map { State(it.matched, it.remaining) })
   }
 }
 
 class DotItem : RegexItem {
   override fun toString(): String = "."
-  override fun findMatch(str: String): AvailableState =
+  override fun findMatch(str: String, position: Int): AvailableState =
-          // .은 임의의 한 문자와 매칭되므로, 첫 문자가 존재하면 매칭 성공
+      // .은 임의의 한 문자와 매칭되므로, 첫 문자가 존재하면 매칭 성공
-          when {
+      when {
-            str.isNotEmpty() ->
+        position < str.length -> sequenceOf(State(str, position, position + 1))
-                    AvailableState(sequenceOf(State(str[0].toString(), str.substring(1))))
+        else -> emptySequence() // 빈 문자열에 대해서는 매칭 실패
-            else -> AvailableState() // 빈 문자열에 대해서는 매칭 실패
+      }
-          }
 }
 
 class AlternationItem(val left: RegexItem, val right: RegexItem) : RegexItem {
-  override fun toString(): String = "(${left}|${right})"
+  override fun toString(): String = "${left}|${right}"
-  override fun findMatch(str: String): AvailableState {
+  override fun findMatch(str: String, position: Int): AvailableState {
     // Alternation은 왼쪽 또는 오른쪽 항목 중 하나와 매칭되므로, 각각 시도해보고 성공하는 경우를 반환
-    val leftMatch = left.findMatch(str)
+    val leftMatch = left.findMatch(str, position)
-    val rightMatch = right.findMatch(str)
+    val rightMatch = right.findMatch(str, position)
 
-    return AvailableState(
+    return leftMatch + rightMatch // 두 매칭 결과를 합쳐서 반환
-            (leftMatch + rightMatch) // 두 매칭 결과를 합쳐서 반환
-    )
   }
-}
+}

lib/src/test/kotlin/org/example/ParserTest.kt

@@ -2,109 +2,171 @@ package org.example
 
 import kotlin.test.Test
 import kotlin.test.assertEquals
-import com.github.h0tk3y.betterParse.grammar.parseToEnd
+import kotlin.test.assertNotNull
+
+private fun checkRegex(
+    pattern: String,
+    block: RegexTestAsserter .() -> Unit
+) {
+    val regex = compileRegex(pattern)
+    // 나중에는 안 같겠지만, 일단은 같다고 가정
+    assertEquals(pattern, regex.toString())
+    block(RegexTestAsserter(regex))
+}
+
+private class RegexTestAsserter(private val regex: RegexItem) {
+    fun String.shouldMatch() {
+        assert(regex.containsMatchIn(this)) { "Expected '$this' to match" }
+    }
+    fun String.shouldNotMatch() {
+        assert(!regex.containsMatchIn(this)) { "Expected '$this' not to match" }
+    }
+}
+
 
 class ParserTest {
 
     @Test
     fun testSimpleCharacter() {
-        val input = "a"
+        checkRegex("a") {
-        val result = RegexParser().parseToEnd(input)
+            "a".shouldMatch()
-        assertEquals("a",result.toString())
+            "b".shouldNotMatch()
+            "".shouldNotMatch()
+        }
     }
     @Test
     fun testCharacterWithPlus() {
-        val input = "a+"
+        checkRegex("a+") {
-        val parser = RegexParser()
+            "a".shouldMatch()
-        val result = parser.parseToEnd(input)
+            "aa".shouldMatch()
-        assertEquals("a+", result.toString())
+            "b".shouldNotMatch()
-        assert(result.match("a").isSuccess)
+            "".shouldNotMatch()
-        assert(result.match("aa").isSuccess)
+        }
-        assert(!result.match("b").isSuccess)
     }
     @Test
     fun testCharacterWithStar() {
-        val input = "b*"
+        checkRegex("b*") {
-        val parser = RegexParser()
+            "b".shouldMatch()
-        val result = parser.parseToEnd(input)
+            "bb".shouldMatch()
-        assertEquals("b*", result.toString())
+            "".shouldMatch() // 빈 문자열도 매칭됨
-        assert(result.match("").isSuccess)
+        }
-        assert(result.match("b").isSuccess)
-        assert(result.match("bb").isSuccess)
-        assert(result.match("a").isSuccess)
     }
     @Test
     fun testCharacterWithQuestion() {
-        val input = "c?"
+        checkRegex("c?") {
-        val parser = RegexParser()
+            "c".shouldMatch()
-        val result = parser.parseToEnd(input)
+            "".shouldMatch() // 빈 문자열도 매칭됨
-        assertEquals("c?", result.toString())
+        }
-        assert(result.match("").isSuccess)
-        assert(result.match("c").isSuccess)
-        assert(result.match("cc").isSuccess)
     }
     @Test
     fun testDot() {
-        val input = "."
+        checkRegex(".") {
-        val parser = RegexParser()
+            "a".shouldMatch()
-        val result = parser.parseToEnd(input)
+            "1".shouldMatch()
-        assertEquals(".", result.toString())
+            "".shouldNotMatch() // 빈 문자열은 매칭되지 않음
-        assert(result.match("a").isSuccess)
+        }
-        assert(result.match("1").isSuccess)
-        assert(!result.match("").isSuccess)
     }
 
     @Test
     fun testAlternation() {
-        val input = "a|b"
+        checkRegex("a|b") {
-        val parser = RegexParser()
+            "a".shouldMatch()
-        val result = parser.parseToEnd(input)
+            "b".shouldMatch()
-        assertEquals("(a|b)", result.toString())
+            "c".shouldNotMatch()
-        assert(result.match("a").isSuccess)
+            "".shouldNotMatch() // 빈 문자열은 매칭되지 않음
-        assert(result.match("b").isSuccess)
+        }
-        assert(!result.match("c").isSuccess)
     }
     @Test
     fun testParentheses() {
-        val input = "(d)"
+        checkRegex("(d)") {
-        val parser = RegexParser()
+            "d".shouldMatch()
-        val result = parser.parseToEnd(input)
+            "e".shouldNotMatch()
-        assertEquals("d", result.toString())
+            "".shouldNotMatch() // 빈 문자열은 매칭되지 않음
-        assert(result.match("d").isSuccess)
+        }
-        assert(!result.match("e").isSuccess)
     }
 
     @Test
     fun testComplexExpression() {
-        val input = "a(b|c)*d+"
+        checkRegex("a(b|c)*d+") {
-        val parser = RegexParser()
+            "ad".shouldMatch()
-        val result = parser.parseToEnd(input)
+            "ab".shouldNotMatch()
-        assertEquals("a(b|c)*d+", result.toString())
+            "acd".shouldMatch()
-        assert(result.match("ad").isSuccess)
+            "abbbd".shouldMatch()
-        assert(!result.match("ab").isSuccess)
+            "a".shouldNotMatch()
-        assert(result.match("acd").isSuccess)
+            "b".shouldNotMatch()
-        assert(result.match("abbbd").isSuccess)
+        }
-        assert(!result.match("a").isSuccess)
-        assert(!result.match("b").isSuccess)
     }
     @Test
     fun testAndThen() {
-        val input = "ab"
+        checkRegex("ab") {
-        val parser = RegexParser()
+            "ab".shouldMatch()
-        val result = parser.parseToEnd(input)
+            "a".shouldNotMatch()
-        assertEquals("ab", result.toString())
+            "b".shouldNotMatch()
-        assert(result.match("ab").isSuccess)
+        }
-        assert(!result.match("a").isSuccess)
-        assert(!result.match("b").isSuccess)
     }
     @Test
     fun testDotAndPlus() {
-        val input = ".+a";
+        checkRegex(".+a") {
-        val parser = RegexParser()
+            "a".shouldNotMatch()
-        val result = parser.parseToEnd(input)
+            "ba".shouldMatch()
-        assertEquals(".+a", result.toString())
+            "bca".shouldMatch()
-        assert(!result.match("a").isSuccess)
+        }
-        assert(result.match("ba").isSuccess)
+    }
-        assert(result.match("bca").isSuccess)
+    @Test
+    fun testEscapedCharacter() {
+        checkRegex("\\+") {
+            "+".shouldMatch()
+            "a".shouldNotMatch()
+        }
+    }
+    @Test
+    fun testBracketContent() {
+        checkRegex("[abc]") {
+            "a".shouldMatch()
+            "b".shouldMatch()
+            "c".shouldMatch()
+            "d".shouldNotMatch()
+        }
+    }
+    @Test
+    fun testNestedGroups() {
+        checkRegex("(a(b|c)d)+") {
+            "ad".shouldNotMatch()
+            "abd".shouldMatch()
+            "acd".shouldMatch()
+            "a".shouldNotMatch()
+        }
+    }
+    @Test
+    fun testAnchorOperators() {
+        checkRegex("^abc$") {
+            "abc".shouldMatch()
+            "ab".shouldNotMatch()
+            "abcd".shouldNotMatch()
+            "xabc".shouldNotMatch()
+        }
+    }
+    @Test
+    fun testCaptureGroups() {
+        val input = "(a)(b)"
+        val result = compileRegex(input)
+        assertEquals("(a)(b)", result.toString())
+        val matchResult = result.find("ab")
+        assertNotNull(matchResult, "Expected match result to be non-null")
+        assertEquals("ab", matchResult.matched)
+        assertEquals(2, matchResult.captures.size)
+        assertEquals("a", matchResult.captures["0"])
+        assertEquals("b", matchResult.captures["1"])
+    }
+    @Test
+    fun testMatchAll() {
+        val input = "a+"
+        val regex = compileRegex(input)
+        val matches = regex.findAll("aaabaaa")
+        val results = matches.toList()
+        println("Matches found: ${results}")
+        // assertEquals(2, results.size)
+        assertEquals("aaa", results[0].matched)
+        assertEquals("aaa", results[1].matched)
     }
 }