Merge pull request #48 from teogor/feature/sealed-tokenizer

Enhance Tokenizer with sealed class structure and improved token handling
teogor · Feb 20, 2024 · 515a882 · 515a882
2 parents 41b9cb0 + 48c6521
commit 515a882
Show file tree

Hide file tree

Showing 5 changed files with 126 additions and 157 deletions.
diff --git a/sudoklify-core/api/sudoklify-core.api b/sudoklify-core/api/sudoklify-core.api
@@ -38,28 +38,29 @@ public final class dev/teogor/sudoklify/core/solver/SudokuSolver {
 	public final fun solve ()Ldev/teogor/sudoklify/common/model/SudokuSolution;
 }
 
-public final class dev/teogor/sudoklify/core/tokenizer/MultiDigitTokenizer : dev/teogor/sudoklify/core/tokenizer/Tokenizer {
-	public fun <init> (I)V
-	public fun populateLayout ([[ILjava/lang/String;)[[Ljava/lang/String;
-	public final fun populateLayout ([[ILjava/lang/String;Ljava/util/Map;)[[Ljava/lang/String;
-	public fun replaceTokens (Ljava/lang/String;Ljava/util/Map;)Ljava/lang/String;
+public abstract class dev/teogor/sudoklify/core/tokenizer/Tokenizer {
+	public static final field Companion Ldev/teogor/sudoklify/core/tokenizer/Tokenizer$Companion;
+	public abstract fun populateLayout ([[ILjava/lang/String;Ljava/util/Map;)[[Ljava/lang/String;
+	public abstract fun replaceTokens (Ljava/lang/String;Ljava/util/Map;)Ljava/lang/String;
 }
 
-public final class dev/teogor/sudoklify/core/tokenizer/SingleDigitTokenizer : dev/teogor/sudoklify/core/tokenizer/Tokenizer {
-	public fun <init> ()V
-	public fun populateLayout ([[ILjava/lang/String;)[[Ljava/lang/String;
-	public fun replaceTokens (Ljava/lang/String;Ljava/util/Map;)Ljava/lang/String;
+public final class dev/teogor/sudoklify/core/tokenizer/Tokenizer$Companion {
+	public final fun create (I)Ldev/teogor/sudoklify/core/tokenizer/Tokenizer;
 }
 
-public class dev/teogor/sudoklify/core/tokenizer/Tokenizer {
-	public static final field Companion Ldev/teogor/sudoklify/core/tokenizer/Tokenizer$Companion;
-	public fun <init> ()V
-	public fun populateLayout ([[ILjava/lang/String;)[[Ljava/lang/String;
+public final class dev/teogor/sudoklify/core/tokenizer/Tokenizer$MultiDigitTokenizer : dev/teogor/sudoklify/core/tokenizer/Tokenizer {
+	public fun <init> (I)V
+	public fun populateLayout ([[ILjava/lang/String;Ljava/util/Map;)[[Ljava/lang/String;
 	public fun replaceTokens (Ljava/lang/String;Ljava/util/Map;)Ljava/lang/String;
 }
 
-public final class dev/teogor/sudoklify/core/tokenizer/Tokenizer$Companion {
-	public final fun create (I)Ldev/teogor/sudoklify/core/tokenizer/Tokenizer;
+public final class dev/teogor/sudoklify/core/tokenizer/Tokenizer$SingleDigitTokenizer : dev/teogor/sudoklify/core/tokenizer/Tokenizer {
+	public static final field INSTANCE Ldev/teogor/sudoklify/core/tokenizer/Tokenizer$SingleDigitTokenizer;
+	public fun equals (Ljava/lang/Object;)Z
+	public fun hashCode ()I
+	public fun populateLayout ([[ILjava/lang/String;Ljava/util/Map;)[[Ljava/lang/String;
+	public fun replaceTokens (Ljava/lang/String;Ljava/util/Map;)Ljava/lang/String;
+	public fun toString ()Ljava/lang/String;
 }
 
 public final class dev/teogor/sudoklify/core/util/BoardConversionsKt {

diff --git a/sudoklify-core/src/main/kotlin/dev/teogor/sudoklify/core/generation/SudokuGenerator.kt b/sudoklify-core/src/main/kotlin/dev/teogor/sudoklify/core/generation/SudokuGenerator.kt
@@ -27,7 +27,6 @@ import dev.teogor.sudoklify.common.types.Layout
 import dev.teogor.sudoklify.common.types.SudokuString
 import dev.teogor.sudoklify.common.types.TokenMap
 import dev.teogor.sudoklify.core.io.toToken
-import dev.teogor.sudoklify.core.tokenizer.MultiDigitTokenizer
 import dev.teogor.sudoklify.core.tokenizer.Tokenizer
 import dev.teogor.sudoklify.core.util.sortRandom
 import dev.teogor.sudoklify.core.util.toBoard
@@ -41,7 +40,7 @@ internal class SudokuGenerator internal constructor(
   private val gameType: GameType,
   private val difficulty: Difficulty,
 ) {
-  private val random: Random
+  private val random: Random = Random(seed)
   private val boxDigits = gameType.gridHeight * gameType.gridWidth
   private val totalDigits = boxDigits * boxDigits
   private val baseLayout: Layout = generateBaseLayout()
@@ -61,10 +60,6 @@ internal class SudokuGenerator internal constructor(
     difficulty: Difficulty,
   ) : this(seeds, 0L, gameType, difficulty)
 
-  init {
-    random = Random(seed)
-  }
-
   private fun generateBaseLayout(): Layout {
     return Array(boxDigits) { i ->
       IntArray(boxDigits) { j -> i * boxDigits + j }
@@ -130,15 +125,7 @@ internal class SudokuGenerator internal constructor(
     layout: Layout,
     seedSequence: String,
     tokenMap: TokenMap,
-  ): Board {
-    val grid =
-      if (tokenizer is MultiDigitTokenizer) {
-        tokenizer.populateLayout(layout, seedSequence, tokenMap)
-      } else {
-        tokenizer.populateLayout(layout, tokenizer.replaceTokens(seedSequence, tokenMap))
-      }
-    return grid
-  }
+  ): Board = tokenizer.populateLayout(layout, seedSequence, tokenMap)
 
   @Deprecated(message = "toSequenceString")
   private fun boardToSequence(board: Board): SudokuString =

diff --git a/sudoklify-core/src/main/kotlin/dev/teogor/sudoklify/core/tokenizer/MultiDigitTokenizer.kt b/sudoklify-core/src/main/kotlin/dev/teogor/sudoklify/core/tokenizer/MultiDigitTokenizer.kt
diff --git a/sudoklify-core/src/main/kotlin/dev/teogor/sudoklify/core/tokenizer/SingleDigitTokenizer.kt b/sudoklify-core/src/main/kotlin/dev/teogor/sudoklify/core/tokenizer/SingleDigitTokenizer.kt
diff --git a/sudoklify-core/src/main/kotlin/dev/teogor/sudoklify/core/tokenizer/Tokenizer.kt b/sudoklify-core/src/main/kotlin/dev/teogor/sudoklify/core/tokenizer/Tokenizer.kt
@@ -20,29 +20,123 @@ import dev.teogor.sudoklify.common.types.Board
 import dev.teogor.sudoklify.common.types.Layout
 import dev.teogor.sudoklify.common.types.TokenMap
 
-// TODO: convert to Sealed Class
-open class Tokenizer {
-  open fun replaceTokens(
+/**
+ * Represents a tokenizer responsible for replacing tokens in sequences and populating
+ * layouts.
+ *
+ * @constructor Creates a Tokenizer instance with the specified number of digits.
+ */
+sealed class Tokenizer {
+  /**
+   * Replaces tokens in a given sequence using a token map.
+   *
+   * @param sequence The input sequence containing tokens.
+   * @param tokenMap A map of tokens to their replacement values.
+   * @return The sequence with replaced tokens.
+   */
+  abstract fun replaceTokens(
     sequence: String,
     tokenMap: TokenMap,
-  ): String {
-    return ""
-  }
+  ): String
 
-  open fun populateLayout(
+  /**
+   * Populates a layout with values from a sequence, replacing tokens as needed.
+   *
+   * @param layout The layout to be populated.
+   * @param sequence The input sequence containing tokens.
+   * @param tokenMap A map of tokens to their replacement values.
+   * @return The populated board.
+   */
+  abstract fun populateLayout(
     layout: Layout,
     sequence: String,
-  ): Board {
-    return emptyArray()
-  }
+    tokenMap: TokenMap,
+  ): Board
 
   companion object {
+    /**
+     * Creates a Tokenizer instance based on the number of digits.
+     *
+     * @param digits The number of digits used for tokenization.
+     * @return A SingleDigitTokenizer for 1-9 digits, or a MultiDigitTokenizer otherwise.
+     */
     fun create(digits: Int): Tokenizer {
-      return if (digits < 10) {
-        SingleDigitTokenizer()
-      } else {
-        MultiDigitTokenizer(digits)
+      return when (digits) {
+        in 1..9 -> SingleDigitTokenizer
+        else -> MultiDigitTokenizer(digits)
+      }
+    }
+  }
+
+  /**
+   * A tokenizer for sequences with single-digit tokens.
+   */
+  data object SingleDigitTokenizer : Tokenizer() {
+    override fun replaceTokens(
+      sequence: String,
+      tokenMap: TokenMap,
+    ): String {
+      val result = StringBuilder()
+      sequence.forEach { char ->
+        result.append(tokenMap[char.toString()] ?: char)
+      }
+      return result.toString()
+    }
+
+    override fun populateLayout(
+      layout: Layout,
+      sequence: String,
+      tokenMap: TokenMap,
+    ): Board {
+      with(replaceTokens(sequence, tokenMap)) {
+        return layout.map { row ->
+          row.map { cell ->
+            this[cell].toString()
+          }.toTypedArray()
+        }.toTypedArray()
+      }
+    }
+  }
+
+  /**
+   * A tokenizer for sequences with multi-digit tokens.
+   */
+  class MultiDigitTokenizer(
+    private val digits: Int,
+  ) : Tokenizer() {
+    override fun replaceTokens(
+      sequence: String,
+      tokenMap: TokenMap,
+    ): String {
+      val regex = Regex("([A-I][a-z]+)|-|[A-I][A-I]+")
+      return regex.replace(sequence) { matchResult ->
+        val token = matchResult.value
+        tokenMap[token] ?: token
       }
     }
+
+    override fun populateLayout(
+      layout: Layout,
+      sequence: String,
+      tokenMap: TokenMap,
+    ): Board {
+      val tokens = extractTokens(sequence, tokenMap)
+      return layout.map { row ->
+        row.map { cell ->
+          val index = if (cell < digits) cell else cell - digits
+          tokens[index]
+        }.toTypedArray()
+      }.toTypedArray()
+    }
+
+    private fun extractTokens(
+      sequence: String,
+      tokenMap: TokenMap,
+    ): List<String> {
+      val regex = Regex("([A-I][a-j]+)|-|[A-I]")
+      return regex.findAll(sequence).map { match ->
+        tokenMap[match.value] ?: match.value
+      }.toList()
+    }
   }
 }