Merge pull request #46 from javecs/feature/best-match

ベストマッチを選択する機能を追加しました。

Author: javecs

build.gradle

@@ -16,7 +16,7 @@ plugins {
 }
 
 group 'xyz.javecs.tools'
-version '0.1.6'
+version '0.1.7'
 
 apply plugin: 'kotlin'
 apply plugin: 'antlr'

src/main/kotlin/xyz/javecs/tools/text2expr/Text2Expr.kt

@@ -18,22 +18,24 @@ class Text2Expr(rulePath: String = "rules") {
     }
 
     fun eval(text: String, rendered: Boolean = false): String {
-        rules.forEach {
-            try {
-                val evaluation = it.eval(text)
-                if (evaluation.value != Double.NaN) return if (rendered) {
-                    val renderer = RuleRenderer(templates.templateOf(it.name))
-                    renderer.add("variables", evaluation.variables.map { Variable(it.key, it.value) }.toList())
-                    renderer.add("expr", evaluation.expr)
-                    renderer.add("text", text)
-                    renderer.add("value", evaluation.value)
-                    renderer.render()
-                } else {
-                    evaluation.value.toString()
-                }
-            } catch (e: Exception) {
+        val evaluated = rules.map { it.eval(text) }
+                .filterNot { it.value == Double.NaN }
+                .sortedByDescending { it.coverage }
+                .toList()
+        return if (evaluated.isEmpty()) {
+            Calculator().eval(normalize(text)).value.toString()
+        } else {
+            val bestMatch = evaluated.first()
+            if (rendered) {
+                val renderer = RuleRenderer(templates.templateOf(bestMatch.rule))
+                renderer.add("variables", bestMatch.variables.map { Variable(it.key, it.value) }.toList())
+                renderer.add("expr", bestMatch.expr)
+                renderer.add("text", text)
+                renderer.add("value", bestMatch.value)
+                renderer.render()
+            } else {
+                bestMatch.value.toString()
             }
         }
-        return Calculator().eval(normalize(text)).value.toString()
     }
 }

src/main/kotlin/xyz/javecs/tools/text2expr/rules/RuleBuilder.kt

@@ -10,9 +10,13 @@ private val NotFound = -2
 private val Optional = -3
 private val tokenizer = Tokenizer()
 
-private fun coverage(scanned: Int, total: Int) = scanned.toDouble() / total.toDouble()
+private fun coverage(matched: Int, total: Int) = matched.toDouble() / total.toDouble()
 
-data class Evaluation(val value: Number = Double.NaN, val expr: List<String> = ArrayList(), val variables: Map<String, Double> = HashMap(), val coverage: Double = 0.0)
+data class Evaluation(val value: Number = Double.NaN,
+                      val expr: List<String> = ArrayList(),
+                      val variables: Map<String, Double> = HashMap(),
+                      val coverage: Double = 0.0,
+                      val rule: String = "")
 
 class RuleBuilder(source: String, val name: String = "") {
     private val parser = RuleParser()
@@ -53,16 +57,14 @@ class RuleBuilder(source: String, val name: String = "") {
         val tokens = tokenizer.tokenize(text)
         var offset = BaseOffset
         var remained = rule().size
-        var coverage = 0.0
         rule().forEach {
             val (index, optional) = indexOf(it, tokens, start = offset + 1, end = Math.min(tokens.size - remained, tokens.lastIndex))
-            if (index == NotFound) return Pair(false, coverage(index, tokens.size))
+            if (index == NotFound) return Pair(false, 0.0)
             if (it.id.isNotEmpty()) recognizedId(Pair(it.id, if (optional) it.optionalValue().toString() else tokens[index].surface))
             if (!optional) offset = index
-            coverage = coverage(index, tokens.size)
             remained--
         }
-        return Pair(true, coverage)
+        return Pair(true, coverage(rule().size, tokens.size))
     }
 
     fun eval(text: String): Evaluation {
@@ -72,9 +74,9 @@ class RuleBuilder(source: String, val name: String = "") {
         return if (matched) {
             args.forEach { calc.eval(it) }
             expr().forEach { calc.eval(it) }
-            Evaluation(calc.value, expr().toList(), calc.variables(), coverage)
+            Evaluation(calc.value, expr().toList(), calc.variables(), coverage, name)
         } else {
-            Evaluation()
+            Evaluation(rule = name)
         }
     }
 

src/test/kotlin/xyz/javecs/tools/text2expr/test/kotlin/RuleBuilderTest.kt

@@ -65,7 +65,7 @@ class RuleBuilderTest {
         val text = "１マイルは何キロメートルですか？"
         val (matched, coverage) = builder.matches(text)
         assertTrue(matched)
-        assertEquals(0.5, coverage)
+        assertEquals(0.375, coverage)
         assertEquals(1.60934, builder.eval(text).value)
     }
 

src/test/kotlin/xyz/javecs/tools/text2expr/test/kotlin/Text2ExprTest.kt

@@ -26,4 +26,12 @@ class Text2ExprTest {
         assertEquals(reply, Text2Expr().eval("1足す1", rendered = true))
     }
 
+    @Test fun eval3() {
+        val text = "1マイルは、ざっくりなんキロメートル？"
+        assertEquals("1.60934", Text2Expr("test-best-match/rule1/rule2/rule3").eval(text))
+        assertEquals("1.6093", Text2Expr("test-best-match/rule1/rule2").eval(text))
+        assertEquals("1.609", Text2Expr("test-best-match/rule1").eval(text))
+        assertEquals("1.6", Text2Expr("test-best-match").eval(text))
+    }
+
 }

src/test/resources/test-best-match/best-match.txt

@@ -0,0 +1,9 @@
+x = P2?数(1)
+SF:マイル|"mile"
+SF:は
+P1:副詞, BF:ざっくり
+SF:なん
+SF:キロメートル|キロ|"km"
+
+x*1.6
+

src/test/resources/test-best-match/rule1/rule1.txt

@@ -0,0 +1,7 @@
+x = P2?数(1)
+SF:マイル|"mile"
+SF:は
+SF:なん
+SF:キロメートル|キロ|"km"
+
+x*1.609

src/test/resources/test-best-match/rule1/rule2/rule2.txt

@@ -0,0 +1,6 @@
+x = P2?数(1)
+SF:マイル|"mile"
+SF:は
+SF:キロメートル|キロ|"km"
+
+x*1.6093

src/test/resources/test-best-match/rule1/rule2/rule3/rule3.txt

@@ -0,0 +1,5 @@
+x = P2?数(1)
+SF:マイル|"mile"
+SF:キロメートル|キロ|"km"
+
+x*1.60934