implement offline recovery

FormalLanguageConstrainedPathQuerying · May 21, 2024 · a25b64a · a25b64a
1 parent b85f892
commit a25b64a
Show file tree

Hide file tree

Showing 14 changed files with 193 additions and 134 deletions.
diff --git a/benchmarks/build.gradle.kts b/benchmarks/build.gradle.kts
@@ -36,35 +36,42 @@ fun getArgs(strFolder: String): Array<String> {
     val files = resourcesDir.listFiles()!!
     return files.map { it.toString() }.sorted().toTypedArray()
 }
-
 benchmark {
     configurations {
-        named("main") {
-            val dataset = "dataset"
-            if (!hasProperty(dataset)) {
-                throw Exception("Error! Set dataset folder by property '$dataset'")
-            }
-            param("fileName", *getArgs(property(dataset).toString()))
-            this.reportFormat = "csv"
-            iterations = 15
-            iterationTime = 2000
-            iterationTimeUnit = "ms"
-            warmups = 5
-            outputTimeUnit = "ms"
-            mode = "avgt"
-            val tools = "toolName"
-            if (hasProperty(tools)) {
-                println("Run benchmarks for: .*${property(tools)}.*")
-                include(".*${property(tools)}.*")
-            }
-
+        named("main") {}
+        targets {
+            register("main")
         }
     }
-    targets {
-        register("main")
-    }
-
 }
+//benchmark {
+//    configurations {
+//        named("main") {
+//            val dataset = "dataset"
+//            if (!hasProperty(dataset)) {
+//                println("Error! Set dataset folder by property '$dataset'")
+//                throw Exception("Error! Set dataset folder by property '$dataset'")
+//            }
+//            param("fileName", *getArgs(property(dataset).toString()))
+//            this.reportFormat = "csv"
+//            iterations = 15
+//            iterationTime = 1000
+//            iterationTimeUnit = "ms"
+//            warmups = 5
+//            outputTimeUnit = "ms"
+//            mode = "avgt"
+//            val tools = "toolName"
+//            if (hasProperty(tools)) {
+//                println("Run benchmarks for: .*${property(tools)}.*")
+//                include(".*${property(tools)}.*")
+//            }
+//
+//        }
+//    }
+//    targets {
+//        register("main")
+//    }
+//}
 
 
 allOpen {

diff --git a/benchmarks/src/main/kotlin/org/AntlrBench.kt → benchmarks/src/main/kotlin/org/Antlr.kt b/benchmarks/src/main/kotlin/org/AntlrBench.kt → benchmarks/src/main/kotlin/org/Antlr.kt
@@ -5,21 +5,10 @@ import org.antlr.Java8Lexer
 import org.antlr.Java8Parser
 import org.antlr.v4.runtime.CharStreams
 import org.antlr.v4.runtime.CommonTokenStream
-import java.io.File
 
 
 @State(Scope.Benchmark)
-class AntlrBench {
-
-    @Param("Throwables.java")
-    var fileName: String = ""
-
-    lateinit var fileContents: String
-
-    @Setup
-    fun prepare() {
-        fileContents = File(fileName).readText()
-    }
+class Antlr : BaseBench(){
 
     @Benchmark
     fun measureAntlr(blackhole: Blackhole) {

diff --git a/benchmarks/src/main/kotlin/org/BaseBench.kt b/benchmarks/src/main/kotlin/org/BaseBench.kt
@@ -0,0 +1,17 @@
+package org
+
+import kotlinx.benchmark.*
+import java.io.File
+@State(Scope.Benchmark)
+abstract class BaseBench {
+    @Param("Throwables.java")
+    var fileName: String = ""
+
+    lateinit var fileContents: String
+
+    @Setup
+    open fun prepare() {
+        fileContents = File(fileName).readText()
+    }
+
+}
diff --git a/benchmarks/src/main/kotlin/org/Main.kt b/benchmarks/src/main/kotlin/org/Main.kt
@@ -52,13 +52,18 @@ fun getAnrltTokens(file: String): Int{
     return stream.numberOfOnChannelTokens
 }
 fun calculateTokens(){
-    val pth = Main::class.java.getResource("/junit")
-    val resourcesDir = File(pth.toURI())
-    val stat = File("junit-tokens.csv").printWriter()
+    val resourcesDir = File("/home/olga/gllgen/dataset_black_box/too_little")
+    val stat = File("too-little-tokens.csv").printWriter()
     stat.println("Param: fileName,Line Count")
     for (file in resourcesDir.listFiles()!!){
-        // stat.println("${file.name},${getTokenStream(file.readText()).edges.size}")
-          stat.println("${file.name},${getAnrltTokens(file.readText())}")
+        try {
+            stat.println("${file.path},${getTokenStream(file.readText()).edges.size}")
+           // stat.println("${file.path},${getAnrltTokens(file.readText())}")
+
+        }
+        catch(e: Error){
+            println(file.name)
+        }
     }
     stat.close()
 }

diff --git a/benchmarks/src/main/kotlin/org/OfflineGll.kt b/benchmarks/src/main/kotlin/org/OfflineGll.kt
@@ -0,0 +1,16 @@
+package org
+
+import kotlinx.benchmark.*
+import org.ucfs.input.LinearInputLabel
+
+
+@State(Scope.Benchmark)
+class OfflineGll : BaseBench() {
+
+    @Benchmark
+    fun measureGll(blackhole: Blackhole) {
+        val parser = org.ucfs.Java8Parser<Int, LinearInputLabel>()
+        parser.input = getTokenStream(fileContents)
+        blackhole.consume(parser.parse())
+    }
+}
diff --git a/benchmarks/src/main/kotlin/org/OfflineGllBench.kt b/benchmarks/src/main/kotlin/org/OfflineGllBench.kt
diff --git a/benchmarks/src/main/kotlin/org/OnlineGll.kt b/benchmarks/src/main/kotlin/org/OnlineGll.kt
@@ -0,0 +1,29 @@
+package org
+
+import kotlinx.benchmark.*
+import org.junit.Before
+import org.ucfs.Java8
+import org.ucfs.input.LinearInput
+import org.ucfs.input.LinearInputLabel
+import org.ucfs.parser.Gll
+
+
+@State(Scope.Benchmark)
+class OnlineGll : BaseBench(){
+
+    val startState = Java8().rsm
+    lateinit var tokens: LinearInput<Int, LinearInputLabel>
+
+    @Setup
+    @Before
+    override fun prepare() {
+        super.prepare()
+        tokens = getTokenStream(fileContents)
+    }
+
+    @Benchmark
+    fun measureGll(blackhole: Blackhole) {
+        val gll = Gll.gll(startState, getTokenStream(fileContents))
+        blackhole.consume(gll.parse())
+    }
+}
diff --git a/benchmarks/src/main/kotlin/org/OnlineGllBench.kt b/benchmarks/src/main/kotlin/org/OnlineGllBench.kt
diff --git a/benchmarks/src/main/kotlin/org/RecoveryOfflineGll.kt b/benchmarks/src/main/kotlin/org/RecoveryOfflineGll.kt
@@ -0,0 +1,31 @@
+package org
+
+import kotlinx.benchmark.*
+import org.ucfs.Java8
+import org.ucfs.input.LinearInputLabel
+import org.ucfs.parser.Gll
+
+
+@State(Scope.Benchmark)
+class RecoveryOfflineGll : BaseBench() {
+
+    @Benchmark
+    fun measureGll(blackhole: Blackhole) {
+        val parser = org.ucfs.Java8Parser<Int, LinearInputLabel>()
+        parser.input = getRecoveryTokenStream(fileContents)
+        blackhole.consume(parser.parse())
+    }
+}
+//fun main(){
+//    val inp = getRecoveryTokenStream(
+//        """            public clas HellowWorldCTY{
+//              }
+//
+//        """.trimIndent()
+//    )
+//    val p = Gll.recoveryGll(Java8().rsm, inp)
+//    assert(p.parse().first != null)
+//    val parser = org.ucfs.Java8Parser<Int, LinearInputLabel>()
+//    parser.input = inp
+//    assert(parser.parse().first != null)
+//}
diff --git a/benchmarks/src/main/kotlin/org/RecoveryOnlineGll.kt b/benchmarks/src/main/kotlin/org/RecoveryOnlineGll.kt
@@ -0,0 +1,29 @@
+package org
+
+import kotlinx.benchmark.*
+import org.junit.Before
+import org.ucfs.Java8
+import org.ucfs.input.LinearInput
+import org.ucfs.input.LinearInputLabel
+import org.ucfs.parser.Gll
+
+
+@State(Scope.Benchmark)
+class RecoveryOnlineGll : BaseBench(){
+
+    val startState = Java8().rsm
+    lateinit var tokens: LinearInput<Int, LinearInputLabel>
+
+    @Setup
+    @Before
+    override fun prepare() {
+        super.prepare()
+        tokens = getRecoveryTokenStream(fileContents)
+    }
+
+    @Benchmark
+    fun measureGll(blackhole: Blackhole) {
+        val gll = Gll.recoveryGll(startState, getRecoveryTokenStream(fileContents))
+        blackhole.consume(gll.parse())
+    }
+}
diff --git a/generator/src/main/kotlin/org/ucfs/parser/GeneratedParser.kt b/generator/src/main/kotlin/org/ucfs/parser/GeneratedParser.kt
@@ -5,7 +5,9 @@ import org.ucfs.grammar.combinator.Grammar
 import org.ucfs.input.Edge
 import org.ucfs.input.IInputGraph
 import org.ucfs.input.ILabel
+import org.ucfs.input.IRecoveryInputGraph
 import org.ucfs.parser.context.Context
+import org.ucfs.parser.context.RecoveryContext
 import org.ucfs.rsm.RsmState
 import org.ucfs.rsm.symbol.ITerminal
 import org.ucfs.rsm.symbol.Nonterminal
@@ -20,7 +22,11 @@ abstract class GeneratedParser<VertexType, LabelType : ILabel> :
             return ctx.input
         }
         set(value) {
-            ctx = Context(grammar.rsm, value)
+            ctx = if(value is IRecoveryInputGraph){
+                RecoveryContext(grammar.rsm, value)
+            } else {
+                Context(grammar.rsm, value)
+            }
         }
 
     //protected abstract val ntFuncs: HashMap<String, (Descriptor<VertexType>, SppfNode<VertexType>?) -> Unit>
@@ -37,18 +43,17 @@ abstract class GeneratedParser<VertexType, LabelType : ILabel> :
 //        val handleEdges = ntFuncs[nt] ?: throw ParsingException("Nonterminal ${nt.name} is absent from the grammar!")
 
         val pos = descriptor.inputPosition
-
-        ctx.descriptors.addToHandled(descriptor)
         val curSppfNode = descriptor.sppfNode
         val epsilonSppfNode = ctx.sppf.getEpsilonSppfNode(descriptor)
-
         val leftExtent = curSppfNode?.leftExtent
         val rightExtent = curSppfNode?.rightExtent
 
         if (state.isFinal) {
             pop(descriptor.gssNode, curSppfNode ?: epsilonSppfNode, pos)
         }
 
+        ctx.descriptors.addToHandled(descriptor)
+
         if (state.isStart && state.isFinal) {
             checkAcceptance(
                 epsilonSppfNode,
@@ -60,6 +65,15 @@ abstract class GeneratedParser<VertexType, LabelType : ILabel> :
         checkAcceptance(curSppfNode, leftExtent, rightExtent, nt)
 
         callNtFuncs(nt, descriptor, curSppfNode)
+        val input = ctx.input
+        if(input is IRecoveryInputGraph){
+            val errorRecoveryEdges = input.createRecoveryEdges(descriptor)
+            input.handleRecoveryEdges(
+                errorRecoveryEdges,
+                this::handleTerminalOrEpsilonEdge,
+                descriptor
+            )
+        }
        // ntFuncs[nt.name]!!(descriptor, curSppfNode)
     }
 

diff --git a/run_bench.sh b/run_bench.sh
@@ -0,0 +1,9 @@
+for dataset in /home/olga/gllgen/dataset_black_box/too_little #/home/olga/gllgen/java7
+do
+  for tool in Antlr Online Offline
+  do
+    echo "running $tool on $dataset, start at $(date)"
+    gradle benchmark -PtoolName=$tool -Pdataset=$dataset >> stdout_$tool.txt 2>> stderr_$tool.txt
+  done
+done
+