LAB-29: split general features provider into 2 separate providers

GitOrigin-RevId: 618f4d8df9c338a7540fc538186f3a9597fed659
2026-09-27 10:03:11 +07:00 · 2020-09-03 11:24:58 +00:00
parent 3af3578c41
commit 6a5aae7b94
5 changed files with 152 additions and 110 deletions
@@ -33,6 +33,7 @@
    <filePrediction.candidateProvider implementation="com.intellij.filePrediction.candidates.FilePredictionRecentSessionsProvider"/>

    <filePrediction.featureProvider implementation="com.intellij.filePrediction.features.FilePredictionGeneralFeatures"/>
+    <filePrediction.featureProvider implementation="com.intellij.filePrediction.features.FilePredictionSimilarityFeatures"/>
    <filePrediction.featureProvider implementation="com.intellij.filePrediction.features.history.FilePredictionHistoryFeatures"/>
    <filePrediction.featureProvider implementation="com.intellij.filePrediction.features.history.context.FilePredictionContextFeatures"/>

@@ -3,39 +3,18 @@ package com.intellij.filePrediction.features

 import com.intellij.filePrediction.features.FilePredictionFeature.Companion.binary
 import com.intellij.filePrediction.features.FilePredictionFeature.Companion.fileType
-import com.intellij.filePrediction.features.FilePredictionFeature.Companion.numerical
 import com.intellij.filePrediction.references.ExternalReferencesResult
 import com.intellij.internal.statistic.collectors.fus.fileTypes.FileTypeUsagesCollector
-import com.intellij.openapi.application.ApplicationManager
 import com.intellij.openapi.project.Project
-import com.intellij.openapi.project.guessProjectDir
-import com.intellij.openapi.roots.FileIndexFacade
-import com.intellij.openapi.util.SystemInfo
-import com.intellij.openapi.util.io.FileUtil
-import com.intellij.openapi.util.text.StringUtil
-import com.intellij.openapi.util.text.StringUtil.toLowerCase
 import com.intellij.openapi.vfs.VirtualFile
-import com.intellij.testFramework.LightVirtualFile
-import com.intellij.util.PathUtil
 import com.intellij.util.ThreeState
-import java.io.File

-class FilePredictionGeneralFeatures: FilePredictionFeatureProvider {
+class FilePredictionGeneralFeatures : FilePredictionFeatureProvider {
  companion object {
    private val FEATURES = arrayListOf(
-      "excluded",
      "file_type",
-      "in_library",
-      "in_project",
      "in_ref",
-      "in_source",
-      "light",
-      "name_prefix",
-      "path_prefix",
-      "prev_file_type",
-      "relative_path_prefix",
-      "same_dir",
-      "same_module"
+      "prev_file_type"
    )
  }

@@ -53,57 +32,10 @@ class FilePredictionGeneralFeatures: FilePredictionFeatureProvider {
      result["in_ref"] = binary(isInRef == ThreeState.YES)
    }

-    addFileTypeFeatures(result, newFile, prevFile)
-    ApplicationManager.getApplication().runReadAction {
-      if (newFile.isValid) {
-        val fileIndex = FileIndexFacade.getInstance(project)
-        result["in_project"] = binary(fileIndex.isInProjectScope(newFile))
-        result["in_source"] = binary(fileIndex.isInSource(newFile))
-        result["in_library"] = binary(fileIndex.isInLibraryClasses(newFile) || fileIndex.isInLibrarySource(newFile))
-        result["excluded"] = binary(fileIndex.isExcludedFile(newFile))
-
-        if (prevFile != null && prevFile.isValid) {
-          val newModule = fileIndex.getModuleForFile(newFile)
-          result["same_module"] = binary(newModule != null && newModule == fileIndex.getModuleForFile(prevFile))
-        }
-        result["light"] = binary(newFile is LightVirtualFile)
-      }
-    }
-
-    if (prevFile != null) {
-      val newFileName = unify(newFile.name)
-      val newFilePath = unify(newFile.path)
-      val prevFileName = unify(prevFile.name)
-      result["name_prefix"] = numerical(StringUtil.commonPrefixLength(newFileName, prevFileName))
-
-      val prevFilePath = unify(prevFile.path)
-      result["path_prefix"] = numerical(StringUtil.commonPrefixLength(newFilePath, prevFilePath))
-
-      if (!project.isDisposed) {
-        val baseDir = project.guessProjectDir()?.path?.let { unify(it) }
-        if (baseDir != null) {
-          val newRelativePath = FileUtil.getRelativePath(baseDir, newFilePath, File.separatorChar, false)
-          val prevRelativePath = FileUtil.getRelativePath(baseDir, prevFilePath, File.separatorChar, false)
-          if (newRelativePath != null && prevRelativePath != null) {
-            result["relative_path_prefix"] = numerical(StringUtil.commonPrefixLength(newRelativePath, prevRelativePath))
-          }
-        }
-      }
-      result["same_dir"] = binary(PathUtil.getParentPath(newFilePath) == PathUtil.getParentPath(prevFilePath))
-    }
-    return result
-  }
-
-  private fun addFileTypeFeatures(result: MutableMap<String, FilePredictionFeature>, newFile: VirtualFile, prevFile: VirtualFile?) {
    result["file_type"] = fileType(FileTypeUsagesCollector.getSafeFileTypeName(newFile.fileType))
-
    if (prevFile != null) {
      result["prev_file_type"] = fileType(FileTypeUsagesCollector.getSafeFileTypeName(prevFile.fileType))
    }
-  }
-
-  private fun unify(path: String) : String {
-    val caseSensitive = SystemInfo.isFileSystemCaseSensitive
-    return if (caseSensitive) FileUtil.getNameWithoutExtension(path) else FileUtil.getNameWithoutExtension(toLowerCase(path))
+    return result
  }
 }
@@ -0,0 +1,89 @@
+// Copyright 2000-2020 JetBrains s.r.o. Use of this source code is governed by the Apache 2.0 license that can be found in the LICENSE file.
+package com.intellij.filePrediction.features
+
+import com.intellij.filePrediction.features.FilePredictionFeature.Companion.binary
+import com.intellij.filePrediction.features.FilePredictionFeature.Companion.numerical
+import com.intellij.filePrediction.references.ExternalReferencesResult
+import com.intellij.openapi.application.ApplicationManager
+import com.intellij.openapi.project.Project
+import com.intellij.openapi.project.guessProjectDir
+import com.intellij.openapi.roots.FileIndexFacade
+import com.intellij.openapi.util.SystemInfo
+import com.intellij.openapi.util.io.FileUtil
+import com.intellij.openapi.util.text.StringUtil
+import com.intellij.openapi.util.text.StringUtil.toLowerCase
+import com.intellij.openapi.vfs.VirtualFile
+import com.intellij.testFramework.LightVirtualFile
+import com.intellij.util.PathUtil
+import java.io.File
+
+class FilePredictionSimilarityFeatures : FilePredictionFeatureProvider {
+  companion object {
+    private val FEATURES = arrayListOf(
+      "excluded",
+      "in_library",
+      "in_project",
+      "in_source",
+      "light",
+      "name_prefix",
+      "path_prefix",
+      "relative_path_prefix",
+      "same_dir",
+      "same_module"
+    )
+  }
+
+  override fun getName(): String = "similarity"
+
+  override fun getFeatures(): List<String> = FEATURES
+
+  override fun calculateFileFeatures(project: Project,
+                                     newFile: VirtualFile,
+                                     prevFile: VirtualFile?,
+                                     refs: ExternalReferencesResult): Map<String, FilePredictionFeature> {
+    val result = HashMap<String, FilePredictionFeature>()
+    ApplicationManager.getApplication().runReadAction {
+      if (newFile.isValid) {
+        val fileIndex = FileIndexFacade.getInstance(project)
+        result["in_project"] = binary(fileIndex.isInProjectScope(newFile))
+        result["in_source"] = binary(fileIndex.isInSource(newFile))
+        result["in_library"] = binary(fileIndex.isInLibraryClasses(newFile) || fileIndex.isInLibrarySource(newFile))
+        result["excluded"] = binary(fileIndex.isExcludedFile(newFile))
+
+        if (prevFile != null && prevFile.isValid) {
+          val newModule = fileIndex.getModuleForFile(newFile)
+          result["same_module"] = binary(newModule != null && newModule == fileIndex.getModuleForFile(prevFile))
+        }
+        result["light"] = binary(newFile is LightVirtualFile)
+      }
+    }
+
+    if (prevFile != null) {
+      val newFileName = unify(newFile.name)
+      val newFilePath = unify(newFile.path)
+      val prevFileName = unify(prevFile.name)
+      result["name_prefix"] = numerical(StringUtil.commonPrefixLength(newFileName, prevFileName))
+
+      val prevFilePath = unify(prevFile.path)
+      result["path_prefix"] = numerical(StringUtil.commonPrefixLength(newFilePath, prevFilePath))
+
+      if (!project.isDisposed) {
+        val baseDir = project.guessProjectDir()?.path?.let { unify(it) }
+        if (baseDir != null) {
+          val newRelativePath = FileUtil.getRelativePath(baseDir, newFilePath, File.separatorChar, false)
+          val prevRelativePath = FileUtil.getRelativePath(baseDir, prevFilePath, File.separatorChar, false)
+          if (newRelativePath != null && prevRelativePath != null) {
+            result["relative_path_prefix"] = numerical(StringUtil.commonPrefixLength(newRelativePath, prevRelativePath))
+          }
+        }
+      }
+      result["same_dir"] = binary(PathUtil.getParentPath(newFilePath) == PathUtil.getParentPath(prevFilePath))
+    }
+    return result
+  }
+
+  private fun unify(path: String): String {
+    val caseSensitive = SystemInfo.isFileSystemCaseSensitive
+    return if (caseSensitive) FileUtil.getNameWithoutExtension(path) else FileUtil.getNameWithoutExtension(toLowerCase(path))
+  }
+}
@@ -18,31 +18,52 @@ import org.jetbrains.jps.model.java.JavaSourceRootType
 import org.jetbrains.jps.model.java.JpsJavaExtensionService
 import java.io.File

-class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtureBuilder<ModuleFixture>>() {
-  private fun doTestGeneralFeatures(prevPath: String, newPath: String, featuresProvider: FileFeaturesProducer) {
+class FilePredictionCommonFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtureBuilder<ModuleFixture>>() {
+  private fun doTestSimilarityFeatures(prevPath: String,
+                                       newPath: String,
+                                       expected: FileFeaturesProducer) {
+    doTestFeatures(
+      prevPath, newPath,
+      listOf(FilePredictionSimilarityFeatures()),
+      expected
+    )
+  }
+
+  private fun doTestFeatures(prevPath: String, newPath: String,
+                             providers: List<FilePredictionFeatureProvider>,
+                             expectedFeaturesProvider: FileFeaturesProducer) {
    val prevFile = myFixture.addFileToProject(prevPath, "PREVIOUS FILE")
    val nextFile = myFixture.addFileToProject(newPath, "NEXT FILE")

-    val provider = FilePredictionGeneralFeatures()
-    val actual = provider.calculateFileFeatures(
-      myFixture.project, nextFile.virtualFile, prevFile.virtualFile, FAILED_COMPUTATION
-    )
-    val expected = featuresProvider.produce(myFixture.project)
+    val actual: MutableMap<String, FilePredictionFeature> = hashMapOf()
+    for (provider in providers) {
+      val features = provider.calculateFileFeatures(
+        myFixture.project, nextFile.virtualFile, prevFile.virtualFile, FAILED_COMPUTATION
+      )
+      actual.putAll(features)
+    }
+
+    val expected = expectedFeaturesProvider.produce(myFixture.project)
    for (feature in expected.entries) {
      assertTrue("Cannot find feature '${feature.key}' in $actual", actual.containsKey(feature.key))
      assertEquals("The value of feature '${feature.key}' is different from expected", feature.value, actual[feature.key])
    }
  }

-  private fun doTestGeneralFeatures(newPath: String, configurator: ProjectConfigurator, featuresProvider: FileFeaturesProducer) {
+  private fun doTestSimilarityFeatures(newPath: String, configurator: ProjectConfigurator, featuresProvider: FileFeaturesProducer) {
+    doTestFeatures(FilePredictionSimilarityFeatures(), newPath, configurator, featuresProvider)
+  }
+
+  private fun doTestFeatures(provider: FilePredictionFeatureProvider,
+                             newPath: String, configurator: ProjectConfigurator,
+                             expectedFeaturesProvider: FileFeaturesProducer) {
    val nextFile = myFixture.addFileToProject(newPath, "NEXT FILE")
    configurator.configure(myFixture.project, myModule)

-    val provider = FilePredictionGeneralFeatures()
    val actual = provider.calculateFileFeatures(
      myFixture.project, nextFile.virtualFile, null, FAILED_COMPUTATION
    )
-    val expected = featuresProvider.produce(myFixture.project)
+    val expected = expectedFeaturesProvider.produce(myFixture.project)
    for (feature in expected.entries) {
      assertTrue("Cannot find feature '${feature.key}' in $actual", actual.containsKey(feature.key))
      assertEquals("The value of feature '${feature.key}' is different from expected", feature.value, actual[feature.key])
@@ -50,7 +71,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file name prefix for completely different files`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "prevFile.txt", "nextFile.txt",
      ConstFileFeaturesProducer(
        "name_prefix" to FilePredictionFeature.numerical(0)
@@ -59,7 +80,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file name prefix for files with common prefix`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "myPrevFile.txt", "myNextFile.txt",
      ConstFileFeaturesProducer(
        "name_prefix" to FilePredictionFeature.numerical(2)
@@ -68,7 +89,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file name prefix for equal files`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "file.txt", "src/file.txt",
      ConstFileFeaturesProducer(
        "name_prefix" to FilePredictionFeature.numerical(4)
@@ -77,7 +98,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file name of different length`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "someFile.txt", "src/file.txt",
      ConstFileFeaturesProducer(
        "name_prefix" to FilePredictionFeature.numerical(0)
@@ -86,7 +107,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file name in child directory`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "src/someFile.txt", "src/file.txt",
      ConstFileFeaturesProducer(
        "name_prefix" to FilePredictionFeature.numerical(0)
@@ -95,7 +116,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file name in neighbour directories`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "src/com/site/ui/someFile.txt", "src/com/site/component/file.txt",
      ConstFileFeaturesProducer(
        "name_prefix" to FilePredictionFeature.numerical(0)
@@ -104,7 +125,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test files path in project root`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "prevFile.txt", "nextFile.txt",
      FileFeaturesByProjectPathProducer(
        "path_prefix" to FilePredictionFeature.numerical(0)
@@ -113,7 +134,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test files path in the same directory`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "src/prevFile.txt", "src/nextFile.txt",
      FileFeaturesByProjectPathProducer(
        "path_prefix" to FilePredictionFeature.numerical(4)
@@ -122,7 +143,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test files path in the neighbour directories`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "src/ui/prevFile.txt", "src/components/nextFile.txt",
      FileFeaturesByProjectPathProducer(
        "path_prefix" to FilePredictionFeature.numerical(4)
@@ -131,7 +152,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test files path of different length`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "firstFile.txt", "another/nextFile.txt",
      FileFeaturesByProjectPathProducer(
        "path_prefix" to FilePredictionFeature.numerical(0)
@@ -140,7 +161,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test files in the root directory`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "prevFile.txt", "nextFile.txt",
      ConstFileFeaturesProducer(
        "same_dir" to FilePredictionFeature.binary(true),
@@ -150,7 +171,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test files in same child directory`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "src/prevFile.txt", "src/nextFile.txt",
      ConstFileFeaturesProducer(
        "same_dir" to FilePredictionFeature.binary(true),
@@ -160,7 +181,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test files in different directories`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "src/prevFile.txt", "test/nextFile.txt",
      ConstFileFeaturesProducer(
        "same_dir" to FilePredictionFeature.binary(false),
@@ -170,7 +191,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file not in a source root`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "file.txt",
      object : ProjectConfigurator {
        override fun configure(project: Project, module: Module) {
@@ -187,7 +208,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file in source root`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "nextFile.txt",
      EmptyProjectConfigurator,
      ConstFileFeaturesProducer(
@@ -200,7 +221,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file not in a custom source root`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "nextFile.txt",
      object : ProjectConfigurator {
        override fun configure(project: Project, module: Module) {
@@ -218,7 +239,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file in a custom source root`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "src/nextFile.txt",
      object : ProjectConfigurator {
        override fun configure(project: Project, module: Module) {
@@ -236,7 +257,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file in a library source root`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "lib/nextFile.txt",
      object : ProjectConfigurator {
        override fun configure(project: Project, module: Module) {
@@ -254,7 +275,7 @@ class FilePredictionGeneralFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtu
  }

  fun `test file in library classes`() {
-    doTestGeneralFeatures(
+    doTestSimilarityFeatures(
      "lib/nextFile.txt",
      object : ProjectConfigurator {
        override fun configure(project: Project, module: Module) {
@@ -8,7 +8,6 @@ import com.intellij.testFramework.builders.ModuleFixtureBuilder
 import com.intellij.testFramework.fixtures.CodeInsightFixtureTestCase
 import com.intellij.testFramework.fixtures.ModuleFixture
 import junit.framework.TestCase
-import org.junit.Test

 /**
 * Smoke tests for a composite features provider, for provider specific checks use dedicated test class
@@ -81,15 +80,15 @@ class FilePredictionFeaturesTest : CodeInsightFixtureTestCase<ModuleFixtureBuild
    doTestFeatures(
      "core_file_type",
      "core_prev_file_type",
-      "core_in_source",
-      "core_in_project",
-      "core_in_library",
-      "core_excluded",
-      "core_same_dir",
-      "core_same_module",
-      "core_name_prefix",
-      "core_path_prefix",
-      "core_relative_path_prefix",
+      "similarity_in_source",
+      "similarity_in_project",
+      "similarity_in_library",
+      "similarity_excluded",
+      "similarity_same_dir",
+      "similarity_same_module",
+      "similarity_name_prefix",
+      "similarity_path_prefix",
+      "similarity_relative_path_prefix",
      "history_size",
      "history_uni_mle",
      "history_bi_mle"