[json] IJ-CR-149952 Prefer allowed values list instead of regexp-based schemaId reporting

(cherry picked from commit 047b36e55b16ec4721f882c879dbb9e415511849)

IJ-CR-149952

GitOrigin-RevId: 6bf17ed7a46cc07e38aad2e3e403ce9315d4b29f

json/intellij.json.iml

@@ -32,5 +32,6 @@
     <orderEntry type="library" name="StreamEx" level="project" />
     <orderEntry type="library" name="kotlinx-collections-immutable" level="project" />
     <orderEntry type="module" module-name="intellij.json.split" exported="" />
+    <orderEntry type="library" name="jackson-module-kotlin" level="project" />
   </component>
 </module>

json/resources/META-INF/plugin.xml

@@ -134,6 +134,7 @@
 
     <pluginSuggestionProvider implementation="com.jetbrains.jsonSchema.wiremock.WireMockSuggestionProvider"/>
     <statistics.counterUsagesCollector implementationClass="com.jetbrains.jsonSchema.fus.JsonFeatureUsageCollector"/>
+    <statistics.validation.customValidationRule implementation="com.jetbrains.jsonSchema.fus.JsonSchemaIdValidationRule"/>
   </extensions>
 
   <extensions defaultExtensionNs="JavaScript">

json/src/com/jetbrains/jsonSchema/fus/JsonSchemaHighlightingFus.kt

@@ -4,7 +4,6 @@ package com.jetbrains.jsonSchema.fus
 import com.intellij.internal.statistic.eventLog.EventLogGroup
 import com.intellij.internal.statistic.eventLog.events.EventField
 import com.intellij.internal.statistic.eventLog.events.EventFields
-import com.intellij.internal.statistic.eventLog.events.EventFields.StringValidatedByRegexpReference
 import com.intellij.internal.statistic.eventLog.events.RoundedIntEventField
 import com.intellij.internal.statistic.eventLog.events.StringEventField
 import com.intellij.internal.statistic.service.fus.collectors.CounterUsagesCollector
@@ -13,7 +12,7 @@ import org.jetbrains.annotations.ApiStatus
 internal object JsonFeatureUsageCollector : CounterUsagesCollector() {
   private val jsonSchemaGroup = EventLogGroup(
     id = "json.schema.features",
-    version = 2,
+    version = 3,
   )
 
   internal val jsonSchemaHighlightingSessionData =
@@ -32,7 +31,7 @@ sealed interface JsonSchemaFusFeature {
   companion object {
     fun getAllRegistered(): List<EventField<*>> {
       return listOf(
-        JsonSchemaFusRegexpFeature.entries,
+        JsonSchemaFusAllowedListFeature.entries,
         JsonSchemaFusCountedUniqueFeature.entries,
         JsonSchemaFusCountedFeature.entries
       ).flatten()
@@ -41,8 +40,8 @@ sealed interface JsonSchemaFusFeature {
   }
 }
 
-enum class JsonSchemaFusRegexpFeature(override val event: StringEventField) : JsonSchemaFusFeature {
-  JsonFusSchemaId(StringValidatedByRegexpReference("schema_id", "^(https?):\\/\\/[^\\s/$.?#].[^\\s]*$", "JSON schema ID"))
+enum class JsonSchemaFusAllowedListFeature(override val event: StringEventField) : JsonSchemaFusFeature {
+  JsonFusSchemaId(EventFields.StringValidatedByCustomRule("schema_id", JsonSchemaIdValidationRule::class.java, "JSON schema ID"))
 }
 
 enum class JsonSchemaFusCountedUniqueFeature(override val event: RoundedIntEventField) : JsonSchemaFusFeature {

json/src/com/jetbrains/jsonSchema/fus/JsonSchemaHighlightingSessionStatisticsCollector.kt

@@ -6,6 +6,7 @@ import com.intellij.openapi.components.Service
 import com.intellij.openapi.components.service
 import com.intellij.openapi.diagnostic.thisLogger
 import com.intellij.psi.util.ReadActionCachedValue
+import com.jetbrains.jsonSchema.impl.JsonSchemaObject
 import org.jetbrains.annotations.ApiStatus
 import java.util.concurrent.atomic.AtomicInteger
 
@@ -42,9 +43,9 @@ class JsonSchemaHighlightingSessionStatisticsCollector {
     return currentHighlightingSession.getCachedOrEvaluate()
   }
 
-  fun reportSchemaType(schemaId: String?) {
+  fun reportSchemaType(schemaRoot: JsonSchemaObject) {
     val currentSession = getOrComputeCurrentSession() ?: return
-    currentSession.schemaType = schemaId
+    currentSession.schemaType = guessBestSchemaId(schemaRoot)
   }
 
   fun reportSchemaUsageFeature(featureKind: JsonSchemaFusFeature) {
@@ -64,7 +65,7 @@ class JsonSchemaHighlightingSessionStatisticsCollector {
       .map { (feature, usagesCount) -> feature.event.with(usagesCount) }
     val uniqueSchemasCount = JsonSchemaFusCountedUniqueFeature.UniqueRemoteUrlDownloadRequest.event.with(sessionData.requestedRemoteSchemas.size)
     val schemaAccessOutsideHighlightingCount = JsonSchemaFusCountedUniqueFeature.SchemaAccessWithoutReadLock.event.with(requestsOutsideHighlightingCounter.getAndSet(0))
-    val schemaId = JsonSchemaFusRegexpFeature.JsonFusSchemaId.event.with(sessionData.schemaType)
+    val schemaId = JsonSchemaFusAllowedListFeature.JsonFusSchemaId.event.with(sessionData.schemaType)
 
     val allDataAccumulated = allCountEventsDuringSession + uniqueSchemasCount + schemaAccessOutsideHighlightingCount + schemaId
     JsonFeatureUsageCollector.jsonSchemaHighlightingSessionData.log(allDataAccumulated)
@@ -74,5 +75,9 @@ class JsonSchemaHighlightingSessionStatisticsCollector {
       thisLogger().debug("JSON schema highlighting session statistics: $printableStatistics")
     }
   }
-}
 
+  private fun guessBestSchemaId(schemaRoot: JsonSchemaObject): String? {
+    val rawSchemaIdentifier = schemaRoot.id ?: schemaRoot.rawFile?.name
+    return rawSchemaIdentifier?.replace("http://", "https://")
+  }
+}

json/src/com/jetbrains/jsonSchema/fus/JsonSchemaIdValidationRule.kt

@@ -0,0 +1,57 @@
+// Copyright 2000-2024 JetBrains s.r.o. and contributors. Use of this source code is governed by the Apache 2.0 license.
+package com.jetbrains.jsonSchema.fus
+
+import com.fasterxml.jackson.annotation.JsonIgnoreProperties
+import com.fasterxml.jackson.core.JsonFactory
+import com.fasterxml.jackson.core.StreamReadFeature
+import com.fasterxml.jackson.databind.ObjectMapper
+import com.intellij.internal.statistic.eventLog.validator.ValidationResultType
+import com.intellij.internal.statistic.eventLog.validator.rules.EventContext
+import com.intellij.internal.statistic.eventLog.validator.rules.impl.CustomValidationRule
+import com.intellij.openapi.diagnostic.thisLogger
+import java.io.IOException
+
+/**
+ * All known schema ids and names downloaded from https://schemastore.org/api/json/catalog.json
+ * The list is quite big, so it is extracted to a separate resource file and must be loaded outside EDT to prevent possible freezes.
+ */
+internal class JsonSchemaIdValidationRule : CustomValidationRule() {
+  override fun getRuleId(): String = "json_schema_id_rule"
+
+  override fun doValidate(data: String, context: EventContext): ValidationResultType {
+    return if (AllowListHolder.allowedNames.contains(data)) ValidationResultType.ACCEPTED else ValidationResultType.REJECTED
+  }
+
+  object AllowListHolder {
+    val allowedNames: Set<String> by lazy {
+      deserialiseBundledAllowedSchemaIds()
+    }
+
+    fun deserialiseBundledAllowedSchemaIds(): Set<String> {
+      val bundledDataStream =
+        try {
+          JsonSchemaIdValidationRule::class.java.getResourceAsStream("KnownSchemaIdentifiers.json").use { stream ->
+            val objectMapper = ObjectMapper(
+              JsonFactory.builder().enable(StreamReadFeature.INCLUDE_SOURCE_IN_LOCATION).build()
+            )
+            objectMapper.readValue<ArrayList<KnownJsonSchemaIdentity>>(
+              stream,
+              objectMapper.typeFactory.constructCollectionType(ArrayList::class.java, KnownJsonSchemaIdentity::class.java)
+            )
+          }
+        }
+        catch (exception: IOException) {
+          thisLogger().warn("Failed to load bundled allowed schema identifiers", exception)
+          return emptySet()
+        }
+      return bundledDataStream
+        .asSequence()
+        .flatMap { sequenceOf(it.url, it.fileName).filterNotNull() }
+        .toSet()
+    }
+
+    @JsonIgnoreProperties(ignoreUnknown = true)
+    data class KnownJsonSchemaIdentity(val url: String? = null, val fileName: String? = null)
+  }
+}
+

json/src/com/jetbrains/jsonSchema/impl/JsonSchemaComplianceChecker.java

@@ -59,7 +59,7 @@ public final class JsonSchemaComplianceChecker {
   }
 
   public void annotate(final @NotNull PsiElement element) {
-    JsonSchemaHighlightingSessionStatisticsCollector.getInstance().reportSchemaType(myRootSchema.getId());
+    JsonSchemaHighlightingSessionStatisticsCollector.getInstance().reportSchemaType(myRootSchema);
     doAnnotate(element);
     JsonSchemaHighlightingSessionStatisticsCollector.getInstance().flushHighlightingSessionDataToFus();
   }