feat: create scanner cli to dispatch the context data to different an…

…alyser

scanner_cli/src/main/kotlin/org/archguard/scanner/ctl/Runner.kt

@@ -0,0 +1,71 @@
+package org.archguard.scanner.ctl
+
+import chapi.domain.core.CodeDataStruct
+import org.archguard.scanner.core.AnalyserSpec
+import org.archguard.scanner.core.context.Context
+import org.archguard.scanner.core.sourcecode.ASTSourceCodeAnalyser
+import org.archguard.scanner.core.sourcecode.FrontierSourceCodeAnalyser
+import org.archguard.scanner.core.sourcecode.SourceCodeContext
+import org.archguard.scanner.ctl.client.ArchGuardHttpClient
+import org.archguard.scanner.ctl.impl.CliSourceCodeContext
+import org.archguard.scanner.ctl.loader.AnalyserDispatcher
+
+// cli main entry
+fun main() {
+    val params = mapOf(
+        "type" to "source_code",
+        "language" to "java",
+        "features" to "db",
+        "path" to "kotlin/org/archguard/scanner/Runner.kt",
+        "systemId" to "1",
+        "withoutStorage" to "false",
+        "archguard_server_url" to "http://localhost:8080/api/v1/",
+        "specs" to listOf(
+            mapOf(
+                "name" to "customized_scanner",
+                "url" to "https://raw.githubusercontent.com/archguard/archguard-scanner/master/official_scanner_specs.json"
+            )
+        )
+    )
+
+    // create source code context when type is source_code
+    val context: Context = CliSourceCodeContext(
+        language = params.getValue("language").toString(),
+        features = params.getValue("features").toString().split(","),
+        path = params.getValue("path").toString(),
+        systemId = params.getValue("systemId").toString(),
+        withoutStorage = params.getValue("systemId").toString().toBoolean(),
+        client = ArchGuardHttpClient(params.getValue("archguard_server_url").toString()),
+    )
+    val customizedScannerSpecs = (params.getValue("specs") as List<Map<String, String>>).map {
+        AnalyserSpec(
+            identifier = "db",
+            host = "host",
+            version = "version",
+            jar = "jar",
+            className = "DBAnalyser",
+        )
+    }
+
+    // execute
+    AnalyserDispatcher(context, customizedScannerSpecs).dispatch()
+}
+
+// move to scan_xxx
+
+class JavaAnalyser(override val context: SourceCodeContext) : FrontierSourceCodeAnalyser {
+    override fun analyse(): List<CodeDataStruct> {
+        println("im in java analyser")
+        println("language: ${context.language}")
+        println("features: ${context.features}")
+        return listOf(CodeDataStruct(NodeName = "java"))
+    }
+}
+
+class DBAnalyser(override val context: SourceCodeContext) : ASTSourceCodeAnalyser {
+    override fun analyse(input: List<CodeDataStruct>): Any? {
+        println("im in db analyser")
+        println("input: ${input.map { it.NodeName }}")
+        return null
+    }
+}

scanner_cli/src/main/kotlin/org/archguard/scanner/ctl/client/ArchGuardGrpcClient.kt

@@ -0,0 +1,6 @@
+package org.archguard.scanner.ctl.client
+
+import org.archguard.scanner.core.context.ArchGuardClient
+
+// maybe?? 通过grpc回写分析数据
+interface ArchGuardGrpcClient : ArchGuardClient

scanner_cli/src/main/kotlin/org/archguard/scanner/client/ArchGuardHttpClient.kt → scanner_cli/src/main/kotlin/org/archguard/scanner/ctl/client/ArchGuardHttpClient.kt

@@ -1,9 +1,10 @@
-package org.archguard.scanner.client
+package org.archguard.scanner.ctl.client
 
 import chapi.domain.core.CodeDataStruct
 import org.archguard.scanner.core.context.ArchGuardClient
 import org.archguard.scanner.core.context.ContainerService
 
+// 通过http api回写分析数据
 class ArchGuardHttpClient(baseUrl: String) : ArchGuardClient {
     override fun saveDataStructure(dataStructure: List<CodeDataStruct>, systemId: String, language: String) {
         TODO("Not yet implemented")

scanner_cli/src/main/kotlin/org/archguard/scanner/ctl/impl/CliSourceCodeContext.kt

@@ -0,0 +1,15 @@
+package org.archguard.scanner.ctl.impl
+
+import org.archguard.scanner.core.context.ArchGuardClient
+import org.archguard.scanner.core.sourcecode.SourceCodeContext
+
+class CliSourceCodeContext(
+    override val language: String,
+    override val features: List<String>,
+    override val client: ArchGuardClient,
+    override val path: String,
+    override val systemId: String,
+    override val withoutStorage: Boolean,
+) : SourceCodeContext {
+    // extend SourceCodeContext to accept cil specific parameters(infra related), like memory limit, queue size, etc.
+}

scanner_cli/src/main/kotlin/org/archguard/scanner/ctl/impl/OfficialAnalyserSpecs.kt

@@ -0,0 +1,20 @@
+package org.archguard.scanner.ctl.impl
+
+import org.archguard.scanner.core.AnalyserSpec
+
+enum class OfficialAnalyserSpecs(val spec: AnalyserSpec) {
+    JAVA(
+        AnalyserSpec(
+            identifier = "java",
+            host = "https://github.com/archguard/scanner/releases/download/v1.5.0",
+            version = "1.5.0",
+            jar = "scan_sourcecode-1.5.0-all.jar",
+            className = "JavaAnalyser",
+        )
+    )
+    ;
+
+    companion object {
+        fun specs() = values().map(OfficialAnalyserSpecs::spec)
+    }
+}

scanner_cli/src/main/kotlin/org/archguard/scanner/ctl/loader/AnalyserDispatcher.kt

@@ -0,0 +1,40 @@
+package org.archguard.scanner.ctl.loader
+
+import kotlinx.coroutines.async
+import kotlinx.coroutines.awaitAll
+import kotlinx.coroutines.runBlocking
+import org.archguard.scanner.core.Analyser
+import org.archguard.scanner.core.AnalyserSpec
+import org.archguard.scanner.core.context.Context
+import org.archguard.scanner.core.sourcecode.SourceCodeAnalyser
+import org.archguard.scanner.core.sourcecode.SourceCodeContext
+import org.archguard.scanner.ctl.impl.OfficialAnalyserSpecs
+
+class AnalyserDispatcher(
+    private val context: Context,
+    customized: List<AnalyserSpec>,
+) {
+    private val specs = customized + OfficialAnalyserSpecs.specs()
+
+    private fun getOrInstall(identifier: String): Analyser<Context> {
+        // specs 会形成一个优先级队列, 对于相同的identifier, 会优先使用用户自定义的. e.g. 自定义java analyser用来覆盖官方的analyser
+        val theOne = specs.find { identifier == it.identifier }
+            ?: throw IllegalArgumentException("No analyser found for identifier: $identifier")
+        return AnalyserLoader.load(context, theOne)
+    }
+
+    fun dispatch() = runBlocking {
+        when (context) {
+            is SourceCodeContext -> {
+                val languageAnalyser = getOrInstall(context.language) as SourceCodeAnalyser
+                val ast = languageAnalyser.analyse(null) ?: return@runBlocking
+                context.features.map {
+                    async {
+                        (getOrInstall(it) as SourceCodeAnalyser).analyse(ast)
+                    }
+                }.awaitAll()
+            }
+            else -> throw IllegalArgumentException("Unsupported context type")
+        }
+    }
+}

scanner_cli/src/main/kotlin/org/archguard/scanner/ctl/loader/AnalyserLoader.kt

@@ -0,0 +1,21 @@
+package org.archguard.scanner.ctl.loader
+
+import org.archguard.scanner.core.Analyser
+import org.archguard.scanner.core.AnalyserSpec
+import org.archguard.scanner.core.context.Context
+
+// TODO load the scanner via classloader
+// 扫描指定包/路径/类
+object AnalyserLoader {
+
+    fun load(context: Context, spec: AnalyserSpec): Analyser<Context> {
+        // isInstalled
+        // install
+        // get with class for name
+
+        // support full class name or short class name
+        return Class.forName("org.archguard.scanner.ctl." + spec.className)
+            .declaredConstructors[0]
+            .newInstance(context) as Analyser<Context>
+    }
+}

scanner_core/src/main/kotlin/org/archguard/scanner/core/Analyser.kt

@@ -0,0 +1,7 @@
+package org.archguard.scanner.core
+
+import org.archguard.scanner.core.context.Context
+
+interface Analyser<C : Context> {
+    val context: C
+}

scanner_core/src/main/kotlin/org/archguard/scanner/core/AnalyserSpec.kt

@@ -0,0 +1,10 @@
+package org.archguard.scanner.core
+
+data class AnalyserSpec(
+    val identifier: String,
+    val host: String,
+    val version: String,
+    val jar: String,
+    val className: String, // calculate via identifier??
+)
+

scanner_core/src/main/kotlin/org/archguard/scanner/core/context/ArchGuardClient.kt

@@ -2,6 +2,9 @@ package org.archguard.scanner.core.context
 
 import chapi.domain.core.CodeDataStruct
 
+// TODO need to declare the structure
+interface ContainerService
+
 // client of the archguard, scanner can communicate to server via this client with limited functions
 interface ArchGuardClient {
     fun saveDataStructure(dataStructure: List<CodeDataStruct>, systemId: String, language: String)