feat(metrics): 初步增加获取运行指标的功能(兼容 Prometheus).

添加兼容 Prometheus 的运行指标收集导出功能, 通过内置 HttpServer 将其导出.
默认关闭.

scalabot-app/build.gradle.kts

@@ -23,6 +23,9 @@ dependencies {
     implementation("org.telegram:telegrambots-abilities:5.6.0")
     implementation("org.telegram:telegrambots:5.6.0")
 
+    implementation("io.prometheus:simpleclient:0.15.0")
+    implementation("io.prometheus:simpleclient_httpserver:0.15.0")
+
     testImplementation(kotlin("test"))
 }
 

scalabot-app/src/main/kotlin/AppConfigs.kt

@@ -60,6 +60,12 @@ internal data class ProxyConfig(
     val port: Int = 1080
 )
 
+internal data class MetricsConfig(
+    val enable: Boolean = false,
+    val port: Int = 9386,
+    val bindAddress: String? = null
+)
+
 /**
  * ScalaBot App 配置.
  *
@@ -68,6 +74,7 @@ internal data class ProxyConfig(
  */
 internal data class AppConfig(
     val proxy: ProxyConfig = ProxyConfig(),
+    val metrics: MetricsConfig = MetricsConfig()
 )
 
 /**

scalabot-app/src/main/kotlin/AppMain.kt

@@ -1,5 +1,6 @@
 package net.lamgc.scalabot
 
+import io.prometheus.client.exporter.HTTPServer
 import kotlinx.coroutines.runBlocking
 import mu.KotlinLogging
 import net.lamgc.scalabot.util.registerShutdownHook
@@ -18,11 +19,30 @@ fun main(args: Array<String>): Unit = runBlocking {
     log.info { "ScalaBot 正在启动中..." }
     log.debug { "启动参数: ${args.joinToString(prefix = "[", postfix = "]")}" }
     initialFiles()
+    if (Const.config.metrics.enable) {
+        startMetricsServer()
+    }
     if (!launcher.launch()) {
         exitProcess(1)
     }
 }
 
+/**
+ * 启动运行指标服务器.
+ * 使用 Prometheus 指标格式.
+ */
+fun startMetricsServer() {
+    val builder = HTTPServer.Builder()
+        .withDaemonThreads(true)
+        .withPort(Const.config.metrics.port)
+        .withHostname(Const.config.metrics.bindAddress)
+
+    val httpServer = builder
+        .build()
+        .registerShutdownHook()
+    log.info { "运行指标服务器已启动. (Port: ${httpServer.port})" }
+}
+
 internal class Launcher : AutoCloseable {
 
     companion object {

scalabot-app/src/main/kotlin/ScalaBot.kt

@@ -1,5 +1,8 @@
 package net.lamgc.scalabot
 
+import io.prometheus.client.Counter
+import io.prometheus.client.Gauge
+import io.prometheus.client.Summary
 import mu.KotlinLogging
 import org.eclipse.aether.artifact.Artifact
 import org.telegram.abilitybots.api.bot.AbilityBot
@@ -7,6 +10,7 @@ import org.telegram.abilitybots.api.db.DBContext
 import org.telegram.abilitybots.api.toggle.BareboneToggle
 import org.telegram.abilitybots.api.toggle.DefaultToggle
 import org.telegram.telegrambots.bots.DefaultBotOptions
+import org.telegram.telegrambots.meta.api.objects.Update
 
 internal class ScalaBot(
     name: String,
@@ -22,6 +26,51 @@ internal class ScalaBot(
     companion object {
         @JvmStatic
         private val log = KotlinLogging.logger { }
+
+        // ------------- Metrics -------------
+
+        @JvmStatic
+        private val botUpdateCounter = Counter.build()
+            .name("updates_total")
+            .help("Total number of updates received by all bots.")
+            .labelNames("bot_name")
+            .subsystem("telegrambots")
+            .register()
+
+        @JvmStatic
+        private val botUpdateGauge = Gauge.build()
+            .name("updates_in_progress")
+            .help("Number of updates in process by all bots.")
+            .labelNames("bot_name")
+            .subsystem("telegrambots")
+            .register()
+
+        @JvmStatic
+        private val onlineBotGauge = Gauge.build()
+            .name("bots_online")
+            .help("Number of bots Online.")
+            .subsystem("telegrambots")
+            .register()
+
+        @JvmStatic
+        private val updateProcessTime = Summary.build()
+            .name("update_process_duration_seconds")
+            .help(
+                "Time to process update. (This indicator includes the pre-processing of update by TelegrammBots, " +
+                        "so it may be different from the actual execution time of ability. " +
+                        "It is not recommended to use it as the accurate execution time of ability)"
+            )
+            .labelNames("bot_name")
+            .subsystem("telegrambots")
+            .register()
+
+        @JvmStatic
+        private val exceptionHandlingCounter = Counter.build()
+            .name("updates_exception_handling")
+            .help("Number of exceptions during processing.")
+            .labelNames("bot_name")
+            .subsystem("telegrambots")
+            .register()
     }
 
     private val extensionLoader = ExtensionLoader(this)
@@ -38,4 +87,30 @@ internal class ScalaBot(
     }
 
     override fun creatorId(): Long = creatorId
+
+    override fun onUpdateReceived(update: Update?) {
+        botUpdateCounter.labels(botUsername).inc()
+        botUpdateGauge.labels(botUsername).inc()
+
+        val timer = updateProcessTime.labels(botUsername).startTimer()
+        try {
+            super.onUpdateReceived(update)
+        } catch (e: Exception) {
+            exceptionHandlingCounter.labels(botUsername).inc()
+            throw e
+        } finally {
+            timer.observeDuration()
+            botUpdateGauge.labels(botUsername).dec()
+        }
+    }
+
+    override fun onRegister() {
+        super.onRegister()
+        onlineBotGauge.inc()
+    }
+
+    override fun onClosing() {
+        super.onClosing()
+        onlineBotGauge.dec()
+    }
 }