All Downloads are FREE. Search and download functionalities are using the official Maven repository.

engine.NlpModelRepository.kt Maven / Gradle / Ivy

There is a newer version: 24.9.4
Show newest version
/*
 * Copyright (C) 2017/2021 e-voyageurs technologies
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 * http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package ai.tock.nlp.model.service.engine

import ai.tock.nlp.core.configuration.NlpApplicationConfiguration
import ai.tock.nlp.model.EntityBuildContext
import ai.tock.nlp.model.EntityContext
import ai.tock.nlp.model.EntityContextKey
import ai.tock.nlp.model.IntentContext
import ai.tock.nlp.model.IntentContext.IntentContextKey
import ai.tock.nlp.model.ModelNotInitializedException
import ai.tock.nlp.model.TokenizerContext
import ai.tock.nlp.model.service.storage.NlpEngineModelDAO
import ai.tock.shared.Executor
import ai.tock.shared.booleanProperty
import ai.tock.shared.debug
import ai.tock.shared.error
import ai.tock.shared.injector
import com.github.salomonbrys.kodein.instance
import com.google.common.cache.Cache
import com.google.common.cache.CacheBuilder
import mu.KotlinLogging
import java.io.InputStream
import java.io.OutputStream
import java.io.PipedInputStream
import java.io.PipedOutputStream
import java.time.Instant
import java.time.Instant.now
import java.util.concurrent.CountDownLatch
import java.util.concurrent.TimeUnit.MINUTES

/**
 *
 */
internal object NlpModelRepository {

    private val logger = KotlinLogging.logger {}

    private data class ConfiguredModel(
        val nativeModel: Any?,
        val lastUpdate: Instant,
        val configuration: NlpApplicationConfiguration
    )

    private val modelDAO: NlpEngineModelDAO by injector.instance()

    private val executor: Executor by injector.instance()

    private val intentModelsCache: Cache =
        CacheBuilder.newBuilder().softValues().build()

    private val entityModelsCache: Cache =
        CacheBuilder.newBuilder().softValues().build()

    init {
        if (booleanProperty("tock_nlp_model_refresh", true)) {
            modelDAO.listenIntentModelChanges { key ->
                intentModelsCache
                    .asMap()
                    .keys
                    .find { it.id() == key }
                    ?.also {
                        logger.info { "refresh intent model for $key" }
                        intentModelsCache.put(
                            it,
                            loadIntentModel(
                                it,
                                NlpEngineRepository.getProvider(it.engineType)
                            )
                        )
                    }
            }
            modelDAO.listenEntityModelChanges { key ->
                entityModelsCache
                    .asMap()
                    .keys
                    .find { it.id() == key }
                    ?.also {
                        logger.info { "refresh entity model for $key" }
                        entityModelsCache.put(
                            it,
                            loadEntityModel(
                                it,
                                NlpEngineRepository.getProvider(it.engineType)
                            )
                        )
                    }
            }
        } else {
            logger.info { "refresh model is disabled" }
        }
    }

    fun getTokenizerModelHolder(context: TokenizerContext, conf: NlpApplicationConfiguration): TokenizerModelHolder {
        return TokenizerModelHolder(context.language, conf)
    }

    fun getConfiguration(context: IntentContext, provider: NlpEngineProvider): NlpApplicationConfiguration =
        getIntentModelHolder(context, provider).configuration

    fun getIntentModelHolder(context: IntentContext, provider: NlpEngineProvider): IntentModelHolder {
        return context
            .key()
            .let { key ->
                intentModelsCache.get(key) {
                    loadIntentModel(key, provider)
                }
            }
            .let { IntentModelHolder(context.application, it.nativeModel!!, it.configuration, it.lastUpdate) }
    }

    private fun NlpEngineProvider.configuration(configuration: NlpApplicationConfiguration? = null): NlpApplicationConfiguration =
        configuration ?: modelBuilder.defaultNlpApplicationConfiguration()

    private fun loadIntentModel(
        contextKey: IntentContextKey,
        provider: NlpEngineProvider
    ): ConfiguredModel {
        val inputStream = modelDAO.getIntentModelInputStream(contextKey)
        if (inputStream != null) {
            logger.debug { "load intent model for $contextKey" }
            val model = provider.modelIo.loadIntentModel(inputStream)
            return ConfiguredModel(
                model,
                inputStream.updatedDate,
                provider.configuration(inputStream.configuration)
            )
        }

        throw ModelNotInitializedException("no intent model found for $contextKey")
    }

    fun getConfiguration(context: EntityContext, provider: NlpEngineProvider): NlpApplicationConfiguration =
        getEntityModelHolder(context, provider)?.configuration
            ?: provider.modelBuilder.defaultNlpApplicationConfiguration()

    fun getEntityModelHolder(context: EntityContext, provider: NlpEngineProvider): EntityModelHolder? {
        return context
            .key()
            .let {
                entityModelsCache.get(it) { loadEntityModel(it, provider) }
            }
            .let { (nativeModel, lastUpdate, conf) ->
                if (nativeModel == null) null else EntityModelHolder(nativeModel, conf, lastUpdate)
            }
    }

    private fun loadEntityModel(contextKey: EntityContextKey, provider: NlpEngineProvider): ConfiguredModel {
        return modelDAO.getEntityModelInputStream(contextKey)
            ?.let { inputStream ->
                logger.debug { "load entity model for $contextKey" }
                val model = provider.modelIo.loadEntityModel(inputStream)
                return ConfiguredModel(
                    model,
                    inputStream.updatedDate,
                    provider.configuration(inputStream.configuration)
                )
            } ?: ConfiguredModel(null, now(), provider.configuration())
    }

    private fun saveModel(copy: (OutputStream) -> Unit, save: (InputStream) -> Unit, retry: Boolean = true) {
        val pipedOutputStream = PipedOutputStream()
        val pipedInputStream = PipedInputStream(pipedOutputStream)
        val latch = CountDownLatch(1)
        executor.executeBlocking {
            pipedOutputStream.use {
                try {
                    logger.debug { "Start copy model" }
                    copy(it)
                } catch (t: Throwable) {
                    if (latch.count == 0L) {
                        logger.debug(t)
                    } else {
                        logger.error(t)
                    }
                } finally {
                    logger.debug { "Copy model end" }
                }
            }
            latch.await(1, MINUTES)
            logger.debug { "latch release" }
        }
        pipedInputStream.use {
            try {
                logger.debug { "Start to save the model" }
                save(it)
            } catch (e: Exception) {
                logger.debug(e)
                if (retry) {
                    saveModel(copy, save, false)
                } else {
                    throw e
                }
            } finally {
                logger.debug { "End persisting model" }
                latch.countDown()
            }
        }
    }

    fun saveIntentModel(
        intentContextKey: IntentContextKey,
        model: IntentModelHolder,
        modelIo: NlpEngineModelIo
    ) {
        saveModel(
            { modelIo.copyIntentModel(model.nativeModel, it) },
            { modelDAO.saveIntentModel(intentContextKey, it) }
        )
    }

    fun saveEntityModel(
        entityContextKey: EntityContextKey,
        model: EntityModelHolder,
        modelIo: NlpEngineModelIo
    ) {
        saveModel(
            { modelIo.copyEntityModel(model.nativeModel, it) },
            { modelDAO.saveEntityModel(entityContextKey, it) }
        )
    }

    fun isIntentModelExist(context: IntentContext): Boolean {
        return modelDAO.getIntentModelLastUpdate(context.key()) != null
    }

    fun isEntityModelExist(context: EntityBuildContext): Boolean {
        return modelDAO.getEntityModelLastUpdate(context.key()) != null
    }

    fun removeEntityModelsNotIn(keys: List) {
        modelDAO.deleteEntityModelsNotIn(keys)
    }

    fun removeIntentModelsNotIn(keys: List) {
        modelDAO.deleteIntentModelsNotIn(keys)
    }
}




© 2015 - 2024 Weber Informatics LLC | Privacy Policy