394 lines
14 KiB
YAML
394 lines
14 KiB
YAML
imports:
|
||
- { resource: 'retriex/runtime.yaml' }
|
||
- { resource: 'retriex/index.yaml' }
|
||
- { resource: 'retriex/vector.yaml' }
|
||
- { resource: 'retriex/commerce.yaml' }
|
||
- { resource: 'retriex/model.yaml' }
|
||
- { resource: 'retriex/prompt.yaml' }
|
||
- { resource: 'retriex/agent.yaml' }
|
||
- { resource: 'retriex/retrieval.yaml' }
|
||
- { resource: 'retriex/language.yaml' }
|
||
- { resource: 'retriex/query_enrichment.yaml' }
|
||
- { resource: 'retriex/search_repair.yaml' }
|
||
- { resource: 'retriex/vocabulary.yaml' }
|
||
- { resource: 'retriex/intent.yaml' }
|
||
- { resource: 'retriex/governance.yaml' }
|
||
|
||
# ------------------------------------------------------------
|
||
# Parameters
|
||
# ------------------------------------------------------------
|
||
|
||
parameters:
|
||
mto.root: '%retriex.root%'
|
||
mto.kernel.dir: '%mto.root%'
|
||
mto.locks.dir: '%retriex.locks.dir%'
|
||
|
||
mto.knowledge.root: '%retriex.knowledge.root%'
|
||
|
||
mto.knowledge.ndjson: '%retriex.knowledge.ndjson%'
|
||
mto.knowledge.index_meta: '%retriex.knowledge.index_meta%'
|
||
mto.knowledge.vector_index: '%retriex.knowledge.vector_index%'
|
||
mto.knowledge.vector_index_meta: '%retriex.knowledge.vector_index_meta%'
|
||
mto.runtime.meta: '%retriex.knowledge.runtime_meta%'
|
||
mto.knowledge.upload: '%retriex.knowledge.upload%'
|
||
|
||
mto.knowledge.tags_ndjson: '%retriex.knowledge.tags_ndjson%'
|
||
mto.knowledge.vector_tags_index: '%retriex.knowledge.vector_tags_index%'
|
||
mto.knowledge.vector_tags_index_meta: '%retriex.knowledge.vector_tags_index_meta%'
|
||
|
||
mto.vector.script_dir: '%retriex.vector.script_dir%'
|
||
|
||
mto.vector.ingest_tags_script: '%retriex.vector.ingest_tags_script%'
|
||
mto.vector.search_tags_script: '%retriex.vector.search_tags_script%'
|
||
|
||
mto.tags.rebuild_lock: '%retriex.tags.rebuild_lock%'
|
||
|
||
mto.vector.data.upload.path: '%mto.knowledge.upload%'
|
||
|
||
mto.index.chunk_size: '%retriex.index.chunk_size%'
|
||
mto.index.chunk_overlap: '%retriex.index.chunk_overlap%'
|
||
mto.index.embedding_model: '%retriex.index.embedding_model%'
|
||
mto.index.embedding_dimension: '%retriex.index.embedding_dimension%'
|
||
mto.index.scoring_version: '%retriex.index.scoring_version%'
|
||
|
||
mto.vector.python_bin: '%retriex.vector.python_bin%'
|
||
mto.vector.control_script: '%retriex.vector.control_script%'
|
||
mto.vector.ingest_script: '%retriex.vector.ingest_script%'
|
||
mto.vector.search_script: '%retriex.vector.search_script%'
|
||
mto.vector.timeout: '%retriex.vector.timeout%'
|
||
mto.vector.service_url: '%retriex.vector.service_url%'
|
||
mto.vector.host: '%retriex.vector.host%'
|
||
mto.vector.port: '%retriex.vector.port%'
|
||
|
||
mto.commerce.enabled: '%retriex.commerce.enabled%'
|
||
mto.commerce.max_shop_results: '%retriex.commerce.max_shop_results%'
|
||
mto.commerce.shop_timeout: '%retriex.commerce.shop_timeout%'
|
||
mto.commerce.store_api_base_url: '%retriex.commerce.store_api_base_url%'
|
||
mto.commerce.sales_channel_access_key: '%retriex.commerce.sales_channel_access_key%'
|
||
|
||
mto.commerce.search_repair.enabled: '%retriex.commerce.search_repair.enabled%'
|
||
mto.commerce.search_repair.max_queries: '%retriex.commerce.search_repair.max_queries%'
|
||
mto.commerce.search_repair.min_primary_results_without_repair: '%retriex.commerce.search_repair.min_primary_results_without_repair%'
|
||
|
||
|
||
|
||
# ------------------------------------------------------------
|
||
# Services
|
||
# ------------------------------------------------------------
|
||
|
||
services:
|
||
|
||
_defaults:
|
||
autowire: true
|
||
autoconfigure: true
|
||
bind:
|
||
Psr\Log\LoggerInterface $agentLogger: '@monolog.logger.agent'
|
||
string $projectDir: '%mto.root%'
|
||
|
||
App\:
|
||
resource: '../src/'
|
||
exclude:
|
||
- '../src/DependencyInjection/'
|
||
- '../src/Entity/'
|
||
- '../src/Kernel.php'
|
||
|
||
App\Controller\:
|
||
resource: '../src/Controller/'
|
||
tags: [ 'controller.service_arguments' ]
|
||
|
||
# ------------------------------------------------------------
|
||
# AI Agent – Infrastructure
|
||
# ------------------------------------------------------------
|
||
|
||
# ------------------------------------------------------------
|
||
# RetrieX Config Facades
|
||
# ------------------------------------------------------------
|
||
|
||
App\Config\ModelGenerationDefaultsConfig:
|
||
arguments:
|
||
$modelName: '%retriex.model.default_name%'
|
||
$stream: '%retriex.model.default_stream%'
|
||
$temperature: '%retriex.model.default_temperature%'
|
||
$topK: '%retriex.model.default_top_k%'
|
||
$topP: '%retriex.model.default_top_p%'
|
||
$repeatPenalty: '%retriex.model.default_repeat_penalty%'
|
||
$numCtx: '%retriex.model.default_num_ctx%'
|
||
$retrievalMaxChunks: '%retriex.model.default_retrieval_max_chunks%'
|
||
$retrievalVectorTopK: '%retriex.model.default_retrieval_vector_top_k%'
|
||
|
||
App\Config\DomainVocabularyConfig:
|
||
arguments:
|
||
$config: '%retriex.vocabulary.config%'
|
||
|
||
App\Config\ContextServiceConfig:
|
||
arguments:
|
||
$config: '%retriex.context.config%'
|
||
|
||
App\Config\CatalogIntentConfig:
|
||
arguments:
|
||
$config: '%retriex.intent.catalog.config%'
|
||
|
||
App\Config\PromptBuilderConfig:
|
||
arguments:
|
||
$config: '%retriex.prompt.config%'
|
||
|
||
App\Config\AgentRunnerConfig:
|
||
arguments:
|
||
$config: '%retriex.agent.config%'
|
||
|
||
App\Config\NdjsonHybridRetrieverConfig:
|
||
arguments:
|
||
$config: '%retriex.retrieval.config%'
|
||
|
||
App\Config\StopWordsConfig:
|
||
arguments:
|
||
$config: '%retriex.stopwords.config%'
|
||
|
||
App\Config\LanguageCleanupConfig:
|
||
arguments:
|
||
$config: '%retriex.stopwords.config%'
|
||
|
||
App\Config\QueryEnricherConfig:
|
||
arguments:
|
||
$config: '%retriex.query_enrichment.config%'
|
||
|
||
App\Config\GovernanceConfig:
|
||
arguments:
|
||
$config: '%retriex.governance.config%'
|
||
|
||
App\Config\ShopServiceConfig:
|
||
arguments:
|
||
$config: '%retriex.shop_matching.config%'
|
||
|
||
App\Infrastructure\OllamaClient:
|
||
arguments:
|
||
$apiUrl: '%env(AI_LLM_API_URL)%'
|
||
$timeoutSeconds: '%retriex.llm.timeout_seconds%'
|
||
|
||
$configProvider: '@App\Service\ModelGenerationConfigProvider'
|
||
|
||
# ------------------------------------------------------------
|
||
# AI Agent – Context & Runner
|
||
# ------------------------------------------------------------
|
||
|
||
App\Context\ContextService:
|
||
arguments:
|
||
$historyDir: '%env(AI_HISTORY_DIR)%'
|
||
$projectDir: '%mto.root%'
|
||
|
||
App\Agent\AgentRunner:
|
||
arguments:
|
||
$debug: '%env(bool:AI_DEBUG)%'
|
||
$logPrompt: '%env(bool:AI_LOG_PROMPT)%'
|
||
$logContext: '%env(bool:AI_LOG_CONTEXT)%'
|
||
|
||
# ------------------------------------------------------------
|
||
# NDJSON Retrieval Stack
|
||
# ------------------------------------------------------------
|
||
|
||
App\Knowledge\Retrieval\NdjsonChunkLookup: ~
|
||
|
||
App\Knowledge\Retrieval\NdjsonHybridRetriever: ~
|
||
|
||
App\Knowledge\Retrieval\RetrieverInterface:
|
||
alias: App\Knowledge\Retrieval\NdjsonHybridRetriever
|
||
|
||
# ------------------------------------------------------------
|
||
# Commerce / Shopware Store API
|
||
# ------------------------------------------------------------
|
||
|
||
App\Intent\CommerceIntentLite: ~
|
||
|
||
App\Config\CommerceIntentConfig:
|
||
arguments:
|
||
$config: '%retriex.intent.commerce.config%'
|
||
|
||
App\Config\CommerceQueryParserConfig:
|
||
arguments:
|
||
$config: '%retriex.commerce_query.config%'
|
||
|
||
App\Config\CommerceReferenceResolverConfig:
|
||
arguments:
|
||
$config: '%retriex.commerce_reference_resolver.config%'
|
||
|
||
App\Commerce\CommerceQueryParser: ~
|
||
|
||
App\Config\SearchRepairConfig:
|
||
arguments:
|
||
$enabled: '%retriex.commerce.search_repair.enabled%'
|
||
$maxRepairQueries: '%retriex.commerce.search_repair.max_queries%'
|
||
$minPrimaryResultsWithoutRepair: '%retriex.commerce.search_repair.min_primary_results_without_repair%'
|
||
$config: '%retriex.search_repair.config%'
|
||
|
||
App\Commerce\SearchRepairService: ~
|
||
|
||
App\Config\IntentLightConfig:
|
||
arguments:
|
||
$config: '%retriex.intent.light.config%'
|
||
|
||
App\Config\SalesIntentConfig:
|
||
arguments:
|
||
$config: '%retriex.intent.sales.config%'
|
||
|
||
App\Shopware\ShopwareCriteriaBuilder: ~
|
||
|
||
App\Shopware\StoreApiClient:
|
||
arguments:
|
||
$baseUrl: '%mto.commerce.store_api_base_url%'
|
||
$salesChannelAccessKey: '%mto.commerce.sales_channel_access_key%'
|
||
$timeoutSeconds: '%mto.commerce.shop_timeout%'
|
||
|
||
App\Commerce\ShopSearchService:
|
||
arguments:
|
||
$enabled: '%mto.commerce.enabled%'
|
||
$maxResults: '%mto.commerce.max_shop_results%'
|
||
$baseUrl: '%mto.commerce.store_api_base_url%'
|
||
|
||
# ------------------------------------------------------------
|
||
# Index Configuration Provider
|
||
# ------------------------------------------------------------
|
||
|
||
App\Index\IndexConfigurationProvider:
|
||
arguments:
|
||
$repository: '@App\Repository\IngestProfileRepository'
|
||
$fallbackChunkSize: '%mto.index.chunk_size%'
|
||
$fallbackChunkOverlap: '%mto.index.chunk_overlap%'
|
||
$fallbackEmbeddingModel: '%mto.index.embedding_model%'
|
||
$fallbackEmbeddingDimension: '%mto.index.embedding_dimension%'
|
||
$fallbackScoringVersion: '%mto.index.scoring_version%'
|
||
|
||
# ------------------------------------------------------------
|
||
# Index Meta Manager
|
||
# ------------------------------------------------------------
|
||
|
||
App\Index\IndexMetaManager:
|
||
arguments:
|
||
$metaPath: '%mto.knowledge.index_meta%'
|
||
$runTimePath: '%mto.runtime.meta%'
|
||
$provider: '@App\Index\IndexConfigurationProvider'
|
||
|
||
# ------------------------------------------------------------
|
||
# Vector Layer
|
||
# ------------------------------------------------------------
|
||
|
||
App\Vector\VectorSearchClient:
|
||
arguments:
|
||
$serviceUrl: '%mto.vector.service_url%'
|
||
$agentLogger: '@monolog.logger.agent'
|
||
$minScore: '%retriex.vector.search.min_score%'
|
||
$maxLimit: '%retriex.vector.search.max_limit%'
|
||
$timeoutSeconds: '%retriex.vector.search.http_timeout%'
|
||
|
||
App\Vector\VectorIndexBuilder:
|
||
arguments:
|
||
$pythonBin: '%mto.vector.python_bin%'
|
||
$scriptPath: '%mto.vector.ingest_script%'
|
||
$indexNdjsonPath: '%mto.knowledge.ndjson%'
|
||
$vectorIndexPath: '%mto.knowledge.vector_index%'
|
||
$timeoutSeconds: '%mto.vector.timeout%'
|
||
$configurationProvider: '@App\Index\IndexConfigurationProvider'
|
||
|
||
# ------------------------------------------------------------
|
||
# Ingest Layer (Phase B Refactor)
|
||
# ------------------------------------------------------------
|
||
|
||
App\Ingest\GuardrailValidator: ~
|
||
|
||
App\Ingest\ChunkWriteService: ~
|
||
|
||
App\Ingest\VectorRebuildService: ~
|
||
|
||
App\Ingest\IngestFlow: ~
|
||
|
||
# ------------------------------------------------------------
|
||
# Tags Export
|
||
# ------------------------------------------------------------
|
||
|
||
App\Tag\TagNdjsonExporter:
|
||
arguments:
|
||
$tagsNdjsonPath: '%mto.knowledge.tags_ndjson%'
|
||
|
||
# ------------------------------------------------------------
|
||
# Tags Vector
|
||
# ------------------------------------------------------------
|
||
|
||
App\Tag\TagVectorIndexBuilder:
|
||
arguments:
|
||
$pythonBin: '%mto.vector.python_bin%'
|
||
$scriptPath: '%mto.vector.ingest_tags_script%'
|
||
$tagsNdjsonPath: '%mto.knowledge.tags_ndjson%'
|
||
$vectorTagsIndexPath: '%mto.knowledge.vector_tags_index%'
|
||
$embeddingModel: '%mto.index.embedding_model%'
|
||
$timeoutSeconds: '%mto.vector.timeout%'
|
||
$agentLogger: '@monolog.logger.agent'
|
||
|
||
App\Tag\TagVectorSearchClient:
|
||
arguments:
|
||
$serviceUrl: '%mto.vector.service_url%'
|
||
$agentLogger: '@monolog.logger.agent'
|
||
$minScore: '%retriex.vector.tags.min_score%'
|
||
$defaultLimit: '%retriex.vector.tags.default_limit%'
|
||
$maxLimit: '%retriex.vector.tags.max_limit%'
|
||
$timeoutSeconds: '%retriex.vector.tags.http_timeout%'
|
||
|
||
App\Tag\TagRoutingService:
|
||
arguments:
|
||
$defaultTopK: '%retriex.vector.tag_routing.default_topk%'
|
||
$minBestScore: '%retriex.vector.tag_routing.min_best_score%'
|
||
$maxScoreDropFromBest: '%retriex.vector.tag_routing.max_score_drop_from_best%'
|
||
$maxRoutingTags: '%retriex.vector.tag_routing.max_routing_tags%'
|
||
$maxCandidateDocs: '%retriex.vector.tag_routing.max_candidate_docs%'
|
||
$multiTagBonusPerExtraTag: '%retriex.vector.tag_routing.multi_tag_bonus_per_extra_tag%'
|
||
$maxMultiTagBonus: '%retriex.vector.tag_routing.max_multi_tag_bonus%'
|
||
|
||
App\Tag\TagVectorIndexHealthService:
|
||
arguments:
|
||
$tagsNdjsonPath: '%mto.knowledge.tags_ndjson%'
|
||
$vectorTagsIndexPath: '%mto.knowledge.vector_tags_index%'
|
||
$vectorTagsMetaPath: '%mto.knowledge.vector_tags_index_meta%'
|
||
|
||
# ------------------------------------------------------------
|
||
# Tag Rebuild Jobs
|
||
# ------------------------------------------------------------
|
||
|
||
App\Service\TagRebuildJobService:
|
||
arguments:
|
||
$projectDir: '%mto.root%'
|
||
$agentLogger: '@monolog.logger.agent'
|
||
|
||
App\Command\TagRebuildRunJobCommand:
|
||
arguments:
|
||
$lockFilePath: '%mto.tags.rebuild_lock%'
|
||
|
||
App\Command\VectorControlCommand:
|
||
arguments:
|
||
$vectorPythonBin: '%mto.vector.python_bin%'
|
||
$vectorControlScript: '%mto.vector.control_script%'
|
||
$defaultHost: '%mto.vector.host%'
|
||
$defaultPort: '%mto.vector.port%'
|
||
$timeoutSeconds: '%mto.vector.timeout%'
|
||
|
||
App\Command\SystemRebuildCommand:
|
||
arguments:
|
||
$projectDir: '%mto.root%'
|
||
$vectorPythonBin: '%mto.vector.python_bin%'
|
||
$vectorControlScript: '%mto.vector.control_script%'
|
||
$vectorHost: '%mto.vector.host%'
|
||
$vectorPort: '%mto.vector.port%'
|
||
$vectorTimeoutSeconds: '%mto.vector.timeout%'
|
||
|
||
# ------------------------------------------------------------
|
||
# Admin Utilities
|
||
# ------------------------------------------------------------
|
||
|
||
App\Service\Admin\IndexNdjsonInspector:
|
||
arguments:
|
||
$ndJsonPath: '%mto.knowledge.ndjson%'
|
||
$indexMetaPath: '%mto.knowledge.index_meta%'
|
||
|
||
App\Vector\VectorIndexHealthService:
|
||
arguments:
|
||
$indexNdjsonPath: '%mto.knowledge.ndjson%'
|
||
$vectorIndexPath: '%mto.knowledge.vector_index%'
|
||
$vectorMetaPath: '%mto.knowledge.vector_index_meta%'
|