@startuml cross_repository_search
!include ../style.puml
title Cross-Repository Search Architecture - Level 4
package "Cross-Repository Search System" {
class CrossRepositorySearchCoordinator <<singleton>> {
- multi_repo_manager: MultiRepoIndexManager
- max_workers: int
- default_result_limit: int
- memory_manager: MemoryAwarePluginManager
- plugin_loader: RepositoryPluginLoader
- _repo_capabilities_cache: Dict[str, Set[str]]
+ search_symbol(symbol, scope): AggregatedResult
+ search_code(query, scope, semantic, limit): AggregatedResult
+ get_search_statistics(): Dict
- _get_target_repositories(scope): List[RepositoryInfo]
- _execute_parallel_symbol_search(): List[CrossRepoSearchResult]
- _execute_parallel_code_search(): List[CrossRepoSearchResult]
- _aggregate_symbol_results(): AggregatedResult
- _aggregate_code_results(): AggregatedResult
- _create_symbol_signature(): str
- _create_content_hash(): str
}
class SearchScope <<dataclass>> {
+ repositories: Optional[List[str]]
+ languages: Optional[List[str]]
+ file_types: Optional[List[str]]
+ max_repositories: int
+ priority_order: bool
}
class AggregatedResult <<dataclass>> {
+ query: str
+ total_results: int
+ repositories_searched: int
+ search_time: float
+ results: List[Dict[str, Any]]
+ repository_stats: Dict[str, int]
+ deduplication_stats: Dict[str, int]
}
}
package "Repository Management" {
class MultiRepoIndexManager {
+ get_active_repositories(): List[RepositoryInfo]
+ search_across_repositories(): List[CrossRepoSearchResult]
}
class RepositoryInfo <<dataclass>> {
+ repository_id: str
+ name: str
+ path: Path
+ index_path: Path
+ language_stats: Dict[str, int]
+ total_files: int
+ total_symbols: int
+ priority: int
}
class CrossRepoSearchResult <<dataclass>> {
+ repository_id: str
+ repository_name: str
+ results: List[Dict[str, Any]]
+ search_time: float
+ error: Optional[str]
}
}
package "Storage Layer" {
class SQLiteStore {
+ search_symbols(symbol): List[Dict]
+ search_content(query): List[Dict]
+ semantic_search(query): List[Dict]
}
}
package "Memory Management" {
class MemoryAwarePluginManager {
+ get_plugin(language): Plugin
+ get_memory_status(): Dict
}
class RepositoryPluginLoader {
+ analyze_repository(path): RepositoryProfile
+ load_plugins_for_repository(): Dict[str, Plugin]
}
}
package "Parallel Processing" {
class ThreadPoolExecutor {
+ submit(fn, *args): Future
+ as_completed(futures): Iterator
}
}
' Relationships
CrossRepositorySearchCoordinator ..> SearchScope : uses
CrossRepositorySearchCoordinator ..> AggregatedResult : creates
CrossRepositorySearchCoordinator ..> MultiRepoIndexManager : coordinates
CrossRepositorySearchCoordinator ..> MemoryAwarePluginManager : manages memory
CrossRepositorySearchCoordinator ..> RepositoryPluginLoader : loads plugins
CrossRepositorySearchCoordinator ..> ThreadPoolExecutor : parallel execution
MultiRepoIndexManager ..> RepositoryInfo : manages
MultiRepoIndexManager ..> CrossRepoSearchResult : creates
MultiRepoIndexManager ..> SQLiteStore : uses
CrossRepositorySearchCoordinator --> SQLiteStore : searches
' Search Flow
note right of CrossRepositorySearchCoordinator
Search Process:
1. Parse search scope and filters
2. Get target repositories by priority
3. Execute parallel searches
4. Aggregate and deduplicate results
5. Apply ranking and limits
6. Return unified results
end note
note left of AggregatedResult
Result Structure:
- Combined results from all repos
- Repository attribution
- Deduplication statistics
- Performance metrics
- Error handling
end note
note bottom of SearchScope
Filtering Options:
- Specific repositories
- Language constraints
- File type filters
- Repository limits
- Priority ordering
end note
' Performance Notes
note top of ThreadPoolExecutor
Parallel Execution:
- Configurable worker pool
- 30s timeout per repo (symbol)
- 60s timeout per repo (code)
- Graceful error handling
end note
note bottom of CrossRepositorySearchCoordinator
Deduplication Strategy:
- Symbol: name + relative path + line
- Content: content hash + relative path
- MD5 signatures for efficiency
- Preserves repository attribution
end note
' Integration Points
package "MCP Interface" {
class MCPSearchTool {
+ cross_repo_symbol_search()
+ cross_repo_code_search()
}
}
MCPSearchTool ..> CrossRepositorySearchCoordinator : uses
' Result Ranking
note right of AggregatedResult
Ranking Strategy:
Symbol Search:
- Exact matches first
- Repository priority second
Code Search:
- Relevance score (if available)
- Repository result count
- Repository priority
end note
@enduml
