@startuml path_management
!include shared_interfaces.puml
package "Path Management System" {
class PathResolver {
- repository_root: Path
--
+ __init__(repository_root: Optional[Path])
+ normalize_path(absolute_path: Union[str, Path]) -> str
+ resolve_path(relative_path: str) -> Path
+ compute_content_hash(file_path: Union[str, Path]) -> str
+ compute_file_hash(file_path: Union[str, Path]) -> str
- _detect_repository_root() -> Path
+ is_within_repository(path: Path) -> bool
}
class FileOperationManager {
- sqlite_store: SQLiteStore
- path_resolver: PathResolver
- semantic_indexer: Optional[SemanticIndexer]
--
+ move_file(old_path: Path, new_path: Path, content_hash: str)
+ remove_file(file_path: Union[Path, str])
+ check_file_changed(file_path: Path, stored_hash: str) -> bool
+ handle_deleted_file(relative_path: str)
- _update_file_references(old_path: str, new_path: str)
}
class MigrationManager {
- path_resolver: PathResolver
- sqlite_store: SQLiteStore
--
+ migrate_to_relative_paths()
+ compute_missing_hashes()
+ remove_duplicate_files()
+ update_repository_roots()
- _convert_path_batch(files: List[Dict])
- _detect_and_merge_duplicates()
}
}
package "Enhanced Storage Layer" {
class EnhancedSQLiteStore {
- path_resolver: PathResolver
{field} content_hash tracking
{field} soft delete support
{field} file move history
--
+ store_file(repository_id: int, file_path: Path, content_hash: str)
+ remove_file(relative_path: str, repository_id: int)
+ mark_file_deleted(relative_path: str, repository_id: int)
+ move_file(old_path: str, new_path: str, repository_id: int, content_hash: str)
+ get_file_by_content_hash(content_hash: str, repository_id: int) -> Optional[Dict]
+ cleanup_deleted_files(days_old: int = 30)
+ get_file_move_history(relative_path: str) -> List[Dict]
- _record_file_move(old_path: str, new_path: str, content_hash: str)
}
database "SQLite Database" {
table files {
* id: INTEGER
* repository_id: INTEGER
* path: TEXT
* relative_path: TEXT
* content_hash: TEXT
* is_deleted: BOOLEAN
* deleted_at: TIMESTAMP
--
UNIQUE(repository_id, relative_path)
INDEX(content_hash)
INDEX(is_deleted)
}
table file_moves {
* id: INTEGER
* repository_id: INTEGER
* old_relative_path: TEXT
* new_relative_path: TEXT
* content_hash: TEXT
* moved_at: TIMESTAMP
* move_type: TEXT
--
INDEX(content_hash)
INDEX(moved_at)
}
}
}
package "File System Integration" {
class EnhancedFileWatcher {
- path_resolver: PathResolver
- file_operation_manager: FileOperationManager
--
+ on_deleted(event: FileSystemEvent)
+ on_moved(event: FileSystemEvent)
+ on_modified(event: FileSystemEvent)
+ on_created(event: FileSystemEvent)
- _should_index_file(path: Path) -> bool
- _compute_and_check_hash(path: Path) -> Tuple[str, bool]
}
class EnhancedDispatcher {
- path_resolver: PathResolver
- file_operation_manager: FileOperationManager
- semantic_indexer: SemanticIndexer
{field} content hash caching
--
+ index_file(file_path: Path)
+ remove_file(file_path: Union[Path, str])
+ move_file(old_path: Path, new_path: Path, content_hash: str)
+ compute_file_hash(file_path: Path) -> str
+ check_file_changed(file_path: Path) -> bool
- _get_or_compute_hash(file_path: Path) -> str
- _invalidate_caches_for_file(relative_path: str)
- _coordinate_storage_operations()
}
}
package "Vector Storage Layer" {
class EnhancedSemanticIndexer {
- path_resolver: PathResolver
- qdrant: QdrantClient
{field} relative path based IDs
{field} content hash tracking
--
+ _symbol_id(relative_path: str, name: str, line: int, content_hash: str) -> int
+ remove_file(relative_path: str)
+ move_file(old_path: str, new_path: str, content_hash: str)
+ get_embeddings_by_content_hash(content_hash: str) -> List[Dict]
+ index_file(path: Path) -> dict
- _update_payload_batch(points: List[PointStruct])
- _migrate_absolute_to_relative()
}
database "Qdrant Vector DB" {
collections {
* code-embeddings-{language}
* contextual-embeddings
* document-chunks
}
payload {
* relative_path: keyword
* content_hash: keyword
* chunk_hash: keyword
* repository_id: integer
* is_deleted: boolean
}
}
}
' Relationships
PathResolver <-- FileOperationManager : uses
PathResolver <-- EnhancedSQLiteStore : uses
PathResolver <-- EnhancedFileWatcher : uses
PathResolver <-- EnhancedDispatcher : uses
PathResolver <-- MigrationManager : uses
PathResolver <-- EnhancedSemanticIndexer : uses
FileOperationManager --> EnhancedSQLiteStore : delegates to
FileOperationManager --> EnhancedSemanticIndexer : coordinates with
FileOperationManager <-- EnhancedDispatcher : uses
FileOperationManager <-- EnhancedFileWatcher : uses
EnhancedSQLiteStore --> files : stores
EnhancedSQLiteStore --> file_moves : tracks
EnhancedSemanticIndexer --> "Qdrant Vector DB" : manages
EnhancedDispatcher --> EnhancedSemanticIndexer : coordinates
EnhancedFileWatcher --> EnhancedDispatcher : notifies
EnhancedDispatcher --> FileOperationManager : uses
' Notes
note top of PathResolver
Central component for all path operations.
Ensures consistency across the system.
Handles both normalization and resolution.
end note
note right of EnhancedSQLiteStore
Backward compatible design:
- Keeps absolute path column
- Uses relative_path as primary
- Supports gradual migration
end note
note bottom of file_moves
Tracks all file movements for:
- Audit trail
- Duplicate detection
- History analysis
end note
note right of EnhancedSemanticIndexer
Vector operations ensure:
- Relative paths in all payloads
- Content hash deduplication
- Coordinated cleanup with SQLite
- Portable embeddings
end note
note bottom of "Qdrant Vector DB"
Migration updates:
- Convert file paths to relative
- Add content/chunk hashes
- Regenerate IDs portably
- Remove duplicates
end note
@enduml
