mirror of
https://github.com/MODSetter/SurfSense.git
synced 2026-07-06 22:12:12 +02:00
7fb0707933
Scoped codemod over surfsense_backend/app (excluding routes/, Wave E): renames search_space_id -> workspace_id, search_space -> workspace, SearchSpace -> Workspace across services, utils, tasks, agents, gateway, event_bus, notifications, podcasts, automations, observability params, and prompt .md files. Also flips the camelCase payload key searchSpaceId -> workspaceId (no backend reader; hard cutover). Preserved carve-outs (verbatim): Celery task names "delete_search_space_background" and "ai_sort_search_space" (wire names), and the OTel/metric key "search_space.id" (dashboards depend on it). Enum values 'SEARCH_SPACE' and SearchSourceConnector untouched.
24 lines
919 B
Python
24 lines
919 B
Python
import hashlib
|
|
|
|
from app.indexing_pipeline.connector_document import ConnectorDocument
|
|
|
|
|
|
def compute_identifier_hash(
|
|
document_type_value: str, unique_id: str, workspace_id: int
|
|
) -> str:
|
|
"""Return a stable SHA-256 hash from raw identity components."""
|
|
combined = f"{document_type_value}:{unique_id}:{workspace_id}"
|
|
return hashlib.sha256(combined.encode("utf-8")).hexdigest()
|
|
|
|
|
|
def compute_unique_identifier_hash(doc: ConnectorDocument) -> str:
|
|
"""Return a stable SHA-256 hash identifying a document by its source identity."""
|
|
return compute_identifier_hash(
|
|
doc.document_type.value, doc.unique_id, doc.workspace_id
|
|
)
|
|
|
|
|
|
def compute_content_hash(doc: ConnectorDocument) -> str:
|
|
"""Return a SHA-256 hash of the document's content scoped to its workspace."""
|
|
combined = f"{doc.workspace_id}:{doc.source_markdown}"
|
|
return hashlib.sha256(combined.encode("utf-8")).hexdigest()
|