trustgraph/trustgraph-flow/trustgraph/gateway/dispatch/manager.py

353 lines
12 KiB
Python
Raw Normal View History

import asyncio
2025-05-29 16:33:21 +01:00
from aiohttp import web
import uuid
import logging
# Module logger
logger = logging.getLogger(__name__)
from . config import ConfigRequestor
from . flow import FlowRequestor
from . librarian import LibrarianRequestor
from . knowledge import KnowledgeRequestor
from . collection_management import CollectionManagementRequestor
from . embeddings import EmbeddingsRequestor
from . agent import AgentRequestor
from . text_completion import TextCompletionRequestor
from . prompt import PromptRequestor
from . graph_rag import GraphRagRequestor
from . document_rag import DocumentRagRequestor
from . triples_query import TriplesQueryRequestor
from . rows_query import RowsQueryRequestor
from . nlp_query import NLPQueryRequestor
from . sparql_query import SparqlQueryRequestor
from . structured_query import StructuredQueryRequestor
from . structured_diag import StructuredDiagRequestor
from . embeddings import EmbeddingsRequestor
from . graph_embeddings_query import GraphEmbeddingsQueryRequestor
from . document_embeddings_query import DocumentEmbeddingsQueryRequestor
from . row_embeddings_query import RowEmbeddingsQueryRequestor
from . mcp_tool import McpToolRequestor
from . text_load import TextLoad
from . document_load import DocumentLoad
from . triples_export import TriplesExport
from . graph_embeddings_export import GraphEmbeddingsExport
from . document_embeddings_export import DocumentEmbeddingsExport
2025-05-17 13:25:09 +01:00
from . entity_contexts_export import EntityContextsExport
from . triples_import import TriplesImport
from . graph_embeddings_import import GraphEmbeddingsImport
from . document_embeddings_import import DocumentEmbeddingsImport
2025-05-17 13:25:09 +01:00
from . entity_contexts_import import EntityContextsImport
from . rows_import import RowsImport
2025-05-29 16:33:21 +01:00
from . core_export import CoreExport
from . core_import import CoreImport
from . document_stream import DocumentStreamExport
2025-05-29 16:33:21 +01:00
from . mux import Mux
request_response_dispatchers = {
"agent": AgentRequestor,
"text-completion": TextCompletionRequestor,
"prompt": PromptRequestor,
"mcp-tool": McpToolRequestor,
"graph-rag": GraphRagRequestor,
"document-rag": DocumentRagRequestor,
"embeddings": EmbeddingsRequestor,
"graph-embeddings": GraphEmbeddingsQueryRequestor,
"document-embeddings": DocumentEmbeddingsQueryRequestor,
"triples": TriplesQueryRequestor,
"rows": RowsQueryRequestor,
"nlp-query": NLPQueryRequestor,
"structured-query": StructuredQueryRequestor,
"structured-diag": StructuredDiagRequestor,
"row-embeddings": RowEmbeddingsQueryRequestor,
"sparql": SparqlQueryRequestor,
}
global_dispatchers = {
"config": ConfigRequestor,
"flow": FlowRequestor,
"librarian": LibrarianRequestor,
"knowledge": KnowledgeRequestor,
"collection-management": CollectionManagementRequestor,
}
sender_dispatchers = {
"text-load": TextLoad,
"document-load": DocumentLoad,
}
export_dispatchers = {
"triples": TriplesExport,
"graph-embeddings": GraphEmbeddingsExport,
"document-embeddings": DocumentEmbeddingsExport,
2025-05-17 13:25:09 +01:00
"entity-contexts": EntityContextsExport,
}
import_dispatchers = {
"triples": TriplesImport,
"graph-embeddings": GraphEmbeddingsImport,
"document-embeddings": DocumentEmbeddingsImport,
2025-05-17 13:25:09 +01:00
"entity-contexts": EntityContextsImport,
"rows": RowsImport,
}
class DispatcherWrapper:
def __init__(self, handler):
self.handler = handler
async def process(self, *args):
return await self.handler(*args)
class DispatcherManager:
def __init__(self, backend, config_receiver, prefix="api-gateway",
2025-12-06 11:01:20 +00:00
queue_overrides=None):
self.backend = backend
self.config_receiver = config_receiver
self.config_receiver.add_handler(self)
self.prefix = prefix
2025-12-06 11:01:20 +00:00
# Store queue overrides for global services
# Format: {"config": {"request": "...", "response": "..."}, ...}
self.queue_overrides = queue_overrides or {}
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
# Flows keyed by (workspace, flow_id)
self.flows = {}
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
# Dispatchers keyed by (workspace, flow_id, kind)
self.dispatchers = {}
self.dispatcher_lock = asyncio.Lock()
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
async def start_flow(self, workspace, id, flow):
logger.info(f"Starting flow {workspace}/{id}")
self.flows[(workspace, id)] = flow
return
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
async def stop_flow(self, workspace, id, flow):
logger.info(f"Stopping flow {workspace}/{id}")
del self.flows[(workspace, id)]
return
def dispatch_global_service(self):
return DispatcherWrapper(self.process_global_service)
2025-05-29 16:33:21 +01:00
def dispatch_core_export(self):
return DispatcherWrapper(self.process_core_export)
def dispatch_core_import(self):
return DispatcherWrapper(self.process_core_import)
def dispatch_document_stream(self):
return DispatcherWrapper(self.process_document_stream)
async def process_document_stream(self, data, error, ok, request):
ds = DocumentStreamExport(self.backend)
return await ds.process(data, error, ok, request)
async def process_core_import(self, data, error, ok, request):
2025-05-29 16:33:21 +01:00
ci = CoreImport(self.backend)
return await ci.process(data, error, ok, request)
2025-05-29 16:33:21 +01:00
async def process_core_export(self, data, error, ok, request):
2025-05-29 16:33:21 +01:00
ce = CoreExport(self.backend)
return await ce.process(data, error, ok, request)
2025-05-29 16:33:21 +01:00
async def process_global_service(self, data, responder, params):
kind = params.get("kind")
return await self.invoke_global_service(data, responder, kind)
async def invoke_global_service(self, data, responder, kind):
key = (None, kind)
if key not in self.dispatchers:
async with self.dispatcher_lock:
if key not in self.dispatchers:
request_queue = None
response_queue = None
if kind in self.queue_overrides:
request_queue = self.queue_overrides[kind].get("request")
response_queue = self.queue_overrides[kind].get("response")
dispatcher = global_dispatchers[kind](
backend = self.backend,
timeout = 120,
consumer = f"{self.prefix}-{kind}-request",
subscriber = f"{self.prefix}-{kind}-request",
request_queue = request_queue,
response_queue = response_queue,
)
await dispatcher.start()
self.dispatchers[key] = dispatcher
return await self.dispatchers[key].process(data, responder)
def dispatch_flow_import(self):
return self.process_flow_import
def dispatch_flow_export(self):
return self.process_flow_export
def dispatch_socket(self):
return self.process_socket
def dispatch_flow_service(self):
return DispatcherWrapper(self.process_flow_service)
async def process_flow_import(self, ws, running, params):
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
workspace = params.get("workspace", "default")
flow = params.get("flow")
kind = params.get("kind")
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
flow_key = (workspace, flow)
if flow_key not in self.flows:
raise RuntimeError(f"Invalid flow {workspace}/{flow}")
if kind not in import_dispatchers:
raise RuntimeError("Invalid kind")
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
key = (workspace, flow, kind)
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
intf_defs = self.flows[flow_key]["interfaces"]
2025-05-17 13:25:09 +01:00
# FIXME: The -store bit, does it make sense?
if kind == "entity-contexts":
int_kind = kind + "-load"
else:
int_kind = kind + "-store"
if int_kind not in intf_defs:
raise RuntimeError("This kind not supported by flow")
# FIXME: The -store bit, does it make sense?
Flow service lifecycle management (#822) feat: separate flow service from config service with explicit queue lifecycle management The flow service is now an independent service that owns the lifecycle of flow and blueprint queues. System services own their own queues. Consumers never create queues. Flow service separation: - New service at trustgraph-flow/trustgraph/flow/service/ - Uses async ConfigClient (RequestResponse pattern) to talk to config service - Config service stripped of all flow handling Queue lifecycle management: - PubSubBackend protocol gains create_queue, delete_queue, queue_exists, ensure_queue — all async - RabbitMQ: implements via pika with asyncio.to_thread internally - Pulsar: stubs for future admin REST API implementation - Consumer _connect() no longer creates queues (passive=True for named queues) - System services call ensure_queue on startup - Flow service creates queues on flow start, deletes on flow stop - Flow service ensures queues for pre-existing flows on startup Two-phase flow stop: - Phase 1: set flow status to "stopping", delete processor config entries - Phase 2: retry queue deletion, then delete flow record Config restructure: - active-flow config replaced with processor:{name} types - Each processor has its own config type, each flow variant is a key - Flow start/stop use batch put/delete — single config push per operation - FlowProcessor subscribes to its own type only Blueprint format: - Processor entries split into topics and parameters dicts - Flow interfaces use {"flow": "topic"} instead of bare strings - Specs (ConsumerSpec, ProducerSpec, etc.) read from definition["topics"] Tests updated
2026-04-16 17:19:39 +01:00
qconfig = intf_defs[int_kind]["flow"]
id = str(uuid.uuid4())
dispatcher = import_dispatchers[kind](
backend = self.backend,
ws = ws,
running = running,
queue = qconfig,
)
2025-05-17 13:01:52 +01:00
await dispatcher.start()
return dispatcher
async def process_flow_export(self, ws, running, params):
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
workspace = params.get("workspace", "default")
flow = params.get("flow")
kind = params.get("kind")
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
flow_key = (workspace, flow)
if flow_key not in self.flows:
raise RuntimeError(f"Invalid flow {workspace}/{flow}")
if kind not in export_dispatchers:
raise RuntimeError("Invalid kind")
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
key = (workspace, flow, kind)
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
intf_defs = self.flows[flow_key]["interfaces"]
2025-05-17 13:25:09 +01:00
# FIXME: The -store bit, does it make sense?
if kind == "entity-contexts":
int_kind = kind + "-load"
else:
int_kind = kind + "-store"
if int_kind not in intf_defs:
raise RuntimeError("This kind not supported by flow")
Flow service lifecycle management (#822) feat: separate flow service from config service with explicit queue lifecycle management The flow service is now an independent service that owns the lifecycle of flow and blueprint queues. System services own their own queues. Consumers never create queues. Flow service separation: - New service at trustgraph-flow/trustgraph/flow/service/ - Uses async ConfigClient (RequestResponse pattern) to talk to config service - Config service stripped of all flow handling Queue lifecycle management: - PubSubBackend protocol gains create_queue, delete_queue, queue_exists, ensure_queue — all async - RabbitMQ: implements via pika with asyncio.to_thread internally - Pulsar: stubs for future admin REST API implementation - Consumer _connect() no longer creates queues (passive=True for named queues) - System services call ensure_queue on startup - Flow service creates queues on flow start, deletes on flow stop - Flow service ensures queues for pre-existing flows on startup Two-phase flow stop: - Phase 1: set flow status to "stopping", delete processor config entries - Phase 2: retry queue deletion, then delete flow record Config restructure: - active-flow config replaced with processor:{name} types - Each processor has its own config type, each flow variant is a key - Flow start/stop use batch put/delete — single config push per operation - FlowProcessor subscribes to its own type only Blueprint format: - Processor entries split into topics and parameters dicts - Flow interfaces use {"flow": "topic"} instead of bare strings - Specs (ConsumerSpec, ProducerSpec, etc.) read from definition["topics"] Tests updated
2026-04-16 17:19:39 +01:00
qconfig = intf_defs[int_kind]["flow"]
id = str(uuid.uuid4())
dispatcher = export_dispatchers[kind](
backend = self.backend,
ws = ws,
running = running,
queue = qconfig,
consumer = f"{self.prefix}-{id}",
subscriber = f"{self.prefix}-{id}",
)
return dispatcher
async def process_socket(self, ws, running, params):
dispatcher = Mux(self, ws, running)
return dispatcher
async def process_flow_service(self, data, responder, params):
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
# Workspace can come from URL or from request body, defaulting
# to "default". Having it in the URL allows gateway routing to
# be workspace-aware without touching the body.
workspace = params.get("workspace")
if not workspace and isinstance(data, dict):
workspace = data.get("workspace")
if not workspace:
workspace = "default"
flow = params.get("flow")
kind = params.get("kind")
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
return await self.invoke_flow_service(
data, responder, workspace, flow, kind,
)
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
async def invoke_flow_service(
self, data, responder, workspace, flow, kind,
):
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
flow_key = (workspace, flow)
if flow_key not in self.flows:
raise RuntimeError(f"Invalid flow {workspace}/{flow}")
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
key = (workspace, flow, kind)
if key not in self.dispatchers:
async with self.dispatcher_lock:
if key not in self.dispatchers:
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
intf_defs = self.flows[flow_key]["interfaces"]
if kind not in intf_defs:
raise RuntimeError("This kind not supported by flow")
qconfig = intf_defs[kind]
if kind in request_response_dispatchers:
dispatcher = request_response_dispatchers[kind](
backend = self.backend,
request_queue = qconfig["request"],
response_queue = qconfig["response"],
timeout = 120,
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840) Introduces `workspace` as the isolation boundary for config, flows, library, and knowledge data. Removes `user` as a schema-level field throughout the code, API specs, and tests; workspace provides the same separation more cleanly at the trusted flow.workspace layer rather than through client-supplied message fields. Design ------ - IAM tech spec (docs/tech-specs/iam.md) documents current state, proposed auth/access model, and migration direction. - Data ownership model (docs/tech-specs/data-ownership-model.md) captures the workspace/collection/flow hierarchy. Schema + messaging ------------------ - Drop `user` field from AgentRequest/Step, GraphRagQuery, DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest, Sparql/Rows/Structured QueryRequest, ToolServiceRequest. - Keep collection/workspace routing via flow.workspace at the service layer. - Translators updated to not serialise/deserialise user. API specs --------- - OpenAPI schemas and path examples cleaned of user fields. - Websocket async-api messages updated. - Removed the unused parameters/User.yaml. Services + base --------------- - Librarian, collection manager, knowledge, config: all operations scoped by workspace. Config client API takes workspace as first positional arg. - `flow.workspace` set at flow start time by the infrastructure; no longer pass-through from clients. - Tool service drops user-personalisation passthrough. CLI + SDK --------- - tg-init-workspace and workspace-aware import/export. - All tg-* commands drop user args; accept --workspace. - Python API/SDK (flow, socket_client, async_*, explainability, library) drop user kwargs from every method signature. MCP server ---------- - All tool endpoints drop user parameters; socket_manager no longer keyed per user. Flow service ------------ - Closure-based topic cleanup on flow stop: only delete topics whose blueprint template was parameterised AND no remaining live flow (across all workspaces) still resolves to that topic. Three scopes fall out naturally from template analysis: * {id} -> per-flow, deleted on stop * {blueprint} -> per-blueprint, kept while any flow of the same blueprint exists * {workspace} -> per-workspace, kept while any flow in the workspace exists * literal -> global, never deleted (e.g. tg.request.librarian) Fixes a bug where stopping a flow silently destroyed the global librarian exchange, wedging all library operations until manual restart. RabbitMQ backend ---------------- - heartbeat=60, blocked_connection_timeout=300. Catches silently dead connections (broker restart, orphaned channels, network partitions) within ~2 heartbeat windows, so the consumer reconnects and re-binds its queue rather than sitting forever on a zombie connection. Tests ----- - Full test refresh: unit, integration, contract, provenance. - Dropped user-field assertions and constructor kwargs across ~100 test files. - Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
consumer = f"{self.prefix}-{workspace}-{flow}-{kind}-request",
subscriber = f"{self.prefix}-{workspace}-{flow}-{kind}-request",
)
elif kind in sender_dispatchers:
dispatcher = sender_dispatchers[kind](
backend = self.backend,
Flow service lifecycle management (#822) feat: separate flow service from config service with explicit queue lifecycle management The flow service is now an independent service that owns the lifecycle of flow and blueprint queues. System services own their own queues. Consumers never create queues. Flow service separation: - New service at trustgraph-flow/trustgraph/flow/service/ - Uses async ConfigClient (RequestResponse pattern) to talk to config service - Config service stripped of all flow handling Queue lifecycle management: - PubSubBackend protocol gains create_queue, delete_queue, queue_exists, ensure_queue — all async - RabbitMQ: implements via pika with asyncio.to_thread internally - Pulsar: stubs for future admin REST API implementation - Consumer _connect() no longer creates queues (passive=True for named queues) - System services call ensure_queue on startup - Flow service creates queues on flow start, deletes on flow stop - Flow service ensures queues for pre-existing flows on startup Two-phase flow stop: - Phase 1: set flow status to "stopping", delete processor config entries - Phase 2: retry queue deletion, then delete flow record Config restructure: - active-flow config replaced with processor:{name} types - Each processor has its own config type, each flow variant is a key - Flow start/stop use batch put/delete — single config push per operation - FlowProcessor subscribes to its own type only Blueprint format: - Processor entries split into topics and parameters dicts - Flow interfaces use {"flow": "topic"} instead of bare strings - Specs (ConsumerSpec, ProducerSpec, etc.) read from definition["topics"] Tests updated
2026-04-16 17:19:39 +01:00
queue = qconfig["flow"],
)
else:
raise RuntimeError("Invalid kind")
await dispatcher.start()
self.dispatchers[key] = dispatcher
return await self.dispatchers[key].process(data, responder)