hindsight-api 0.4.6__tar.gz → 0.4.7__tar.gz

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hindsight-api
-Version: 0.4.6
+Version: 0.4.7
 Summary: Hindsight: Agent Memory That Works Like Human Memory
 Requires-Python: >=3.11
 Requires-Dist: aiohttp>=3.13.3

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/hindsight_api/__init__.py

@@ -46,4 +46,4 @@ __all__ = [
     "RemoteTEICrossEncoder",
     "LLMConfig",
 ]
-__version__ = "0.4.6"
+__version__ = "0.4.7"

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/hindsight_api/api/http.py

@@ -1398,13 +1398,18 @@ def create_app(
 
         # Start worker poller if enabled (standalone mode)
         if config.worker_enabled and memory._pool is not None:
+            from ..config import DEFAULT_DATABASE_SCHEMA
+
             worker_id = config.worker_id or socket.gethostname()
+            # Convert default schema to None for SQL compatibility (no schema prefix)
+            schema = None if config.database_schema == DEFAULT_DATABASE_SCHEMA else config.database_schema
             poller = WorkerPoller(
                 pool=memory._pool,
                 worker_id=worker_id,
                 executor=memory.execute_task,
                 poll_interval_ms=config.worker_poll_interval_ms,
                 max_retries=config.worker_max_retries,
+                schema=schema,
                 tenant_extension=getattr(memory, "_tenant_extension", None),
                 max_slots=config.worker_max_slots,
                 consolidation_max_slots=config.worker_consolidation_max_slots,
@@ -2285,6 +2290,23 @@ def _register_routes(app: FastAPI):
     ):
         """Get a mental model by ID."""
         try:
+            # Pre-operation validation hook
+            validator = app.state.memory._operation_validator
+            if validator:
+                from hindsight_api.extensions.operation_validator import MentalModelGetContext
+
+                ctx = MentalModelGetContext(
+                    bank_id=bank_id,
+                    mental_model_id=mental_model_id,
+                    request_context=request_context,
+                )
+                validation = await validator.validate_mental_model_get(ctx)
+                if not validation.allowed:
+                    raise OperationValidationError(
+                        validation.reason or "Operation not allowed",
+                        status_code=validation.status_code,
+                    )
+
             mental_model = await app.state.memory.get_mental_model(
                 bank_id=bank_id,
                 mental_model_id=mental_model_id,
@@ -2292,9 +2314,31 @@ def _register_routes(app: FastAPI):
             )
             if mental_model is None:
                 raise HTTPException(status_code=404, detail=f"Mental model '{mental_model_id}' not found")
+
+            # Post-operation hook
+            if validator:
+                from hindsight_api.extensions.operation_validator import MentalModelGetResult
+
+                content = mental_model.get("content", "")
+                output_tokens = len(content) // 4 if content else 0
+
+                result_ctx = MentalModelGetResult(
+                    bank_id=bank_id,
+                    mental_model_id=mental_model_id,
+                    request_context=request_context,
+                    output_tokens=output_tokens,
+                    success=True,
+                )
+                try:
+                    await validator.on_mental_model_get_complete(result_ctx)
+                except Exception as hook_err:
+                    logger.warning(f"Post-mental-model-get hook error (non-fatal): {hook_err}")
+
             return MentalModelResponse(**mental_model)
         except (AuthenticationError, HTTPException):
             raise
+        except OperationValidationError as e:
+            raise HTTPException(status_code=e.status_code, detail=e.reason)
         except Exception as e:
             import traceback
 

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/hindsight_api/daemon.py

@@ -15,7 +15,7 @@ from pathlib import Path
 logger = logging.getLogger(__name__)
 
 # Default daemon configuration
-DEFAULT_DAEMON_PORT = 8889
+DEFAULT_DAEMON_PORT = 8888
 DEFAULT_IDLE_TIMEOUT = 0  # 0 = no auto-exit (hindsight-embed passes its own timeout)
 LOCKFILE_PATH = Path.home() / ".hindsight" / "daemon.lock"
 DAEMON_LOG_PATH = Path.home() / ".hindsight" / "daemon.log"

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/hindsight_api/engine/memory_engine.py

@@ -597,7 +597,13 @@ class MemoryEngine(MemoryEngineInterface):
 
         from hindsight_api.models import RequestContext
 
-        internal_context = RequestContext(internal=True)
+        # Restore tenant_id/api_key_id from task payload so extensions can
+        # attribute the mental_model_refresh operation to the correct org.
+        internal_context = RequestContext(
+            internal=True,
+            tenant_id=task_dict.get("_tenant_id"),
+            api_key_id=task_dict.get("_api_key_id"),
+        )
 
         # Get the current mental model to get source_query
         mental_model = await self.get_mental_model(bank_id, mental_model_id, request_context=internal_context)
@@ -641,6 +647,42 @@ class MemoryEngine(MemoryEngineInterface):
             request_context=internal_context,
         )
 
+        # Call post-operation hook if validator is configured
+        if self._operation_validator:
+            from hindsight_api.extensions.operation_validator import MentalModelRefreshResult
+
+            # Count facts and mental models from based_on
+            facts_used = 0
+            mental_models_used = 0
+            if reflect_result.based_on:
+                for fact_type, facts in reflect_result.based_on.items():
+                    if facts:
+                        if fact_type == "mental_models":
+                            mental_models_used += len(facts)
+                        else:
+                            facts_used += len(facts)
+
+            # Estimate tokens
+            query_tokens = len(source_query) // 4 if source_query else 0
+            output_tokens = len(generated_content) // 4 if generated_content else 0
+            context_tokens = 0  # refresh doesn't use additional context
+
+            result_ctx = MentalModelRefreshResult(
+                bank_id=bank_id,
+                mental_model_id=mental_model_id,
+                request_context=internal_context,
+                query_tokens=query_tokens,
+                output_tokens=output_tokens,
+                context_tokens=context_tokens,
+                facts_used=facts_used,
+                mental_models_used=mental_models_used,
+                success=True,
+            )
+            try:
+                await self._operation_validator.on_mental_model_refresh_complete(result_ctx)
+            except Exception as hook_err:
+                logger.warning(f"Post-mental-model-refresh hook error (non-fatal): {hook_err}")
+
         logger.info(f"[REFRESH_MENTAL_MODEL_TASK] Completed for bank_id={bank_id}, mental_model_id={mental_model_id}")
 
     async def execute_task(self, task_dict: dict[str, Any]):
@@ -5482,13 +5524,21 @@ class MemoryEngine(MemoryEngineInterface):
         if not mental_model:
             raise ValueError(f"Mental model {mental_model_id} not found in bank {bank_id}")
 
+        # Pass tenant_id and api_key_id through task payload so the worker
+        # can provide request context to extension hooks.
+        task_payload: dict[str, Any] = {
+            "mental_model_id": mental_model_id,
+        }
+        if request_context.tenant_id:
+            task_payload["_tenant_id"] = request_context.tenant_id
+        if request_context.api_key_id:
+            task_payload["_api_key_id"] = request_context.api_key_id
+
         return await self._submit_async_operation(
             bank_id=bank_id,
             operation_type="refresh_mental_model",
             task_type="refresh_mental_model",
-            task_payload={
-                "mental_model_id": mental_model_id,
-            },
+            task_payload=task_payload,
             result_metadata={"mental_model_id": mental_model_id, "name": mental_model["name"]},
             dedupe_by_bank=False,
         )

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/hindsight_api/engine/retain/fact_extraction.py

@@ -57,21 +57,25 @@ def _infer_temporal_date(fact_text: str, event_date: datetime) -> str | None:
     return None
 
 
-def _sanitize_text(text: str) -> str:
+def _sanitize_text(text: str | None) -> str | None:
     """
-    Sanitize text by removing invalid Unicode surrogate characters.
+    Sanitize text by removing characters that break downstream systems.
 
-    Surrogate characters (U+D800 to U+DFFF) are used in UTF-16 encoding
-    but cannot be encoded in UTF-8. They can appear in Python strings
-    from improperly decoded data (e.g., from JavaScript or broken files).
+    Removes:
+    - Null bytes (\\x00): Invalid in PostgreSQL UTF-8 encoding
+    - Unicode surrogates (U+D800-U+DFFF): Invalid in UTF-8, break LLM APIs
 
-    This function removes unpaired surrogates to prevent UnicodeEncodeError
-    when the text is sent to the LLM API.
+    Surrogate characters are used in UTF-16 encoding but cannot be encoded
+    in UTF-8. They can appear in Python strings from improperly decoded data
+    (e.g., from JavaScript or broken files). Null bytes commonly appear in
+    OCR output, PDF extraction, or copy-paste from binary sources.
     """
+    if text is None:
+        return None
     if not text:
         return text
-    # Remove surrogate characters (U+D800 to U+DFFF) using regex
-    # These are invalid in UTF-8 and cause encoding errors
+    # Remove null bytes and surrogate characters
+    text = text.replace("\x00", "")
     return re.sub(r"[\ud800-\udfff]", "", text)
 
 

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/hindsight_api/engine/retain/fact_storage.py

@@ -8,6 +8,7 @@ import json
 import logging
 
 from ..memory_engine import fq_table
+from .fact_extraction import _sanitize_text
 from .types import ProcessedFact
 
 logger = logging.getLogger(__name__)
@@ -47,7 +48,7 @@ async def insert_facts_batch(
     tags_list = []
 
     for fact in facts:
-        fact_texts.append(fact.fact_text)
+        fact_texts.append(_sanitize_text(fact.fact_text))
         # Convert embedding to string for asyncpg vector type
         embeddings.append(str(fact.embedding))
         # event_date: Use occurred_start if available, otherwise use mentioned_at
@@ -56,7 +57,7 @@ async def insert_facts_batch(
         occurred_starts.append(fact.occurred_start)
         occurred_ends.append(fact.occurred_end)
         mentioned_ats.append(fact.mentioned_at)
-        contexts.append(fact.context)
+        contexts.append(_sanitize_text(fact.context))
         fact_types.append(fact.fact_type)
         # confidence_score is only for opinion facts
         confidence_scores.append(1.0 if fact.fact_type == "opinion" else None)
@@ -157,7 +158,8 @@ async def handle_document_tracking(
     """
     import hashlib
 
-    # Calculate content hash
+    # Sanitize and calculate content hash
+    combined_content = _sanitize_text(combined_content) or ""
     content_hash = hashlib.sha256(combined_content.encode()).hexdigest()
 
     # Always delete old document first if it exists (cascades to units and links)

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/hindsight_api/extensions/__init__.py

@@ -24,6 +24,10 @@ from hindsight_api.extensions.operation_validator import (
     # Consolidation operation
     ConsolidateContext,
     ConsolidateResult,
+    # Mental Model operations
+    MentalModelGetContext,
+    MentalModelGetResult,
+    MentalModelRefreshResult,
     # Core operations
     OperationValidationError,
     OperationValidatorExtension,
@@ -65,6 +69,10 @@ __all__ = [
     # Operation Validator - Consolidation
     "ConsolidateContext",
     "ConsolidateResult",
+    # Operation Validator - Mental Model
+    "MentalModelGetContext",
+    "MentalModelGetResult",
+    "MentalModelRefreshResult",
     # Tenant/Auth
     "ApiKeyTenantExtension",
     "AuthenticationError",

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/hindsight_api/extensions/operation_validator.py

@@ -196,6 +196,48 @@ class ConsolidateResult:
     error: str | None = None
 
 
+# =============================================================================
+# Mental Model Contexts
+# =============================================================================
+
+
+@dataclass
+class MentalModelGetContext:
+    """Context for a mental model GET operation validation (pre-operation)."""
+
+    bank_id: str
+    mental_model_id: str
+    request_context: "RequestContext"
+
+
+@dataclass
+class MentalModelGetResult:
+    """Result context for post-mental-model-GET hook."""
+
+    bank_id: str
+    mental_model_id: str
+    request_context: "RequestContext"
+    output_tokens: int  # tokens in the returned content
+    success: bool = True
+    error: str | None = None
+
+
+@dataclass
+class MentalModelRefreshResult:
+    """Result context for post-mental-model-refresh hook."""
+
+    bank_id: str
+    mental_model_id: str
+    request_context: "RequestContext"
+    query_tokens: int  # tokens in source_query
+    output_tokens: int  # tokens in generated content
+    context_tokens: int  # tokens in context (if any)
+    facts_used: int  # facts referenced in based_on
+    mental_models_used: int  # mental models referenced in based_on
+    success: bool = True
+    error: str | None = None
+
+
 class OperationValidatorExtension(Extension, ABC):
     """
     Validates and hooks into retain/recall/reflect/consolidate operations.
@@ -402,3 +444,64 @@ class OperationValidatorExtension(Extension, ABC):
                 - error: Error message (if failed)
         """
         pass
+
+    # =========================================================================
+    # Mental Model - Pre-operation validation hook (optional - override to implement)
+    # =========================================================================
+
+    async def validate_mental_model_get(self, ctx: MentalModelGetContext) -> ValidationResult:
+        """
+        Validate a mental model GET operation before execution.
+
+        Override to implement custom validation logic for mental model retrieval.
+
+        Args:
+            ctx: Context containing:
+                - bank_id: Bank identifier
+                - mental_model_id: Mental model identifier
+                - request_context: Request context with auth info
+
+        Returns:
+            ValidationResult indicating whether the operation is allowed.
+        """
+        return ValidationResult.accept()
+
+    # =========================================================================
+    # Mental Model - Post-operation hooks (optional - override to implement)
+    # =========================================================================
+
+    async def on_mental_model_get_complete(self, result: MentalModelGetResult) -> None:
+        """
+        Called after a mental model GET operation completes (success or failure).
+
+        Override to implement post-operation logic such as tracking or audit logging.
+
+        Args:
+            result: Result context containing:
+                - bank_id: Bank identifier
+                - mental_model_id: Mental model identifier
+                - output_tokens: Token count of the returned content
+                - success: Whether the operation succeeded
+                - error: Error message (if failed)
+        """
+        pass
+
+    async def on_mental_model_refresh_complete(self, result: MentalModelRefreshResult) -> None:
+        """
+        Called after a mental model refresh operation completes (success or failure).
+
+        Override to implement post-operation logic such as tracking or audit logging.
+
+        Args:
+            result: Result context containing:
+                - bank_id: Bank identifier
+                - mental_model_id: Mental model identifier
+                - query_tokens: Tokens in source_query
+                - output_tokens: Tokens in generated content
+                - context_tokens: Tokens in context
+                - facts_used: Number of facts referenced
+                - mental_models_used: Number of mental models referenced
+                - success: Whether the operation succeeded
+                - error: Error message (if failed)
+        """
+        pass

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/hindsight_api/worker/main.py

@@ -200,15 +200,20 @@ def main():
         if tenant_extension:
             print("Tenant extension loaded - schemas will be discovered dynamically on each poll")
         else:
-            print("No tenant extension configured, using public schema only")
+            print(f"No tenant extension configured, using schema: {config.database_schema}")
 
         # Create a single poller that handles all schemas dynamically
+        # Convert default schema to None for SQL compatibility (no schema prefix)
+        from hindsight_api.config import DEFAULT_DATABASE_SCHEMA
+
+        schema = None if config.database_schema == DEFAULT_DATABASE_SCHEMA else config.database_schema
         poller = WorkerPoller(
             pool=memory._pool,
             worker_id=args.worker_id,
             executor=memory.execute_task,
             poll_interval_ms=args.poll_interval,
             max_retries=args.max_retries,
+            schema=schema,
             tenant_extension=tenant_extension,
             max_slots=config.worker_max_slots,
             consolidation_max_slots=config.worker_consolidation_max_slots,

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/hindsight_api/worker/poller.py

@@ -99,11 +99,13 @@ class WorkerPoller:
         self._in_flight_by_type: dict[str, int] = {}
 
     async def _get_schemas(self) -> list[str | None]:
-        """Get list of schemas to poll. Returns [None] for public schema."""
+        """Get list of schemas to poll. Returns [None] for default schema (no prefix)."""
         if self._tenant_extension is not None:
+            from ..config import DEFAULT_DATABASE_SCHEMA
+
             tenants = await self._tenant_extension.list_tenants()
-            # Convert "public" to None for SQL compatibility, keep others as-is
-            return [t.schema if t.schema != "public" else None for t in tenants]
+            # Convert default schema to None for SQL compatibility (no prefix), keep others as-is
+            return [t.schema if t.schema != DEFAULT_DATABASE_SCHEMA else None for t in tenants]
         # Single schema mode
         return [self._schema]
 
@@ -194,7 +196,9 @@ class WorkerPoller:
         try:
             return await self._claim_batch_for_schema_inner(schema, limit, consolidation_limit)
         except Exception as e:
-            logger.warning(f"Worker {self._worker_id} failed to claim tasks for schema {schema or 'public'}: {e}")
+            # Format schema for logging: custom schemas in quotes, None as-is
+            schema_display = f'"{schema}"' if schema else str(schema)
+            logger.warning(f"Worker {self._worker_id} failed to claim tasks for schema {schema_display}: {e}")
             return []
 
     async def _claim_batch_for_schema_inner(
@@ -418,7 +422,9 @@ class WorkerPoller:
                 count = int(result.split()[-1]) if result else 0
                 total_count += count
             except Exception as e:
-                logger.warning(f"Worker {self._worker_id} failed to recover tasks for schema {schema or 'public'}: {e}")
+                # Format schema for logging: custom schemas in quotes, None as-is
+                schema_display = f'"{schema}"' if schema else str(schema)
+                logger.warning(f"Worker {self._worker_id} failed to recover tasks for schema {schema_display}: {e}")
 
         if total_count > 0:
             logger.info(f"Worker {self._worker_id} recovered {total_count} stale tasks from previous run")
@@ -457,7 +463,8 @@ class WorkerPoller:
                             consolidation_count += 1
 
                     types_str = ", ".join(f"{k}:{v}" for k, v in task_types.items())
-                    schemas_str = ", ".join(s or "public" for s in schemas_seen)
+                    # Display None as "default" in logs
+                    schemas_str = ", ".join(s if s else "default" for s in schemas_seen)
                     logger.info(
                         f"Worker {self._worker_id} claimed {len(tasks)} tasks "
                         f"({consolidation_count} consolidation): {types_str} (schemas: {schemas_str})"
@@ -591,7 +598,8 @@ class WorkerPoller:
                     other_workers.append(f"{wid}:{cnt}")
             others_str = ", ".join(other_workers) if other_workers else "none"
 
-            schemas_str = ", ".join(s or "public" for s in schemas)
+            # Display None as "default" in logs
+            schemas_str = ", ".join(s if s else "default" for s in schemas)
             logger.info(
                 f"[WORKER_STATS] worker={self._worker_id} "
                 f"slots={in_flight}/{self._max_slots} (consolidation={consolidation_count}/{self._consolidation_max_slots}) | "

{hindsight_api-0.4.6 → hindsight_api-0.4.7}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "hindsight-api"
-version = "0.4.6"
+version = "0.4.7"
 description = "Hindsight: Agent Memory That Works Like Human Memory"
 readme = "README.md"
 requires-python = ">=3.11"