aws-python-helper 0.33.0__tar.gz → 0.34.0__tar.gz

{aws_python_helper-0.33.0 → aws_python_helper-0.34.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aws-python-helper
-Version: 0.33.0
+Version: 0.34.0
 Summary: AWS Python Helper Framework
 Author-email: Fabian Claros <neufabiae@gmail.com>
 License-Expression: MIT
@@ -657,7 +657,7 @@ The framework propagates a `Session` object automatically across the entire asyn
 |-------------|-------------------------|
 | **API Gateway** | `constitution-state` header → `session.state` (when `AUTHORIZATION` includes `state`); auth middleware → `session.user` (when includes `user`). Returns `400` if required header is missing |
 | **Standalone Lambda** | `session` dict in the event payload — **required** (must include `state`), raises `ValueError` if missing |
-| **SQS Consumer (single mode)** | Per-record: reads `session` from SNS `MessageAttributes` (JSON), falls back to legacy `body.constitution_state` |
+| **SQS Consumer (single mode)** | Per-record: reads `session` from SNS `MessageAttributes` (Base64-encoded JSON) |
 | **SQS Consumer (batch mode)** | Groups records by `session.state`; calls `process_batch()` once per group with the correct session in context |
 | **Fargate Task** | `SESSION` env var (JSON) — auto-injected by `FargateExecutor` |
 
@@ -665,7 +665,7 @@ The framework propagates a `Session` object automatically across the entire asyn
 
 | Downstream service | Propagation mechanism |
 |--------------------|-----------------------|
-| **SNS Publisher** | Auto-injects the full session as a `session` `MessageAttribute` (JSON string) on every published message |
+| **SNS Publisher** | Auto-injects the full session as a `session` `MessageAttribute` (Base64-encoded JSON) on every published message |
 | **FargateExecutor** | Auto-injects `SESSION` as a JSON env var when launching Fargate containers |
 
 This means that an API call with `constitution-state: connecticut` will automatically carry the full session (state + user) through SNS → SQS → Fargate without any code changes in your consumers or tasks.
@@ -1131,7 +1131,7 @@ environment_variables = {
 By default, consumers process messages one by one (`"single"` mode). Use `"batch"` mode when you need to group or bulk-process messages.
 
 **Constitution-state handling in SQS:**
-- **Single mode**: the framework extracts `constitution-state` from each record automatically (from SNS `MessageAttributes`, then from `body.constitution_state`) and sets it in context before calling `process_record()`. You do not need to extract it yourself.
+- **Single mode**: the framework extracts the session from each record automatically (from SNS `MessageAttributes`, Base64-decoded) and sets it in context before calling `process_record()`. You do not need to extract it yourself.
 - **Batch mode**: the framework groups the incoming records by `constitution-state` and calls `process_batch()` once per group, with the correct state in context for each group. This ensures that state-scoped repositories resolve to the right database even when a batch contains records from different states.
 
 ```python
@@ -1181,7 +1181,7 @@ class OrderConsumer(SQSConsumer):
 
 ### SNS Publisher - Batch Publishing
 
-The `SNSPublisher` automatically injects the current `constitution-state` as a `MessageAttribute` on every published message. SQS consumers built with this framework will then extract it automatically, ensuring the state flows end-to-end through the SNS → SQS chain without any manual code.
+The `SNSPublisher` automatically injects the current session as a Base64-encoded `MessageAttribute` on every published message. Base64 encoding is used to avoid SNS filter policy issues with raw JSON string values in attributes. SQS consumers built with this framework will decode it automatically, ensuring the session flows end-to-end through the SNS → SQS chain without any manual code.
 
 ```python
 topic = TitleIndexedTopic()

{aws_python_helper-0.33.0 → aws_python_helper-0.34.0}/README.md

@@ -638,7 +638,7 @@ The framework propagates a `Session` object automatically across the entire asyn
 |-------------|-------------------------|
 | **API Gateway** | `constitution-state` header → `session.state` (when `AUTHORIZATION` includes `state`); auth middleware → `session.user` (when includes `user`). Returns `400` if required header is missing |
 | **Standalone Lambda** | `session` dict in the event payload — **required** (must include `state`), raises `ValueError` if missing |
-| **SQS Consumer (single mode)** | Per-record: reads `session` from SNS `MessageAttributes` (JSON), falls back to legacy `body.constitution_state` |
+| **SQS Consumer (single mode)** | Per-record: reads `session` from SNS `MessageAttributes` (Base64-encoded JSON) |
 | **SQS Consumer (batch mode)** | Groups records by `session.state`; calls `process_batch()` once per group with the correct session in context |
 | **Fargate Task** | `SESSION` env var (JSON) — auto-injected by `FargateExecutor` |
 
@@ -646,7 +646,7 @@ The framework propagates a `Session` object automatically across the entire asyn
 
 | Downstream service | Propagation mechanism |
 |--------------------|-----------------------|
-| **SNS Publisher** | Auto-injects the full session as a `session` `MessageAttribute` (JSON string) on every published message |
+| **SNS Publisher** | Auto-injects the full session as a `session` `MessageAttribute` (Base64-encoded JSON) on every published message |
 | **FargateExecutor** | Auto-injects `SESSION` as a JSON env var when launching Fargate containers |
 
 This means that an API call with `constitution-state: connecticut` will automatically carry the full session (state + user) through SNS → SQS → Fargate without any code changes in your consumers or tasks.
@@ -1112,7 +1112,7 @@ environment_variables = {
 By default, consumers process messages one by one (`"single"` mode). Use `"batch"` mode when you need to group or bulk-process messages.
 
 **Constitution-state handling in SQS:**
-- **Single mode**: the framework extracts `constitution-state` from each record automatically (from SNS `MessageAttributes`, then from `body.constitution_state`) and sets it in context before calling `process_record()`. You do not need to extract it yourself.
+- **Single mode**: the framework extracts the session from each record automatically (from SNS `MessageAttributes`, Base64-decoded) and sets it in context before calling `process_record()`. You do not need to extract it yourself.
 - **Batch mode**: the framework groups the incoming records by `constitution-state` and calls `process_batch()` once per group, with the correct state in context for each group. This ensures that state-scoped repositories resolve to the right database even when a batch contains records from different states.
 
 ```python
@@ -1162,7 +1162,7 @@ class OrderConsumer(SQSConsumer):
 
 ### SNS Publisher - Batch Publishing
 
-The `SNSPublisher` automatically injects the current `constitution-state` as a `MessageAttribute` on every published message. SQS consumers built with this framework will then extract it automatically, ensuring the state flows end-to-end through the SNS → SQS chain without any manual code.
+The `SNSPublisher` automatically injects the current session as a Base64-encoded `MessageAttribute` on every published message. Base64 encoding is used to avoid SNS filter policy issues with raw JSON string values in attributes. SQS consumers built with this framework will decode it automatically, ensuring the session flows end-to-end through the SNS → SQS chain without any manual code.
 
 ```python
 topic = TitleIndexedTopic()

{aws_python_helper-0.33.0 → aws_python_helper-0.34.0}/aws_python_helper/__init__.py

@@ -29,6 +29,7 @@ from .repository.base import Repository
 
 # Context
 from .context.session import Session, get_session, set_session
+from .context.state_validator import StateValidator, InvalidStateError
 
 
 __all__ = [
@@ -51,5 +52,7 @@ __all__ = [
     'Session',
     'get_session',
     'set_session',
+    'StateValidator',
+    'InvalidStateError',
 ]
 

{aws_python_helper-0.33.0 → aws_python_helper-0.34.0}/aws_python_helper/api/dispatcher.py

@@ -13,6 +13,7 @@ from .exceptions import UnauthorizedError, ForbiddenError, AuthenticationError
 from .auth_middleware import AuthMiddleware
 from .auth_validators import TokenValidator
 from ..context.session import get_session
+from ..context.state_validator import StateValidator, InvalidStateError
 
 logger = logging.getLogger(__name__)
 
@@ -96,6 +97,17 @@ class Dispatcher:
                         },
                         'headers': {}
                     }
+                try:
+                    await StateValidator.validate(state)
+                except InvalidStateError as e:
+                    return {
+                        'code': 403,
+                        'body': {
+                            'error': 'Forbidden',
+                            'message': str(e)
+                        },
+                        'headers': {}
+                    }
                 session = get_session()
                 session.state = state
 

aws_python_helper-0.34.0/aws_python_helper/context/__init__.py

@@ -0,0 +1,4 @@
+from .session import Session, get_session, set_session
+from .state_validator import StateValidator, InvalidStateError
+
+__all__ = ['Session', 'get_session', 'set_session', 'StateValidator', 'InvalidStateError']

aws_python_helper-0.34.0/aws_python_helper/context/state_validator.py

@@ -0,0 +1,54 @@
+import time
+from typing import Dict, Tuple
+
+
+class InvalidStateError(Exception):
+    """Raised when a state is not found or not active in the database."""
+    pass
+
+
+class StateValidator:
+    """
+    Validates that a state exists and is active in the core.states collection.
+
+    Uses an in-memory cache with a 5-minute TTL to avoid querying the database
+    on every request. Only valid (active) states are cached — invalid states
+    always hit the database so that recently activated states are picked up
+    immediately.
+    """
+
+    _cache: Dict[str, float] = {}  # state -> timestamp when cached
+    _CACHE_TTL: int = 300  # 5 minutes
+
+    @classmethod
+    async def validate(cls, state: str):
+        """
+        Validate that state exists in core.states with is_active=True.
+
+        Args:
+            state: The state name to validate
+
+        Raises:
+            InvalidStateError: If the state doesn't exist or is not active
+        """
+        now = time.time()
+
+        if state in cls._cache:
+            if now - cls._cache[state] < cls._CACHE_TTL:
+                return
+            del cls._cache[state]
+
+        from ..database.mongo_manager import MongoManager
+        db = MongoManager.get_database('core')
+        doc = await db.states.find_one({"name": state, "is_active": True})
+
+        if doc:
+            cls._cache[state] = now
+            return
+
+        raise InvalidStateError(f"State '{state}' is not valid or not active")
+
+    @classmethod
+    def clear_cache(cls):
+        """Clear the state validation cache."""
+        cls._cache.clear()

{aws_python_helper-0.33.0 → aws_python_helper-0.34.0}/aws_python_helper/lambda_standalone/base.py

@@ -7,6 +7,7 @@ from typing import Dict, Any
 import logging
 
 from ..context.session import Session, set_session, get_session
+from ..context.state_validator import StateValidator
 from ..database.mongo_manager import MongoManager
 from ..database.database_proxy import DatabaseProxy
 from ..database.external_mongo_manager import ExternalMongoManager
@@ -169,6 +170,7 @@ class Lambda(ABC):
             if not session_data or not session_data.get('state'):
                 raise ValueError("'session' with 'state' is required in the event")
             session = Session.from_dict(session_data)
+            await StateValidator.validate(session.state)
             set_session(session)
 
             # Step 1: Validate

{aws_python_helper-0.33.0 → aws_python_helper-0.34.0}/aws_python_helper/sns/publisher.py

@@ -7,6 +7,7 @@ without complex validations, only basic JSON serialization.
 
 import json
 import logging
+import base64
 from typing import Dict, Any, List, Optional, Union
 import boto3
 from abc import ABC
@@ -185,14 +186,27 @@ class SNSPublisher(ABC):
 
         params['PublishBatchRequestEntries'] = message_to_publish
 
-        result = self.sns_client.publish_batch(**params)
+        try:
+            result = self.sns_client.publish_batch(**params)
+        except Exception as e:
+            self.logger.error(f"Error calling publish_batch to {self.topic_arn}: {e}")
+            raise
 
-        for message in result['Successful']:
+        for message in result.get('Successful', []):
             message_ids_success.append(message['MessageId'])
 
-        for message in result['Failed']:
+        for message in result.get('Failed', []):
+            self.logger.error(
+                f"Failed to publish message {message.get('Id')} to {self.topic_arn}: "
+                f"Code={message.get('Code')}, Message={message.get('Message')}"
+            )
             message_ids_failed.append(None)
-        
+
+        self.logger.info(
+            f"Batch publish to {self.topic_arn}: "
+            f"{len(message_ids_success)} successful, {len(message_ids_failed)} failed"
+        )
+
         return message_ids_success, message_ids_failed
     
     def _serialize_message(self, content: Dict[str, Any]) -> str:
@@ -258,14 +272,16 @@ class SNSPublisher(ABC):
         attributes = message.get("attributes")
         subject = message.get("subject")
 
-        # Auto-inject session as a message attribute
+        # Auto-inject session as a message attribute (Base64-encoded to avoid
+        # SNS filter policy issues with JSON string values)
         session = get_session()
         session_dict = session.to_dict()
         if session_dict:
             if not attributes:
                 attributes = {}
             if 'session' not in attributes:
-                attributes['session'] = self._serialize_message(session_dict)
+                session_json = self._serialize_message(session_dict)
+                attributes['session'] = base64.b64encode(session_json.encode('utf-8')).decode('utf-8')
 
         params = {
             'Id': str(index),

{aws_python_helper-0.33.0 → aws_python_helper-0.34.0}/aws_python_helper/sqs/consumer_base.py

@@ -3,6 +3,7 @@ SQS Consumer Base - Base class for all SQS consumers
 """
 
 import os
+import base64
 from abc import ABC, abstractmethod
 from typing import Dict, Any, List, Optional
 import logging
@@ -277,9 +278,8 @@ class SQSConsumer(ABC):
         """
         Extract Session from an SQS record without setting it in the context.
 
-        Looks first in SNS MessageAttributes for 'session' (injected automatically
-        by SNSPublisher), then falls back to the parsed message body content for
-        legacy 'constitution_state' field.
+        Looks in SNS MessageAttributes for 'session' (Base64-encoded, injected
+        automatically by SNSPublisher).
 
         Args:
             record: SQS record
@@ -287,7 +287,6 @@ class SQSConsumer(ABC):
         Returns:
             Session instance if found, None otherwise
         """
-        # Primary: SNS MessageAttributes (auto-injected by SNSPublisher)
         raw_body = record.get('body', '{}')
         if isinstance(raw_body, str):
             try:
@@ -298,20 +297,14 @@ class SQSConsumer(ABC):
             raw_body_json = raw_body or {}
 
         if isinstance(raw_body_json, dict) and 'MessageAttributes' in raw_body_json:
-            # Try full session attribute first
             session_attr = raw_body_json['MessageAttributes'].get('session', {})
             session_str = session_attr.get('Value')
             if session_str:
                 try:
-                    return Session.from_dict(json.loads(session_str))
-                except (json.JSONDecodeError, TypeError):
+                    decoded = base64.b64decode(session_str).decode('utf-8')
+                    return Session.from_dict(json.loads(decoded))
+                except (json.JSONDecodeError, TypeError, Exception) as e:
                     self.logger.warning(f"Could not parse session attribute: {session_str}")
-
-        # Fallback: parsed body content (for direct SQS messages without SNS wrapping)
-        body = self.extract_content_message(record)
-        state = body.get("constitution_state") or body.get("content", {}).get("constitution_state")
-        if state:
-            return Session(state=state)
         return None
 
     def _extract_and_set_session(self, record: Dict[str, Any]) -> Optional[Session]:

{aws_python_helper-0.33.0 → aws_python_helper-0.34.0}/aws_python_helper.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aws-python-helper
-Version: 0.33.0
+Version: 0.34.0
 Summary: AWS Python Helper Framework
 Author-email: Fabian Claros <neufabiae@gmail.com>
 License-Expression: MIT
@@ -657,7 +657,7 @@ The framework propagates a `Session` object automatically across the entire asyn
 |-------------|-------------------------|
 | **API Gateway** | `constitution-state` header → `session.state` (when `AUTHORIZATION` includes `state`); auth middleware → `session.user` (when includes `user`). Returns `400` if required header is missing |
 | **Standalone Lambda** | `session` dict in the event payload — **required** (must include `state`), raises `ValueError` if missing |
-| **SQS Consumer (single mode)** | Per-record: reads `session` from SNS `MessageAttributes` (JSON), falls back to legacy `body.constitution_state` |
+| **SQS Consumer (single mode)** | Per-record: reads `session` from SNS `MessageAttributes` (Base64-encoded JSON) |
 | **SQS Consumer (batch mode)** | Groups records by `session.state`; calls `process_batch()` once per group with the correct session in context |
 | **Fargate Task** | `SESSION` env var (JSON) — auto-injected by `FargateExecutor` |
 
@@ -665,7 +665,7 @@ The framework propagates a `Session` object automatically across the entire asyn
 
 | Downstream service | Propagation mechanism |
 |--------------------|-----------------------|
-| **SNS Publisher** | Auto-injects the full session as a `session` `MessageAttribute` (JSON string) on every published message |
+| **SNS Publisher** | Auto-injects the full session as a `session` `MessageAttribute` (Base64-encoded JSON) on every published message |
 | **FargateExecutor** | Auto-injects `SESSION` as a JSON env var when launching Fargate containers |
 
 This means that an API call with `constitution-state: connecticut` will automatically carry the full session (state + user) through SNS → SQS → Fargate without any code changes in your consumers or tasks.
@@ -1131,7 +1131,7 @@ environment_variables = {
 By default, consumers process messages one by one (`"single"` mode). Use `"batch"` mode when you need to group or bulk-process messages.
 
 **Constitution-state handling in SQS:**
-- **Single mode**: the framework extracts `constitution-state` from each record automatically (from SNS `MessageAttributes`, then from `body.constitution_state`) and sets it in context before calling `process_record()`. You do not need to extract it yourself.
+- **Single mode**: the framework extracts the session from each record automatically (from SNS `MessageAttributes`, Base64-decoded) and sets it in context before calling `process_record()`. You do not need to extract it yourself.
 - **Batch mode**: the framework groups the incoming records by `constitution-state` and calls `process_batch()` once per group, with the correct state in context for each group. This ensures that state-scoped repositories resolve to the right database even when a batch contains records from different states.
 
 ```python
@@ -1181,7 +1181,7 @@ class OrderConsumer(SQSConsumer):
 
 ### SNS Publisher - Batch Publishing
 
-The `SNSPublisher` automatically injects the current `constitution-state` as a `MessageAttribute` on every published message. SQS consumers built with this framework will then extract it automatically, ensuring the state flows end-to-end through the SNS → SQS chain without any manual code.
+The `SNSPublisher` automatically injects the current session as a Base64-encoded `MessageAttribute` on every published message. Base64 encoding is used to avoid SNS filter policy issues with raw JSON string values in attributes. SQS consumers built with this framework will decode it automatically, ensuring the session flows end-to-end through the SNS → SQS chain without any manual code.
 
 ```python
 topic = TitleIndexedTopic()

{aws_python_helper-0.33.0 → aws_python_helper-0.34.0}/aws_python_helper.egg-info/SOURCES.txt

@@ -16,6 +16,7 @@ aws_python_helper/api/fetcher.py
 aws_python_helper/api/handler.py
 aws_python_helper/context/__init__.py
 aws_python_helper/context/session.py
+aws_python_helper/context/state_validator.py
 aws_python_helper/database/__init__.py
 aws_python_helper/database/database_proxy.py
 aws_python_helper/database/external_database_proxy.py

{aws_python_helper-0.33.0 → aws_python_helper-0.34.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "aws-python-helper"
-version = "0.33.0"
+version = "0.34.0"
 description = "AWS Python Helper Framework"
 readme = "README.md"
 requires-python = ">=3.9"

aws_python_helper-0.33.0/aws_python_helper/context/__init__.py

@@ -1,3 +0,0 @@
-from .session import Session, get_session, set_session
-
-__all__ = ['Session', 'get_session', 'set_session']