rasa-pro 3.13.7__py3-none-any.whl → 3.14.0a1__py3-none-any.whl

rasa/shared/core/flows/flow_step.py

@@ -52,7 +52,13 @@ def step_from_json(flow_id: Text, data: Dict[Text, Any]) -> FlowStep:
         return SetSlotsFlowStep.from_json(flow_id, data)
     if "noop" in data:
         return NoOperationFlowStep.from_json(flow_id, data)
-    raise RasaException(f"Failed to parse step from json. Unknown type for {data}.")
+
+    required_properties = ["action", "collect", "link", "call", "set_slots", "noop"]
+    raise RasaException(
+        f"Failed to parse step from json. Unknown type for {data}. "
+        f"At least one of the following properties is required: "
+        f"{', '.join(required_properties)}"
+    )
 
 
 @dataclass

rasa/shared/core/flows/steps/call.py

@@ -50,7 +50,14 @@ class CallFlowStep(FlowStep):
 
         if should_resolve_calls:
             if not self.called_flow_reference:
-                raise ValueError("Call flow reference not set.")
+                raise ValueError(
+                    f"Flow step '{self.id}' in flow '{self.flow_id}' is trying "
+                    f"to call flow '{self.call}', but the flow reference could "
+                    f"not be resolved. Please ensure that:\n"
+                    f"1. A flow named '{self.call}' is defined in your domain\n"
+                    f"2. The flow name is spelled correctly (case-sensitive)\n"
+                    f"3. The called flow is properly formatted with valid YAML syntax"
+                )
 
             yield from self.called_flow_reference.steps_with_calls_resolved
 

rasa/shared/core/flows/yaml_flows_io.py

@@ -262,12 +262,9 @@ class YamlFlowsWriter:
         Returns:
             The dumped YAML.
         """
-        dump = {}
-        for flow in flows:
-            dumped_flow = get_flow_as_json(flow, should_clean_json)
-            del dumped_flow["id"]
-            dump[flow.id] = dumped_flow
-        return dump_obj_as_yaml_to_string({KEY_FLOWS: dump})
+        return dump_obj_as_yaml_to_string(
+            {KEY_FLOWS: get_flows_as_json(flows, should_clean_json)}
+        )
 
     @staticmethod
     def dump(
@@ -424,9 +421,20 @@ def process_yaml_content(yaml_content: Dict[str, Any]) -> Dict[str, Any]:
     return yaml_content
 
 
+def get_flows_as_json(
+    flows: FlowsList, should_clean_json: bool = False
+) -> Dict[str, Any]:
+    """Get the flows as a JSON dictionary."""
+    dump = {}
+    for flow in flows:
+        dumped_flow = get_flow_as_json(flow, should_clean_json)
+        del dumped_flow["id"]
+        dump[flow.id] = dumped_flow
+    return dump
+
+
 def get_flow_as_json(flow: Flow, should_clean_json: bool = False) -> Dict[str, Any]:
-    """
-    Clean the Flow JSON by removing default values and empty fields.
+    """Clean the Flow JSON by removing default values and empty fields.
 
     Args:
         flow: The Flow object to clean.

rasa/shared/core/slots.py

@@ -273,10 +273,14 @@ class Slot(ABC):
         try:
             return rasa.shared.utils.common.class_from_module_path(type_name)
         except (ImportError, AttributeError):
+            known_types = [
+                cls.type_name for cls in rasa.shared.utils.common.all_subclasses(Slot)
+            ]
             raise InvalidSlotTypeException(
                 f"Failed to find slot type, '{type_name}' is neither a known type nor "
                 f"user-defined. If you are creating your own slot type, make "
                 f"sure its module path is correct. "
+                f"Known types: {', '.join(known_types)} "
                 f"You can find all build in types at {DOCS_URL_SLOTS}"
             )
 

rasa/shared/importers/importer.py

@@ -207,6 +207,12 @@ class TrainingDataImporter(ABC):
                 )
             ]
 
+        return TrainingDataImporter.wrap_in_builtins(importers)
+
+    @staticmethod
+    def wrap_in_builtins(
+        importers: List["TrainingDataImporter"],
+    ) -> "TrainingDataImporter":
         return LanguageImporter(
             E2EImporter(
                 FlowSyncImporter(ResponsesSyncImporter(CombinedDataImporter(importers)))

rasa/shared/importers/utils.py

@@ -1,9 +1,37 @@
-from typing import Iterable, List, Optional, Text
+from typing import Any, Dict, Iterable, List, Optional, Text
+
+from pydantic import BaseModel, Field
 
 from rasa.shared.core.domain import Domain
 from rasa.shared.core.flows import FlowsList
+from rasa.shared.core.flows.yaml_flows_io import KEY_FLOWS, get_flows_as_json
 from rasa.shared.core.training_data.structures import StoryGraph
+from rasa.shared.importers.importer import TrainingDataImporter
+from rasa.shared.nlu.training_data.formats.rasa_yaml import RasaYAMLWriter
 from rasa.shared.nlu.training_data.training_data import TrainingData
+from rasa.utils.json_utils import extract_values
+
+
+class CALMUserData(BaseModel):
+    """All pieces that will be uploaded to Rasa Studio."""
+
+    flows: Dict[str, Any] = Field(default_factory=dict)
+    domain: Dict[str, Any] = Field(default_factory=dict)
+    config: Dict[str, Any] = Field(default_factory=dict)
+    endpoints: Dict[str, Any] = Field(default_factory=dict)
+    nlu: Dict[str, Any] = Field(default_factory=dict)
+
+
+DOMAIN_KEYS = [
+    "version",
+    "actions",
+    "responses",
+    "slots",
+    "intents",
+    "entities",
+    "forms",
+    "session_config",
+]
 
 
 def training_data_from_paths(paths: Iterable[Text], language: Text) -> TrainingData:
@@ -34,3 +62,51 @@ def flows_from_paths(files: List[Text]) -> FlowsList:
         )
     flows.validate()
     return flows
+
+
+def extract_calm_import_parts_from_importer(
+    importer: TrainingDataImporter,
+    config: Optional[Dict[str, Any]] = None,
+    endpoints: Optional[Dict[str, Any]] = None,
+) -> CALMUserData:
+    """Extracts CALMUserData from a TrainingDataImporter.
+
+    Args:
+        importer: The training data importer
+        data_paths: The path(s) to the training data for flows
+        config: Optional config dict, if not provided will use importer.get_config()
+        endpoints: Optional endpoints dict, defaults to empty dict
+
+    Returns:
+        CALMUserData containing flows, domain, config, endpoints, and nlu data
+    """
+    # Extract config
+    if config is None:
+        config = importer.get_config()
+
+    # Extract domain
+    domain_from_files = importer.get_user_domain().as_dict()
+    domain = extract_values(domain_from_files, DOMAIN_KEYS)
+
+    # Extract flows
+    flows = importer.get_user_flows()
+    flows_dict = {KEY_FLOWS: get_flows_as_json(flows)}
+
+    # Extract NLU data
+    nlu_data = importer.get_nlu_data()
+    nlu_examples = nlu_data.filter_training_examples(
+        lambda ex: ex.get("intent") in nlu_data.intents
+    )
+    nlu_dict = RasaYAMLWriter().training_data_to_dict(nlu_examples)
+
+    # Use provided endpoints or default to empty dict
+    if endpoints is None:
+        endpoints = {}
+
+    return CALMUserData(
+        flows=flows_dict or {},
+        domain=domain or {},
+        config=config or {},
+        endpoints=endpoints or {},
+        nlu=nlu_dict or {},
+    )

rasa/shared/providers/_utils.py

@@ -1,87 +1,103 @@
 from typing import Any, Dict, Optional
 
+import boto3
 import structlog
-from litellm import validate_environment
+from botocore.exceptions import BotoCoreError, ClientError
 
 from rasa.shared.constants import (
     API_BASE_CONFIG_KEY,
     API_VERSION_CONFIG_KEY,
     AWS_ACCESS_KEY_ID_CONFIG_KEY,
-    AWS_ACCESS_KEY_ID_ENV_VAR,
+    AWS_BEDROCK_PROVIDER,
     AWS_REGION_NAME_CONFIG_KEY,
-    AWS_REGION_NAME_ENV_VAR,
+    AWS_SAGEMAKER_CHAT_PROVIDER,
+    AWS_SAGEMAKER_PROVIDER,
     AWS_SECRET_ACCESS_KEY_CONFIG_KEY,
-    AWS_SECRET_ACCESS_KEY_ENV_VAR,
     AWS_SESSION_TOKEN_CONFIG_KEY,
-    AWS_SESSION_TOKEN_ENV_VAR,
     AZURE_API_BASE_ENV_VAR,
     AZURE_API_VERSION_ENV_VAR,
     DEPLOYMENT_CONFIG_KEY,
 )
 from rasa.shared.exceptions import ProviderClientValidationError
-from rasa.shared.providers.embedding._base_litellm_embedding_client import (
-    _VALIDATE_ENVIRONMENT_MISSING_KEYS_KEY,
-)
+from rasa.shared.utils.io import resolve_environment_variables
 
 structlogger = structlog.get_logger()
 
 
 def validate_aws_setup_for_litellm_clients(
-    litellm_model_name: str, litellm_call_kwargs: dict, source_log: str
+    litellm_model_name: str, litellm_call_kwargs: Dict, source_log: str, provider: str
 ) -> None:
-    """Validates the AWS setup for LiteLLM clients to ensure all required
-    environment variables or corresponding call kwargs are set.
+    """Validates the AWS setup for LiteLLM clients to ensure credentials are set.
 
     Args:
         litellm_model_name (str): The name of the LiteLLM model being validated.
         litellm_call_kwargs (dict): Additional keyword arguments passed to the client,
             which may include configuration values for AWS credentials.
         source_log (str): The source log identifier for structured logging.
+        provider (str): The provider for which the validation is being performed.
 
     Raises:
         ProviderClientValidationError: If any required AWS environment variable
             or corresponding configuration key is missing.
     """
-
-    # Mapping of environment variable names to their corresponding config keys
-    envs_to_args = {
-        AWS_ACCESS_KEY_ID_ENV_VAR: AWS_ACCESS_KEY_ID_CONFIG_KEY,
-        AWS_SECRET_ACCESS_KEY_ENV_VAR: AWS_SECRET_ACCESS_KEY_CONFIG_KEY,
-        AWS_REGION_NAME_ENV_VAR: AWS_REGION_NAME_CONFIG_KEY,
-        AWS_SESSION_TOKEN_ENV_VAR: AWS_SESSION_TOKEN_CONFIG_KEY,
-    }
-
-    # Validate the environment setup for the model
-    validation_info = validate_environment(litellm_model_name)
-    missing_environment_variables = validation_info.get(
-        _VALIDATE_ENVIRONMENT_MISSING_KEYS_KEY, []
+    # expand environment variables if referenced in the config
+    resolved_litellm_call_kwargs: Dict = resolve_environment_variables(
+        litellm_call_kwargs
+    )  # type: ignore[assignment]
+
+    # boto3 only accepts bedrock and sagemaker as valid clients
+    # therefore we need to convert the provider name if it is defined
+    # as sagemaker_chat
+    provider = (
+        AWS_SAGEMAKER_PROVIDER if provider == AWS_SAGEMAKER_CHAT_PROVIDER else provider
     )
-    # Filter out missing environment variables that have been set trough arguments
-    # in extra parameters
-    missing_environment_variables = [
-        missing_env_var
-        for missing_env_var in missing_environment_variables
-        if litellm_call_kwargs.get(envs_to_args.get(missing_env_var)) is None
-    ]
 
-    if missing_environment_variables:
-        missing_environment_details = [
-            (
-                f"'{missing_env_var}' environment variable or "
-                f"'{envs_to_args.get(missing_env_var)}' config key"
-            )
-            for missing_env_var in missing_environment_variables
+    # if the AWS credentials are defined in the endpoints yaml model config,
+    # either as referenced secret env vars or direct values, we need to pass them
+    # to the boto3 client to ensure that the client can connect to the AWS service.
+    additional_kwargs: Dict[str, Any] = {}
+    if AWS_ACCESS_KEY_ID_CONFIG_KEY in resolved_litellm_call_kwargs:
+        additional_kwargs[AWS_ACCESS_KEY_ID_CONFIG_KEY] = resolved_litellm_call_kwargs[
+            AWS_ACCESS_KEY_ID_CONFIG_KEY
         ]
+    if AWS_SECRET_ACCESS_KEY_CONFIG_KEY in resolved_litellm_call_kwargs:
+        additional_kwargs[AWS_SECRET_ACCESS_KEY_CONFIG_KEY] = (
+            resolved_litellm_call_kwargs[AWS_SECRET_ACCESS_KEY_CONFIG_KEY]
+        )
+    if AWS_SESSION_TOKEN_CONFIG_KEY in resolved_litellm_call_kwargs:
+        additional_kwargs[AWS_SESSION_TOKEN_CONFIG_KEY] = resolved_litellm_call_kwargs[
+            AWS_SESSION_TOKEN_CONFIG_KEY
+        ]
+    if AWS_REGION_NAME_CONFIG_KEY in resolved_litellm_call_kwargs:
+        additional_kwargs["region_name"] = resolved_litellm_call_kwargs[
+            AWS_REGION_NAME_CONFIG_KEY
+        ]
+
+    try:
+        # We are using the boto3 client because it can discover the AWS credentials
+        # from the environment variables, credentials file, or IAM roles.
+        # This is necessary to ensure that the client can connect to the AWS service.
+        aws_client = boto3.client(provider, **additional_kwargs)
+
+        # Using different method calls available to different AWS clients
+        # to test the connection
+        if provider == AWS_SAGEMAKER_PROVIDER:
+            aws_client.list_models()
+        elif provider == AWS_BEDROCK_PROVIDER:
+            aws_client.get_model_invocation_logging_configuration()
+
+    except (ClientError, BotoCoreError) as exc:
         event_info = (
-            f"The following environment variables or configuration keys are "
-            f"missing: "
-            f"{', '.join(missing_environment_details)}. "
-            f"These settings are required for API calls."
+            f"Failed to validate AWS setup for LiteLLM clients: {exc}. "
+            f"Ensure that you are using one of the available authentication methods:"
+            f"credentials file, environment variables, or IAM roles. "
+            f"Also, ensure that the AWS region is set correctly. "
         )
         structlogger.error(
-            f"{source_log}.validate_aws_environment_variables",
+            f"{source_log}.validate_aws_credentials_for_litellm_clients",
             event_info=event_info,
-            missing_environment_variables=missing_environment_variables,
+            exception=str(exc),
+            model_name=litellm_model_name,
         )
         raise ProviderClientValidationError(event_info)
 

rasa/shared/providers/embedding/default_litellm_embedding_client.py

@@ -37,6 +37,7 @@ class DefaultLiteLLMEmbeddingClient(_BaseLiteLLMEmbeddingClient):
 
     @classmethod
     def from_config(cls, config: Dict[str, Any]) -> "DefaultLiteLLMEmbeddingClient":
+        """Creates a DefaultLiteLLMEmbeddingClient instance from a config dict."""
         default_config = DefaultLiteLLMClientConfig.from_dict(config)
         return cls(
             model=default_config.model,
@@ -121,6 +122,7 @@ class DefaultLiteLLMEmbeddingClient(_BaseLiteLLMEmbeddingClient):
                 self._litellm_model_name,
                 self._litellm_extra_parameters,
                 "default_litellm_embedding_client",
+                provider=self.provider,
             )
         else:
             super().validate_client_setup()

rasa/shared/providers/llm/default_litellm_llm_client.py

@@ -39,6 +39,7 @@ class DefaultLiteLLMClient(_BaseLiteLLMClient):
 
     @classmethod
     def from_config(cls, config: Dict[str, Any]) -> DefaultLiteLLMClient:
+        """Creates a DefaultLiteLLMClient instance from a configuration dictionary."""
         default_config = DefaultLiteLLMClientConfig.from_dict(config)
         return cls(
             model=default_config.model,
@@ -110,6 +111,7 @@ class DefaultLiteLLMClient(_BaseLiteLLMClient):
                 self._litellm_model_name,
                 self._litellm_extra_parameters,
                 "default_litellm_llm_client",
+                provider=self.provider,
             )
         else:
             super().validate_client_setup()

rasa/shared/utils/common.py

@@ -373,7 +373,8 @@ def validate_environment(
             importlib.import_module(p)
         except ImportError:
             raise MissingDependencyException(
-                f"Missing package for {component_name}: {p}"
+                f"Missing dependency for {component_name}: {p}. "
+                f"Please ensure the correct package is installed."
             )
 
 

rasa/studio/upload.py

@@ -7,7 +7,6 @@ from typing import Any, Dict, Iterable, List, Optional, Set, Text, Tuple, Union
 import questionary
 import requests
 import structlog
-from pydantic import BaseModel, Field
 
 import rasa.cli.telemetry
 import rasa.cli.utils
@@ -24,9 +23,13 @@ from rasa.shared.constants import (
     DEFAULT_DOMAIN_PATHS,
 )
 from rasa.shared.core.domain import Domain
-from rasa.shared.core.flows.yaml_flows_io import YAMLFlowsReader, YamlFlowsWriter
+from rasa.shared.core.flows.yaml_flows_io import YAMLFlowsReader
 from rasa.shared.exceptions import RasaException
-from rasa.shared.importers.importer import FlowSyncImporter, TrainingDataImporter
+from rasa.shared.importers.importer import TrainingDataImporter
+from rasa.shared.importers.utils import (
+    CALMUserData,
+    extract_calm_import_parts_from_importer,
+)
 from rasa.shared.nlu.training_data.formats.rasa_yaml import (
     RasaYAMLReader,
     RasaYAMLWriter,
@@ -34,7 +37,6 @@ from rasa.shared.nlu.training_data.formats.rasa_yaml import (
 from rasa.shared.utils.llm import collect_custom_prompts
 from rasa.shared.utils.yaml import (
     dump_obj_as_yaml_to_string,
-    read_yaml,
     read_yaml_file,
 )
 from rasa.studio import results_logger
@@ -43,6 +45,7 @@ from rasa.studio.config import StudioConfig
 from rasa.studio.results_logger import StudioResult, with_studio_error_handler
 from rasa.studio.utils import validate_argument_paths
 from rasa.telemetry import track_upload_to_studio_failed
+from rasa.utils.json_utils import extract_values
 
 structlogger = structlog.get_logger()
 
@@ -68,16 +71,6 @@ DOMAIN_KEYS = [
 ]
 
 
-class CALMImportParts(BaseModel):
-    """All pieces that will be uploaded to Rasa Studio."""
-
-    flows: Dict[str, Any]
-    domain: Dict[str, Any]
-    config: Dict[str, Any]
-    endpoints: Dict[str, Any]
-    nlu: Dict[str, Any] = Field(default_factory=dict)
-
-
 def _get_selected_entities_and_intents(
     args: argparse.Namespace,
     intents_from_files: Set[Text],
@@ -195,11 +188,6 @@ config_keys = [
 ]
 
 
-def extract_values(data: Dict, keys: List[Text]) -> Dict:
-    """Extracts values for given keys from a dictionary."""
-    return {key: data.get(key) for key in keys if data.get(key)}
-
-
 def _get_assistant_name(config: Dict[Text, Any]) -> str:
     config_assistant_id = config.get("assistant_id", "")
     assistant_name = questionary.text(
@@ -238,7 +226,7 @@ def build_calm_import_parts(
     config_path: Text,
     endpoints_path: Optional[Text] = None,
     assistant_name: Optional[Text] = None,
-) -> Tuple[str, CALMImportParts]:
+) -> Tuple[str, CALMUserData]:
     """Builds the parts of the assistant to be uploaded to Studio.
 
     Args:
@@ -251,9 +239,11 @@ def build_calm_import_parts(
     Returns:
         The assistant name and the parts to be uploaded
     """
+    training_data_paths = data_path if isinstance(data_path, list) else [str(data_path)]
     importer = TrainingDataImporter.load_from_dict(
         domain_path=domain_path,
         config_path=config_path,
+        training_data_paths=training_data_paths,
         expand_env_vars=False,
     )
 
@@ -261,34 +251,10 @@ def build_calm_import_parts(
     endpoints = read_yaml_file(endpoints_path, expand_env_vars=False)
     assistant_name = assistant_name or _get_assistant_name(config)
 
-    domain_from_files = importer.get_user_domain().as_dict()
-    domain = extract_values(domain_from_files, DOMAIN_KEYS)
-
-    training_data_paths = data_path if isinstance(data_path, list) else [str(data_path)]
-    flow_importer = FlowSyncImporter.load_from_dict(
-        training_data_paths=training_data_paths, expand_env_vars=False
-    )
-
-    flows = list(flow_importer.get_user_flows())
-    flows_yaml = YamlFlowsWriter().dumps(flows)
-    flows = read_yaml(flows_yaml, expand_env_vars=False)
-
-    nlu_importer = TrainingDataImporter.load_from_dict(
-        training_data_paths=training_data_paths, expand_env_vars=False
-    )
-    nlu_data = nlu_importer.get_nlu_data()
-    nlu_examples = nlu_data.filter_training_examples(
-        lambda ex: ex.get("intent") in nlu_data.intents
-    )
-    nlu_examples_yaml = RasaYAMLWriter().dumps(nlu_examples)
-    nlu = read_yaml(nlu_examples_yaml, expand_env_vars=False)
-
-    parts = CALMImportParts(
-        flows=flows,
-        domain=domain,
+    parts = extract_calm_import_parts_from_importer(
+        importer=importer,
         config=config,
         endpoints=endpoints,
-        nlu=nlu,
     )
 
     return assistant_name, parts