rasa-pro 3.14.0a20__py3-none-any.whl → 3.14.0a23__py3-none-any.whl

rasa/engine/recipes/default_recipe.py

@@ -5,6 +5,7 @@ import dataclasses
 import enum
 import logging
 import math
+import sys
 from enum import Enum
 from typing import Any, Callable, Dict, List, Optional, Set, Text, Tuple, Type, Union
 
@@ -180,7 +181,28 @@ class DefaultV1Recipe(Recipe):
     @classmethod
     def _from_registry(cls, name: Text) -> RegisteredComponent:
         # Importing all the default Rasa components will automatically register them
-        from rasa.engine.recipes.default_components import DEFAULT_COMPONENTS  # noqa
+        from rasa.engine.recipes.default_components import (
+            CONVERT_FEATURIZER_AVAILABLE,
+            CRF_ENTITY_EXTRACTOR_AVAILABLE,
+            DIET_CLASSIFIER_AVAILABLE,
+            JIEBA_TOKENIZER_AVAILABLE,
+            LANGUAGE_MODEL_FEATURIZER_AVAILABLE,
+            LEXICAL_SYNTACTIC_FEATURIZER_AVAILABLE,
+            LOGISTIC_REGRESSION_CLASSIFIER_AVAILABLE,
+            MITIE_ENTITY_EXTRACTOR_AVAILABLE,
+            MITIE_FEATURIZER_AVAILABLE,
+            MITIE_INTENT_CLASSIFIER_AVAILABLE,
+            MITIE_NLP_AVAILABLE,
+            MITIE_TOKENIZER_AVAILABLE,
+            RESPONSE_SELECTOR_AVAILABLE,
+            SKLEARN_INTENT_CLASSIFIER_AVAILABLE,
+            SPACY_ENTITY_EXTRACTOR_AVAILABLE,
+            SPACY_FEATURIZER_AVAILABLE,
+            SPACY_NLP_AVAILABLE,
+            SPACY_TOKENIZER_AVAILABLE,
+            TED_POLICY_AVAILABLE,
+            UNEXPECTED_INTENT_POLICY_AVAILABLE,
+        )
 
         if name in cls._registered_components:
             return cls._registered_components[name]
@@ -190,6 +212,56 @@ class DefaultV1Recipe(Recipe):
             if clazz.__name__ in cls._registered_components:
                 return cls._registered_components[clazz.__name__]
 
+        # Check Python version compatibility for TensorFlow components
+        # TensorFlow support was removed for Python 3.12+ due to compatibility issues
+        if sys.version_info >= (3, 12):
+            raise InvalidConfigException(
+                f"{name} is not available on Python 3.12+. "
+                "TensorFlow support has been removed for Python 3.12+ versions. "
+                "Please use Python 3.11 or earlier for TensorFlow-based components. "
+            )
+
+        # Check for nlu components that failed to load
+        # because conditional module import of those components did not succeed
+        nlu_components = {
+            # tensorflow-dependent components
+            "DIETClassifier": DIET_CLASSIFIER_AVAILABLE,
+            "TEDPolicy": TED_POLICY_AVAILABLE,
+            "UnexpecTEDIntentPolicy": UNEXPECTED_INTENT_POLICY_AVAILABLE,
+            "ResponseSelector": RESPONSE_SELECTOR_AVAILABLE,
+            "ConveRTFeaturizer": CONVERT_FEATURIZER_AVAILABLE,
+            "LanguageModelFeaturizer": LANGUAGE_MODEL_FEATURIZER_AVAILABLE,
+            # nlu components dependent on other dependencies than tensorflow
+            "LogisticRegressionClassifier": LOGISTIC_REGRESSION_CLASSIFIER_AVAILABLE,
+            "SklearnIntentClassifier": SKLEARN_INTENT_CLASSIFIER_AVAILABLE,
+            "LexicalSyntacticFeaturizer": LEXICAL_SYNTACTIC_FEATURIZER_AVAILABLE,
+            "MitieFeaturizer": MITIE_FEATURIZER_AVAILABLE,
+            "SpacyFeaturizer": SPACY_FEATURIZER_AVAILABLE,
+            "JiebaTokenizer": JIEBA_TOKENIZER_AVAILABLE,
+            "MitieTokenizer": MITIE_TOKENIZER_AVAILABLE,
+            "SpacyTokenizer": SPACY_TOKENIZER_AVAILABLE,
+            "MitieIntentClassifier": MITIE_INTENT_CLASSIFIER_AVAILABLE,
+            "MitieEntityExtractor": MITIE_ENTITY_EXTRACTOR_AVAILABLE,
+            "SpacyEntityExtractor": SPACY_ENTITY_EXTRACTOR_AVAILABLE,
+            "MitieNLP": MITIE_NLP_AVAILABLE,
+            "SpacyNLP": SPACY_NLP_AVAILABLE,
+            "CRFEntityExtractor": CRF_ENTITY_EXTRACTOR_AVAILABLE,
+        }
+
+        if name in nlu_components:
+            # nlu-dependent component failed to load
+            # because nlu dependency group not installed
+            if not nlu_components[name]:
+                from rasa.exceptions import MissingDependencyException
+
+                raise MissingDependencyException(
+                    f"The {name} component requires additional dependencies "
+                    f"which are not installed. "
+                    f"Please install the required extra by running: "
+                    f"pip install 'rasa-pro[nlu]' OR "
+                    f"poetry add 'rasa-pro[nlu]'"
+                )
+
         raise InvalidConfigException(
             f"Can't load class for name '{name}'. Please make sure to provide "
             f"a valid name or module path and to register it using the "
@@ -485,17 +557,42 @@ class DefaultV1Recipe(Recipe):
         component: Type[GraphComponent],
         component_config: Dict[Text, Any],
     ) -> Dict[Text, Any]:
-        from rasa.nlu.classifiers.mitie_intent_classifier import MitieIntentClassifier
-        from rasa.nlu.classifiers.sklearn_intent_classifier import (
-            SklearnIntentClassifier,
-        )
-        from rasa.nlu.extractors.mitie_entity_extractor import MitieEntityExtractor
+        # Try to import MitieIntentClassifier, but handle ImportError gracefully
+        # since the dependencies for it may not be installed
+        try:
+            from rasa.nlu.classifiers.mitie_intent_classifier import (
+                MitieIntentClassifier,
+            )
+        except ImportError:
+            MitieIntentClassifier = None  # type: ignore
 
-        cli_args_mapping: Dict[Type[GraphComponent], List[Text]] = {
-            MitieIntentClassifier: ["num_threads"],
-            MitieEntityExtractor: ["num_threads"],
-            SklearnIntentClassifier: ["num_threads"],
-        }
+        # Try to import MitieEntityExtractor, but handle ImportError gracefully
+        # since the dependencies for it may not be installed
+        try:
+            from rasa.nlu.extractors.mitie_entity_extractor import MitieEntityExtractor
+        except ImportError:
+            MitieEntityExtractor = None  # type: ignore
+
+        # Try to import SklearnIntentClassifier, but handle ImportError gracefully
+        # since the dependencies for it may not be installed
+        try:
+            from rasa.nlu.classifiers.sklearn_intent_classifier import (
+                SklearnIntentClassifier,
+            )
+        except ImportError:
+            SklearnIntentClassifier = None  # type: ignore
+
+        cli_args_mapping: Dict[Type[GraphComponent], List[Text]] = {}
+
+        # Only add components if they were successfully imported
+        if MitieIntentClassifier is not None:
+            cli_args_mapping[MitieIntentClassifier] = ["num_threads"]
+
+        if MitieEntityExtractor is not None:
+            cli_args_mapping[MitieEntityExtractor] = ["num_threads"]
+
+        if SklearnIntentClassifier is not None:
+            cli_args_mapping[SklearnIntentClassifier] = ["num_threads"]
 
         config_from_cli = {
             param: cli_parameters[param]

rasa/engine/runner/dask.py

@@ -1,18 +1,18 @@
 from __future__ import annotations
 
 import asyncio
-import logging
 from typing import Any, Dict, List, Optional, Text
 
 import dask.core
 import dask.local
+import structlog
 
 from rasa.engine.exceptions import GraphRunError
 from rasa.engine.graph import ExecutionContext, GraphNode, GraphNodeHook, GraphSchema
 from rasa.engine.runner.interface import GraphRunner
 from rasa.engine.storage.storage import ModelStorage
 
-logger = logging.getLogger(__name__)
+structlogger = structlog.get_logger()
 
 
 class DaskGraphRunner(GraphRunner):
@@ -94,9 +94,12 @@ class DaskGraphRunner(GraphRunner):
         if inputs:
             self._add_inputs_to_graph(inputs, run_graph)
 
-        logger.debug(
-            f"Running graph with inputs: {inputs}, targets: {targets} "
-            f"and {self._execution_context}."
+        structlogger.debug(
+            "rasa.engine.runner.dask.run",
+            event_info="Running graph with inputs, targets and execution context.",
+            inputs=inputs,
+            targets=targets,
+            execution_context=self._execution_context,
         )
 
         try:

rasa/engine/validation.py

@@ -23,7 +23,8 @@ import structlog
 import typing_utils
 
 import rasa.utils.common
-from rasa.core.available_endpoints import AvailableEndpoints
+from rasa.core.config.available_endpoints import AvailableEndpoints
+from rasa.core.config.configuration import Configuration
 from rasa.core.nlg.contextual_response_rephraser import ContextualResponseRephraser
 from rasa.core.policies.intentless_policy import IntentlessPolicy
 from rasa.core.policies.policy import PolicyPrediction
@@ -1039,7 +1040,7 @@ def validate_model_client_configuration_setup_during_training_time(
                     )
 
     # also include the ContextualResponseRephraser component
-    endpoints = AvailableEndpoints.get_instance()
+    endpoints = Configuration.get_instance().endpoints
     if endpoints.nlg is not None:
         _validate_component_model_client_config(
             endpoints.nlg.kwargs,
@@ -1075,7 +1076,7 @@ def validate_model_client_configuration_setup_during_training_time(
             ),
         )
 
-    endpoints = AvailableEndpoints.get_instance()
+    endpoints = Configuration.get_instance().endpoints
     if len(model_group_ids) > 0 and endpoints.model_groups is None:
         raise ValidationError(
             code="engine.validation.validate_model_client_configuration_setup"
@@ -1138,7 +1139,7 @@ def _validate_component_model_client_config_has_references_to_endpoints(
         # no llm/embeddings configuration present
         return
 
-    endpoints = AvailableEndpoints.get_instance()
+    endpoints = Configuration.get_instance().endpoints
 
     if MODEL_GROUP_CONFIG_KEY in component_config[key]:
         referencing_model_group_id = component_config[key][MODEL_GROUP_CONFIG_KEY]
@@ -1213,7 +1214,7 @@ def validate_model_client_configuration_setup_during_inference_time(
                     )
 
     # also include the ContextualResponseRephraser component
-    endpoints = AvailableEndpoints.get_instance()
+    endpoints = Configuration.get_instance().endpoints
     if endpoints.nlg is not None:
         _validate_component_model_client_config_has_references_to_endpoints(
             component_config=endpoints.nlg.kwargs,
@@ -1224,6 +1225,12 @@ def validate_model_client_configuration_setup_during_inference_time(
 
 def _validate_unique_model_group_ids(model_groups: List[Dict[str, Any]]) -> None:
     # Each model id must be unique within the model_groups
+    structlogger.debug(
+        "engine.validation.validate_unique_model_group_ids",
+        event_info="Validating that model group IDs are unique.",
+        model_groups=model_groups,
+    )
+
     model_ids = [model_group[MODEL_GROUP_ID_CONFIG_KEY] for model_group in model_groups]
     if len(model_ids) != len(set(model_ids)):
         counts = Counter(model_ids)
@@ -1410,11 +1417,17 @@ def _validate_sensitive_keys_are_an_environment_variables_for_model_groups(
 
 def validate_model_group_configuration_setup() -> None:
     """Validates the model group configuration setup in endpoints.yml."""
-    endpoints = AvailableEndpoints.get_instance()
+    endpoints = Configuration.get_instance().endpoints
 
     if endpoints.model_groups is None:
         return
 
+    structlogger.debug(
+        "engine.validation.validate_model_group_configuration_setup",
+        event_info="Validating the model group configuration setup.",
+        model_groups=endpoints.model_groups,
+        path=endpoints.config_file_path,
+    )
     _validate_unique_model_group_ids(endpoints.model_groups)
     _validate_model_group_with_multiple_models(endpoints.model_groups)
     _validate_usage_of_environment_variables_in_model_group_config(

rasa/graph_components/validators/default_recipe_validator.py

@@ -10,22 +10,13 @@ from rasa.core.policies.flow_policy import FlowPolicy
 from rasa.core.policies.memoization import AugmentedMemoizationPolicy, MemoizationPolicy
 from rasa.core.policies.policy import Policy, SupportedData
 from rasa.core.policies.rule_policy import RulePolicy
-from rasa.core.policies.ted_policy import TEDPolicy
-from rasa.core.policies.unexpected_intent_policy import UnexpecTEDIntentPolicy
 from rasa.engine.graph import ExecutionContext, GraphComponent, GraphSchema, SchemaNode
 from rasa.engine.storage.resource import Resource
 from rasa.engine.storage.storage import ModelStorage
-from rasa.nlu.classifiers.diet_classifier import DIETClassifier
-from rasa.nlu.extractors.crf_entity_extractor import (
-    CRFEntityExtractor,
-    CRFEntityExtractorOptions,
-)
 from rasa.nlu.extractors.entity_synonyms import EntitySynonymMapper
-from rasa.nlu.extractors.mitie_entity_extractor import MitieEntityExtractor
 from rasa.nlu.extractors.regex_entity_extractor import RegexEntityExtractor
 from rasa.nlu.featurizers.featurizer import Featurizer
 from rasa.nlu.featurizers.sparse_featurizer.regex_featurizer import RegexFeaturizer
-from rasa.nlu.selectors.response_selector import ResponseSelector
 from rasa.nlu.tokenizers.tokenizer import Tokenizer
 from rasa.shared.constants import (
     DEFAULT_CONFIG_PATH,
@@ -45,11 +36,45 @@ from rasa.shared.core.training_data.structures import RuleStep, StoryGraph
 from rasa.shared.exceptions import InvalidConfigException
 from rasa.shared.importers.importer import TrainingDataImporter
 from rasa.shared.nlu.training_data.training_data import TrainingData
+from rasa.shared.utils.common import conditional_import
 
-# TODO: Can we replace this with the registered types from the regitry?
-TRAINABLE_EXTRACTORS = [MitieEntityExtractor, CRFEntityExtractor, DIETClassifier]
+# Conditional imports for TensorFlow-dependent components
+TEDPolicy, TED_POLICY_AVAILABLE = conditional_import(
+    "rasa.core.policies.ted_policy", "TEDPolicy"
+)
+UnexpecTEDIntentPolicy, UNEXPECTED_INTENT_POLICY_AVAILABLE = conditional_import(
+    "rasa.core.policies.unexpected_intent_policy", "UnexpecTEDIntentPolicy"
+)
+DIETClassifier, DIET_CLASSIFIER_AVAILABLE = conditional_import(
+    "rasa.nlu.classifiers.diet_classifier", "DIETClassifier"
+)
+ResponseSelector, RESPONSE_SELECTOR_AVAILABLE = conditional_import(
+    "rasa.nlu.selectors.response_selector", "ResponseSelector"
+)
+
+# Conditional imports for nlu components requiring other dependencies than tensorflow
+MitieEntityExtractor, MITIE_ENTITY_EXTRACTOR_AVAILABLE = conditional_import(
+    "rasa.nlu.extractors.mitie_entity_extractor", "MitieEntityExtractor"
+)
+CRFEntityExtractor, CRF_ENTITY_EXTRACTOR_AVAILABLE = conditional_import(
+    "rasa.nlu.extractors.crf_entity_extractor", "CRFEntityExtractor"
+)
+CRFEntityExtractorOptions, _ = conditional_import(
+    "rasa.nlu.extractors.crf_entity_extractor", "CRFEntityExtractorOptions"
+)
+
+# TODO: Can we replace this with the registered types from the registry?
+TRAINABLE_EXTRACTORS = []
+if CRF_ENTITY_EXTRACTOR_AVAILABLE:
+    TRAINABLE_EXTRACTORS.append(CRFEntityExtractor)
+if MITIE_ENTITY_EXTRACTOR_AVAILABLE:
+    TRAINABLE_EXTRACTORS.append(MitieEntityExtractor)
+if DIET_CLASSIFIER_AVAILABLE:
+    TRAINABLE_EXTRACTORS.append(DIETClassifier)
 # TODO: replace these once the Recipe is merged (used in tests)
-POLICY_CLASSSES = {TEDPolicy, MemoizationPolicy, RulePolicy}
+POLICY_CLASSSES = {MemoizationPolicy, RulePolicy}
+if TED_POLICY_AVAILABLE:
+    POLICY_CLASSSES.add(TEDPolicy)
 
 
 def _types_to_str(types: Iterable[Type]) -> Text:
@@ -158,18 +183,27 @@ class DefaultV1RecipeValidator(GraphComponent):
                 docs=DOCS_URL_COMPONENTS,
             )
 
+        available_extractors = {DIETClassifier}
+        if CRF_ENTITY_EXTRACTOR_AVAILABLE:
+            available_extractors.add(CRFEntityExtractor)
+
         if training_data.entity_examples and self._component_types.isdisjoint(
-            {DIETClassifier, CRFEntityExtractor}
+            available_extractors
         ):
             if training_data.entity_roles_groups_used():
+                crf_name = (
+                    CRFEntityExtractor.__name__
+                    if CRFEntityExtractor
+                    else "CRFEntityExtractor"
+                )
                 rasa.shared.utils.io.raise_warning(
                     f"You have defined training data with entities that "
                     f"have roles/groups, but your NLU configuration does not "
                     f"include a '{DIETClassifier.__name__}' "
-                    f"or a '{CRFEntityExtractor.__name__}'. "
+                    f"or a '{crf_name}'. "
                     f"To train entities that have roles/groups, "
                     f"add either '{DIETClassifier.__name__}' "
-                    f"or '{CRFEntityExtractor.__name__}' to your "
+                    f"or '{crf_name}' to your "
                     f"configuration.",
                     docs=DOCS_URL_COMPONENTS,
                 )
@@ -203,7 +237,16 @@ class DefaultV1RecipeValidator(GraphComponent):
             )
 
         if training_data.lookup_tables:
-            if self._component_types.isdisjoint([CRFEntityExtractor, DIETClassifier]):
+            lookup_extractors = [DIETClassifier]
+            if CRF_ENTITY_EXTRACTOR_AVAILABLE:
+                lookup_extractors.append(CRFEntityExtractor)
+
+            if self._component_types.isdisjoint(lookup_extractors):
+                crf_name = (
+                    CRFEntityExtractor.__name__
+                    if CRFEntityExtractor
+                    else "CRFEntityExtractor"
+                )
                 rasa.shared.utils.io.raise_warning(
                     f"You have defined training data consisting of lookup tables, but "
                     f"your NLU configuration does not include any components "
@@ -211,33 +254,45 @@ class DefaultV1RecipeValidator(GraphComponent):
                     f"To make use of the features that are created with the "
                     f"help of the lookup tables, "
                     f"add a '{DIETClassifier.__name__}' or a "
-                    f"'{CRFEntityExtractor.__name__}' "
+                    f"'{crf_name}' "
                     f"with the 'pattern' feature "
                     f"to your configuration.",
                     docs=DOCS_URL_COMPONENTS,
                 )
 
-            elif CRFEntityExtractor in self._component_types:
+            elif (
+                CRF_ENTITY_EXTRACTOR_AVAILABLE
+                and CRFEntityExtractor in self._component_types
+            ):
                 crf_schema_nodes = [
                     schema_node
                     for schema_node in self._graph_schema.nodes.values()
                     if schema_node.uses == CRFEntityExtractor
                 ]
-                has_pattern_feature = any(
-                    CRFEntityExtractorOptions.PATTERN in feature_list
-                    for crf in crf_schema_nodes
-                    for feature_list in crf.config.get("features", [])
+                has_pattern_feature = (
+                    any(
+                        CRFEntityExtractorOptions.PATTERN in feature_list
+                        for crf in crf_schema_nodes
+                        for feature_list in crf.config.get("features", [])
+                    )
+                    if CRFEntityExtractorOptions
+                    else False
                 )
 
                 if not has_pattern_feature:
+                    crf_name = (
+                        CRFEntityExtractor.__name__
+                        if CRFEntityExtractor
+                        else "CRFEntityExtractor"
+                    )
                     rasa.shared.utils.io.raise_warning(
                         f"You have defined training data consisting of "
                         f"lookup tables, but your NLU configuration's "
-                        f"'{CRFEntityExtractor.__name__}' "
+                        f"'{crf_name}' "
                         f"does not include the "
                         f"'pattern' feature. To featurize lookup tables, "
                         f"add the 'pattern' feature to the "
-                        f"'{CRFEntityExtractor.__name__}' "
+                        f"'{crf_name}' "
                         "in your configuration.",
                         docs=DOCS_URL_COMPONENTS,
                     )
@@ -446,13 +501,16 @@ class DefaultV1RecipeValidator(GraphComponent):
             return
 
         # if we are using a flow policy, there shouldn't be any dm1 policies
-        dm1_policies = (
+        dm1_policies_list = [
             RulePolicy,
-            TEDPolicy,
             MemoizationPolicy,
             AugmentedMemoizationPolicy,
-            UnexpecTEDIntentPolicy,
-        )
+        ]
+        if TED_POLICY_AVAILABLE:
+            dm1_policies_list.append(TEDPolicy)
+        if UNEXPECTED_INTENT_POLICY_AVAILABLE:
+            dm1_policies_list.append(UnexpecTEDIntentPolicy)
+        dm1_policies = tuple(dm1_policies_list)
 
         for schema_node in self._graph_schema.nodes.values():
             if issubclass(schema_node.uses, dm1_policies):

rasa/hooks.py

@@ -19,7 +19,7 @@ hookimpl = pluggy.HookimplMarker("rasa")
 logger = logging.getLogger(__name__)
 
 
-@hookimpl  # type: ignore[misc]
+@hookimpl
 def refine_cli(
     subparsers: "SubParsersAction",
     parent_parsers: List[argparse.ArgumentParser],
@@ -37,7 +37,7 @@ def refine_cli(
     return None
 
 
-@hookimpl  # type: ignore[misc]
+@hookimpl
 def configure_commandline(cmdline_arguments: argparse.Namespace) -> Optional[Text]:
     from rasa.cli import x as rasa_x
     from rasa.tracing import config
@@ -59,21 +59,21 @@ def configure_commandline(cmdline_arguments: argparse.Namespace) -> Optional[Tex
     return endpoints_file
 
 
-@hookimpl  # type: ignore[misc]
+@hookimpl
 def init_telemetry(endpoints_file: Optional[Text]) -> None:
     import rasa.telemetry
 
     rasa.telemetry.identify_endpoint_config_traits(endpoints_file)
 
 
-@hookimpl  # type: ignore[misc]
+@hookimpl
 def init_managers(endpoints_file: Optional[Text]) -> None:
     from rasa.core.secrets_manager.factory import load_secret_manager
 
     load_secret_manager(endpoints_file)
 
 
-@hookimpl  # type: ignore[misc]
+@hookimpl
 def create_tracker_store(
     endpoint_config: Union["TrackerStore", "EndpointConfig"],
     domain: "Domain",

rasa/llm_fine_tuning/utils.py

@@ -61,9 +61,9 @@ def patch_invoke_llm_in_generators(mock_impl: Callable) -> Generator:
     try:
         # apply the monkey-patch everywhere
         for cls in originals:
-            cls.invoke_llm = mock_impl  # type: ignore[assignment]
+            cls.invoke_llm = mock_impl  # type: ignore[method-assign]
         yield
     finally:
         # restore originals (even if an exception happened)
         for cls, orig in originals.items():
-            cls.invoke_llm = orig  # type: ignore[assignment]
+            cls.invoke_llm = orig  # type: ignore[method-assign]