rasa-pro 3.14.0.dev7__py3-none-any.whl → 3.14.0.dev9__py3-none-any.whl

rasa/engine/recipes/default_components.py

@@ -1,7 +1,8 @@
 from rasa.core.policies.enterprise_search_policy import EnterpriseSearchPolicy
 from rasa.core.policies.flow_policy import FlowPolicy
 from rasa.core.policies.intentless_policy import IntentlessPolicy
-from rasa.core.policies.memoization import AugmentedMemoizationPolicy
+from rasa.core.policies.memoization import AugmentedMemoizationPolicy, MemoizationPolicy
+from rasa.core.policies.rule_policy import RulePolicy
 from rasa.dialogue_understanding.coexistence.intent_based_router import (
     IntentBasedRouter,
 )
@@ -10,6 +11,7 @@ from rasa.dialogue_understanding.generator import (
     LLMCommandGenerator,
 )
 from rasa.dialogue_understanding.generator.nlu_command_adapter import NLUCommandAdapter
+from rasa.engine.graph import GraphComponent
 from rasa.nlu.classifiers.fallback_classifier import FallbackClassifier
 from rasa.nlu.classifiers.keyword_intent_classifier import KeywordIntentClassifier
 from rasa.nlu.classifiers.logistic_regression_classifier import (
@@ -17,6 +19,7 @@ from rasa.nlu.classifiers.logistic_regression_classifier import (
 )
 from rasa.nlu.classifiers.mitie_intent_classifier import MitieIntentClassifier
 from rasa.nlu.classifiers.sklearn_intent_classifier import SklearnIntentClassifier
+from rasa.nlu.extractors.crf_entity_extractor import CRFEntityExtractor
 from rasa.nlu.extractors.duckling_entity_extractor import DucklingEntityExtractor
 from rasa.nlu.extractors.entity_synonyms import EntitySynonymMapper
 from rasa.nlu.extractors.mitie_entity_extractor import MitieEntityExtractor
@@ -37,10 +40,30 @@ from rasa.nlu.tokenizers.spacy_tokenizer import SpacyTokenizer
 from rasa.nlu.tokenizers.whitespace_tokenizer import WhitespaceTokenizer
 from rasa.nlu.utils.mitie_utils import MitieNLP
 from rasa.nlu.utils.spacy_utils import SpacyNLP
+from rasa.shared.utils.common import conditional_import
+
+# 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"
+)
+ConveRTFeaturizer, CONVERT_FEATURIZER_AVAILABLE = conditional_import(
+    "rasa.nlu.featurizers.dense_featurizer.convert_featurizer", "ConveRTFeaturizer"
+)
+LanguageModelFeaturizer, LANGUAGE_MODEL_FEATURIZER_AVAILABLE = conditional_import(
+    "rasa.nlu.featurizers.dense_featurizer.lm_featurizer", "LanguageModelFeaturizer"
+)
+ResponseSelector, RESPONSE_SELECTOR_AVAILABLE = conditional_import(
+    "rasa.nlu.selectors.response_selector", "ResponseSelector"
+)
 
 DEFAULT_COMPONENTS = [
     # Message Classifiers
-    # DIETClassifier,
     FallbackClassifier,
     KeywordIntentClassifier,
     MitieIntentClassifier,
@@ -51,9 +74,8 @@ DEFAULT_COMPONENTS = [
     LLMBasedRouter,
     IntentBasedRouter,
     # Response Selectors
-    # ResponseSelector,
     # Message Entity Extractors
-    # CRFEntityExtractor,
+    CRFEntityExtractor,
     DucklingEntityExtractor,
     EntitySynonymMapper,
     MitieEntityExtractor,
@@ -61,11 +83,9 @@ DEFAULT_COMPONENTS = [
     RegexEntityExtractor,
     # Message Feauturizers
     LexicalSyntacticFeaturizer,
-    # ConveRTFeaturizer,
     MitieFeaturizer,
     SpacyFeaturizer,
     CountVectorsFeaturizer,
-    # LanguageModelFeaturizer,
     RegexFeaturizer,
     # Tokenizers
     JiebaTokenizer,
@@ -76,12 +96,60 @@ DEFAULT_COMPONENTS = [
     MitieNLP,
     SpacyNLP,
     # Dialogue Management Policies
-    # TEDPolicy,
-    # UnexpecTEDIntentPolicy,
-    # RulePolicy,
-    # MemoizationPolicy,
+    RulePolicy,
+    MemoizationPolicy,
     AugmentedMemoizationPolicy,
     FlowPolicy,
     EnterpriseSearchPolicy,
     IntentlessPolicy,
 ]
+
+
+class ComponentInserter:
+    """Manages insertion of components at specific positions with index adjustment."""
+
+    def __init__(self, insertion_points: dict[str, int]):
+        self.insertion_points = insertion_points
+        self.insertion_counts = {category: 0 for category in insertion_points}
+
+    def insert_if_available(
+        self, component: GraphComponent, is_available: bool, category: str
+    ) -> None:
+        """Insert a component at the appropriate position if available."""
+        if is_available and component is not None:
+            # Calculate the adjusted index based on insertions in earlier categories
+            adjusted_index = self.insertion_points[category]
+            for cat, count in self.insertion_counts.items():
+                if self.insertion_points[cat] < self.insertion_points[category]:
+                    adjusted_index += count
+
+            DEFAULT_COMPONENTS.insert(adjusted_index, component)
+            self.insertion_counts[category] += 1
+
+
+# Define insertion points by component category to preserve order
+INSERTION_POINTS = {
+    "classifiers": 1,
+    "response_selectors": 11,
+    "featurizers": 20,
+    "policies": 35,
+}
+
+# Create inserter instance
+inserter = ComponentInserter(INSERTION_POINTS)
+
+# Conditionally add TensorFlow-dependent components
+inserter.insert_if_available(DIETClassifier, DIET_CLASSIFIER_AVAILABLE, "classifiers")
+inserter.insert_if_available(
+    ResponseSelector, RESPONSE_SELECTOR_AVAILABLE, "response_selectors"
+)
+inserter.insert_if_available(
+    ConveRTFeaturizer, CONVERT_FEATURIZER_AVAILABLE, "featurizers"
+)
+inserter.insert_if_available(
+    LanguageModelFeaturizer, LANGUAGE_MODEL_FEATURIZER_AVAILABLE, "featurizers"
+)
+inserter.insert_if_available(TEDPolicy, TED_POLICY_AVAILABLE, "policies")
+inserter.insert_if_available(
+    UnexpecTEDIntentPolicy, UNEXPECTED_INTENT_POLICY_AVAILABLE, "policies"
+)

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 +181,14 @@ 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,
+            DIET_CLASSIFIER_AVAILABLE,
+            LANGUAGE_MODEL_FEATURIZER_AVAILABLE,
+            RESPONSE_SELECTOR_AVAILABLE,
+            TED_POLICY_AVAILABLE,
+            UNEXPECTED_INTENT_POLICY_AVAILABLE,
+        )
 
         if name in cls._registered_components:
             return cls._registered_components[name]
@@ -190,6 +198,38 @@ 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 TensorFlow components that failed to load
+        # because conditional module import of those components did not succeed
+        tensorflow_component_availability = {
+            "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,
+        }
+
+        if name in tensorflow_component_availability:
+            # TensorFlow-dependent component failed to load
+            # if TensorFlow is not available
+            if not tensorflow_component_availability[name]:
+                raise InvalidConfigException(
+                    f"{name} requires TensorFlow. "
+                    "Please install it with: "
+                    "`pip install rasa[tensorflow]`. "
+                    "If using poetry, install it with "
+                    "`poetry install --extras tensorflow`."
+                )
+
         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 "

rasa/engine/storage/local_model_storage.py

@@ -1,14 +1,16 @@
 from __future__ import annotations
 
 import logging
+import os
 import shutil
 import sys
+import tarfile
 import tempfile
 import uuid
 from contextlib import contextmanager
 from datetime import datetime
 from pathlib import Path
-from typing import Generator, Optional, Text, Tuple, Union
+from typing import Callable, Generator, List, Optional, Text, Tuple, Union
 
 from tarsafe import TarSafe
 
@@ -57,6 +59,35 @@ def windows_safe_temporary_directory(
             yield temporary_directory
 
 
+def filter_normpath(member: tarfile.TarInfo, dest_path: str) -> tarfile.TarInfo:
+    """Normalize tar member paths for safe extraction"""
+    if member.name:
+        member.name = os.path.normpath(member.name)
+    return member
+
+
+FilterFunction = Callable[[tarfile.TarInfo, str], Optional[tarfile.TarInfo]]
+
+
+def create_combined_filter(existing_filter: Optional[FilterFunction]) -> FilterFunction:
+    """Create a filter that combines existing filter with path normalization"""
+
+    def combined_filter(
+        member: tarfile.TarInfo, dest_path: str
+    ) -> Optional[tarfile.TarInfo]:
+        """Apply existing filter first, then path normalization"""
+        if existing_filter is not None:
+            filtered_member = existing_filter(member, dest_path)
+            if filtered_member is None:
+                return None  # Rejected by existing filter
+            member = filtered_member  # Use the filtered result
+
+        # Apply our path normalization
+        return filter_normpath(member, dest_path)
+
+    return combined_filter
+
+
 class LocalModelStorage(ModelStorage):
     """Stores and provides output of `GraphComponents` on local disk."""
 
@@ -122,9 +153,29 @@ class LocalModelStorage(ModelStorage):
                 # this restriction in environments where it's not possible
                 # to override this behavior, mostly for internal policy reasons
                 # reference: https://stackoverflow.com/a/49102229
-                tar.extractall(f"\\\\?\\{temporary_directory}")
+                try:
+                    # Use extraction filter to normalize paths for compatibility
+                    # before trying the \\?\ prefix approach first
+                    prev_filter = getattr(tar, "extraction_filter", None)
+                    tar.extraction_filter = create_combined_filter(prev_filter)
+                    tar.extractall(
+                        f"\\\\?\\{temporary_directory}",
+                        members=yield_safe_members(tar.getmembers()),
+                    )
+                except Exception:
+                    # Fallback for Python versions with tarfile security fix
+                    logger.warning(
+                        "Failed to extract model archive with long path support. "
+                        "Falling back to regular extraction."
+                    )
+                    tar.extractall(
+                        temporary_directory,
+                        members=yield_safe_members(tar.getmembers()),
+                    )
             else:
-                tar.extractall(temporary_directory)
+                tar.extractall(
+                    temporary_directory, members=yield_safe_members(tar.getmembers())
+                )
         LocalModelStorage._assert_not_rasa2_archive(temporary_directory)
 
     @staticmethod
@@ -244,3 +295,32 @@ class LocalModelStorage(ModelStorage):
             core_target=model_configuration.core_target,
             nlu_target=model_configuration.nlu_target,
         )
+
+
+def yield_safe_members(
+    members: List[tarfile.TarInfo],
+) -> Generator[tarfile.TarInfo, None, None]:
+    """
+    Filter function for tar.extractall members parameter.
+    Validates each member and yields only safe ones.
+
+    Args:
+        members: Iterator of TarInfo objects from tar.getmembers()
+
+    Yields:
+        TarInfo: Safe members to extract
+    """
+    for member in members:
+        # Skip absolute paths
+        if Path(member.name).is_absolute():
+            continue
+
+        # Skip paths with directory traversal sequences
+        if ".." in member.name or "\\.." in member.name:
+            continue
+
+        # Skip special file types unless you need them
+        if member.isdev() or member.issym():
+            continue
+
+        yield member