rasa-pro 3.13.0.dev2__py3-none-any.whl → 3.13.0.dev3__py3-none-any.whl

rasa/cli/run.py

@@ -64,12 +64,16 @@ def run_actions(args: argparse.Namespace) -> None:
 
 def _validate_model_path(model_path: Text, parameter: Text, default: Text) -> Text:
     if model_path is not None and not os.path.exists(model_path):
-        reason_str = f"'{model_path}' not found."
-        if model_path is None:
-            reason_str = f"Parameter '{parameter}' not set."
-
-        logger.debug(f"{reason_str} Using default location '{default}' instead.")
-
+        raise ModelNotFound(
+            f"The provided model path '{model_path}' could not be found. "
+            "Provide an existing model path."
+        )
+
+    if model_path is None:
+        logger.debug(
+            f"Parameter '{parameter}' not set. "
+            "Using default location '{default}' instead."
+        )
         os.makedirs(default, exist_ok=True)
         model_path = default
 

rasa/cli/utils.py

@@ -14,6 +14,7 @@ import structlog
 import rasa.shared.utils.cli
 import rasa.shared.utils.io
 from rasa import telemetry
+from rasa.exceptions import ModelNotFound
 from rasa.shared.constants import (
     ASSISTANT_ID_DEFAULT_VALUE,
     ASSISTANT_ID_KEY,
@@ -77,6 +78,12 @@ def get_validated_path(
     if current and os.path.exists(current):
         return current
 
+    if parameter == "model":
+        raise ModelNotFound(
+            f"The provided model path '{current}' could not be found. "
+            "Provide an existing model path."
+        )
+
     # try to find a valid option among the defaults
     if isinstance(default, str) or isinstance(default, Path):
         default_options = [str(default)]

rasa/core/channels/channel.py

@@ -1,6 +1,9 @@
+import hmac
 import json
 import logging
 import uuid
+from base64 import b64encode
+from functools import wraps
 from typing import (
     Any,
     Awaitable,
@@ -15,6 +18,7 @@ from typing import (
 
 import jwt
 from sanic import Blueprint, Sanic
+from sanic.exceptions import Unauthorized
 from sanic.request import Request
 
 from rasa.cli import utils as cli_utils
@@ -454,3 +458,29 @@ class CollectingOutputChannel(OutputChannel):
         self, recipient_id: Text, json_message: Dict[Text, Any], **kwargs: Any
     ) -> None:
         await self._persist_message(self._message(recipient_id, custom=json_message))
+
+
+def requires_basic_auth(username: Optional[Text], password: Optional[Text]) -> Callable:
+    """Decorator to require basic auth for a route."""
+
+    def decorator(func: Callable) -> Callable:
+        @wraps(func)
+        async def wrapper(request: Request, *args: Any, **kwargs: Any) -> Any:
+            if not username or not password:
+                return await func(request, *args, **kwargs)
+
+            auth = request.headers.get("Authorization")
+            if not auth or not auth.startswith("Basic "):
+                logger.error("Missing or invalid authorization header.")
+                raise Unauthorized("Missing or invalid authorization header.")  # type: ignore[no-untyped-call]
+
+            encoded = b64encode(f"{username}:{password}".encode()).decode()
+            if not hmac.compare_digest(auth[6:], encoded):
+                logger.error("Invalid username or password.")
+                raise Unauthorized("Invalid username or password.")  # type: ignore[no-untyped-call]
+
+            return await func(request, *args, **kwargs)
+
+        return wrapper
+
+    return decorator

rasa/core/channels/voice_ready/jambonz.py

@@ -5,8 +5,14 @@ from sanic import Blueprint, Websocket, response  # type: ignore[attr-defined]
 from sanic.request import Request
 from sanic.response import HTTPResponse
 
-from rasa.core.channels.channel import InputChannel, OutputChannel, UserMessage
+from rasa.core.channels.channel import (
+    InputChannel,
+    OutputChannel,
+    UserMessage,
+    requires_basic_auth,
+)
 from rasa.core.channels.voice_ready.jambonz_protocol import (
+    CHANNEL_NAME,
     send_ws_hangup_message,
     send_ws_text_message,
     websocket_message_handler,
@@ -18,8 +24,6 @@ from rasa.utils.io import remove_emojis
 
 structlogger = structlog.get_logger()
 
-CHANNEL_NAME = "jambonz"
-
 DEFAULT_HANGUP_DELAY_SECONDS = 1
 
 
@@ -32,12 +36,27 @@ class JambonzVoiceReadyInput(InputChannel):
 
     @classmethod
     def from_credentials(cls, credentials: Optional[Dict[Text, Any]]) -> InputChannel:
-        return cls()
+        if not credentials:
+            return cls()
+
+        username = credentials.get("username")
+        password = credentials.get("password")
+        if (username is None) != (password is None):
+            raise RasaException(
+                "In Jambonz channel, either both username and password "
+                "or neither should be provided. "
+            )
 
-    def __init__(self) -> None:
+        return cls(username, password)
+
+    def __init__(
+        self, username: Optional[Text] = None, password: Optional[Text] = None
+    ) -> None:
         """Initializes the JambonzVoiceReadyInput channel."""
         mark_as_beta_feature("Jambonz Channel")
         validate_voice_license_scope()
+        self.username = username
+        self.password = password
 
     def blueprint(
         self, on_new_message: Callable[[UserMessage], Awaitable[Any]]
@@ -50,6 +69,7 @@ class JambonzVoiceReadyInput(InputChannel):
             return response.json({"status": "ok"})
 
         @jambonz_webhook.websocket("/websocket", subprotocols=["ws.jambonz.org"])  # type: ignore
+        @requires_basic_auth(self.username, self.password)
         async def websocket(request: Request, ws: Websocket) -> None:
             """Triggered on new websocket connection."""
             async for message in ws:

rasa/core/channels/voice_ready/jambonz_protocol.py

@@ -10,6 +10,7 @@ from rasa.core.channels.channel import UserMessage
 from rasa.core.channels.voice_ready.utils import CallParameters
 
 structlogger = structlog.get_logger()
+CHANNEL_NAME = "jambonz"
 
 
 @dataclass
@@ -206,6 +207,7 @@ async def handle_new_session(
         output_channel=output_channel,
         sender_id=message.call_sid,
         metadata=asdict(message.call_params),
+        input_channel=CHANNEL_NAME,
     )
     await send_config_ack(message.message_id, ws)
     await on_new_message(user_msg)
@@ -238,6 +240,7 @@ async def handle_gather_completed(
         output_channel = JambonzWebsocketOutput(ws, transcript_result.call_sid)
         user_msg = UserMessage(
             text=most_likely_transcript.text,
+            input_channel=CHANNEL_NAME,
             output_channel=output_channel,
             sender_id=transcript_result.call_sid,
             metadata={},
@@ -288,6 +291,7 @@ async def handle_call_status(
         output_channel = JambonzWebsocketOutput(ws, call_status.call_sid)
         user_msg = UserMessage(
             text="/session_end",
+            input_channel=CHANNEL_NAME,
             output_channel=output_channel,
             sender_id=call_status.call_sid,
             metadata={},

rasa/core/information_retrieval/faiss.py

@@ -31,12 +31,10 @@ class FAISS_Store(InformationRetrieval):
         index_path: str,
         docs_folder: Optional[str],
         create_index: Optional[bool] = False,
-        use_llm: bool = False,
     ):
         """Initializes the FAISS Store."""
         self.chunk_size = 1000
         self.chunk_overlap = 20
-        self.use_llm = use_llm
 
         path = Path(index_path) / "documents_faiss"
         if create_index:
@@ -73,57 +71,6 @@ class FAISS_Store(InformationRetrieval):
 
         return loader.load()
 
-    def _format_faqs(self, docs: List["Document"]) -> List["Document"]:
-        """Splits each loaded file into individual FAQs.
-
-        Args:
-            docs: Documents representing whole files containing FAQs.
-
-        Returns:
-            List of Document objects, each containing a separate FAQ.
-
-        Examples:
-            An example of a file containing FAQs:
-
-            Q: Who is Finley?
-            A: Finley is your smart assistant for the FinX App. You can add him to your
-               favorite messenger and tell him what you need help with.
-
-            Q: How does Finley work?
-            A: Finley is powered by the latest chatbot technology leveraging a unique
-               interplay of large language models and secure logic.
-
-        More details in documentation: https://rasa.com/docs/reference/config/policies/extractive-search/
-        """
-        structured_faqs = []
-        from langchain.schema import Document
-
-        for doc in docs:
-            faq_chunks = doc.page_content.strip().split("\n\n")
-
-            for chunk in faq_chunks:
-                lines = chunk.strip().split("\n")
-                if len(lines) < 2:
-                    continue  # Skip if something unexpected
-
-                question_line = lines[0].strip()
-                answer_line = lines[1].strip()
-
-                question = question_line.replace("Q: ", "").strip()
-                answer = answer_line.replace("A: ", "").strip()
-
-                doc_obj = Document(
-                    page_content=question,
-                    metadata={
-                        "title": question.lower().replace(" ", "_")[:-1],
-                        "type": "faq",
-                        "answer": answer,
-                    },
-                )
-
-                structured_faqs.append(doc_obj)
-        return structured_faqs
-
     def _create_document_index(
         self, docs_folder: Optional[str], embedding: "Embeddings"
     ) -> FAISS:
@@ -140,15 +87,12 @@ class FAISS_Store(InformationRetrieval):
             raise ValueError("parameter `docs_folder` needs to be specified")
 
         docs = self.load_documents(docs_folder)
-        if self.use_llm:
-            splitter = RecursiveCharacterTextSplitter(
-                chunk_size=self.chunk_size,
-                chunk_overlap=self.chunk_overlap,
-                length_function=len,
-            )
-            doc_chunks = splitter.split_documents(docs)
-        else:
-            doc_chunks = self._format_faqs(docs)
+        splitter = RecursiveCharacterTextSplitter(
+            chunk_size=self.chunk_size,
+            chunk_overlap=self.chunk_overlap,
+            length_function=len,
+        )
+        doc_chunks = splitter.split_documents(docs)
 
         logger.info(
             "information_retrieval.faiss_store._create_document_index",
@@ -169,15 +113,10 @@ class FAISS_Store(InformationRetrieval):
         pass
 
     async def search(
-        self,
-        query: Text,
-        tracker_state: Dict[str, Any],
-        threshold: float = 0.0,
-        k: int = 1,
+        self, query: Text, tracker_state: Dict[str, Any], threshold: float = 0.0
     ) -> SearchResultList:
         logger.debug("information_retrieval.faiss_store.search", query=query)
         try:
-            # TODO: make use of k
             documents = await self.index.as_retriever().ainvoke(query)
         except Exception as exc:
             raise InformationRetrievalException from exc

rasa/core/information_retrieval/information_retrieval.py

@@ -36,19 +36,6 @@ class SearchResult:
         """Construct a SearchResult object from Langchain Document object."""
         return cls(text=document.page_content, metadata=document.metadata)
 
-    @classmethod
-    def from_dict(cls, data: dict[str, Any]) -> "SearchResult":
-        """Construct a SearchResult object from a JSON object."""
-        return cls(text=data["text"], metadata=data["metadata"], score=data["score"])
-
-    def to_dict(self) -> dict[str, Any]:
-        """Convert the SearchResult object to a dictionary."""
-        return {
-            "text": self.text,
-            "metadata": self.metadata,
-            "score": self.score,
-        }
-
 
 @dataclass
 class SearchResultList:
@@ -57,7 +44,8 @@ class SearchResultList:
 
     @classmethod
     def from_document_list(cls, documents: List["Document"]) -> "SearchResultList":
-        """Convert a list of Langchain Documents to a SearchResultList object.
+        """
+        Convert a list of Langchain Documents to a SearchResultList object.
 
         Args:
             documents: List of Langchain Documents.
@@ -70,31 +58,6 @@ class SearchResultList:
             metadata={"total_results": len(documents)},
         )
 
-    @classmethod
-    def from_dict(cls, data: dict[str, Any]) -> "SearchResultList":
-        """Convert a JSON object to a SearchResultList object.
-
-        Args:
-            data: JSON object.
-
-        Returns:
-            SearchResultList object.
-        """
-        if not data:
-            return cls(results=[], metadata={})
-
-        return cls(
-            results=[SearchResult.from_dict(result) for result in data["results"]],
-            metadata=data["metadata"],
-        )
-
-    def to_dict(self) -> dict[str, Any]:
-        """Convert the SearchResultList object to a dictionary."""
-        return {
-            "results": [result.to_dict() for result in self.results],
-            "metadata": self.metadata,
-        }
-
 
 class InformationRetrievalException(RasaException):
     """Base class for exceptions raised by InformationRetrieval operations."""
@@ -126,7 +89,6 @@ class InformationRetrieval:
         query: Text,
         tracker_state: dict[str, Any],
         threshold: float = 0.0,
-        k: int = 1,
     ) -> SearchResultList:
         """Search for a document in the InformationRetrieval system."""
         raise NotImplementedError(

rasa/core/information_retrieval/milvus.py

@@ -31,25 +31,20 @@ class Milvus_Store(InformationRetrieval):
         )
 
     async def search(
-        self,
-        query: Text,
-        tracker_state: Dict[str, Any],
-        threshold: float = 0.0,
-        k: int = 1,
+        self, query: Text, tracker_state: Dict[str, Any], threshold: float = 0.0
     ) -> SearchResultList:
         """Search for documents in the Milvus store.
 
         Args:
             query: The query to search for.
             threshold: minimum similarity score to consider a document a match.
-            k: number of results to return.
 
         Returns:
         A list of documents that match the query.
         """
         logger.debug("information_retrieval.milvus_store.search", query=query)
         try:
-            hits = await self.client.asimilarity_search_with_score(query, k=k)
+            hits = await self.client.asimilarity_search_with_score(query, k=4)
         except Exception as exc:
             raise InformationRetrievalException from exc
 

rasa/core/information_retrieval/qdrant.py

@@ -66,18 +66,13 @@ class Qdrant_Store(InformationRetrieval):
         )
 
     async def search(
-        self,
-        query: Text,
-        tracker_state: Dict[str, Any],
-        threshold: float = 0.0,
-        k: int = 1,
+        self, query: Text, tracker_state: Dict[str, Any], threshold: float = 0.0
     ) -> SearchResultList:
         """Search for a document in the Qdrant vector store.
 
         Args:
             query: The query to search for.
             threshold: minimum similarity score to consider a document a match.
-            k: number of results to return.
 
         Returns:
         A list of documents that match the query.
@@ -85,7 +80,7 @@ class Qdrant_Store(InformationRetrieval):
         logger.debug("information_retrieval.qdrant_store.search", query=query)
         try:
             hits = await self.client.asimilarity_search(
-                query, k=k, score_threshold=threshold
+                query, k=4, score_threshold=threshold
             )
         except ValidationError as e:
             raise PayloadNotFoundException(

rasa/core/nlg/contextual_response_rephraser.py

@@ -27,6 +27,7 @@ from rasa.shared.nlu.constants import (
     PROMPTS,
 )
 from rasa.shared.providers.llm.llm_response import LLMResponse, measure_llm_latency
+from rasa.shared.utils.constants import LOG_COMPONENT_SOURCE_METHOD_INIT
 from rasa.shared.utils.health_check.llm_health_check_mixin import LLMHealthCheckMixin
 from rasa.shared.utils.llm import (
     DEFAULT_OPENAI_GENERATE_MODEL_NAME,
@@ -105,6 +106,8 @@ class ContextualResponseRephraser(
         self.prompt_template = get_prompt_template(
             self.nlg_endpoint.kwargs.get(PROMPT_CONFIG_KEY),
             DEFAULT_RESPONSE_VARIATION_PROMPT_TEMPLATE,
+            log_source_component=ContextualResponseRephraser.__name__,
+            log_source_method=LOG_COMPONENT_SOURCE_METHOD_INIT,
         )
         self.rephrase_all = self.nlg_endpoint.kwargs.get(
             "rephrase_all", DEFAULT_REPHRASE_ALL