unique_toolkit 0.7.7__py3-none-any.whl → 1.23.0__py3-none-any.whl

unique_toolkit/content/service.py

@@ -1,11 +1,15 @@
 import logging
 from pathlib import Path
+from typing import Any, overload
 
+import unique_sdk
 from requests import Response
 from typing_extensions import deprecated
 
+from unique_toolkit._common.utils.files import is_file_content, is_image_content
 from unique_toolkit._common.validate_required_values import validate_required_values
 from unique_toolkit.app.schemas import BaseEvent, ChatEvent, Event
+from unique_toolkit.app.unique_settings import UniqueSettings
 from unique_toolkit.content import DOMAIN_NAME
 from unique_toolkit.content.constants import DEFAULT_SEARCH_LANGUAGE
 from unique_toolkit.content.functions import (
@@ -30,16 +34,34 @@ from unique_toolkit.content.schemas import (
 logger = logging.getLogger(f"toolkit.{DOMAIN_NAME}.{__name__}")
 
 
+@deprecated("Use KnowledgeBaseService instead")
 class ContentService:
     """
     Provides methods for searching, downloading and uploading content in the knowledge base.
+    """
+
+    @deprecated(
+        "Use __init__ with company_id, user_id and chat_id instead or use the classmethod `from_event`"
+    )
+    @overload
+    def __init__(self, event: Event | ChatEvent | BaseEvent): ...
+
+    """
+        Initialize the ContentService with an event (deprecated)
+    """
 
-    Attributes:
-        event: BaseEvent | Event, this can be None ONLY if company_id and user_id are provided.
-        company_id (str): The company ID.
-        user_id (str): The user ID.
-        chat_id (str): The chat ID. Defaults to None
-        metadata_filter (dict | None): is only initialised from an Event(Deprecated) or ChatEvent.
+    @overload
+    def __init__(
+        self,
+        *,
+        company_id: str,
+        user_id: str,
+        chat_id: str | None = None,
+        metadata_filter: dict | None = None,
+    ): ...
+
+    """
+        Initialize the ContentService with a company_id, user_id and chat_id and metadata_filter.
     """
 
     def __init__(
@@ -48,20 +70,66 @@ class ContentService:
         company_id: str | None = None,
         user_id: str | None = None,
         chat_id: str | None = None,
+        metadata_filter: dict | None = None,
     ):
+        """
+        Initialize the ContentService with a company_id, user_id and chat_id.
+        """
+
         self._event = event  # Changed to protected attribute
-        self.metadata_filter = None
+        self._metadata_filter = None
         if event:
-            self.company_id = event.company_id
-            self.user_id = event.user_id
+            self._company_id: str = event.company_id
+            self._user_id: str = event.user_id
             if isinstance(event, (ChatEvent, Event)):
-                self.metadata_filter = event.payload.metadata_filter
-                self.chat_id = event.payload.chat_id
+                self._metadata_filter = event.payload.metadata_filter
+                self._chat_id: str | None = event.payload.chat_id
         else:
             [company_id, user_id] = validate_required_values([company_id, user_id])
-            self.company_id = company_id
-            self.user_id = user_id
-            self.chat_id = chat_id
+            self._company_id: str = company_id
+            self._user_id: str = user_id
+            self._chat_id: str | None = chat_id
+            self._metadata_filter = metadata_filter
+
+    @classmethod
+    def from_event(cls, event: Event | ChatEvent | BaseEvent):
+        """
+        Initialize the ContentService with an event.
+        """
+        chat_id = None
+        metadata_filter = None
+
+        if isinstance(event, (ChatEvent | Event)):
+            chat_id = event.payload.chat_id
+            metadata_filter = event.payload.metadata_filter
+
+        return cls(
+            company_id=event.company_id,
+            user_id=event.user_id,
+            chat_id=chat_id,
+            metadata_filter=metadata_filter,
+        )
+
+    @classmethod
+    def from_settings(
+        cls,
+        settings: UniqueSettings | str | None = None,
+        metadata_filter: dict | None = None,
+    ):
+        """
+        Initialize the ContentService with a settings object and metadata filter.
+        """
+
+        if settings is None:
+            settings = UniqueSettings.from_env_auto_with_sdk_init()
+        elif isinstance(settings, str):
+            settings = UniqueSettings.from_env_auto_with_sdk_init(filename=settings)
+
+        return cls(
+            company_id=settings.auth.company_id.get_secret_value(),
+            user_id=settings.auth.user_id.get_secret_value(),
+            metadata_filter=metadata_filter,
+        )
 
     @property
     @deprecated(
@@ -76,6 +144,110 @@ class ContentService:
         """
         return self._event
 
+    @property
+    @deprecated(
+        "The company_id property is deprecated and will be removed in a future version."
+    )
+    def company_id(self) -> str | None:
+        """
+        Get the company identifier (deprecated).
+
+        Returns:
+            str | None: The company identifier.
+        """
+        return self._company_id
+
+    @company_id.setter
+    @deprecated(
+        "The company_id setter is deprecated and will be removed in a future version."
+    )
+    def company_id(self, value: str) -> None:
+        """
+        Set the company identifier (deprecated).
+
+        Args:
+            value (str | None): The company identifier.
+        """
+        self._company_id = value
+
+    @property
+    @deprecated(
+        "The user_id property is deprecated and will be removed in a future version."
+    )
+    def user_id(self) -> str | None:
+        """
+        Get the user identifier (deprecated).
+
+        Returns:
+            str | None: The user identifier.
+        """
+        return self._user_id
+
+    @user_id.setter
+    @deprecated(
+        "The user_id setter is deprecated and will be removed in a future version."
+    )
+    def user_id(self, value: str) -> None:
+        """
+        Set the user identifier (deprecated).
+
+        Args:
+            value (str | None): The user identifier.
+        """
+        self._user_id = value
+
+    @property
+    @deprecated(
+        "The chat_id property is deprecated and will be removed in a future version."
+    )
+    def chat_id(self) -> str | None:
+        """
+        Get the chat identifier (deprecated).
+
+        Returns:
+            str | None: The chat identifier.
+        """
+        return self._chat_id
+
+    @chat_id.setter
+    @deprecated(
+        "The chat_id setter is deprecated and will be removed in a future version."
+    )
+    def chat_id(self, value: str | None) -> None:
+        """
+        Set the chat identifier (deprecated).
+
+        Args:
+            value (str | None): The chat identifier.
+        """
+        self._chat_id = value
+
+    @property
+    @deprecated(
+        "The metadata_filter property is deprecated and will be removed in a future version."
+    )
+    def metadata_filter(self) -> dict | None:
+        """
+        Get the metadata filter (deprecated).
+
+        Returns:
+            dict | None: The metadata filter.
+        """
+        return self._metadata_filter
+
+    @metadata_filter.setter
+    @deprecated(
+        "The metadata_filter setter is deprecated and will be removed in a future version."
+    )
+    def metadata_filter(self, value: dict | None) -> None:
+        """
+        Set the metadata filter (deprecated).
+
+        Args:
+            value (dict | None): The metadata filter.
+        """
+        self._metadata_filter = value
+
     def search_content_chunks(
         self,
         search_string: str,
@@ -88,6 +260,7 @@ class ContentService:
         chat_only: bool | None = None,
         metadata_filter: dict | None = None,
         content_ids: list[str] | None = None,
+        score_threshold: float | None = None,
     ) -> list[ContentChunk]:
         """
         Performs a synchronous search for content chunks in the knowledge base.
@@ -103,6 +276,7 @@ class ContentService:
             chat_only (bool | None, optional): Whether to search only in the current chat. Defaults to None.
             metadata_filter (dict | None, optional): UniqueQL metadata filter. If unspecified/None, it tries to use the metadata filter from the event. Defaults to None.
             content_ids (list[str] | None, optional): The content IDs to search within. Defaults to None.
+            score_threshold (float | None, optional): Sets the minimum similarity score for search results to be considered. Defaults to 0.
 
         Returns:
             list[ContentChunk]: The search results.
@@ -112,17 +286,17 @@ class ContentService:
         """
 
         if metadata_filter is None:
-            metadata_filter = self.metadata_filter
+            metadata_filter = self._metadata_filter
 
-        chat_id = chat_id or self.chat_id  # type: ignore
+        chat_id = chat_id or self._chat_id  # type: ignore
 
         if chat_only and not chat_id:
             raise ValueError("Please provide chat_id when limiting with chat_only")
 
         try:
             searches = search_content_chunks(
-                user_id=self.user_id,
-                company_id=self.company_id,
+                user_id=self._user_id,
+                company_id=self._company_id,
                 chat_id=chat_id,
                 search_string=search_string,
                 search_type=search_type,
@@ -133,12 +307,14 @@ class ContentService:
                 chat_only=chat_only,
                 metadata_filter=metadata_filter,
                 content_ids=content_ids,
+                score_threshold=score_threshold,
             )
             return searches
         except Exception as e:
             logger.error(f"Error while searching content chunks: {e}")
             raise e
 
+    @deprecated("Use search_chunks_async instead")
     async def search_content_chunks_async(
         self,
         search_string: str,
@@ -151,6 +327,7 @@ class ContentService:
         chat_only: bool | None = None,
         metadata_filter: dict | None = None,
         content_ids: list[str] | None = None,
+        score_threshold: float | None = None,
     ):
         """
         Performs an asynchronous search for content chunks in the knowledge base.
@@ -166,6 +343,7 @@ class ContentService:
             chat_only (bool | None, optional): Whether to search only in the current chat. Defaults to None.
             metadata_filter (dict | None, optional): UniqueQL metadata filter. If unspecified/None, it tries to use the metadata filter from the event. Defaults to None.
             content_ids (list[str] | None, optional): The content IDs to search within. Defaults to None.
+            score_threshold (float | None, optional): Sets the minimum similarity score for search results to be considered. Defaults to 0.
 
         Returns:
             list[ContentChunk]: The search results.
@@ -174,17 +352,17 @@ class ContentService:
             Exception: If there's an error during the search operation.
         """
         if metadata_filter is None:
-            metadata_filter = self.metadata_filter
+            metadata_filter = self._metadata_filter
 
-        chat_id = chat_id or self.chat_id  # type: ignore
+        chat_id = chat_id or self._chat_id  # type: ignore
 
         if chat_only and not chat_id:
             raise ValueError("Please provide chat_id when limiting with chat_only.")
 
         try:
             searches = await search_content_chunks_async(
-                user_id=self.user_id,
-                company_id=self.company_id,
+                user_id=self._user_id,
+                company_id=self._company_id,
                 chat_id=chat_id,
                 search_string=search_string,
                 search_type=search_type,
@@ -195,6 +373,7 @@ class ContentService:
                 chat_only=chat_only,
                 metadata_filter=metadata_filter,
                 content_ids=content_ids,
+                score_threshold=score_threshold,
             )
             return searches
         except Exception as e:
@@ -216,11 +395,11 @@ class ContentService:
         Returns:
             list[Content]: The search results.
         """
-        chat_id = chat_id or self.chat_id  # type: ignore
+        chat_id = chat_id or self._chat_id  # type: ignore
 
         return search_contents(
-            user_id=self.user_id,
-            company_id=self.company_id,
+            user_id=self._user_id,
+            company_id=self._company_id,
             chat_id=chat_id,
             where=where,
         )
@@ -239,11 +418,11 @@ class ContentService:
         Returns:
             list[Content]: The search results.
         """
-        chat_id = chat_id or self.chat_id  # type: ignore
+        chat_id = chat_id or self._chat_id  # type: ignore
 
         return await search_contents_async(
-            user_id=self.user_id,
-            company_id=self.company_id,
+            user_id=self._user_id,
+            company_id=self._company_id,
             chat_id=chat_id,
             where=where,
         )
@@ -261,6 +440,9 @@ class ContentService:
         scope_id: str | None = None,
         chat_id: str | None = None,
         skip_ingestion: bool = False,
+        skip_excel_ingestion: bool = False,
+        ingestion_config: unique_sdk.Content.IngestionConfig | None = None,
+        metadata: dict | None = None,
     ) -> Content:
         """
         Uploads content to the knowledge base.
@@ -272,20 +454,25 @@ class ContentService:
             scope_id (str | None): The scope ID. Defaults to None.
             chat_id (str | None): The chat ID. Defaults to None.
             skip_ingestion (bool): Whether to skip ingestion. Defaults to False.
+            skip_excel_ingestion (bool): Whether to skip excel ingestion. Defaults to False.
+            ingestion_config (unique_sdk.Content.IngestionConfig | None): The ingestion configuration. Defaults to None.
+            metadata (dict | None): The metadata to associate with the content. Defaults to None.
 
         Returns:
             Content: The uploaded content.
         """
 
         return upload_content_from_bytes(
-            user_id=self.user_id,
-            company_id=self.company_id,
+            user_id=self._user_id,
+            company_id=self._company_id,
             content=content,
             content_name=content_name,
             mime_type=mime_type,
             scope_id=scope_id,
             chat_id=chat_id,
             skip_ingestion=skip_ingestion,
+            ingestion_config=ingestion_config,
+            metadata=metadata,
         )
 
     def upload_content(
@@ -296,6 +483,9 @@ class ContentService:
         scope_id: str | None = None,
         chat_id: str | None = None,
         skip_ingestion: bool = False,
+        skip_excel_ingestion: bool = False,
+        ingestion_config: unique_sdk.Content.IngestionConfig | None = None,
+        metadata: dict[str, Any] | None = None,
     ):
         """
         Uploads content to the knowledge base.
@@ -307,20 +497,26 @@ class ContentService:
             scope_id (str | None): The scope ID. Defaults to None.
             chat_id (str | None): The chat ID. Defaults to None.
             skip_ingestion (bool): Whether to skip ingestion. Defaults to False.
+            skip_excel_ingestion (bool): Whether to skip excel ingestion. Defaults to False.
+            ingestion_config (unique_sdk.Content.IngestionConfig | None): The ingestion configuration. Defaults to None.
+            metadata (dict[str, Any] | None): The metadata to associate with the content. Defaults to None.
 
         Returns:
             Content: The uploaded content.
         """
 
         return upload_content(
-            user_id=self.user_id,
-            company_id=self.company_id,
+            user_id=self._user_id,
+            company_id=self._company_id,
             path_to_content=path_to_content,
             content_name=content_name,
             mime_type=mime_type,
             scope_id=scope_id,
             chat_id=chat_id,
             skip_ingestion=skip_ingestion,
+            skip_excel_ingestion=skip_excel_ingestion,
+            ingestion_config=ingestion_config,
+            metadata=metadata,
         )
 
     def request_content_by_id(
@@ -339,11 +535,11 @@ class ContentService:
             requests.Response: The response object containing the downloaded content.
 
         """
-        chat_id = chat_id or self.chat_id  # type: ignore
+        chat_id = chat_id or self._chat_id  # type: ignore
 
         return request_content_by_id(
-            user_id=self.user_id,
-            company_id=self.company_id,
+            user_id=self._user_id,
+            company_id=self._company_id,
             content_id=content_id,
             chat_id=chat_id,
         )
@@ -371,11 +567,11 @@ class ContentService:
             Exception: If the download fails or the filename cannot be determined.
         """
 
-        chat_id = chat_id or self.chat_id  # type: ignore
+        chat_id = chat_id or self._chat_id  # type: ignore
 
         return download_content_to_file_by_id(
-            user_id=self.user_id,
-            company_id=self.company_id,
+            user_id=self._user_id,
+            company_id=self._company_id,
             content_id=content_id,
             chat_id=chat_id,
             filename=filename,
@@ -406,11 +602,11 @@ class ContentService:
             Exception: If the download fails.
         """
 
-        chat_id = chat_id or self.chat_id  # type: ignore
+        chat_id = chat_id or self._chat_id  # type: ignore
 
         return download_content(
-            user_id=self.user_id,
-            company_id=self.company_id,
+            user_id=self._user_id,
+            company_id=self._company_id,
             content_id=content_id,
             content_name=content_name,
             chat_id=chat_id,
@@ -435,10 +631,48 @@ class ContentService:
         Raises:
             Exception: If the download fails.
         """
-        chat_id = chat_id or self.chat_id  # type: ignore
+        chat_id = chat_id or self._chat_id  # type: ignore
         return download_content_to_bytes(
-            user_id=self.user_id,
-            company_id=self.company_id,
+            user_id=self._user_id,
+            company_id=self._company_id,
             content_id=content_id,
             chat_id=chat_id,
         )
+
+    def get_documents_uploaded_to_chat(self) -> list[Content]:
+        chat_contents = self.search_contents(
+            where={
+                "ownerId": {
+                    "equals": self._chat_id,
+                },
+            },
+        )
+
+        content: list[Content] = []
+        for c in chat_contents:
+            if self.is_file_content(c.key):
+                content.append(c)
+
+        return content
+
+    def get_images_uploaded_to_chat(self) -> list[Content]:
+        chat_contents = self.search_contents(
+            where={
+                "ownerId": {
+                    "equals": self._chat_id,
+                },
+            },
+        )
+
+        content: list[Content] = []
+        for c in chat_contents:
+            if self.is_image_content(c.key):
+                content.append(c)
+
+        return content
+
+    def is_file_content(self, filename: str) -> bool:
+        return is_file_content(filename=filename)
+
+    def is_image_content(self, filename: str) -> bool:
+        return is_image_content(filename=filename)