vector-vault 7.3.2__tar.gz → 7.3.4__tar.gz

{vector_vault-7.3.2 → vector_vault-7.3.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vector_vault
-Version: 7.3.2
+Version: 7.3.4
 Summary: Quickly create RAG apps, Agents, and Unleash the full power of AI with Vector Vault
 Home-page: https://github.com/John-Rood/VectorVault
 Author: VectorVault.io

{vector_vault-7.3.2 → vector_vault-7.3.4}/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 setup(
     name="vector_vault",
-    version="7.3.2",
+    version="7.3.4",
     packages=find_packages(),
     author="VectorVault.io",
     author_email="john@johnrood.com",

{vector_vault-7.3.2 → vector_vault-7.3.4}/vector_vault.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vector-vault
-Version: 7.3.2
+Version: 7.3.4
 Summary: Quickly create RAG apps, Agents, and Unleash the full power of AI with Vector Vault
 Home-page: https://github.com/John-Rood/VectorVault
 Author: VectorVault.io

{vector_vault-7.3.2 → vector_vault-7.3.4}/vectorvault/ai.py

@@ -1158,7 +1158,7 @@ class LLMClient:
 
         return {'text': text, 'history': history, 'context': context}
 
-    def llm(self, user_input: str = '', history: str = '', model=None, custom_prompt=False, temperature=0, timeout=None, max_retries=5):
+    def text_llm(self, user_input: str = '', history: str = '', model=None, custom_prompt=False, temperature=0, timeout=None, max_retries=5):
         timeout = self.timeout if not timeout else timeout
         prompt_template = custom_prompt if custom_prompt else self.prompt
 
@@ -1205,7 +1205,35 @@ class LLMClient:
             )
         else:
             # Otherwise use regular text LLM
-            return self.llm(
+            return self.text_llm(
+                user_input=user_input,
+                history=history,
+                model=model,
+                custom_prompt=custom_prompt,
+                temperature=temperature,
+                timeout=timeout,
+                max_retries=max_retries
+            )
+        
+    def llm(self, user_input: str = '', history: str = '', model=None, custom_prompt=False, temperature=0, timeout=None, max_retries=5, image_path=None, image_url=None):
+        """
+        Smart LLM method that automatically switches between text-only LLM and image inference
+        based on whether image parameters are provided.
+        """
+        # If image parameters are provided, use image inference
+        if image_path or image_url:
+            return self.image_inference(
+                image_path=image_path,
+                image_url=image_url,
+                user_text=user_input,
+                model=model,
+                stream=False,
+                temperature=temperature,
+                timeout=timeout
+            )
+        else:
+            # Otherwise use regular text LLM
+            return self.text_llm(
                 user_input=user_input,
                 history=history,
                 model=model,

{vector_vault-7.3.2 → vector_vault-7.3.4}/vectorvault/vault.py

@@ -25,7 +25,7 @@ import traceback
 import random
 from threading import Thread as T
 from datetime import datetime, timedelta
-from typing import List, Union
+from typing import List, Union, Dict
 from .ai import openai, OpenAIPlatform, AnthropicPlatform, GrokPlatform, GeminiPlatform, CerebrasPlatform, LLMClient, get_all_models
 from .cloud_api import call_cloud_save, run_flow, run_flow_stream
 from .cloudmanager import CloudManager, VaultStorageManager, as_completed, ThreadPoolExecutor
@@ -124,6 +124,64 @@ class Vault:
     Question: {content}
     """    
         self.personality_message = personality_message if personality_message else ""
+        
+        # Async prefetch for custom prompts and personality message
+        self._prefetch_executor = ThreadPoolExecutor(max_workers=3)
+        self._personality_future = None
+        self._custom_prompt_future = None
+        self._custom_prompt_no_context_future = None
+        
+        # Start async prefetch if user and api_key are provided
+        if user and api_key:
+            self._start_prefetch()
+
+    def _start_prefetch(self):
+        """Start async prefetch of personality message and custom prompts"""
+        try:
+            # Submit async tasks to prefetch data
+            self._personality_future = self._prefetch_executor.submit(self._fetch_personality_from_cloud)
+            self._custom_prompt_future = self._prefetch_executor.submit(self._fetch_custom_prompt_from_cloud, True)
+            self._custom_prompt_no_context_future = self._prefetch_executor.submit(self._fetch_custom_prompt_from_cloud, False)
+        except Exception as e:
+            if self.verbose:
+                print(f"Could not start prefetch: {e}")
+
+    def prefetch_prompts(self):
+        """
+        Manually trigger prefetch of personality message and custom prompts.
+        Useful if you want to explicitly control when the async fetch begins.
+        """
+        if not self._personality_future or not self._custom_prompt_future:
+            self._start_prefetch()
+            if self.verbose:
+                print("Prefetch started for personality and custom prompts")
+        elif self.verbose:
+            print("Prefetch already in progress")
+
+    def __del__(self):
+        """Cleanup executor on deletion"""
+        try:
+            if hasattr(self, '_prefetch_executor'):
+                self._prefetch_executor.shutdown(wait=False)
+        except:
+            pass
+
+    def _fetch_personality_from_cloud(self):
+        """Internal method to fetch personality message from cloud"""
+        try:
+            return self.cloud_manager.download_text_from_cloud(f'{self.vault}/personality_message')
+        except:
+            return None
+
+    def _fetch_custom_prompt_from_cloud(self, context=True):
+        """Internal method to fetch custom prompt from cloud"""
+        try:
+            if context:
+                return self.cloud_manager.download_text_from_cloud(f'{self.vault}/prompt')
+            else:
+                return self.cloud_manager.download_text_from_cloud(f'{self.vault}/no_context_prompt')
+        except:
+            return None
 
     @property
     def cloud_manager(self):
@@ -364,8 +422,20 @@ class Vault:
 
     def fetch_personality_message(self):
         '''
-            Retrieves personality_message from the vault if it is there or else use the defualt
+            Retrieves personality_message from the vault if it is there or else use the defualt.
+            Uses prefetched result if available for faster access.
         '''
+        # Check if we have a prefetch future and get the result
+        if self._personality_future is not None:
+            try:
+                personality_message = self._personality_future.result(timeout=10)
+                if personality_message is not None:
+                    return personality_message
+            except Exception as e:
+                if self.verbose:
+                    print(f"Could not retrieve prefetched personality: {e}")
+        
+        # Fallback to synchronous fetch
         try:
             personality_message = self.cloud_manager.download_text_from_cloud(f'{self.vault}/personality_message')
         except:
@@ -390,8 +460,21 @@ class Vault:
     def fetch_custom_prompt(self, context=True):
         '''
             Retrieves custom_prompt from the vault if there or eles use defualt - (used for get_context = True responses)
-            context == False will return custom prompt for not context situations
+            context == False will return custom prompt for not context situations.
+            Uses prefetched result if available for faster access.
         '''
+        # Check if we have a prefetch future and get the result
+        future = self._custom_prompt_future if context else self._custom_prompt_no_context_future
+        if future is not None:
+            try:
+                prompt = future.result(timeout=10)
+                if prompt is not None:
+                    return prompt
+            except Exception as e:
+                if self.verbose:
+                    print(f"Could not retrieve prefetched custom prompt: {e}")
+        
+        # Fallback to synchronous fetch
         try:
             prompt = self.cloud_manager.download_text_from_cloud(f'{self.vault}/prompt') if context else self.cloud_manager.download_text_from_cloud(f'{self.vault}/no_context_prompt')
         except:
@@ -1402,22 +1485,22 @@ class Vault:
             
 
     def get_chat(self, 
-            text: str = None, 
-            history: str = '', 
-            summary: bool = False, 
-            get_context: bool = False, 
-            n_context: int = 4, 
-            return_context: bool = False, 
-            history_search: bool = False, 
-            smart_history_search: bool = False, 
-            model: str = None, 
-            include_context_meta: bool = False, 
-            custom_prompt: bool = False, 
-            temperature: int = 0, 
-            timeout: int = 300,
-            image_path: str = None,
-            image_url: str = None,
-            vaults: List[str] = None,
+            text: str = None, # The text to send to the LLM
+            history: str = '', # The chat history to send to the LLM
+            summary: bool = False, # Whether or not the LLM should return a summary of the text
+            get_context: bool = False, # Whether or not the LLM should get context from the vault before responding
+            n_context: int = 4, # How many items to return from the vault
+            return_context: bool = False, # Whether or not the LLM should return the context in the response
+            history_search: bool = False, # Whether or not the LLM should search the vector database with chat history as well as the text input
+            smart_history_search: bool = False, # Whether or not the LLM should search the chat history for context using a custom prompt
+            model: str = None, # The model to use for the LLM
+            include_context_meta: bool = False, # Whether or not the LLM should see the context meta data in the context
+            custom_prompt: bool = False, # Optionally inject your own custom prompt here
+            temperature: int = 0, # The temperature to use for the LLM
+            timeout: int = 300, # How long to wait for the LLM to respond
+            image_path: str = None, # The path to an image to send to the LLM
+            image_url: str = None, # The url of an image to send to the LLM
+            vaults: Union[None, str, List[str], Dict] = None, # A vault name, list of vaults, or dict of vaults to search for context in
             ):
         '''
             Chat get response from OpenAI's ChatGPT. 
@@ -1544,25 +1627,25 @@ class Vault:
         
 
     def get_chat_stream(self, 
-            text: str = None, 
-            history: str = '', 
-            summary: bool = False, 
-            get_context: bool = False,
-            n_context: int = 4, 
-            return_context: bool = False, 
-            history_search: bool = False, 
-            smart_history_search: bool = False,
-            model: str = None, 
-            include_context_meta: bool = False, 
-            metatag: bool = False,
-            metatag_prefixes: bool = False, 
-            metatag_suffixes: bool = False, 
-            custom_prompt: bool = False, 
-            temperature: int = 0, 
-            timeout: int = 300,
-            image_path: str = None,
-            image_url: str = None,
-            vaults: List[str] = None,
+            text: str = None, # The text to send to the LLM
+            history: str = '', # The chat history to send to the LLM
+            summary: bool = False, # Whether or not the LLM should return a summary of the text
+            get_context: bool = False, # Whether or not the LLM should get context from the vault before responding
+            n_context: int = 4, # How many items to return from the vault
+            return_context: bool = False, # Whether or not the LLM should return the context in the response
+            history_search: bool = False, # Whether or not the LLM should search the vector database with chat history as well as the text input
+            smart_history_search: bool = False, # Whether or not the LLM should search the chat history for context using a custom prompt
+            model: str = None, # The model to use for the LLM
+            include_context_meta: bool = False, # Whether or not the LLM should see the context meta data in the context
+            metatag: bool = False, # Legacy parameter, do not use
+            metatag_prefixes: bool = False, # Legacy parameter, do not use
+            metatag_suffixes: bool = False, # Legacy parameter, do not use
+            custom_prompt: bool = False, # Optionally inject your own custom prompt here
+            temperature: int = 0, # The temperature to use for the LLM
+            timeout: int = 300, # How long to wait for the LLM to respond
+            image_path: str = None, # The path to an image to send to the LLM
+            image_url: str = None, # The url of an image to send to the LLM
+            vaults: Union[None, str, List[str], Dict] = None, # A vault name, list of vaults, or dict of vaults to search for context in
             ):
         '''
             Always use this get_chat_stream() wrapped by either print_stream(), or cloud_stream()