khoj 1.20.4.dev13__py3-none-any.whl → 1.20.5.dev17__py3-none-any.whl

khoj/processor/conversation/offline/chat_model.py

@@ -7,7 +7,7 @@ from typing import Any, Iterator, List, Union
 from langchain.schema import ChatMessage
 from llama_cpp import Llama
 
-from khoj.database.models import Agent
+from khoj.database.models import Agent, KhojUser
 from khoj.processor.conversation import prompts
 from khoj.processor.conversation.offline.utils import download_model
 from khoj.processor.conversation.utils import (
@@ -30,7 +30,9 @@ def extract_questions_offline(
     use_history: bool = True,
     should_extract_questions: bool = True,
     location_data: LocationData = None,
+    user: KhojUser = None,
     max_prompt_size: int = None,
+    temperature: float = 0.7,
 ) -> List[str]:
     """
     Infer search queries to retrieve relevant notes to answer user query
@@ -45,6 +47,7 @@ def extract_questions_offline(
     offline_chat_model = loaded_model or download_model(model, max_tokens=max_prompt_size)
 
     location = f"{location_data.city}, {location_data.region}, {location_data.country}" if location_data else "Unknown"
+    username = prompts.user_name.format(name=user.get_full_name()) if user and user.get_full_name() else ""
 
     # Extract Past User Message and Inferred Questions from Conversation Log
     chat_history = ""
@@ -64,10 +67,12 @@ def extract_questions_offline(
         chat_history=chat_history,
         current_date=today.strftime("%Y-%m-%d"),
         day_of_week=today.strftime("%A"),
+        current_month=today.strftime("%Y-%m"),
         yesterday_date=yesterday,
         last_year=last_year,
         this_year=today.year,
         location=location,
+        username=username,
     )
 
     messages = generate_chatml_messages_with_context(
@@ -77,7 +82,11 @@ def extract_questions_offline(
     state.chat_lock.acquire()
     try:
         response = send_message_to_model_offline(
-            messages, loaded_model=offline_chat_model, model=model, max_prompt_size=max_prompt_size
+            messages,
+            loaded_model=offline_chat_model,
+            model=model,
+            max_prompt_size=max_prompt_size,
+            temperature=temperature,
         )
     finally:
         state.chat_lock.release()
@@ -229,6 +238,7 @@ def send_message_to_model_offline(
     messages: List[ChatMessage],
     loaded_model=None,
     model="NousResearch/Hermes-2-Pro-Mistral-7B-GGUF",
+    temperature: float = 0.2,
     streaming=False,
     stop=[],
     max_prompt_size: int = None,
@@ -236,7 +246,9 @@ def send_message_to_model_offline(
     assert loaded_model is None or isinstance(loaded_model, Llama), "loaded_model must be of type Llama, if configured"
     offline_chat_model = loaded_model or download_model(model, max_tokens=max_prompt_size)
     messages_dict = [{"role": message.role, "content": message.content} for message in messages]
-    response = offline_chat_model.create_chat_completion(messages_dict, stop=stop, stream=streaming)
+    response = offline_chat_model.create_chat_completion(
+        messages_dict, stop=stop, stream=streaming, temperature=temperature
+    )
     if streaming:
         return response
     else:

khoj/processor/conversation/openai/gpt.py

@@ -5,7 +5,7 @@ from typing import Dict, Optional
 
 from langchain.schema import ChatMessage
 
-from khoj.database.models import Agent
+from khoj.database.models import Agent, KhojUser
 from khoj.processor.conversation import prompts
 from khoj.processor.conversation.openai.utils import (
     chat_completion_with_backoff,
@@ -24,14 +24,15 @@ def extract_questions(
     conversation_log={},
     api_key=None,
     api_base_url=None,
-    temperature=0,
-    max_tokens=100,
+    temperature=0.7,
     location_data: LocationData = None,
+    user: KhojUser = None,
 ):
     """
     Infer search queries to retrieve relevant notes to answer user query
     """
     location = f"{location_data.city}, {location_data.region}, {location_data.country}" if location_data else "Unknown"
+    username = prompts.user_name.format(name=user.get_full_name()) if user and user.get_full_name() else ""
 
     # Extract Past User Message and Inferred Questions from Conversation Log
     chat_history = "".join(
@@ -50,6 +51,7 @@ def extract_questions(
     prompt = prompts.extract_questions.format(
         current_date=today.strftime("%Y-%m-%d"),
         day_of_week=today.strftime("%A"),
+        current_month=today.strftime("%Y-%m"),
         last_new_year=last_new_year.strftime("%Y"),
         last_new_year_date=last_new_year.strftime("%Y-%m-%d"),
         current_new_year_date=current_new_year.strftime("%Y-%m-%d"),
@@ -59,6 +61,7 @@ def extract_questions(
         text=text,
         yesterday_date=(today - timedelta(days=1)).strftime("%Y-%m-%d"),
         location=location,
+        username=username,
     )
     messages = [ChatMessage(content=prompt, role="user")]
 

khoj/processor/conversation/openai/utils.py

@@ -36,7 +36,7 @@ def completion_with_backoff(
     messages, model, temperature=0, openai_api_key=None, api_base_url=None, model_kwargs=None
 ) -> str:
     client_key = f"{openai_api_key}--{api_base_url}"
-    client: openai.OpenAI = openai_clients.get(client_key)
+    client: openai.OpenAI | None = openai_clients.get(client_key)
     if not client:
         client = openai.OpenAI(
             api_key=openai_api_key,

khoj/processor/conversation/prompts.py

@@ -208,10 +208,12 @@ Construct search queries to retrieve relevant information to answer the user's q
 - Add as much context from the previous questions and answers as required into your search queries.
 - Break messages into multiple search queries when required to retrieve the relevant information.
 - Add date filters to your search queries from questions and answers when required to retrieve the relevant information.
+- When asked a meta, vague or random questions, search for a variety of broad topics to answer the user's question.
 - Share relevant search queries as a JSON list of strings. Do not say anything else.
 
 Current Date: {day_of_week}, {current_date}
 User's Location: {location}
+{username}
 
 Examples:
 Q: How was my trip to Cambodia?
@@ -238,6 +240,9 @@ Khoj: ["What kind of plants do I have?", "What issues do my plants have?"]
 Q: Who all did I meet here yesterday?
 Khoj: ["Met in {location} on {yesterday_date} dt>='{yesterday_date}' dt<'{current_date}'"]
 
+Q: Share some random, interesting experiences from this month
+Khoj: ["Exciting travel adventures from {current_month}", "Fun social events dt>='{current_month}-01' dt<'{current_date}'", "Intense emotional experiences in {current_month}"]
+
 Chat History:
 {chat_history}
 What searches will you perform to answer the following question, using the chat history as reference? Respond only with relevant search queries as a valid JSON list of strings.
@@ -254,10 +259,12 @@ Construct search queries to retrieve relevant information to answer the user's q
 - Add as much context from the previous questions and answers as required into your search queries.
 - Break messages into multiple search queries when required to retrieve the relevant information.
 - Add date filters to your search queries from questions and answers when required to retrieve the relevant information.
+- When asked a meta, vague or random questions, search for a variety of broad topics to answer the user's question.
 
 What searches will you perform to answer the users question? Respond with search queries as list of strings in a JSON object.
 Current Date: {day_of_week}, {current_date}
 User's Location: {location}
+{username}
 
 Q: How was my trip to Cambodia?
 Khoj: {{"queries": ["How was my trip to Cambodia?"]}}
@@ -279,6 +286,10 @@ Q: How many tennis balls fit in the back of a 2002 Honda Civic?
 Khoj: {{"queries": ["What is the size of a tennis ball?", "What is the trunk size of a 2002 Honda Civic?"]}}
 A: 1085 tennis balls will fit in the trunk of a Honda Civic
 
+Q: Share some random, interesting experiences from this month
+Khoj: {{"queries": ["Exciting travel adventures from {current_month}", "Fun social events dt>='{current_month}-01' dt<'{current_date}'", "Intense emotional experiences in {current_month}"]}}
+A: You had a great time at the local beach with your friends, attended a music concert and had a deep conversation with your friend, Khalid.
+
 Q: Is Bob older than Tom?
 Khoj: {{"queries": ["When was Bob born?", "What is Tom's age?"]}}
 A: Yes, Bob is older than Tom. As Bob was born on 1984-01-01 and Tom is 30 years old.
@@ -305,11 +316,13 @@ Construct search queries to retrieve relevant information to answer the user's q
 - Add as much context from the previous questions and answers as required into your search queries.
 - Break messages into multiple search queries when required to retrieve the relevant information.
 - Add date filters to your search queries from questions and answers when required to retrieve the relevant information.
+- When asked a meta, vague or random questions, search for a variety of broad topics to answer the user's question.
 
 What searches will you perform to answer the users question? Respond with a JSON object with the key "queries" mapping to a list of searches you would perform on the user's knowledge base. Just return the queries and nothing else.
 
 Current Date: {day_of_week}, {current_date}
 User's Location: {location}
+{username}
 
 Here are some examples of how you can construct search queries to answer the user's question:
 
@@ -328,6 +341,11 @@ A: I can help you live healthier and happier across work and personal life
 User: Who all did I meet here yesterday?
 Assistant: {{"queries": ["Met in {location} on {yesterday_date} dt>='{yesterday_date}' dt<'{current_date}'"]}}
 A: Yesterday's note mentions your visit to your local beach with Ram and Shyam.
+
+User: Share some random, interesting experiences from this month
+Assistant: {{"queries": ["Exciting travel adventures from {current_month}", "Fun social events dt>='{current_month}-01' dt<'{current_date}'", "Intense emotional experiences in {current_month}"]}}
+A: You had a great time at the local beach with your friends, attended a music concert and had a deep conversation with your friend, Khalid.
+
 """.strip()
 )
 
@@ -525,6 +543,7 @@ Which webpages will you need to read to answer the user's question?
 Provide web page links as a list of strings in a JSON object.
 Current Date: {current_date}
 User's Location: {location}
+{username}
 
 Here are some examples:
 History:
@@ -571,6 +590,7 @@ What Google searches, if any, will you need to perform to answer the user's ques
 Provide search queries as a list of strings in a JSON object. Do not wrap the json in a codeblock.
 Current Date: {current_date}
 User's Location: {location}
+{username}
 
 Here are some examples:
 History:

khoj/processor/embeddings.py

@@ -95,11 +95,13 @@ class CrossEncoderModel:
         model_name: str = "mixedbread-ai/mxbai-rerank-xsmall-v1",
         cross_encoder_inference_endpoint: str = None,
         cross_encoder_inference_endpoint_api_key: str = None,
+        model_kwargs: dict = {},
     ):
         self.model_name = model_name
-        self.cross_encoder_model = CrossEncoder(model_name=self.model_name, device=get_device())
         self.inference_endpoint = cross_encoder_inference_endpoint
         self.api_key = cross_encoder_inference_endpoint_api_key
+        self.model_kwargs = merge_dicts(model_kwargs, {"device": get_device()})
+        self.cross_encoder_model = CrossEncoder(model_name=self.model_name, **self.model_kwargs)
 
     def inference_server_enabled(self) -> bool:
         return self.api_key is not None and self.inference_endpoint is not None

khoj/processor/tools/online_search.py

@@ -10,6 +10,7 @@ import aiohttp
 from bs4 import BeautifulSoup
 from markdownify import markdownify
 
+from khoj.database.models import KhojUser
 from khoj.routers.helpers import (
     ChatEvent,
     extract_relevant_info,
@@ -51,6 +52,8 @@ async def search_online(
     query: str,
     conversation_history: dict,
     location: LocationData,
+    user: KhojUser,
+    subscribed: bool = False,
     send_status_func: Optional[Callable] = None,
     custom_filters: List[str] = [],
 ):
@@ -61,7 +64,7 @@ async def search_online(
         return
 
     # Breakdown the query into subqueries to get the correct answer
-    subqueries = await generate_online_subqueries(query, conversation_history, location)
+    subqueries = await generate_online_subqueries(query, conversation_history, location, user)
     response_dict = {}
 
     if subqueries:
@@ -89,12 +92,15 @@ async def search_online(
     # Read, extract relevant info from the retrieved web pages
     if webpages:
         webpage_links = [link for link, _, _ in webpages]
-        logger.info(f"🌐👀 Reading web pages at: {list(webpage_links)}")
+        logger.info(f"Reading web pages at: {list(webpage_links)}")
         if send_status_func:
             webpage_links_str = "\n- " + "\n- ".join(list(webpage_links))
             async for event in send_status_func(f"**Reading web pages**: {webpage_links_str}"):
                 yield {ChatEvent.STATUS: event}
-    tasks = [read_webpage_and_extract_content(subquery, link, content) for link, subquery, content in webpages]
+    tasks = [
+        read_webpage_and_extract_content(subquery, link, content, subscribed=subscribed)
+        for link, subquery, content in webpages
+    ]
     results = await asyncio.gather(*tasks)
 
     # Collect extracted info from the retrieved web pages
@@ -126,21 +132,26 @@ async def search_with_google(query: str) -> Tuple[str, Dict[str, List[Dict]]]:
 
 
 async def read_webpages(
-    query: str, conversation_history: dict, location: LocationData, send_status_func: Optional[Callable] = None
+    query: str,
+    conversation_history: dict,
+    location: LocationData,
+    user: KhojUser,
+    subscribed: bool = False,
+    send_status_func: Optional[Callable] = None,
 ):
     "Infer web pages to read from the query and extract relevant information from them"
     logger.info(f"Inferring web pages to read")
     if send_status_func:
         async for event in send_status_func(f"**Inferring web pages to read**"):
             yield {ChatEvent.STATUS: event}
-    urls = await infer_webpage_urls(query, conversation_history, location)
+    urls = await infer_webpage_urls(query, conversation_history, location, user)
 
     logger.info(f"Reading web pages at: {urls}")
     if send_status_func:
         webpage_links_str = "\n- " + "\n- ".join(list(urls))
         async for event in send_status_func(f"**Reading web pages**: {webpage_links_str}"):
             yield {ChatEvent.STATUS: event}
-    tasks = [read_webpage_and_extract_content(query, url) for url in urls]
+    tasks = [read_webpage_and_extract_content(query, url, subscribed=subscribed) for url in urls]
     results = await asyncio.gather(*tasks)
 
     response: Dict[str, Dict] = defaultdict(dict)
@@ -151,14 +162,14 @@ async def read_webpages(
 
 
 async def read_webpage_and_extract_content(
-    subquery: str, url: str, content: str = None
+    subquery: str, url: str, content: str = None, subscribed: bool = False
 ) -> Tuple[str, Union[None, str], str]:
     try:
         if is_none_or_empty(content):
             with timer(f"Reading web page at '{url}' took", logger):
                 content = await read_webpage_with_olostep(url) if OLOSTEP_API_KEY else await read_webpage_with_jina(url)
         with timer(f"Extracting relevant information from web page at '{url}' took", logger):
-            extracted_info = await extract_relevant_info(subquery, content)
+            extracted_info = await extract_relevant_info(subquery, content, subscribed=subscribed)
         return subquery, extracted_info, url
     except Exception as e:
         logger.error(f"Failed to read web page at '{url}' with {e}")

khoj/routers/api.py

@@ -388,6 +388,7 @@ async def extract_references_and_questions(
                 conversation_log=meta_log,
                 should_extract_questions=True,
                 location_data=location_data,
+                user=user,
                 max_prompt_size=conversation_config.max_prompt_size,
             )
         elif conversation_config.model_type == ChatModelOptions.ModelType.OPENAI:
@@ -402,7 +403,7 @@ async def extract_references_and_questions(
                 api_base_url=base_url,
                 conversation_log=meta_log,
                 location_data=location_data,
-                max_tokens=conversation_config.max_prompt_size,
+                user=user,
             )
         elif conversation_config.model_type == ChatModelOptions.ModelType.ANTHROPIC:
             api_key = conversation_config.openai_config.api_key
@@ -413,6 +414,7 @@ async def extract_references_and_questions(
                 api_key=api_key,
                 conversation_log=meta_log,
                 location_data=location_data,
+                user=user,
             )
 
     # Collate search results as context for GPT

khoj/routers/api_chat.py

@@ -4,14 +4,14 @@ import logging
 import time
 from datetime import datetime
 from functools import partial
-from typing import Any, Dict, List, Optional
+from typing import Dict, Optional
 from urllib.parse import unquote
 
 from asgiref.sync import sync_to_async
 from fastapi import APIRouter, Depends, HTTPException, Request
 from fastapi.requests import Request
 from fastapi.responses import Response, StreamingResponse
-from starlette.authentication import requires
+from starlette.authentication import has_required_scope, requires
 
 from khoj.app.settings import ALLOWED_HOSTS
 from khoj.database.adapters import (
@@ -59,7 +59,7 @@ from khoj.utils.rawconfig import FileFilterRequest, FilesFilterRequest, Location
 # Initialize Router
 logger = logging.getLogger(__name__)
 conversation_command_rate_limiter = ConversationCommandRateLimiter(
-    trial_rate_limit=2, subscribed_rate_limit=100, slug="command"
+    trial_rate_limit=100, subscribed_rate_limit=100, slug="command"
 )
 
 
@@ -532,10 +532,10 @@ async def chat(
     country: Optional[str] = None,
     timezone: Optional[str] = None,
     rate_limiter_per_minute=Depends(
-        ApiUserRateLimiter(requests=5, subscribed_requests=60, window=60, slug="chat_minute")
+        ApiUserRateLimiter(requests=60, subscribed_requests=60, window=60, slug="chat_minute")
     ),
     rate_limiter_per_day=Depends(
-        ApiUserRateLimiter(requests=5, subscribed_requests=600, window=60 * 60 * 24, slug="chat_day")
+        ApiUserRateLimiter(requests=600, subscribed_requests=600, window=60 * 60 * 24, slug="chat_day")
     ),
 ):
     async def event_generator(q: str):
@@ -544,6 +544,7 @@ async def chat(
         chat_metadata: dict = {}
         connection_alive = True
         user: KhojUser = request.user.object
+        subscribed: bool = has_required_scope(request, ["premium"])
         event_delimiter = "␃🔚␗"
         q = unquote(q)
 
@@ -632,7 +633,9 @@ async def chat(
         is_automated_task = conversation_commands == [ConversationCommand.AutomatedTask]
 
         if conversation_commands == [ConversationCommand.Default] or is_automated_task:
-            conversation_commands = await aget_relevant_information_sources(q, meta_log, is_automated_task)
+            conversation_commands = await aget_relevant_information_sources(
+                q, meta_log, is_automated_task, subscribed=subscribed
+            )
             conversation_commands_str = ", ".join([cmd.value for cmd in conversation_commands])
             async for result in send_event(
                 ChatEvent.STATUS, f"**Chose Data Sources to Search:** {conversation_commands_str}"
@@ -687,7 +690,7 @@ async def chat(
                     ):
                         yield result
 
-                    response = await extract_relevant_summary(q, contextual_data)
+                    response = await extract_relevant_summary(q, contextual_data, subscribed=subscribed)
                     response_log = str(response)
                     async for result in send_llm_response(response_log):
                         yield result
@@ -775,6 +778,8 @@ async def chat(
 
         if not is_none_or_empty(compiled_references):
             headings = "\n- " + "\n- ".join(set([c.get("compiled", c).split("\n")[0] for c in compiled_references]))
+            # Strip only leading # from headings
+            headings = headings.replace("#", "")
             async for result in send_event(ChatEvent.STATUS, f"**Found Relevant Notes**: {headings}"):
                 yield result
 
@@ -792,7 +797,13 @@ async def chat(
         if ConversationCommand.Online in conversation_commands:
             try:
                 async for result in search_online(
-                    defiltered_query, meta_log, location, partial(send_event, ChatEvent.STATUS), custom_filters
+                    defiltered_query,
+                    meta_log,
+                    location,
+                    user,
+                    subscribed,
+                    partial(send_event, ChatEvent.STATUS),
+                    custom_filters,
                 ):
                     if isinstance(result, dict) and ChatEvent.STATUS in result:
                         yield result[ChatEvent.STATUS]
@@ -809,7 +820,7 @@ async def chat(
         if ConversationCommand.Webpage in conversation_commands:
             try:
                 async for result in read_webpages(
-                    defiltered_query, meta_log, location, partial(send_event, ChatEvent.STATUS)
+                    defiltered_query, meta_log, location, user, subscribed, partial(send_event, ChatEvent.STATUS)
                 ):
                     if isinstance(result, dict) and ChatEvent.STATUS in result:
                         yield result[ChatEvent.STATUS]
@@ -853,6 +864,7 @@ async def chat(
                 location_data=location,
                 references=compiled_references,
                 online_results=online_results,
+                subscribed=subscribed,
                 send_status_func=partial(send_event, ChatEvent.STATUS),
             ):
                 if isinstance(result, dict) and ChatEvent.STATUS in result:

khoj/routers/helpers.py

@@ -252,7 +252,7 @@ async def acreate_title_from_query(query: str) -> str:
     return response.strip()
 
 
-async def aget_relevant_information_sources(query: str, conversation_history: dict, is_task: bool):
+async def aget_relevant_information_sources(query: str, conversation_history: dict, is_task: bool, subscribed: bool):
     """
     Given a query, determine which of the available tools the agent should use in order to answer appropriately.
     """
@@ -273,7 +273,9 @@ async def aget_relevant_information_sources(query: str, conversation_history: di
     )
 
     with timer("Chat actor: Infer information sources to refer", logger):
-        response = await send_message_to_model_wrapper(relevant_tools_prompt, response_type="json_object")
+        response = await send_message_to_model_wrapper(
+            relevant_tools_prompt, response_type="json_object", subscribed=subscribed
+        )
 
     try:
         response = response.strip()
@@ -340,11 +342,14 @@ async def aget_relevant_output_modes(query: str, conversation_history: dict, is_
         return ConversationCommand.Text
 
 
-async def infer_webpage_urls(q: str, conversation_history: dict, location_data: LocationData) -> List[str]:
+async def infer_webpage_urls(
+    q: str, conversation_history: dict, location_data: LocationData, user: KhojUser
+) -> List[str]:
     """
     Infer webpage links from the given query
     """
     location = f"{location_data.city}, {location_data.region}, {location_data.country}" if location_data else "Unknown"
+    username = prompts.user_name.format(name=user.get_full_name()) if user.get_full_name() else ""
     chat_history = construct_chat_history(conversation_history)
 
     utc_date = datetime.utcnow().strftime("%Y-%m-%d")
@@ -353,6 +358,7 @@ async def infer_webpage_urls(q: str, conversation_history: dict, location_data:
         query=q,
         chat_history=chat_history,
         location=location,
+        username=username,
     )
 
     with timer("Chat actor: Infer webpage urls to read", logger):
@@ -370,11 +376,14 @@ async def infer_webpage_urls(q: str, conversation_history: dict, location_data:
         raise ValueError(f"Invalid list of urls: {response}")
 
 
-async def generate_online_subqueries(q: str, conversation_history: dict, location_data: LocationData) -> List[str]:
+async def generate_online_subqueries(
+    q: str, conversation_history: dict, location_data: LocationData, user: KhojUser
+) -> List[str]:
     """
     Generate subqueries from the given query
     """
     location = f"{location_data.city}, {location_data.region}, {location_data.country}" if location_data else "Unknown"
+    username = prompts.user_name.format(name=user.get_full_name()) if user.get_full_name() else ""
     chat_history = construct_chat_history(conversation_history)
 
     utc_date = datetime.utcnow().strftime("%Y-%m-%d")
@@ -383,6 +392,7 @@ async def generate_online_subqueries(q: str, conversation_history: dict, locatio
         query=q,
         chat_history=chat_history,
         location=location,
+        username=username,
     )
 
     with timer("Chat actor: Generate online search subqueries", logger):
@@ -426,7 +436,7 @@ async def schedule_query(q: str, conversation_history: dict) -> Tuple[str, ...]:
         raise AssertionError(f"Invalid response for scheduling query: {raw_response}")
 
 
-async def extract_relevant_info(q: str, corpus: str) -> Union[str, None]:
+async def extract_relevant_info(q: str, corpus: str, subscribed: bool) -> Union[str, None]:
     """
     Extract relevant information for a given query from the target corpus
     """
@@ -439,18 +449,19 @@ async def extract_relevant_info(q: str, corpus: str) -> Union[str, None]:
         corpus=corpus.strip(),
     )
 
-    summarizer_model: ChatModelOptions = await ConversationAdapters.aget_summarizer_conversation_config()
+    chat_model: ChatModelOptions = await ConversationAdapters.aget_default_conversation_config()
 
     with timer("Chat actor: Extract relevant information from data", logger):
         response = await send_message_to_model_wrapper(
             extract_relevant_information,
             prompts.system_prompt_extract_relevant_information,
-            chat_model_option=summarizer_model,
+            chat_model_option=chat_model,
+            subscribed=subscribed,
         )
     return response.strip()
 
 
-async def extract_relevant_summary(q: str, corpus: str) -> Union[str, None]:
+async def extract_relevant_summary(q: str, corpus: str, subscribed: bool = False) -> Union[str, None]:
     """
     Extract relevant information for a given query from the target corpus
     """
@@ -463,13 +474,14 @@ async def extract_relevant_summary(q: str, corpus: str) -> Union[str, None]:
         corpus=corpus.strip(),
     )
 
-    summarizer_model: ChatModelOptions = await ConversationAdapters.aget_summarizer_conversation_config()
+    chat_model: ChatModelOptions = await ConversationAdapters.aget_default_conversation_config()
 
     with timer("Chat actor: Extract relevant information from data", logger):
         response = await send_message_to_model_wrapper(
             extract_relevant_information,
             prompts.system_prompt_extract_relevant_summary,
-            chat_model_option=summarizer_model,
+            chat_model_option=chat_model,
+            subscribed=subscribed,
         )
     return response.strip()
 
@@ -481,6 +493,7 @@ async def generate_better_image_prompt(
     note_references: List[Dict[str, Any]],
     online_results: Optional[dict] = None,
     model_type: Optional[str] = None,
+    subscribed: bool = False,
 ) -> str:
     """
     Generate a better image prompt from the given query
@@ -525,10 +538,12 @@ async def generate_better_image_prompt(
             online_results=simplified_online_results,
         )
 
-    summarizer_model: ChatModelOptions = await ConversationAdapters.aget_summarizer_conversation_config()
+    chat_model: ChatModelOptions = await ConversationAdapters.aget_default_conversation_config()
 
     with timer("Chat actor: Generate contextual image prompt", logger):
-        response = await send_message_to_model_wrapper(image_prompt, chat_model_option=summarizer_model)
+        response = await send_message_to_model_wrapper(
+            image_prompt, chat_model_option=chat_model, subscribed=subscribed
+        )
         response = response.strip()
         if response.startswith(('"', "'")) and response.endswith(('"', "'")):
             response = response[1:-1]
@@ -541,13 +556,18 @@ async def send_message_to_model_wrapper(
     system_message: str = "",
     response_type: str = "text",
     chat_model_option: ChatModelOptions = None,
+    subscribed: bool = False,
 ):
     conversation_config: ChatModelOptions = (
         chat_model_option or await ConversationAdapters.aget_default_conversation_config()
     )
 
     chat_model = conversation_config.chat_model
-    max_tokens = conversation_config.max_prompt_size
+    max_tokens = (
+        conversation_config.subscribed_max_prompt_size
+        if subscribed and conversation_config.subscribed_max_prompt_size
+        else conversation_config.max_prompt_size
+    )
     tokenizer = conversation_config.tokenizer
 
     if conversation_config.model_type == "offline":
@@ -778,6 +798,7 @@ async def text_to_image(
     location_data: LocationData,
     references: List[Dict[str, Any]],
     online_results: Dict[str, Any],
+    subscribed: bool = False,
     send_status_func: Optional[Callable] = None,
 ):
     status_code = 200
@@ -814,6 +835,7 @@ async def text_to_image(
         note_references=references,
         online_results=online_results,
         model_type=text_to_image_config.model_type,
+        subscribed=subscribed,
     )
 
     if send_status_func:
@@ -1351,7 +1373,9 @@ def get_user_config(user: KhojUser, request: Request, is_detailed: bool = False)
     current_notion_config = get_user_notion_config(user)
     notion_token = current_notion_config.token if current_notion_config else ""
 
-    selected_chat_model_config = ConversationAdapters.get_conversation_config(user)
+    selected_chat_model_config = (
+        ConversationAdapters.get_conversation_config(user) or ConversationAdapters.get_default_conversation_config()
+    )
     chat_models = ConversationAdapters.get_conversation_processor_options().all()
     chat_model_options = list()
     for chat_model in chat_models:

khoj/search_filter/file_filter.py

@@ -11,7 +11,8 @@ logger = logging.getLogger(__name__)
 
 
 class FileFilter(BaseFilter):
-    file_filter_regex = r'file:"(.+?)" ?'
+    file_filter_regex = r'(?<!-)file:"(.+?)" ?'
+    excluded_file_filter_regex = r'-file:"(.+?)" ?'
 
     def __init__(self, entry_key="file"):
         self.entry_key = entry_key
@@ -20,7 +21,9 @@ class FileFilter(BaseFilter):
 
     def get_filter_terms(self, query: str) -> List[str]:
         "Get all filter terms in query"
-        return [f"{self.convert_to_regex(term)}" for term in re.findall(self.file_filter_regex, query)]
+        required_files = [f"{required_file}" for required_file in re.findall(self.file_filter_regex, query)]
+        excluded_files = [f"-{excluded_file}" for excluded_file in re.findall(self.excluded_file_filter_regex, query)]
+        return required_files + excluded_files
 
     def convert_to_regex(self, file_filter: str) -> str:
         "Convert file filter to regex"

{khoj-1.20.4.dev13.dist-info → khoj-1.20.5.dev17.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: khoj
-Version: 1.20.4.dev13
+Version: 1.20.5.dev17
 Summary: Your Second Brain
 Project-URL: Homepage, https://khoj.dev
 Project-URL: Documentation, https://docs.khoj.dev