Add vision support for Gemini models in Khoj

src/khoj/processor/conversation/google/gemini_chat.py

@@ -13,7 +13,10 @@ from khoj.processor.conversation.google.utils import (
     gemini_chat_completion_with_backoff,
     gemini_completion_with_backoff,
 )
-from khoj.processor.conversation.utils import generate_chatml_messages_with_context
+from khoj.processor.conversation.utils import (
+    construct_structured_message,
+    generate_chatml_messages_with_context,
+)
 from khoj.utils.helpers import ConversationCommand, is_none_or_empty
 from khoj.utils.rawconfig import LocationData
 
@@ -29,6 +32,8 @@ def extract_questions_gemini(
     max_tokens=None,
     location_data: LocationData = None,
     user: KhojUser = None,
+    query_images: Optional[list[str]] = None,
+    vision_enabled: bool = False,
     personality_context: Optional[str] = None,
 ):
     """
@@ -70,17 +75,17 @@ def extract_questions_gemini(
         text=text,
     )
 
-    messages = [ChatMessage(content=prompt, role="user")]
+    prompt = construct_structured_message(
+        message=prompt,
+        images=query_images,
+        model_type=ChatModelOptions.ModelType.GOOGLE,
+        vision_enabled=vision_enabled,
+    )
 
-    model_kwargs = {"response_mime_type": "application/json"}
+    messages = [ChatMessage(content=prompt, role="user"), ChatMessage(content=system_prompt, role="system")]
 
-    response = gemini_completion_with_backoff(
-        messages=messages,
-        system_prompt=system_prompt,
-        model_name=model,
-        temperature=temperature,
-        api_key=api_key,
-        model_kwargs=model_kwargs,
+    response = gemini_send_message_to_model(
+        messages, api_key, model, response_type="json_object", temperature=temperature
     )
 
     # Extract, Clean Message from Gemini's Response
@@ -102,7 +107,7 @@ def extract_questions_gemini(
     return questions
 
 
-def gemini_send_message_to_model(messages, api_key, model, response_type="text"):
+def gemini_send_message_to_model(messages, api_key, model, response_type="text", temperature=0, model_kwargs=None):
     """
     Send message to model
     """
@@ -114,7 +119,12 @@ def gemini_send_message_to_model(messages, api_key, model, response_type="text")
 
     # Get Response from Gemini
     return gemini_completion_with_backoff(
-        messages=messages, system_prompt=system_prompt, model_name=model, api_key=api_key, model_kwargs=model_kwargs
+        messages=messages,
+        system_prompt=system_prompt,
+        model_name=model,
+        api_key=api_key,
+        temperature=temperature,
+        model_kwargs=model_kwargs,
     )
 
 
@@ -133,6 +143,8 @@ def converse_gemini(
     location_data: LocationData = None,
     user_name: str = None,
     agent: Agent = None,
+    query_images: Optional[list[str]] = None,
+    vision_available: bool = False,
 ):
     """
     Converse with user using Google's Gemini
@@ -187,6 +199,8 @@ def converse_gemini(
         model_name=model,
         max_prompt_size=max_prompt_size,
         tokenizer_name=tokenizer_name,
+        query_images=query_images,
+        vision_enabled=vision_available,
         model_type=ChatModelOptions.ModelType.GOOGLE,
     )
 

src/khoj/processor/conversation/google/utils.py

@@ -1,8 +1,11 @@
 import logging
 import random
+from io import BytesIO
 from threading import Thread
 
 import google.generativeai as genai
+import PIL.Image
+import requests
 from google.generativeai.types.answer_types import FinishReason
 from google.generativeai.types.generation_types import StopCandidateException
 from google.generativeai.types.safety_types import (
@@ -53,14 +56,14 @@ def gemini_completion_with_backoff(
         },
     )
 
-    formatted_messages = [{"role": message.role, "parts": [message.content]} for message in messages]
+    formatted_messages = [{"role": message.role, "parts": message.content} for message in messages]
 
     # Start chat session. All messages up to the last are considered to be part of the chat history
     chat_session = model.start_chat(history=formatted_messages[0:-1])
 
     try:
         # Generate the response. The last message is considered to be the current prompt
-        aggregated_response = chat_session.send_message(formatted_messages[-1]["parts"][0])
+        aggregated_response = chat_session.send_message(formatted_messages[-1]["parts"])
         return aggregated_response.text
     except StopCandidateException as e:
         response_message, _ = handle_gemini_response(e.args)
@@ -117,11 +120,11 @@ def gemini_llm_thread(g, messages, system_prompt, model_name, temperature, api_k
             },
         )
 
-        formatted_messages = [{"role": message.role, "parts": [message.content]} for message in messages]
+        formatted_messages = [{"role": message.role, "parts": message.content} for message in messages]
         # all messages up to the last are considered to be part of the chat history
         chat_session = model.start_chat(history=formatted_messages[0:-1])
         # the last message is considered to be the current prompt
-        for chunk in chat_session.send_message(formatted_messages[-1]["parts"][0], stream=True):
+        for chunk in chat_session.send_message(formatted_messages[-1]["parts"], stream=True):
             message, stopped = handle_gemini_response(chunk.candidates, chunk.prompt_feedback)
             message = message or chunk.text
             g.send(message)
@@ -191,14 +194,6 @@ def generate_safety_response(safety_ratings):
 
 
 def format_messages_for_gemini(messages: list[ChatMessage], system_prompt: str = None) -> tuple[list[str], str]:
-    if len(messages) == 1:
-        messages[0].role = "user"
-        return messages, system_prompt
-
-    for message in messages:
-        if message.role == "assistant":
-            message.role = "model"
-
     # Extract system message
     system_prompt = system_prompt or ""
     for message in messages.copy():
@@ -207,4 +202,31 @@ def format_messages_for_gemini(messages: list[ChatMessage], system_prompt: str =
             messages.remove(message)
     system_prompt = None if is_none_or_empty(system_prompt) else system_prompt
 
+    for message in messages:
+        # Convert message content to string list from chatml dictionary list
+        if isinstance(message.content, list):
+            # Convert image_urls to PIL.Image and place them at beginning of list (better for Gemini)
+            message.content = [
+                get_image_from_url(item["image_url"]["url"]) if item["type"] == "image_url" else item["text"]
+                for item in sorted(message.content, key=lambda x: 0 if x["type"] == "image_url" else 1)
+            ]
+        elif isinstance(message.content, str):
+            message.content = [message.content]
+
+        if message.role == "assistant":
+            message.role = "model"
+
+    if len(messages) == 1:
+        messages[0].role = "user"
+
     return messages, system_prompt
+
+
+def get_image_from_url(image_url: str) -> PIL.Image:
+    try:
+        response = requests.get(image_url)
+        response.raise_for_status()  # Check if the request was successful
+        return PIL.Image.open(BytesIO(response.content))
+    except requests.exceptions.RequestException as e:
+        logger.error(f"Failed to get image from URL {image_url}: {e}")
+        return None

src/khoj/processor/conversation/utils.py

@@ -152,7 +152,7 @@ def construct_structured_message(message: str, images: list[str], model_type: st
     if not images or not vision_enabled:
         return message
 
-    if model_type == ChatModelOptions.ModelType.OPENAI:
+    if model_type in [ChatModelOptions.ModelType.OPENAI, ChatModelOptions.ModelType.GOOGLE]:
         return [
             {"type": "text", "text": message},
             *[{"type": "image_url", "image_url": {"url": image}} for image in images],

src/khoj/routers/api.py

@@ -452,12 +452,14 @@ async def extract_references_and_questions(
             chat_model = conversation_config.chat_model
             inferred_queries = extract_questions_gemini(
                 defiltered_query,
+                query_images=query_images,
                 model=chat_model,
                 api_key=api_key,
                 conversation_log=meta_log,
                 location_data=location_data,
                 max_tokens=conversation_config.max_prompt_size,
                 user=user,
+                vision_enabled=vision_enabled,
                 personality_context=personality_context,
             )
 

src/khoj/routers/helpers.py

@@ -995,8 +995,9 @@ def generate_chat_response(
             chat_response = converse_gemini(
                 compiled_references,
                 q,
-                online_results,
-                meta_log,
+                query_images=query_images,
+                online_results=online_results,
+                conversation_log=meta_log,
                 model=conversation_config.chat_model,
                 api_key=api_key,
                 completion_func=partial_completion,
@@ -1006,6 +1007,7 @@ def generate_chat_response(
                 location_data=location_data,
                 user_name=user_name,
                 agent=agent,
+                vision_available=vision_available,
             )
 
         metadata.update({"chat_model": conversation_config.chat_model})