Update app.py

app.py

@@ -1,8 +1,8 @@
-
+import gradio as gr
 import chainlit as cl
-from langchain import OpenAI, LLMChain 
+from langchain import OpenAI, LLMChain
 from langchain.agents import Tool, AgentExecutor, LLMSingleActionAgent, AgentOutputParser
-from langchain.memory import ConversationBufferWindowMemory 
+from langchain.memory import ConversationBufferWindowMemory
 from langchain.prompts import StringPromptTemplate
 from langchain.tools import DuckDuckGoSearchRun
 from typing import List, Union
@@ -15,8 +15,7 @@ from main1 import predict
 from PIL import Image
 import io
 
-
-OPENAI_API_KEY='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'
+OPENAI_API_KEY = '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'
 
 search = DuckDuckGoSearchRun()
 
@@ -26,26 +25,25 @@ def duck_wrapper(input_text):
 
 tools = [
     Tool(
-        name = "Search",
+        name="Search",
         func=duck_wrapper,
-        description="useful for when you need to answer medical and pharmalogical questions"
+        description="useful for when you need to answer medical and pharmaceutical questions"
     )
 ]
 
-
 def call_detection_model(index):
     results = [
         {
-            "has_cancer":False, 
-            "chances_of_having_cancer":8.64
+            "has_cancer": False,
+            "chances_of_having_cancer": 8.64
         },
         {
-            "has_cancer":True, 
-            "chances_of_having_cancer":97.89
+            "has_cancer": True,
+            "chances_of_having_cancer": 97.89
         },
         {
-            "has_cancer":False,
-            "chances_of_having_cancer":2.78
+            "has_cancer": False,
+            "chances_of_having_cancer": 2.78
         }
     ]
     return results[index]
@@ -55,7 +53,7 @@ class CustomPromptTemplate(StringPromptTemplate):
     template: str
     # The list of tools available
     tools: List[Tool]
-    
+
     def format(self, **kwargs) -> str:
         # Get the intermediate steps (AgentAction, Observation tuples)
         # Format them in a particular way
@@ -77,7 +75,7 @@ class CustomOutputParser(AgentOutputParser):
         # Check if agent should finish
         if "Final Answer:" in llm_output:
             return AgentFinish(
-                # Return values is generally always a dictionary with a single `output` key
+                # Return values are generally always a dictionary with a single `output` key
                 # It is not recommended to try anything else at the moment :)
                 return_values={"output": llm_output.split("Final Answer:")[-1].strip()},
                 log=llm_output,
@@ -94,74 +92,74 @@ class CustomOutputParser(AgentOutputParser):
 
 template = None
 prompt_with_history = None
-from contextlib import redirect_stdout
-
 
 @cl.on_chat_start
 async def main():
     cl.user_session.set("index", 0)
     cl.user_session.set("has_uploaded_image", False)
-    await cl.Message("Upload image of your condition").send()
 
-@cl.on_file_upload(accept=['image/png'])
-async def main(file:any):
-    index = cl.user_session.get("index")
-    file = file[0]["content"]
-    image_stream = io.BytesIO(file)
+def analyze_image(file, user_session):
+    index = user_session.get("index")
+    image_stream = io.BytesIO(file.read())
     image = Image.open(image_stream)
     image = image.convert('RGB')
     image = image.resize((150, 150))
     image.save("image.png", 'png')
     results = call_detection_model(index)
-    cl.user_session.set("index", index+1)
+    user_session.set("index", index + 1)
     image.close()
-    cl.user_session.set("results", results)
+    user_session.set("results", results)
     if results["has_cancer"]:
-        cl.user_session.set("template", template_for_has_cancer)
+        user_session.set("template", template_for_has_cancer)
     else:
-        cl.user_session.set("template", template_for_does_not_have_cancer)
+        user_session.set("template", template_for_does_not_have_cancer)
     prompt_with_history = CustomPromptTemplate(
-        template=cl.user_session.get("template"),
+        template=user_session.get("template"),
         tools=tools,
         input_variables=["input", "intermediate_steps", "history"]
     )
-    llm_chain = LLMChain(prompt = prompt_with_history,llm=OpenAI(temperature=1.2,streaming=True),verbose=True)
+    llm_chain = LLMChain(prompt=prompt_with_history, llm=OpenAI(temperature=1.2, streaming=True), verbose=True)
     tool_names = [tool.name for tool in tools]
     output_parser = CustomOutputParser()
     agent = LLMSingleActionAgent(
-        llm_chain=llm_chain, 
+        llm_chain=llm_chain,
         output_parser=output_parser,
-        stop=["\nObservation:"], 
+        stop=["\nObservation:"],
         allowed_tools=tool_names
     )
-    memory=ConversationBufferWindowMemory(k=2)
+    memory = ConversationBufferWindowMemory(k=2)
     agent_executor = AgentExecutor.from_agent_and_tools(
-        agent=agent, 
-        tools=tools, 
-        verbose=True, 
+        agent=agent,
+        tools=tools,
+        verbose=True,
         memory=memory
     )
-    cl.user_session.set("agent_executor", agent_executor)
-    cl.user_session.set("has_uploaded_image", True)
-    await cl.Message("Image has been uploaded and analyzing...\n").send()
-    
-
+    user_session.set("agent_executor", agent_executor)
+    user_session.set("has_uploaded_image", True)
 
-@cl.on_message
-async def main(message : str):
-    has_uploaded_image = cl.user_session.get("has_uploaded_image")
-    results = cl.user_session.get("results")
+def get_result(user_session, message):
+    has_uploaded_image = user_session.get("has_uploaded_image")
+    results = user_session.get("results")
     if has_uploaded_image == False:
-        await cl.Message("Please upload a relevent image to proceed with this conversation").send()
-        return
-    if "result" in message or "results" in message:     
-        msg = f"These results are a good estimation but its not meant to replace human medical intervention and should be taken with a grain of salt. According to the image uploaded, your chances of having skin cancer are {results['chances_of_having_cancer']}% and your condition lies in the {cancer_category(results['chances_of_having_cancer'])} range. "
+        return "Please upload a relevant image to proceed with this conversation"
+    if "result" in message or "results" in message:
+        msg = f"These results are a good estimation but it's not meant to replace human medical intervention and should be taken with a grain of salt. According to the image uploaded, your chances of having skin cancer are {results['chances_of_having_cancer']}% and your condition lies in the {cancer_category(results['chances_of_having_cancer'])} range. "
         if cancer_category(results["chances_of_having_cancer"]) != "Pre Benign":
-            msg += "You should consider vising the doctor for a complete checkup."
-        await cl.Message(msg).send()
-        return
-    agent_executor = cl.user_session.get("agent_executor")
+            msg += "You should consider visiting the doctor for a complete checkup."
+        return msg
+    agent_executor = user_session.get("agent_executor")
     res = agent_executor.run(message)
-    await cl.Message(res).send()
-
-
+    return res
+
+iface = gr.Interface(
+    analyze_image,
+    inputs=gr.inputs.File(label="Upload image of your condition", type="file", accept="image/png"),
+    outputs=gr.outputs.Textbox(),
+    live=True,
+    capture_session=True,
+    fn_kwargs={"user_session": cl.user_session},
+    title="Skin Condition Analyzer",
+    description="Upload an image of your skin condition and interact with the AI to get information about it."
+)
+
+iface.launch()