doing summary with AI

.gitignore

@@ -2,4 +2,6 @@ node_modules/
 package-lock.json
 rec/*
 */__pycache__/*
-__pycache__
+__pycache__
+agent-py-bot/scrape/raw/summary_log.txt
+agent-py-bot/scrape/raw/*

agent-py-bot/agent.py

@@ -11,6 +11,7 @@ from selenium import webdriver
 from selenium.webdriver.chrome.options import Options
 from io import BytesIO
 from PIL import Image
+from datetime import datetime, timedelta
 
 # Apply nest_asyncio
 nest_asyncio.apply()
@@ -78,13 +79,18 @@ import re
 from agents.runner import execute_python_code
 
 #https://github.com/jmorganca/ollama/blob/main/docs/api.md#generate-a-completion
-async def query_llm(user_message):
+async def query_llm(user_message, model=None):
     """Query the LLM with the user's message."""
+    # use the model if provided, otherwise use the default llama2
+    if model is None:
+        model = "llama2:latest"
+
     data = {
-        "model": "llama2:latest",
+        "model": model,
         "messages": [{"role": "user", "content": user_message}],
         "stream": False
     }
+    
     response = requests.post(LLM_ENDPOINT, json=data)
     if response.status_code == 200:
             response_data = response.json()
@@ -99,15 +105,15 @@ async def query_llm(user_message):
 
             # Find and execute all code blocks
             code_blocks = re.findall(r"```(.*?)```", content, re.DOTALL)
-            for code in code_blocks:
-                execution_result = execute_python_code(code.strip())
-            if APPEND_RESULTS:
-                # Append the result after the code block
-                content = content.replace(f"```{code}```", f"```{code}```\n```{execution_result}```")
-            else:
-                # Replace the code block with its result
-                content = content.replace(f"```{code}```", f"```{execution_result}```")
-
+            if code_blocks:
+                for code in code_blocks:
+                    execution_result = execute_python_code(code.strip())
+                    if APPEND_RESULTS:
+                        # Append the result after the code block
+                        content = content.replace(f"```{code}```", f"```{code}```\n```{execution_result}```")
+                    else:
+                        # Replace the code block with its result
+                        content = content.replace(f"```{code}```", f"```{execution_result}```")
 
             return content
     else:    
@@ -193,50 +199,35 @@ import time
 from agents.webagent import run_web_agent, save_data
 
 async def run_web_agent_and_process_result(topic, folder):
-    print(f"Running web agent for topic: {topic}")
     news_data = run_web_agent(topic, folder)
-    save_data(news_data, folder)
     # summary = summarize_data(news_data)
     
     # with open(os.path.join(folder, "summary_log.txt"), 'a') as log_file:
     #     log_file.write(f"{datetime.now()}: {summary}\n")
     
     # Process the result immediately after obtaining it
-    user_message = f"New data received: {news_data}"
-    query_result = await query_llm(user_message)
+    user_message = f"Summarize these news and make sentiment analysis on each news and one overall: {news_data}"
+    
+    print(f"[{datetime.now()}] Doing sentiment analysis with AI model.")
+    start = time.time()
+    query_result = await query_llm(user_message, "openhermes")
+    print(f"[{datetime.now()}] AI call returned in {time.time() - start} seconds.")
+    news_data["summary"] = query_result
+    save_data(news_data, folder)
+    
+    with open(os.path.join(folder, "summary_log.txt"), 'a') as log_file:
+        log_file.write(f"{datetime.now()}: {query_result}\n")
+
     # Process the query_result as needed
 
 async def async_main():
-    # Assuming this is your asynchronous main function with its full details
-    loop = asyncio.get_event_loop()
-    if loop.is_running():
-        loop.create_task(main())
-    else:
-        await main()
-
-def sync_main():
-    # Synchronous part for scheduling
     topic = "tesla news"
-    interval = 8  # in minutes
+    interval = 1  # in hours
     folder = "agent-py-bot/scrape/raw"
 
-    # schedule.every(interval).minutes.do(run_web_agent_and_process_result, topic=topic, folder=folder)
-    schedule.every(interval).hours.do(run_web_agent_and_process_result, topic=topic, folder=folder)
-    # Run once at the start
-    news_json = await run_web_agent_and_process_result(topic=topic, folder=folder)
-
     while True:
-        schedule.run_pending()
-        time.sleep(1)
+        await run_web_agent_and_process_result(topic=topic, folder=folder)
+        await asyncio.sleep(interval * 60)  # Convert hours to seconds
 
 if __name__ == '__main__':
-    loop = asyncio.get_event_loop()
-
-    # Run the asynchronous part
-    if loop.is_running():
-        loop.create_task(async_main())
-    else:
-        loop.run_until_complete(async_main())
-
-    # Run the synchronous part
-    sync_main()
+    asyncio.run(async_main())

agent-py-bot/agents/webagent.py

@@ -203,7 +203,7 @@ def summarize_data(data):
     return summary
 
 def run_web_agent(topic, folder):
-    print(f"Running web agent for topic: {topic}")
+    print(f"[{datetime.now()}] Running web agent for topic: {topic}")
     news_data = search_news(topic)
     # save_data(news_data, folder)
     # summary = summarize_data(news_data)

agent-py-bot/scrape/raw/summary_log.txt

@@ -1,11 +0,0 @@
-2023-12-23 01:18:42.922812: Summarized information
-2023-12-25 17:02:01.477567: Summarized information
-2024-01-08 13:12:04.190959: Summarized information
-2024-01-08 13:13:03.437567: Summarized information
-2024-01-08 13:14:04.749784: Summarized information
-2024-01-08 13:15:06.100403: Summarized information
-2024-01-08 13:16:07.387491: Summarized information
-2024-01-08 13:17:09.016139: Summarized information
-2024-01-08 13:18:10.384559: Summarized information
-2024-01-08 13:19:12.129203: Summarized information
-2024-01-08 13:20:13.569597: Summarized information