feat: Implement logging functionality; add logger configuration and retrieval

ai/compute.py

@@ -52,7 +52,6 @@ def reprompt(nextsteps: str, processor) -> None:
     scr = screenshot_to_base64(take_screenshot())
     return processor.process(nextsteps, img_data=scr)
 
-
 def _execute(name, args, processor):
     if name == "click_button":
         press_mouse(MouseInput(**args))

ai/processor.py

@@ -4,6 +4,10 @@ import openai
 from flask import jsonify
 from objects import aic
 import ai.compute
+from objects import logger as logger_module
+import logging
+
+logger: logging.Logger = logger_module.get_logger(__name__)
 
 class AIProcessor:
     def __init__(self, api_key: str, model: str = "gpt-4.1"):
@@ -62,6 +66,12 @@ class AIProcessor:
                         processor=self,
                     )
                     outputs.append(r) if r else None
+                # Make sure the two last messages from user has an image, but set disable_image to True for the others
+                for msg in self.session.messages[-2:]:
+                    msg.disable_image = True
+                logger.debug(
+                    self.session.messages
+                )
                 if reexec:
                     img = ai.compute.screenshot_to_base64(
                         ai.compute.take_screenshot()
@@ -86,15 +96,6 @@ class AIProcessor:
                 aic.Message(role="assistant", content=output_text)
             )
 
-            # re-execute if needed
-            if reexec:
-                img = ai.compute.screenshot_to_base64(
-                    ai.compute.take_screenshot()
-                )
-
-                outputs.append(
-                    *self.process(nextsteps, img)
-                )
             return outputs
         except Exception as e:
             traceback.print_exc()

objects/logger.py

@@ -0,0 +1,42 @@
+import logging
+import os
+from logging.handlers import RotatingFileHandler
+
+# Configuration values
+LOG_LEVEL = os.getenv("LOG_LEVEL", "INFO").upper()
+LOG_FORMAT = "%(asctime)s - %(name)s - %(levelname)s - %(message)s"
+LOG_DIR = os.getenv("LOG_DIR", os.path.join(os.getcwd(), "logs"))
+
+# Ensure log directory exists
+os.makedirs(LOG_DIR, exist_ok=True)
+
+# Log file path
+LOG_FILE = os.path.join(LOG_DIR, "app.log")
+
+# Create root logger
+logger = logging.getLogger("gpt-agent")
+logger.setLevel(LOG_LEVEL)
+
+# Formatter
+formatter = logging.Formatter(LOG_FORMAT)
+
+# Console handler
+console_handler = logging.StreamHandler()
+console_handler.setLevel(LOG_LEVEL)
+console_handler.setFormatter(formatter)
+logger.addHandler(console_handler)
+
+# Rotating file handler
+file_handler = RotatingFileHandler(LOG_FILE, maxBytes=5*1024*1024, backupCount=5)
+file_handler.setLevel(LOG_LEVEL)
+file_handler.setFormatter(formatter)
+logger.addHandler(file_handler)
+
+def get_logger(name: str | None = None) -> logging.Logger:
+    """
+    Retrieve a configured logger instance. If name is provided,
+    returns a child logger of the configured root logger.
+    """
+    if name:
+        return logger.getChild(name)
+    return logger