feat: process log file using python script

Author: lingjiefeng

.gitignore

@@ -22,4 +22,5 @@ public/manifest.json
 # benchmarking
 /results
 tasks_test.jsonl
-webwand_test_log.txt
+webwand_test_log.txt
+tasks_status.txt

benchmark.py

@@ -174,7 +174,7 @@ def click_extensions_icon(driver):
     top = window_position['y']
     right = window_position['x'] + window_position['width']
     # click Extensions icon
-    pyautogui.click(right - 150, top + 50)
+    pyautogui.click(right - 165, top + 50)
     # click webwand
     pyautogui.click(right - 300, top + 210)
 
@@ -196,10 +196,10 @@ def main():
             result = run_webwand_task(driver, task_id, task['ques'])
             logging.info(f'Task {task_id} status: {result}')
             # Optional: if the previous task timed out, reset the driver after each task to ensure proper state for the next task
-            # if result == "js-script-timeout":
-            #     driver.quit()
-            #     driver = setup_driver()
-            #     initial_load = True
+            if result == "js-script-timeout":
+                driver.quit()
+                driver = setup_driver()
+                initial_load = True
     driver.quit()
 
 if __name__ == "__main__":

process_log.py

@@ -0,0 +1,32 @@
+import re
+
+def process_log_file(log_file_path):
+    with open(log_file_path, 'r') as file:
+        log_data = file.readlines()
+
+    task_status_pattern = re.compile(r"Task ([\w\s-]+--\d+) status: (script-error|success|fail|doc-unload-max-retry|js-script-timeout|webdriver-error|python-script-error)")
+    tasks = []
+
+    for line in log_data:
+        match = task_status_pattern.search(line)
+        if match:
+            task_id = match.group(1)
+            status = match.group(2)
+            tasks.append((task_id, status))
+
+    return tasks
+
+def write_task_status(tasks):
+    header = "Task_id\tTask_status"
+    rows = [f"{task_id}\t{status}" for task_id, status in tasks]
+    return header + "\n" + "\n".join(rows)
+
+log_file_path = 'webwand_test_log.txt'
+tasks = process_log_file(log_file_path)
+formatted_output = write_task_status(tasks)
+
+output_file_path = 'tasks_status.txt'
+with open(output_file_path, 'w') as output_file:
+    output_file.write(formatted_output)
+
+print(f"Tasks results are saved to {output_file_path}")