更新 date non farm 抓取方法

main.py

@@ -32,7 +32,23 @@ def broadcast_message(message:str,chat_id:str):
         },
     )
 
+def find_indices(lst):
+    # 找到所有空字串的索引
+    empty_indices = [i for i, x in enumerate(lst) if x == '']
+    # 找到第二個連續兩個空字串的索引
+    start_index = None
+    for i in range(1, len(empty_indices)):
+        if empty_indices[i] - empty_indices[i-1] == 1 :
+            if start_index is not None:  # 如果已經找到第二個連續的空字串，則跳出迴圈
+                start_index = empty_indices[i-1]
+                break
+            start_index = empty_indices[i-1]
+    # 找到下一個空字串的索引
+    end_index = None
 
+    end_index = empty_indices[empty_indices.index(start_index)+2]
+
+    return start_index+2, end_index
 def find_cpi(driver):
     try:
         driver.get(cpi_url)
@@ -59,18 +75,14 @@ def find_fomc(driver, date:str):
 def find_non_farm(driver):
     try:
         driver.get(nonfarm_url)
-        for _ in range(100):
-            driver.execute_script("window.scrollTo(0, document.body.scrollHeight);")
-        # print all of the page source that was loaded
-        print(driver.page_source.encode("utf-8"))
         print("Open")
-        # 等待元素出現
-        WebDriverWait(driver, 10).until(
-            EC.presence_of_element_located((By.XPATH, '/html/body/div[2]/div[5]/div/div[1]/pre')))
+        date = driver.title.split(" ")[5]
+        value = driver.find_element(By.XPATH, '/html/body/div[2]/div[5]/div/div[1]/pre').text.split('\n')
+        start , end = find_indices(value)
+        result = value[start:end]
 
-        date = driver.find_element(By.XPATH, '/html/body/div[2]/div[5]/div/div[1]/pre').text.split('\n')[6]
-        value = driver.find_element(By.XPATH, '/html/body/div[2]/div[5]/div/div[1]/pre').text.split('\n')[8:12]
-        return driver, date, value
+
+        return driver, date, result
     except Exception as e:
         print(f"Error in find_non_farm: {e}")
         return driver, None, None
@@ -276,14 +288,14 @@ schedules = {
     convert_to_utc("2024/10/31", "20:30", 5): {"function": wrapper_function_pce, "args": ["September", "september-2024"]},
     convert_to_utc("2024/11/27", "21:30", 5): {"function": wrapper_function_pce, "args": ["October", "october-2024"]},
     convert_to_utc("2024/12/20", "21:30", 5): {"function": wrapper_function_pce, "args": ["November", "november-2024"]},
-    convert_to_utc("2024/06/05", "17:30", 5): {"function": wrapper_function_non_farm, "args": ["APRIL"]},
-    convert_to_utc("2024/06/07", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["MAY"]},
-    convert_to_utc("2024/07/05", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["JUNE"]},
-    convert_to_utc("2024/08/02", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["JULY"]},
-    convert_to_utc("2024/09/06", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["AUGUST"]},
-    convert_to_utc("2024/10/04", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["SEPTEMBER"]},
-    convert_to_utc("2024/11/01", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["OCTOBER"]},
-    convert_to_utc("2024/12/06", "21:30", 5): {"function": wrapper_function_non_farm, "args": ["NOVEMBER"]},
+    convert_to_utc("2024/06/06", "09:33", 5): {"function": wrapper_function_non_farm, "args": ["M04"]},
+    convert_to_utc("2024/06/07", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["M05"]},
+    convert_to_utc("2024/07/05", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["M06"]},
+    convert_to_utc("2024/08/02", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["M07"]},
+    convert_to_utc("2024/09/06", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["M08"]},
+    convert_to_utc("2024/10/04", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["M09"]},
+    convert_to_utc("2024/11/01", "20:30", 5): {"function": wrapper_function_non_farm, "args": ["M10"]},
+    convert_to_utc("2024/12/06", "21:30", 5): {"function": wrapper_function_non_farm, "args": ["M11"]},
     convert_to_utc("2024/06/13", "02:00", 5): {"function": wrapper_function_fomc, "args": ["June", "20240613a"]},
     convert_to_utc("2024/08/01", "02:00", 5): {"function": wrapper_function_fomc, "args": ["August", "20240801a"]},
     convert_to_utc("2024/09/19", "02:00", 5): {"function": wrapper_function_fomc, "args": ["September", "20240919a"]},
@@ -294,7 +306,7 @@ schedules = {
 if __name__ == "__main__":
     global nonfarm_url, cpi_url, fomc_url, pce_url, options
     options = Options()
-    # options.add_argument('--headless')
+    options.add_argument('--headless')
     options.add_argument('--no-sandbox')
     options.add_argument('--disable-dev-shm-usage')
     options.add_argument("--window-size=1920,1080")  # 可以根據需要調整這個大小
@@ -308,7 +320,6 @@ if __name__ == "__main__":
 
     print("Start Time:" , datetime.fromtimestamp(time.time()))
 
-    # schedule.every().day.at("10:44").do(wrapper_function_fomc, "March", "20240320a")
     for times, task in schedules.items():
         func = task["function"]
         args = task["args"]
@@ -321,10 +332,13 @@ if __name__ == "__main__":
             has_broadcasted = False
         time.sleep(0.1)  # Check every 0.1 seconds
 
-    # NonFarm
+    #NonFarm
+
     # driver = webdriver.Chrome(options=options)
     # driver, date, message = find_non_farm(driver)
+    # print("Non Farm Date")
     # print(date)
+    # print("Non Farm Message")
     # print(message)
     #CPI
     # find_cpi("FEBRUARY")