'commit'

WeiXin/T5_SendText_ActionTest.py

@@ -74,18 +74,33 @@ def main():
     #   - Input text
     #   - Find and click Send button
     
-    logger.info("Step 1 & 2: Finding input box and writing text...")
+    logger.info("Step 1: Finding input box and writing text...")
     
-    # We can get a rough position first for logging, though perform_input_action is more robust
-    screenshot_path = os.path.join(WxUtil.OUTPUT_DIR, "T5_Initial_Check.jpg")
-    d.screenshot(screenshot_path)
-    input_pos, _ = WxUtil.find_input_box_center(screenshot_path)
-    logger.info(f"Initial target input position: {input_pos}")
+    # We pass debug_prefix="T5" to perform_input_action to get intermediate screenshots
+    # The function will save:
+    #   - T5_1_check_mode.jpg
+    #   - T5_2_after_mode.jpg
+    #   - T5_3_input_box_found.jpg (if template matched)
+    #   - T5_4_after_input.jpg
+    #   - T5_5_after_send.jpg
+    
+    # Initial state screenshot
+    initial_path = os.path.join(WxUtil.OUTPUT_DIR, "T5_0_Initial.jpg")
+    d.screenshot(initial_path)
+    logger.info(f"Initial state screenshot saved: {initial_path}")
 
-    success = WxUtil.perform_input_action(d, input_pos, test_text, auto_send=True)
+    # Use a more robust way to find input_pos for logging
+    input_pos, _ = WxUtil.find_input_box_center(initial_path)
+    logger.info(f"Target input position (estimated): {input_pos}")
+
+    success = WxUtil.perform_input_action(d, input_pos, test_text, auto_send=True, debug_prefix="T5")
     
     if success:
         logger.info("✅ Step 3: Send action completed successfully.")
+        # Final state screenshot
+        final_path = os.path.join(WxUtil.OUTPUT_DIR, "T5_6_Final.jpg")
+        d.screenshot(final_path)
+        logger.info(f"Final state screenshot saved: {final_path}")
     else:
         logger.error("❌ Send action failed.")
 

WeiXin/WxUtil.py

@@ -1186,13 +1186,21 @@ def find_all_template_matches(screen_path, template_path, threshold=0.8):
         logger.error(f"find_all_template_matches failed: {e}")
         return []
 
-def perform_input_action(d, center_point, text, auto_send=True):
+def perform_input_action(d, center_point, text, auto_send=True, debug_prefix=None):
     """
     执行输入操作
+    :param debug_prefix: 如果提供，将在关键步骤保存截图，如 {debug_prefix}_before_mode.jpg
     """
     try:
+        def save_debug_shot(name):
+            if debug_prefix:
+                shot_path = os.path.join(OUTPUT_DIR, f"{debug_prefix}_{name}.jpg")
+                d.screenshot(shot_path)
+                logger.info(f"保存中间过程截图: {shot_path}")
+
         # --- 新增逻辑：确保处于文字输入模式 ---
         logger.info("正在检查输入模式...")
+        save_debug_shot("1_check_mode")
         
         # 优先使用 uiautomator2 的属性检测（比图像识别更稳）
         # 1. 检查是否有 "切换到键盘" 按钮（说明当前是语音模式）
@@ -1230,6 +1238,7 @@ def perform_input_action(d, center_point, text, auto_send=True):
                 except:
                     pass
         # --- 新增逻辑结束 ---
+        save_debug_shot("2_after_mode")
 
         # 1. 尝试找到原生输入框并输入
         # 增加多种查找方式
@@ -1252,6 +1261,7 @@ def perform_input_action(d, center_point, text, auto_send=True):
                 
                 if input_pos:
                     logger.info(f"✅ [Template] 通过 input_text.jpg 找到输入框: {input_pos}")
+                    save_debug_shot("3_input_box_found")
                     
                     # 绘制调试图 (蓝框)
                     try:
@@ -1335,6 +1345,7 @@ def perform_input_action(d, center_point, text, auto_send=True):
                 # 尝试粘贴
                 d.press("paste")
         
+        save_debug_shot("4_after_input")
         time.sleep(1.0)
         
         # 3. 发送
@@ -1360,6 +1371,7 @@ def perform_input_action(d, center_point, text, auto_send=True):
                     d.press("enter")
                     logger.info("Pressed Enter")
             
+            save_debug_shot("5_after_send")
             # 清理临时文件
             if os.path.exists(tmp_screen):
                 try: