ota with 4g

main.py

@@ -156,13 +156,33 @@ class ATClient:
             except:
                 return str(self._resp)
 
+    def _find_urc_tag(self, tag: bytes):
+        """
+        只在“真正的 URC 边界”查找 tag，避免误命中 HTTP payload 内容。
+        规则：tag 必须出现在 buffer 开头，或紧跟在 b"\\r\\n" 后面。
+        """
+        try:
+            i = 0
+            rx = self._rx
+            while True:
+                j = rx.find(tag, i)
+                if j < 0:
+                    return -1
+                if j == 0:
+                    return 0
+                if j >= 2 and rx[j - 2:j] == b"\r\n":
+                    return j
+                i = j + 1
+        except:
+            return -1
+
     def _parse_mipurc_rtcp(self):
         """
         解析：+MIPURC: "rtcp",<link_id>,<len>,<payload...>
         之前硬编码 link_id=0 会导致在多连接/重连场景下收不到数据。
         """
         prefix = b'+MIPURC: "rtcp",'
-        i = self._rx.find(prefix)
+        i = self._find_urc_tag(prefix)
         if i < 0:
             return False
         # 丢掉前置噪声
@@ -215,7 +235,7 @@ class ATClient:
 
     def _parse_mhttpurc_header(self):
         tag = b'+MHTTPURC: "header",'
-        i = self._rx.find(tag)
+        i = self._find_urc_tag(tag)
         if i < 0:
             return False
         if i > 0:
@@ -254,7 +274,7 @@ class ATClient:
 
     def _parse_mhttpurc_content(self):
         tag = b'+MHTTPURC: "content",'
-        i = self._rx.find(tag)
+        i = self._find_urc_tag(tag)
         if i < 0:
             return False
         if i > 0:
@@ -294,19 +314,26 @@ class ATClient:
 
     def _reader_loop(self):
         while self._running:
-            d = self.uart.read(4096)
+            # 关键：UART 驱动偶发 read failed，必须兜住，否则线程挂了 OTA/TCP 都会卡死
+            try:
+                d = self.uart.read(4096)  # 8192 在一些驱动上更容易触发 read failed
+            except Exception as e:
+                try:
+                    print("[ATClient] uart read failed:", e)
+                except:
+                    pass
+                time.sleep_ms(50)
+                continue
+
             if not d:
-                time.sleep_ms(2)
+                time.sleep_ms(1)
                 continue
 
             with self._q_lock:
-                # 统一累积到内部 buffer（用于 URC 解析）
                 self._rx += d
-                # 命令等待期间，把原始字节流复制到响应缓冲（不影响 URC 解析）
                 if self._waiting:
                     self._resp += d
 
-                # 解析 URC：尽可能多地从 _rx 中剥离完整 URC，避免丢包
                 while True:
                     progressed = (
                         self._parse_mipurc_rtcp()
@@ -316,17 +343,16 @@ class ATClient:
                     if not progressed:
                         break
 
-                # 防止 _rx 因为"非 URC 文本/回显"无限增长：保留尾部即可
+                try:
-                # 关键修复：如果 buffer 中有 HTTP URC 标签（说明 OTA 在进行），完全禁用截断
+                    ota_flag = int(globals().get("ota_in_progress", 0)) > 0
-                # 避免在 OTA 下载时截断 buffer 导致数据丢失（之前 16KB 限制太小，导致数据被截断）
+                except:
-                has_http_urc = (b'+MHTTPURC: "content"' in self._rx or 
+                    ota_flag = False
-                               b'+MHTTPURC: "header"' in self._rx)
+
-                if has_http_urc:
+                has_http_hint = (b"+MHTTP" in self._rx) or (b"+MHTTPURC" in self._rx)
-                    # OTA 下载中：完全禁用 buffer 截断，避免数据丢失
+                if ota_flag or has_http_hint:
-                    # 通常 OTA 文件不会超过几百 KB，即使不截断也不会导致内存问题
+                    if len(self._rx) > 512 * 1024:
-                    pass  # 不截断
+                        self._rx = self._rx[-256 * 1024:]
                 else:
-                    # 非 OTA 状态：使用较小的 buffer 限制（16KB）
                     if len(self._rx) > 16384:
                         self._rx = self._rx[-4096:]
 
@@ -435,7 +461,9 @@ normal_send_queue = [] # 普通发送队列：电量/校准结果/状态等
 queue_lock = _Mutex()  # 互斥锁，保护队列
 uart4g_lock = _Mutex() # 互斥锁，保护 4G 串口 AT 发送流程（防并发）
 update_thread_started = False  # 防止 OTA 更新线程重复启动
-ota_in_progress = False  # OTA(4G HTTP URC) 期间暂停 tcp_main 读取 uart4g，避免吞掉 +MHTTPURC
+# OTA(4G HTTP URC) 期间“暂停 TCP 活动/让读线程用大 buffer”：
+# 用计数器而不是 bool，避免“外层 OTA 还没结束，内层 downloader finally 又把它置 False”。
+ota_in_progress = 0
 
 # ==================== 工具函数 ====================
 
@@ -457,6 +485,121 @@ def download_file(url, filename):
         return f"下载失败！发生未知错误: {e}"
 
 
+def _download_file_system_bytes(url, filename, timeout_s=25):
+    """
+    走“系统网络栈”的下载（RNDIS/ECM/eth0 等），用 TCP 自带重传，适合 OTA 大文件。
+    注意：这里写 bytes，不做编码假设。
+    """
+    try:
+        print(f"[NET] system download: {url} -> {filename}")
+        r = requests.get(url, timeout=timeout_s)
+        r.raise_for_status()
+        data = r.content
+        with open(filename, "wb") as f:
+            f.write(data)
+        return True, f"OK size={len(data)}"
+    except Exception as e:
+        return False, f"system_download_failed: {e}"
+
+
+def _has_default_route():
+    try:
+        out = os.popen("ip route 2>/dev/null").read()
+        for line in out.splitlines():
+            if line.strip().startswith("default "):
+                return True
+        return False
+    except:
+        return False
+
+
+def _get_if_ipv4(ifname: str):
+    try:
+        out = os.popen(f"ip -4 addr show {ifname} 2>/dev/null").read()
+        m = re.search(r"\binet\s+(\d+\.\d+\.\d+\.\d+)/(\d+)", out)
+        if not m:
+            return None
+        return m.group(1)
+    except:
+        return None
+
+
+def _ping_once(ip: str, ifname=None, timeout_s=1):
+    # busybox ping 可能不支持 -W；做两套尝试
+    if ifname:
+        cmd1 = f"ping -I {ifname} -c 1 -W {timeout_s} {ip} >/dev/null 2>&1"
+        cmd2 = f"ping -I {ifname} -c 1 {ip} >/dev/null 2>&1"
+    else:
+        cmd1 = f"ping -c 1 -W {timeout_s} {ip} >/dev/null 2>&1"
+        cmd2 = f"ping -c 1 {ip} >/dev/null 2>&1"
+    rc = os.system(cmd1)
+    if rc == 0:
+        return True
+    rc = os.system(cmd2)
+    return rc == 0
+
+
+def ensure_ml307r_dialup_and_route(prefer_if=("usb0", "usb1"), metric=200, debug=True):
+    """
+    用 ML307R 的 RNDIS/ECM 方式把“系统网络”拉起来：
+    - AT+MDIALUP=1,1（拨号）
+    - 为 usb0/usb1 添加 default route（不改 IP，尽量不影响现有管理网段）
+    目标：让 requests.get() 能直接下载 OTA，避免 UART URC 丢包。
+    """
+    def _dlog(*a):
+        if debug:
+            print("[DIALUP]", *a)
+
+    # 1) 拨号（重复执行也安全）
+    try:
+        with uart4g_lock:
+            at('AT+MDIALUPCFG="mode"', "OK", 2000)
+            r = at("AT+MDIALUP=1,1", "OK", 20000)
+        _dlog("MDIALUP resp:", r)
+    except Exception as e:
+        _dlog("MDIALUP exception:", e)
+
+    # 2) 如果已经有 default route，直接返回
+    if _has_default_route():
+        _dlog("default route already exists")
+        return True
+
+    # 3) 尝试在 usb0/usb1 上猜测网关并加默认路由
+    for ifname in prefer_if:
+        ip = _get_if_ipv4(ifname)
+        if not ip:
+            continue
+
+        parts = ip.split(".")
+        if len(parts) != 4:
+            continue
+        base = ".".join(parts[:3])
+        last = int(parts[3])
+
+        # 常见网关候选：.2 / .254 / .1(跳过自己)
+        candidates = []
+        if last != 2:
+            candidates.append(f"{base}.2")
+        if last != 254:
+            candidates.append(f"{base}.254")
+        if last != 1:
+            candidates.append(f"{base}.1")
+
+        for gw in candidates:
+            if gw == ip:
+                continue
+            # ping 一下让 ARP/neigh 建立（不一定通，但很多系统会因此学到邻居）
+            _ping_once(gw, ifname=ifname, timeout_s=1)
+            # 直接尝试加默认路由（若已存在会失败但无害）
+            os.system(f"ip route add default via {gw} dev {ifname} metric {metric} 2>/dev/null")
+            if _has_default_route():
+                _dlog("default route set:", gw, "dev", ifname)
+                _dlog(os.popen("ip route 2>/dev/null").read().strip())
+                return True
+
+    _dlog("failed to set default route. ip route:", os.popen("ip route 2>/dev/null").read().strip())
+    return False
+
 def is_server_reachable(host, port=80, timeout=5):
     """检查目标主机端口是否可达（用于 OTA 前网络检测）"""
     try:
@@ -1023,6 +1166,14 @@ def tcp_main():
     global tcp_connected, high_send_queue, normal_send_queue, queue_lock, laser_calibration_active, laser_calibration_result, laser_calibration_lock, update_thread_started
     send_hartbeat_fail_count = 0
     while not app.need_exit():
+        # OTA 期间不要 connect/登录/心跳/发送，避免与 HTTP URC 下载抢 uart4g_lock 导致心跳超时被服务器断开
+        try:
+            if ota_in_progress:
+                time.sleep_ms(200)
+                continue
+        except:
+            pass
+
         if not connect_server():
             time.sleep_ms(5000)
             continue
@@ -1039,6 +1190,14 @@ def tcp_main():
         last_heartbeat_ack_time = time.ticks_ms()
         last_heartbeat_send_time = time.ticks_ms()
         while True:
+            # OTA 期间暂停 TCP 活动（不发心跳、不发业务），让下载独占 4G 串口
+            try:
+                if ota_in_progress:
+                    time.sleep_ms(200)
+                    continue
+            except:
+                pass
+
             # 接收数据（唯一来源：ATClient 解析后的 TCP payload 队列）
             item = at_client.pop_tcp_payload()
             if item:
@@ -1257,6 +1416,14 @@ def laser_calibration_worker():
     """后台线程：持续检测是否需要执行激光校准"""
     global laser_calibration_active, laser_calibration_result, laser_calibration_lock
     while True:
+        # OTA 期间尽量省电：暂停后台校准（会占用 Camera）
+        try:
+            if int(globals().get("ota_in_progress", 0)) > 0:
+                time.sleep_ms(200)
+                continue
+        except:
+            pass
+
         if laser_calibration_active:
             # 关键：不要在每次尝试里反复 new Camera（会导致 MMF 反复初始化刷屏）
             cam = None
@@ -1296,7 +1463,7 @@ def laser_calibration_worker():
 
 
 
-def download_file_via_4g(url, filename,
+def download_file_via_4g_legacy(url, filename,
                          total_timeout_ms=30000,
                          retries=3,
                          debug=False):
@@ -1444,17 +1611,34 @@ def download_file_via_4g(url, filename,
         if got_ranges is None:
             got_ranges = set()
         filled_new_bytes = 0
+        # last_sum/resp_total 用于判断“本次 HTTP 响应体”是否接收完成（尤其是 Range 场景）
         last_sum = 0
+        resp_total = None
         no_progress_count = 0  # 连续没有进展的次数
         last_print_ms = time.ticks_ms()
         last_print_sum = 0
 
+        # Range 补洞不需要等太久，避免卡死；全量下载用总超时
+        attempt_timeout_ms = total_timeout_ms
+        if range_start is not None and range_end is not None:
+            attempt_timeout_ms = min(total_timeout_ms, 8000)
+
         t0 = time.ticks_ms()
-        while time.ticks_ms() - t0 < total_timeout_ms:
+        while time.ticks_ms() - t0 < attempt_timeout_ms:
             ev = at_client.pop_http_event()
             if not ev:
                 # 如果 sum 已经达到 total_len，但仍有 gaps，等待更长时间（有些分片可能延迟到达）
-                if total_len and last_sum >= total_len:
+                # 对 Range：last_sum 只会到 resp_total（比如 686/774），不能拿 total_len(59776) 比
+                if resp_total and last_sum >= resp_total:
+                    # 本次响应体应该收齐了，继续等一小会儿（防止最后一个 URC 延迟），然后退出循环
+                    time.sleep_ms(30)
+                    no_progress_count += 1
+                    if no_progress_count > 30:
+                        break
+                    continue
+
+                # 全量模式：如果模块宣称 sum 已经达到 total_len，但仍有 gaps，稍微多等
+                if (range_start is None and range_end is None) and total_len and last_sum >= total_len:
                     gaps_now, merged_now = _compute_gaps(total_len, got_ranges)
                     if gaps_now and not (len(gaps_now) == 1 and gaps_now[0] == (0, 0)):
                         time.sleep_ms(50)
@@ -1463,7 +1647,10 @@ def download_file_via_4g(url, filename,
                 else:
                     time.sleep_ms(5)
                 no_progress_count += 1
-                # 如果长时间没有新事件，且 sum 已经达到 total_len，认为接收完成（可能有丢包）
+                # Range：如果长时间没有事件也结束（让上层重试）
+                if range_start is not None and range_end is not None and no_progress_count > 200:
+                    break
+                # 全量：如果长时间没有新事件，且 sum 已经达到 total_len，认为接收完成（可能有丢包）
                 if no_progress_count > 100 and total_len and last_sum >= total_len:
                     break
                 continue
@@ -1495,6 +1682,10 @@ def download_file_via_4g(url, filename,
                     if body_buf is None and total_len:
                         body_buf = bytearray(total_len)
 
+                # 对 Range 响应：优先使用 Content-Length 作为本次响应体长度
+                if expect_len is not None:
+                    resp_total = expect_len
+
                 _log(f"[HDR] id={hid} code={code} len={expect_len} md5={expect_md5_b64}")
                 continue
 
@@ -1515,6 +1706,10 @@ def download_file_via_4g(url, filename,
                 if body_buf is None or total_len is None:
                     continue
 
+                # 若 header 没给 Content-Length，就用 content 的 _total 作为本次响应体长度（Range 场景下通常是这次 body 的长度）
+                if resp_total is None:
+                    resp_total = _total
+
                 rel_start = _sum - _cur
                 rel_end = _sum
                 abs_start = offset_base + rel_start
@@ -1553,6 +1748,12 @@ def download_file_via_4g(url, filename,
                     # 不要用 filled_new_bytes 判断是否完整（可能有重叠）
                     pass
 
+                # Range：本次响应体已收齐，退出，交给上层判断是否补上了缺口
+                if resp_total is not None and last_sum >= resp_total:
+                    # 给一点时间让可能的尾部事件入队，然后退出
+                    time.sleep_ms(10)
+                    break
+
         # 5) 清理实例
         at(f"AT+MHTTPDEL={httpid}", "OK", 3000)
 
@@ -1567,7 +1768,10 @@ def download_file_via_4g(url, filename,
         return True, f"PARTIAL ok +{filled_new_bytes} ip={ip} code={status_code}", body_buf, got_ranges, total_len, expect_md5_b64
 
     global ota_in_progress
-    ota_in_progress = True
+    try:
+        ota_in_progress = int(ota_in_progress) + 1
+    except:
+        ota_in_progress = 1
     with uart4g_lock:
         try:
             # -------- Phase 1: 全量 GET（允许不完整，后面用 Range 补洞）--------
@@ -1604,12 +1808,21 @@ def download_file_via_4g(url, filename,
                 gaps = []
 
             # -------- Phase 2: Range 补洞 --------
-            # 限制单次 Range 的最大长度（太大也会触发同样的 UART 压力）
+            # 说明：
-            MAX_RANGE_BYTES = 8192
+            # - “全量 GET 多次重试 + 合并已收到分片”我们已经在 Phase1 做了（got_ranges/body_buf 会跨 attempt 累积）。
-            RANGE_RETRIES_EACH = 2
+            # - 仍存在 gaps 说明：这些字节段在全量阶段始终没收到，需要靠 Range 反复补洞。
-            MAX_HOLE_ROUNDS = 10
+            #
+            # 策略：
+            # - Range 分块更小（更稳），失败时继续“二分缩小”到 MIN_RANGE_BYTES；
+            # - 不要因为某一轮 no_progress 就立刻退出（UART 偶发丢 URC，需要多轮撞上一次成功）。
+            MAX_RANGE_BYTES = 1024
+            MIN_RANGE_BYTES = 128
+            RANGE_RETRIES_EACH = 8
+            MAX_HOLE_ROUNDS = 50
+            NO_PROGRESS_ROUNDS_LIMIT = 8
 
             round_i = 0
+            no_progress_rounds = 0
             while gaps and round_i < MAX_HOLE_ROUNDS:
                 round_i += 1
                 # 优先补最大的洞（通常只丢中间一两段）
@@ -1620,8 +1833,9 @@ def download_file_via_4g(url, filename,
                 # 每轮最多补前 5 个洞，避免无限循环
                 for (gs, ge) in gaps[:5]:
                     cur = gs
+                    chunk = MAX_RANGE_BYTES
                     while cur < ge:
-                        sub_end = min(ge, cur + MAX_RANGE_BYTES)
+                        sub_end = min(ge, cur + chunk)
                         # HTTP Range end is inclusive
                         rs = cur
                         re_incl = sub_end - 1
@@ -1641,19 +1855,28 @@ def download_file_via_4g(url, filename,
                                 sub_ok = True
                                 break
                             _log(f"[RETRY] range {rs}-{re_incl} try={k} failed={msg2}")
-                            time.sleep_ms(120)
+                            time.sleep_ms(150)
 
                         after_gaps, after_merged = _compute_gaps(total_len, got_ranges)
                         after_filled = sum(e - s for s, e in after_merged)
                         if after_filled > before_filled:
                             progress_any = True
-
+                            # 成功推进：恢复到较大 chunk，加快补洞
-                        if not sub_ok:
+                            chunk = MAX_RANGE_BYTES
-                            _log(f"[WARN] range {rs}-{re_incl} failed={sub_err}")
+                            cur = sub_end
+                        else:
+                            # 没推进：缩小 chunk，继续在同一位置重试；不要前进 cur
+                            if chunk > MIN_RANGE_BYTES:
+                                chunk = max(MIN_RANGE_BYTES, chunk // 2)
+                                _log(f"[RANGE] shrink chunk -> {chunk} at pos={cur}")
+                            else:
+                                # 已经很小还不行：本轮先放弃这个位置，留给下一轮再撞
+                                if not sub_ok:
+                                    _log(f"[WARN] range {rs}-{re_incl} failed={sub_err}")
+                                break
 
                         # 小歇一下，给读线程喘息
-                        time.sleep_ms(80)
+                        time.sleep_ms(120)
-                        cur = sub_end
 
                 gaps, merged = _compute_gaps(total_len, got_ranges)
                 if gaps and gaps[0] == (0, 0):
@@ -1663,9 +1886,16 @@ def download_file_via_4g(url, filename,
                 if not gaps:
                     break
                 if not progress_any:
-                    # 本轮没有推进，退出避免死循环
+                    no_progress_rounds += 1
-                    _log(f"[RANGE] no progress in round={round_i}, stop. filled={filled_total}/{total_len}")
+                    _log(f"[RANGE] no progress in round={round_i} ({no_progress_rounds}/{NO_PROGRESS_ROUNDS_LIMIT}) filled={filled_total}/{total_len}")
-                    break
+                    # 多轮无进展才退出（避免偶发“只 header 无 content URC”导致过早退出）
+                    if no_progress_rounds >= NO_PROGRESS_ROUNDS_LIMIT:
+                        break
+                    # 退避等待一下再继续下一轮
+                    time.sleep_ms(500)
+                    continue
+                else:
+                    no_progress_rounds = 0
                 _log(f"[RANGE] round={round_i} filled={filled_total}/{total_len} gaps={gaps[:3]}")
 
             # 完整性检查
@@ -1690,34 +1920,493 @@ def download_file_via_4g(url, filename,
 
             return True, f"OK size={len(data)} ip={_get_ip()} md5={expect_md5_b64 or ''}"
         finally:
-            ota_in_progress = False
+            try:
+                ota_in_progress = max(0, int(ota_in_progress) - 1)
+            except:
+                ota_in_progress = 0
+
+
+def download_file_via_4g(url, filename,
+                         total_timeout_ms=600000,
+                         retries=3,
+                         debug=True):
+    """
+    ML307R HTTP 下载（更稳的“固定小块 Range 顺序下载”）：
+    - 只依赖 +MHTTPURC:"header"/"content"（不依赖 MHTTPREAD/cached）
+    - 每次只请求一个小块 Range（默认 1024B），失败就重试同一块，必要时缩小块大小
+    - 每个 chunk 都重新 MHTTPCREATE/MHTTPREQUEST，避免卡在“206 header 但不吐 content”的坏状态
+    """
+    from urllib.parse import urlparse
+
+    # 小块策略（可按现场再调）
+    # - chunk 越小越稳（URC 压力更小），代价是请求次数更多
+    CHUNK_MAX = 10240
+    CHUNK_MIN = 128
+    CHUNK_RETRIES = 12
+    FRAG_SIZE = 1024     # 0-1024
+    FRAG_DELAY = 10    # 0-2000 ms
+
+    t_func0 = time.ticks_ms()
+
+    parsed = urlparse(url)
+    host = parsed.hostname
+    path = parsed.path or "/"
+    if not host:
+        return False, "bad_url (no host)"
+
+    # 很多 ML307R 的 MHTTP 对 https 不稳定；对已知域名做降级
+    if isinstance(url, str) and url.startswith("https://static.shelingxingqiu.com/"):
+        base_url = "http://static.shelingxingqiu.com"
+    else:
+        base_url = f"http://{host}"
+
+    def _log(*a):
+        if debug:
+            print(*a)
+
+    def _pwr_log(prefix=""):
+        """debug 用：输出电压/电量，用于判断是否掉压导致 4G/USB 异常"""
+        if not debug:
+            return
+        try:
+            v = get_bus_voltage()
+            p = voltage_to_percent(v)
+            print(f"[PWR]{prefix} v={v:.3f}V p={p}%")
+        except Exception as e:
+            try:
+                print(f"[PWR]{prefix} read_failed: {e}")
+            except:
+                pass
+
+    def _clear_http_events():
+        while at_client.pop_http_event() is not None:
+            pass
+
+    def _parse_httpid(resp: str):
+        m = re.search(r"\+MHTTPCREATE:\s*(\d+)", resp)
+        return int(m.group(1)) if m else None
+
+    def _get_ip():
+        r = at("AT+CGPADDR=1", "OK", 3000)
+        m = re.search(r'\+CGPADDR:\s*1,"([^"]+)"', r)
+        return m.group(1) if m else ""
+
+    def _ensure_pdp():
+        ip = _get_ip()
+        if ip and ip != "0.0.0.0":
+            return True, ip
+        at("AT+MIPCALL=1,1", "OK", 15000)
+        for _ in range(10):
+            ip = _get_ip()
+            if ip and ip != "0.0.0.0":
+                return True, ip
+            time.sleep(1)
+        return False, ip
+
+    def _extract_hdr_fields(hdr_text: str):
+        mlen = re.search(r"Content-Length:\s*(\d+)", hdr_text, re.IGNORECASE)
+        clen = int(mlen.group(1)) if mlen else None
+        mmd5 = re.search(r"Content-Md5:\s*([A-Za-z0-9+/=]+)", hdr_text, re.IGNORECASE)
+        md5_b64 = mmd5.group(1).strip() if mmd5 else None
+        return clen, md5_b64
+
+    def _extract_content_range(hdr_text: str):
+        m = re.search(r"Content-Range:\s*bytes\s*(\d+)\s*-\s*(\d+)\s*/\s*(\d+)", hdr_text, re.IGNORECASE)
+        if not m:
+            return None, None, None
+        try:
+            return int(m.group(1)), int(m.group(2)), int(m.group(3))
+        except:
+            return None, None, None
+
+    def _hard_reset_http():
+        """
+        模块进入“坏状态”时的保守清场：
+        - 清空 ATClient 的事件队列，避免串台
+        - 删除 0..5 的 httpid（常见固件槽位范围），尽量把内部 HTTP 状态机拉回干净
+        注意：很慢，所以只在连续异常时调用。
+        """
+        _clear_http_events()
+        for i in range(0, 6):
+            try:
+                at(f"AT+MHTTPDEL={i}", "OK", 1200)
+            except:
+                pass
+        _clear_http_events()
+
+    def _create_httpid(full_reset=False):
+        _clear_http_events()
+        if full_reset:
+            _hard_reset_http()
+        resp = at(f'AT+MHTTPCREATE="{base_url}"', "OK", 8000)
+        hid = _parse_httpid(resp)
+        return hid, resp
+
+    def _fetch_range_into_buf(start, want_len, out_buf, full_reset=False):
+        """
+        请求 Range [start, start+want_len) ，写入 out_buf（bytearray，长度=want_len）
+        返回 (ok, msg, total_len, md5_b64, got_len)
+        """
+        end_incl = start + want_len - 1
+        hid, cresp = _create_httpid(full_reset=full_reset)
+        if hid is None:
+            return False, f"MHTTPCREATE failed: {cresp}", None, None, 0
+
+        # 降低 URC 压力（分片/延迟）
+        at(f'AT+MHTTPCFG="fragment",{hid},{FRAG_SIZE},{FRAG_DELAY}', "OK", 1500)
+        # 设置 Range header（inclusive）
+        at(f'AT+MHTTPCFG="header",{hid},"Range: bytes={start}-{end_incl}"', "OK", 3000)
+
+        req = at(f'AT+MHTTPREQUEST={hid},1,0,"{path}"', "OK", 15000)
+        if "ERROR" in req or "CME ERROR" in req:
+            at(f"AT+MHTTPDEL={hid}", "OK", 2000)
+            return False, f"MHTTPREQUEST failed: {req}", None, None, 0
+
+        # 等 header + content
+        # 注意：部分 ML307R 固件会把 header 分成多条 +MHTTPURC:"header" 分片吐出来，
+        # 其中有的分片只有 Content-Length，有的只有 Content-Range。
+        # 因此这里需要做“累积解析”，否则会出现 resp_total=None -> no_header_or_total。
+        hdr_text = None
+        hdr_accum = ""
+        code = None
+        resp_total = None
+        total_len = None
+        md5_b64 = None
+
+        got_ranges = set()
+        last_sum = 0
+        t0 = time.ticks_ms()
+        # Range 场景不宜等待太久，卡住就换 hid 重来
+        timeout_ms = 9000
+        logged_hdr = False
+
+        while time.ticks_ms() - t0 < timeout_ms:
+            ev = at_client.pop_http_event()
+            if not ev:
+                time.sleep_ms(5)
+                continue
+
+            if ev[0] == "header":
+                _, ehid, ecode, ehdr = ev
+                if ehid != hid:
+                    continue
+                code = ecode
+                hdr_text = ehdr
+                # 累积 header 文本并从累积内容里提取字段（避免 split header 丢字段）
+                if ehdr:
+                    hdr_accum = (hdr_accum + "\n" + ehdr) if hdr_accum else ehdr
+
+                resp_total_tmp, md5_tmp = _extract_hdr_fields(hdr_accum)
+                if md5_tmp:
+                    md5_b64 = md5_tmp
+                cr_s, cr_e, cr_total = _extract_content_range(hdr_accum)
+                if cr_total is not None:
+                    total_len = cr_total
+                # 有些 header 没有 Content-Length，但有 Content-Range（206），可由 range 计算出本次 body 长度
+                if resp_total_tmp is not None:
+                    resp_total = resp_total_tmp
+                elif resp_total is None and (cr_s is not None) and (cr_e is not None) and (cr_e >= cr_s):
+                    resp_total = (cr_e - cr_s + 1)
+                # 206 才是 Range 正常响应；部分服务器可能忽略 Range 返回 200
+                # 节流：每个 hid 只打一次 header（否则你会看到连续 3-4 条 [HDR]，且很多 cr=None）
+                if (not logged_hdr) and (resp_total is not None or total_len is not None):
+                    _log(f"[HDR] id={hid} code={code} clen={resp_total} cr={cr_s}-{cr_e}/{cr_total}")
+                    logged_hdr = True
+                continue
+
+            if ev[0] == "content":
+                _, ehid, _total, _sum, _cur, payload = ev
+                if ehid != hid:
+                    continue
+                if resp_total is None:
+                    resp_total = _total
+                if resp_total is None or resp_total <= 0:
+                    continue
+                start_rel = _sum - _cur
+                end_rel = _sum
+                if start_rel < 0 or start_rel >= resp_total:
+                    continue
+                if end_rel > resp_total:
+                    end_rel = resp_total
+                actual_len = min(len(payload), end_rel - start_rel)
+                if actual_len <= 0:
+                    continue
+                out_buf[start_rel:start_rel + actual_len] = payload[:actual_len]
+                got_ranges.add((start_rel, start_rel + actual_len))
+                if _sum > last_sum:
+                    last_sum = _sum
+                # 进度日志节流
+                if debug and (last_sum >= resp_total or (last_sum % 512 == 0)):
+                    _log(f"[CHUNK] {start}+{last_sum}/{resp_total}")
+
+                # 收齐就退出
+                if last_sum >= resp_total:
+                    break
+
+        # 清理实例（快路径：只删当前 hid）
+        try:
+            at(f"AT+MHTTPDEL={hid}", "OK", 2000)
+        except:
+            pass
+
+        if resp_total is None:
+            return False, "no_header_or_total", total_len, md5_b64, 0
+
+        # 计算实际填充长度
+        merged = sorted(got_ranges)
+        filled = 0
+        prev = 0
+        for s, e in merged:
+            if e <= s:
+                continue
+            if s > prev:
+                # 有洞
+                pass
+            prev = max(prev, e)
+        # 重新合并算 filled
+        merged2 = []
+        for s, e in merged:
+            if not merged2 or s > merged2[-1][1]:
+                merged2.append((s, e))
+            else:
+                merged2[-1] = (merged2[-1][0], max(merged2[-1][1], e))
+        filled = sum(e - s for s, e in merged2)
+
+        if filled < resp_total:
+            return False, f"incomplete_chunk got={filled} expected={resp_total} code={code}", total_len, md5_b64, filled
+
+        got_len = resp_total
+        # 如果服务器忽略 Range 返回 200，resp_total 可能是整文件，这里允许 want_len 不匹配
+        return True, "OK", total_len, md5_b64, got_len
+
+    global ota_in_progress
+    try:
+        ota_in_progress = int(ota_in_progress) + 1
+    except:
+        ota_in_progress = 1
+    with uart4g_lock:
+        try:
+            ok_pdp, ip = _ensure_pdp()
+            if not ok_pdp:
+                return False, f"PDP not ready (ip={ip})"
+
+            # 先清空旧事件，避免串台
+            _clear_http_events()
+
+            # 为了支持随机写入，先创建空文件
+            try:
+                with open(filename, "wb") as f:
+                    f.write(b"")
+            except Exception as e:
+                return False, f"open_file_failed: {e}"
+
+            total_len = None
+            expect_md5_b64 = None
+
+            offset = 0
+            chunk = CHUNK_MAX
+            t_start = time.ticks_ms()
+            last_progress_ms = t_start
+            STALL_TIMEOUT_MS = 60000  # 60s 没有任何 offset 推进则判定卡死
+            last_pwr_ms = t_start
+            _pwr_log(prefix=" ota_start")
+            bad_http_state = 0  # 连续“疑似模块 HTTP 坏状态”的计数，达到阈值才做 full reset
+
+            while True:
+                now = time.ticks_ms()
+                # debug：每 5 秒打印一次电压/电量 + 进度
+                if debug and time.ticks_diff(now, last_pwr_ms) >= 5000:
+                    last_pwr_ms = now
+                    _pwr_log(prefix=f" off={offset}/{total_len or '?'}")
+                # 超时保护：整体（很宽，避免“慢但在推进”的情况误判失败）
+                if time.ticks_diff(now, t_start) > total_timeout_ms:
+                    return False, f"timeout overall after {total_timeout_ms}ms offset={offset} total={total_len}"
+
+                # 超时保护：无进展（关键）
+                if time.ticks_diff(now, last_progress_ms) > STALL_TIMEOUT_MS:
+                    return False, f"timeout stalled {STALL_TIMEOUT_MS}ms offset={offset} total={total_len}"
+
+                if total_len is not None and offset >= total_len:
+                    break
+
+                want = chunk
+                if total_len is not None:
+                    remain = total_len - offset
+                    if remain <= 0:
+                        break
+                    if want > remain:
+                        want = remain
+
+                # 本 chunk 的 buffer（长度=want）
+                buf = bytearray(want)
+
+                success = False
+                last_err = "unknown"
+                md5_seen = None
+                got_len = 0
+                for k in range(1, CHUNK_RETRIES + 1):
+                    # 只有在连续坏状态时才 full reset，否则只删当前 hid（更快）
+                    do_full_reset = (bad_http_state >= 2)
+                    ok, msg, tlen, md5_b64, got = _fetch_range_into_buf(offset, want, buf, full_reset=do_full_reset)
+                    last_err = msg
+                    if tlen is not None and total_len is None:
+                        total_len = tlen
+                    if md5_b64 and not expect_md5_b64:
+                        expect_md5_b64 = md5_b64
+                    if ok:
+                        success = True
+                        got_len = got
+                        bad_http_state = 0
+                        break
+
+                    # 判定是否属于“模块 HTTP 坏状态”（header-only/no header/request err 等）
+                    try:
+                        if ("no_header_or_total" in msg) or ("MHTTPREQUEST failed" in msg) or ("MHTTPCREATE failed" in msg):
+                            bad_http_state += 1
+                        else:
+                            bad_http_state = max(0, bad_http_state - 1)
+                    except:
+                        pass
+
+                    # 失败：缩小 chunk，提高成功率
+                    if chunk > CHUNK_MIN:
+                        chunk = max(CHUNK_MIN, chunk // 2)
+                        want = min(chunk, want)
+                        buf = bytearray(want)
+                    _log(f"[RETRY] off={offset} want={want} try={k} err={msg}")
+                    _pwr_log(prefix=f" retry{k} off={offset}")
+                    time.sleep_ms(120)
+
+                if not success:
+                    return False, f"chunk_failed off={offset} want={want} err={last_err} total={total_len}"
+
+                # 写入文件（注意：got_len 可能 > want（服务器忽略 Range 返回 200））
+                # 只写入当前请求的 want 字节（buf），避免越界
+                try:
+                    with open(filename, "r+b") as f:
+                        f.seek(offset)
+                        f.write(bytes(buf))
+                except Exception as e:
+                    return False, f"write_failed off={offset}: {e}"
+
+                offset += len(buf)
+                last_progress_ms = time.ticks_ms()
+                # 成功推进后恢复 chunk
+                chunk = CHUNK_MAX
+                if debug:
+                    _log(f"[OK] offset={offset}/{total_len or '?'}")
+
+            # 可选：如果有 Content-Md5 且 hashlib 可用，做校验（Range 响应未必会提供 md5）
+            if expect_md5_b64 and hashlib is not None:
+                try:
+                    with open(filename, "rb") as f:
+                        data = f.read()
+                    digest = hashlib.md5(data).digest()
+                    got_b64 = binascii.b2a_base64(digest).decode().strip()
+                    if got_b64 != expect_md5_b64:
+                        return False, f"md5_mismatch got={got_b64} expected={expect_md5_b64}"
+                except Exception as e:
+                    return False, f"md5_check_failed: {e}"
+
+            t_cost = time.ticks_diff(time.ticks_ms(), t_func0)
+            return True, f"OK size={offset} ip={ip} cost_ms={t_cost}"
+        finally:
+            try:
+                ota_in_progress = max(0, int(ota_in_progress) - 1)
+            except:
+                ota_in_progress = 0
 
 
 def direct_ota_download_via_4g(ota_url):
     """通过 4G 模块下载 OTA（不需要 Wi-Fi）"""
-    global update_thread_started
+    global update_thread_started, ota_in_progress, tcp_connected
     try:
+        t_ota0 = time.ticks_ms()
         if not ota_url:
             safe_enqueue({"result": "ota_failed", "reason": "missing_url"}, 2)
             return
 
+        # OTA 全程暂停 TCP（避免心跳/重连抢占 uart4g_lock，导致 server 断链 + HTTP URC 更容易丢）
+        try:
+            ota_in_progress = int(ota_in_progress) + 1
+        except:
+            ota_in_progress = 1
+
+        # 主动断开 AT TCP，减少 +MIPURC 噪声干扰 HTTP URC 下载
+        tcp_connected = False
+        try:
+            with uart4g_lock:
+                at("AT+MIPCLOSE=0", "OK", 1500)
+        except:
+            pass
+
         print(f"[OTA-4G] 开始通过 4G 下载: {ota_url}")
-        success, msg = download_file_via_4g(ota_url, local_filename, debug=True)
+        # 重要说明：
-        print(f"[OTA-4G] {msg}")
+        # - AT+MDIALUP / RNDIS 是“USB 主机拨号上网”模式，在不少 ML307R 固件上会占用/切换内部网络栈，
-        
+        #   从而导致 AT+MIPOPEN / +MIPURC 这套 TCP 连接无法工作（你会看到一直“连接到服务器...”）。
-        if success and "OK" in msg:
+        # - 这个设备当前 4G 是走 UART + AT Socket（MIPOPEN），并没有把 4G 变成系统网卡（如 ppp0）。
-            # 下载成功：备份+替换+重启
+        # 因此这里不再自动拨号/改路由；只有当系统本来就有 default route（例如 eth0 已联网）时，才尝试走 requests 下载。
+
+        ok_sys = False
+        msg_sys = ""
+        try:
+            if _has_default_route():
+                # 1) 先试原始 URL（可能是 https）
+                ok_sys, msg_sys = _download_file_system_bytes(ota_url, local_filename, timeout_s=30)
+                if (not ok_sys) and isinstance(ota_url, str) and ota_url.startswith("https://static.shelingxingqiu.com/"):
+                    # 2) 部分系统 SSL 不完整：对固定域名降级到 http 再试一次
+                    http_url = "http://" + ota_url[len("https://"):]
+                    ok_sys, msg_sys = _download_file_system_bytes(http_url, local_filename, timeout_s=30)
+            else:
+                ok_sys = False
+                msg_sys = "no_default_route (system network not available)"
+        except Exception as e:
+            ok_sys = False
+            msg_sys = f"system_download_exception: {e}"
+
+        if ok_sys:
+            print(f"[OTA-4G] system {msg_sys}")
             if apply_ota_and_reboot(ota_url):
-                return  # 会重启，不会执行到 finally
+                return
+
+        print(f"[OTA-4G] system failed: {msg_sys} -> fallback to URC download")
+        # debug：进入 URC 下载前打印一次电压/电量（防止下载太快看不到 [PWR] 周期日志）
+        try:
+            v = get_bus_voltage()
+            p = voltage_to_percent(v)
+            print(f"[OTA-4G][PWR] before_urc v={v:.3f}V p={p}%")
+        except Exception as e:
+            print(f"[OTA-4G][PWR] before_urc read_failed: {e}")
+
+        t_dl0 = time.ticks_ms()
+        success, msg = download_file_via_4g(ota_url, local_filename, debug=True)
+        t_dl_cost = time.ticks_diff(time.ticks_ms(), t_dl0)
+        print(f"[OTA-4G] {msg}")
+        print(f"[OTA-4G] download_cost_ms={t_dl_cost}")
+
+        if success and "OK" in msg:
+            if apply_ota_and_reboot(ota_url):
+                return
         else:
-            safe_enqueue({"result": msg}, 2)
+            safe_enqueue({"result": msg_sys or msg}, 2)
             
     except Exception as e:
         error_msg = f"OTA-4G 异常: {str(e)}"
         print(error_msg)
         safe_enqueue({"result": "ota_failed", "reason": error_msg}, 2)
     finally:
+        # 总耗时（注意：若成功并 reboot，这行可能来不及打印）
+        try:
+            t_cost = time.ticks_diff(time.ticks_ms(), t_ota0)
+            print(f"[OTA-4G] total_cost_ms={t_cost}")
+        except:
+            pass
         update_thread_started = False
+        # 对应上面的 ota_in_progress +1
+        try:
+            ota_in_progress = max(0, int(ota_in_progress) - 1)
+        except:
+            ota_in_progress = 0
 
 
 # ==================== 主程序入口 ====================
@@ -1754,6 +2443,14 @@ def cmd_str():
     # 主循环：检测扳机触发 → 拍照 → 分析 → 上报
     while not app.need_exit():
         current_time = time.ticks_ms()
+        # OTA 期间尽量省电：暂停相机预览/拍照/分析，只保留最低频率的循环
+        try:
+            if int(globals().get("ota_in_progress", 0)) > 0:
+                time.sleep_ms(250)
+                continue
+        except:
+            pass
+
         # print("压力传感器数值： ", adc_obj.read())
         adc_val = adc_obj.read()
         # if adc_val > 2400: