Files

1142 lines
34 KiB
Markdown
Raw Permalink Normal View History

2026-07-09 05:08:16 +08:00
# Mt5Bridge API 使用指南
> 本文档面向**开发者**,假设 Bridge 已在云端部署运行。直接复制代码即可使用。
---
## 连接信息
| 项目 | 值 |
|------|-----|
| 地址 | `http://61.164.252.86:13485` |
| 认证 | `X-API-Key` Header 或 `?key=` URL 参数 |
| 格式 | 所有返回均为 JSON |
| WebSocket | `ws://61.164.252.86:13485`,握手时用 Header `X-API-Key` |
| SSE | `http://.../stream/ticks-sse/{symbol}`Header 或 `?key=` 均可 |
> **浏览器注意**:原生 `WebSocket` 不支持自定义 Header,须用 `?key=` URL 参数;SSE 用 `new EventSource(url + '?key=...')` 同理。Node/Python/Java 客户端用 Header 更干净。
---
## 快速开始(Python
```python
import requests
BRIDGE = "http://61.164.252.86:13485"
KEY = "your-api-key"
def api(path, params=None):
"""统一请求封装"""
resp = requests.get(f"{BRIDGE}{path}", params=params, headers={"X-API-Key": KEY})
resp.raise_for_status()
return resp.json()
def api_post(path, data):
"""POST 请求封装"""
resp = requests.post(f"{BRIDGE}{path}", json=data, headers={"X-API-Key": KEY})
resp.raise_for_status()
return resp.json()
# 测试连接
print(api("/health"))
```
---
## API 接口速查
| 分类 | 方法 | 端点 | 用途 |
|------|------|------|------|
| **系统** | `GET` | `/health` | 健康检查 + MT5 连接状态 |
| **账户** | `GET` | `/account` | 余额/净值/保证金/杠杆 |
| **行情** | `GET` | `/symbols/{symbol}` | 品种信息(点值/手数/合约) |
| | `GET` | `/symbols/{symbol}/tick` | 拉取一次 tick |
| | `WS` | `/stream/ticks/{symbol}` | **实时 tick 推送**(推荐) |
| | `GET` | `/stream/ticks-sse/{symbol}` | SSE 推送(浏览器/内网代理友好) |
| **K 线** | `GET` | `/rates/from-pos` | K 线(按偏移量) |
| | `GET` | `/rates/from-date` | **K 线(按时间范围)⭐** |
| **持仓** | `GET` | `/positions[?symbol=]` | 当前持仓列表 |
| | `POST` | `/position/close` | 平仓(全/部分) |
| | `POST` | `/position/modify` | 改 SL/TP |
| | `POST` | `/position/close-by` | 对冲平仓(节省点差) |
| | `POST` | `/positions/close-batch` | 批量平仓(按 magic/symbol |
| **挂单** | `GET` | `/orders[?symbol=]` | 挂单列表 |
| | `POST` | `/order/cancel` | 撤单 |
| | `POST` | `/order/modify` | 改挂单价/SL/TP |
| **下单** | `POST` | `/order/check` | 预检(不成交) |
| | `POST` | `/order/send` | 实际下单 |
| **历史** | `GET` | `/history/deals` | 历史成交 |
| **信号** | `GET/POST/DELETE` | `/gvar[/{name}]` | MQL5 全局变量(指标信号桥) |
> 推送类端点(WS/SSE):每品种最多 50 个订阅者,超出返回 `429`;每 30 秒发一条心跳包用于穿透 NAT 保持连接。
---
## ⚠️ 隐含约定与已知陷阱(先读)
下面这些坑都是踩过的,**不读这节直接调接口几乎必踩**:
### P1. `/history/deals` 的 `date_to` 是 **EXCLUSIVE**(不含当天)
```bash
# ❌ 0 deals — 07-08 当天全部丢失
GET /history/deals?date_from=2026-07-06&date_to=2026-07-08
# ✅ 42 deals — date_to 设成"明天"才能取到 07-08 当天
GET /history/deals?date_from=2026-07-08&date_to=2026-07-09
```
**规则**:永远把 `date_to` 设为"目标日期的下一天"。
### P2. `/history/deals` 的 `entry` 字段语义跟 MT5 标准 **相反**
```
bridge entry = 1 ⇒ OUT(关仓)—— profit 字段是已实现 P&LUSD
bridge entry = 0 ⇒ IN (开仓)—— profit 固定为 0
```
MT5 MQL5 原生约定是 `DEAL_ENTRY_IN=0 / DEAL_ENTRY_OUT=1`,这个 bridge 的 C# 实现把语义反过来了。
**如果不验证就用 close 路径过滤 deal,会一个都匹配不到**(结果 P&L 永远是 0)。
```python
# ✅ 正确:找关仓 deal
exits = [d for d in deals if d.get('entry') == 1 and d.get('magic') == 88001]
```
### P3. `/order/send` 只返回 `{retcode, order, comment}`**没有成交价、没有 deal ticket**
```json
// 实际响应(成功)
{"data": {"retcode": 10009, "order": 1797395084, "comment": "Request executed"}}
```
bridge 不返回 `price` 也不返回 `deal` 字段。**所以拿真实 fill 价格和实现 P&L,唯一办法是 close 之后查 `/history/deals`**。
```python
# ❌ 永远拿不到正确价格
result = api_post('/order/send', {...})['data']
result.get('price') # None
# ✅ 正确:成交后从 history/deals 拿
deals = api('/history/deals', params={'date_from': today, 'date_to': tomorrow})['data']
exit_deal = next(d for d in deals if d['entry'] == 1 and d['symbol'] == sym)
realized_pnl_usd = exit_deal['profit'] # broker 已经换算成 deposit currency
actual_fill_price = exit_deal['price']
```
### P4. `/positions.profit` **是 deposit currencyUSD),不是 quote currency**
```python
# USDCAD SELL 当前浮动 -0.17
# 这是 USD 真实值 (-0.24 CAD ÷ 1.417 USD/CAD = -0.169 USD ≈ -0.17)
# 不是 CAD
```
MT5 `POSITION_PROFIT` 的官方语义就是 in deposit currencybridge 严格遵循。
**如果手动算 USDCAD / USDJPY P&L 时按 quote currency 处理,会差一个汇率倍数**USDCAD 大约 1.4×)。
### P5. bridge 拿到的 tick 跟 broker 实际 fill 差 1-4 ticks
`/positions``price_open` / `price_current``/order/send` 时看到的 tick 跟 broker 服务器**真实成交价**有几毫秒级的时间差,导致值差 0.0001-0.0004(约 0.1-0.4 pip)。
```
broker 实际 fill: 1.41715
/positions.price_open: 1.41711 ← 差 0.00004 (0.4 pip)
```
**永远以 `/history/deals` 里的 `price` 为准做对账,不要用 `/positions` 的 price_open**
### P6. `/position/close` 与 `/order/send` 的成功 retcode 含义不同
| 端点 | 成功 retcode | 失败 retcode | 来源 |
|------|--------------|--------------|------|
| `/order/send` | MT5 原生(通常 `10009` | MT5 原生 | 直接透传 `result.Retcode` |
| `/position/close` | **合成** `10009` | **合成** `10004` | C# 代码 `ok ? 10009u : 10004u` |
两者都是 `10009 = success`,但**别假设 0 是 success**。建议统一判 `retcode == 10009`
2026-07-09 06:12:37 +00:00
### P7. `data: []` 与"默认值数据"的语义混淆
Bridge 的所有 list 端点(`/account`, `/positions`, `/orders`, `/history/deals`, `/symbols` 等)都遵循这个统一约定:
- **"有数据"** ⇒ `{"data": [{...}], "count": 1, ...}`
- **"无数据"** ⇒ `{"data": [], "count": 0, ...}`**HTTP 仍 200**,不是错误)
客户端很容易把"无数据"误判成"默认值数据"。例:
```python
# ❌ 错:data:[] 时 fallback 到 {},默认零值看着像合法数据
info = api('/account')
acc = (info.get('data') or [{}])[0]
if not acc.get('trade_allowed'):
raise PermissionError('trading disabled')
# ↑ 实际可能是"账户未登录",不是"Algo Trading 真关闭"
# ❌ 错:bool({})==True,让健康检查通过
if api('/account'):
mark_healthy()
```
**正确做法**:用 `data` 长度 + 身份字段(如 `login``ticket`)作为"真实数据就绪"的标志:
```python
def data_ready(info, key='login'):
"""data 非空 + 身份字段 > 0 ⇒ 真实数据"""
items = info.get('data') or []
return bool(items) and (items[0].get(key, 0) if items else 0) not in (0, None, '')
# ✅
info = api('/account')
if not data_ready(info, 'login'):
raise ConnectionError('account not loaded yet')
```
**适用所有 list 端点**。判 `positions``orders``deals` 同理。
2026-07-09 05:08:16 +08:00
---
## 1. 健康检查
```
GET /health
```
```python
status = api("/health")
# {"status": "healthy", "mt5_connected": true, "api_version": "1.0.0"}
```
---
## 2. 账户信息
```
GET /account
```
```python
acc = api("/account")["data"][0]
print(f"余额: {acc['balance']}, 净值: {acc['equity']}, 浮动盈亏: {acc['profit']}")
print(f"保证金: {acc['margin']}, 可用保证金: {acc['margin_free']}, 比例: {acc['margin_level']}%")
print(f"杠杆: 1:{acc['leverage']}, 币种: {acc['currency']}")
```
**返回字段:**
| 字段 | 含义 |
|------|------|
| login | 账户号 |
| balance | 余额 |
| equity | 净值 |
| profit | 浮动盈亏 |
| margin | 已用保证金 |
| margin_free | 可用保证金 |
| margin_level | 保证金比例 |
| leverage | 杠杆 |
| currency | 账户币种 |
| trade_allowed | 是否允许交易 |
| trade_expert | 是否允许 EA 交易 |
---
## 3. 实时行情
#### 3-1. 主动拉取(一次性)
```
GET /symbols/{symbol}/tick
```
自动将品种加入 MT5 Market Watch,并等待最多 3 秒获取真实 tick 数据(解决品种未订阅时返回空值的问题)。
```python
def get_tick(symbol):
data = api(f"/symbols/{symbol}/tick")["data"][0]
return data["bid"], data["ask"]
bid, ask = get_tick("XAUUSD")
print(f"XAUUSD Bid: {bid} Ask: {ask} Spread: {ask - bid}")
```
**返回字段:**
| 字段 | 含义 |
|------|------|
| bid | 卖价 |
| ask | 买价 |
| last | 最新成交价 |
| volume | 成交量 |
| time | 时间 |
#### 3-2. 订阅推送(WebSocket 流式)⭐ 推荐
```
WS /stream/ticks/{symbol}
```
MT5 每收到一个 tick 就立即推给所有订阅者,省去轮询。
-`X-API-Key` 认证(Header `X-API-Key: your-key`WebSocket 客户端在握手 Header 里带)
- 连上时自动把品种加入 MT5 Market Watch;最后一个订阅者断开时自动移除
- 每个品种最多 50 个订阅者(含 WS + SSE 总数),超出返回 `429`
- 每 30 秒发一条心跳(无 tick 时也发),客户端可用于保活与断线检测
**推送格式(每条 tick 一帧 JSON 文本):**
```json
{
"type": "tick",
"symbol": "XAUUSDc",
"time": "2026-07-08T10:30:45",
"bid": 4180.0,
"ask": 4180.5,
"last": 4180.2,
"volume": 100,
"time_msc": "2026-07-08T10:30:45.123000",
"flags": 6
}
```
**心跳包:**
```json
{"type":"heartbeat","time":"2026-07-08T10:31:15"}
```
#### 3-3. SSE 推送(不能用 WebSocket 的环境)
```
GET /stream/ticks-sse/{symbol} → Content-Type: text/event-stream
```
面向无法建 WebSocket 的客户端(部分老浏览器、内网代理、curl 测试等)。语义同 3-2,每条 tick 一帧 SSE
```
data: {"type":"tick","symbol":"XAUUSDc","bid":4180.0,...}
data: {"type":"tick","symbol":"XAUUSDc","bid":4181.0,...}
```
**curl 测试:**
```bash
curl -N -H "X-API-Key: your-api-key" \
http://61.164.252.86:13485/stream/ticks-sse/XAUUSDc
```
**Python SSE 客户端(`sseclient-py`):**
```python
from sseclient import SSEClient
import json
messages = SSEClient("http://61.164.252.86:13485/stream/ticks-sse/XAUUSDc",
headers={"X-API-Key": "your-api-key"})
for msg in messages:
data = json.loads(msg.data)
if data.get("type") == "heartbeat":
continue
print(data["symbol"], data["bid"], data["ask"])
```
**Python 示例(需安装 `websocket-client`):**
```python
import websocket
import threading
def on_message(ws, msg):
tick = eval(msg) # 或 json.loads(msg)
print(f"{tick['symbol']} Bid:{tick['bid']} Ask:{tick['ask']}")
def on_open(ws):
print("connected")
def on_close(ws, code, reason):
print(f"disconnected: {code} {reason}")
ws = websocket.WebSocketApp(
f"ws://61.164.252.86:13485/stream/ticks/XAUUSDc",
header=[f"X-API-Key: your-api-key"],
on_message=on_message,
on_open=on_open,
on_close=on_close,
)
ws.run_forever()
```
**`websockets` 库(asyncio 版):**
```python
import asyncio
import websockets
import json
async def watch_ticks():
headers = {"X-API-Key": "your-api-key"}
async with websockets.connect(
"ws://61.164.252.86:13485/stream/ticks/XAUUSDc",
additional_headers=headers,
) as ws:
async for raw in ws:
tick = json.loads(raw)
print(tick["symbol"], tick["bid"], tick["ask"])
asyncio.run(watch_ticks())
```
**浏览器控制台测试:**
```js
const ws = new WebSocket("ws://61.164.252.86:13485/stream/ticks/XAUUSDc", {
headers: { "X-API-Key": "your-api-key" } // 浏览器原生 WS 不支持自定义 Header,需走 ?key= 参数,见下
});
// 浏览器场景:用 query 参数传 key
const ws2 = new WebSocket("ws://61.164.252.86:13485/stream/ticks/XAUUSDc?key=your-api-key");
ws2.onmessage = (e) => console.log(JSON.parse(e.data));
```
---
## 4. 品种信息
```
GET /symbols/{symbol}
```
bid/ask 从实时 tick 数据获取(自动等待最多 3 秒),避免品种刚加入 Market Watch 时返回 0 的问题。
```python
def get_symbol_info(symbol):
info = api(f"/symbols/{symbol}")["data"][0]
print(f"品种: {info['name']}, 描述: {info['description']}")
print(f"小数位: {info['digits']}, 点值: {info['point']}")
print(f"最小手数: {info['volume_min']}, 最大: {info['volume_max']}, 步长: {info['volume_step']}")
print(f"合约大小: {info['trade_contract_size']}")
return info
```
---
## 5. 历史 K 线(按偏移量)
```
GET /rates/from-pos?symbol={symbol}&timeframe={timeframe}&start_pos={start}&count={count}
```
| 参数 | 可选值 |
|------|--------|
| timeframe | `TIMEFRAME_M1` / `M5` / `M15` / `M30` / `H1` / `H4` / `D1` |
| start_pos | 0 = 最新,1 = 前一根,以此类推 |
| count | 获取数量,最大 10000 |
```python
import pandas as pd
def get_rates(symbol, timeframe, count):
"""获取 K 线并转为 DataFrame"""
data = api("/rates/from-pos", params={
"symbol": symbol,
"timeframe": f"TIMEFRAME_{timeframe}",
"start_pos": 0,
"count": count
})["data"]
df = pd.DataFrame(data)
df["time"] = pd.to_datetime(df["time"])
df.set_index("time", inplace=True)
return df
# 获取最近 100 根 H1 K 线
df = get_rates("XAUUSD", "H1", 100)
print(df.head())
```
**返回字段:** `time`, `open`, `high`, `low`, `close`, `tick_volume`, `spread`, `real_volume`
---
### 5-2. 历史 K 线(按时间范围)⭐ 推荐
```
GET /rates/from-date?symbol={symbol}&timeframe={timeframe}&date_from={date_from}&date_to={date_to}
```
| 参数 | 可选值 |
|------|--------|
| timeframe | `TIMEFRAME_M1` / `M5` / `M15` / `M30` / `H1` / `H4` / `D1` |
| date_from | 起始日期,ISO-8601 或 `yyyy-MM-dd` |
| date_to | 结束日期,ISO-8601 或 `yyyy-MM-dd` |
```python
def get_rates_by_date(symbol, timeframe, date_from, date_to):
"""按时间范围获取 K 线"""
data = api("/rates/from-date", params={
"symbol": symbol,
"timeframe": f"TIMEFRAME_{timeframe}",
"date_from": date_from,
"date_to": date_to,
})["data"]
df = pd.DataFrame(data)
df["time"] = pd.to_datetime(df["time"])
df.set_index("time", inplace=True)
return df
# 获取 2026年7月1日 ~ 7月3日 的 H1 K 线
df = get_rates_by_date("XAUUSD", "H1", "2026-07-01", "2026-07-03")
print(df.head())
```
**返回字段:**`/rates/from-pos`
---
## 6. 当前持仓
```
GET /positions[?symbol={symbol}]
```
symbol 可选过滤。
```python
def get_positions(symbol=None):
return api("/positions", params={"symbol": symbol} if symbol else None)["data"]
positions = get_positions()
for pos in positions:
print(f"{pos['ticket']} {pos['symbol']} "
f"{'买' if pos['type'] == 0 else '卖'} "
f"手数:{pos['volume']} 盈亏:{pos['profit']}")
```
**返回字段:** `ticket`, `symbol`, `type`(0=买,1=卖), `volume`, `price_open`, `sl`, `tp`, `price_current`, `swap`, `profit`, `comment`, `magic`
### 6-1. 平仓
```
POST /position/close
```
```json
// 全平
{ "ticket": 12345678 }
// 部分平仓
{ "ticket": 12345678, "volume": 0.05 }
```
| 字段 | 必填 | 说明 |
|------|------|------|
| ticket | ✅ | 持仓编号 |
| volume | ❌ | 平仓手数;不传/0 = 全平;>0 = 部分平仓 |
| deviation | ❌ | 允许滑点(默认 10) |
```python
def close_position(ticket, volume=None):
body = {"ticket": ticket}
if volume: body["volume"] = volume
return api_post("/position/close", body)["data"]
close_position(12345678) # 全平
close_position(12345678, volume=0.05) # 部分平
```
### 6-2. 改持仓 SL/TP
```
POST /position/modify
```
```json
{ "ticket": 12345678, "sl": 4170.0, "tp": 4190.0 }
```
SL/TP 设为 0 表示清除对应止损/止盈。
```python
def modify_position(ticket, sl=0, tp=0):
return api_post("/position/modify", {"ticket": ticket, "sl": sl, "tp": tp})["data"]
```
### 6-3. 对冲平仓(节省点差)
```
POST /position/close-by
```
用一张反向持仓对冲平仓,只收一次点差(MT5 净额结算),适合双向网格 / 锁仓策略快速离场。
```json
{ "position": 111, "position_by": 222 }
```
要求:两张持仓 **同品种 + 反向**
```python
def close_by(ticket_a, ticket_b):
return api_post("/position/close-by", {"position": ticket_a, "position_by": ticket_b})["data"]
```
### 6-4. 移动止损(客户端轮询模式)
Bridge 不在服务端跑轮询,暴露 `/position/modify` 由客户端自己做:
```python
def trail_stop(ticket, distance, step):
"""distance: 跟踪距离(如 50 点);step: 最小推进步长(如 10 点)"""
pos = next((p for p in bridge.positions() if p["ticket"] == ticket), None)
if not pos:
return
current = pos["price_current"]
if pos["type"] == 0: # 多单
new_sl = current - distance
if new_sl - pos["sl"] >= step:
bridge.modify_position(ticket, sl=new_sl)
else: # 空单
new_sl = current + distance
if pos["sl"] - new_sl >= step:
bridge.modify_position(ticket, sl=new_sl)
# 每秒跑一次
while True:
for p in bridge.positions():
trail_stop(p["ticket"], distance=50, step=10)
time.sleep(1)
```
也可以用 WebSocket tick 流推送驱动,把 `time.sleep(1)` 换成 tick 回调,反应更快。
### 6-5. 批量平仓
```
POST /positions/close-batch
```
`symbol` 和/或 `magic` 批量平仓。**至少传一个**过滤条件,避免误清整个账户。
```json
{ "magic": 123456 }
{ "symbol": "XAUUSDc", "magic": 123456 }
{ "symbol": "XAUUSDc" }
```
| 字段 | 必填 | 说明 |
|------|------|------|
| symbol | 一 | 品种过滤 |
| magic | 一 | Magic Number 过滤 |
| deviation | ❌ | 允许滑点(默认 10) |
```python
def close_by_magic(magic):
return api_post("/positions/close-batch", {"magic": magic})["data"]
result = close_by_magic(123456)
print(f"已平 {result['closed']} 单,失败 {result['failed']} 单")
# data 数组里有每张单的 ticket / symbol / retcode / comment
```
---
## 7. 挂单
```
GET /orders?symbol={symbol}
```
symbol 可选,不传返回全部。
```python
orders = api("/orders")["data"]
for o in orders:
print(f"{o['ticket']} {o['symbol']} 类型:{o['type']} 手数:{o['volume_initial']}")
```
### 7-1. 撤单
```
POST /order/cancel
```
```json
{ "ticket": 87654321 }
```
```python
def cancel_order(ticket):
return api_post("/order/cancel", {"ticket": ticket})["data"]
cancel_order(87654321)
```
### 7-2. 改挂单
```
POST /order/modify
```
```json
{ "ticket": 87654321, "price": 4180.0, "sl": 4170.0, "tp": 4190.0 }
```
sl/tp 设为 0 表示清除。
```python
def modify_order(ticket, price, sl=0, tp=0):
return api_post("/order/modify", {"ticket": ticket, "price": price, "sl": sl, "tp": tp})["data"]
```
---
## 8. 订单预检
```
POST /order/check
```
下单前验证,不会真正执行。检查保证金是否足够、价格是否有效等。
**填充模式自动适配**:服务端会根据品种的 `SYMBOL_FILLING_MODE` 自动选择经纪商支持的填充模式。如果请求的 `type_filling` 不被支持,会按 IOC(1) → FOK(0) → RETURN(2) 顺序降级,无需客户端手动判断。
```python
def check_order(symbol, volume, order_type, price, sl=None, tp=None, magic=0, comment=""):
"""预检订单"""
data = {
"action": 1, # 1=即时成交
"symbol": symbol,
"volume": volume,
"order_type": order_type, # 0=市价买, 1=市价卖
"price": price,
"sl": sl or 0,
"tp": tp or 0,
"magic": magic,
"comment": comment,
"deviation": 10,
"type_filling": 0 # 0=FOK, 1=IOC, 2=RETURN — 服务端自动适配,不传也行
}
result = api_post("/order/check", data)["data"]
print(f"预检结果: retcode={result['retcode']}, comment={result['comment']}")
if result['retcode'] == 0:
print("✅ 可以下单")
else:
print("❌ 不可下单")
return result
check_order("XAUUSD", 0.01, 0, 4180.0, sl=4170.0, tp=4190.0)
```
**type_filling 说明:**
| 值 | 含义 | 说明 |
|------|------|------|
| 0 | FOK (Fill or Kill) | 必须全部成交,否则取消 |
| 1 | IOC (Immediate or Cancel) | 能成交多少成交多少 |
| 2 | RETURN | 剩余部分留在订单簿 |
> 不同经纪商支持的填充模式不同(如 ICMarkets 只支持 IOC),服务端会自动降级,客户端无需关心。
---
## 9. 下单
```
POST /order/send
```
`/order/check` 相同,`type_filling` 会自动适配经纪商支持的填充模式。
```python
def send_order(symbol, volume, order_type, price, sl=None, tp=None, magic=0, comment=""):
"""下单"""
data = {
"request": {
"action": 1,
"symbol": symbol,
"volume": volume,
"order_type": order_type,
"price": price,
"sl": sl or 0,
"tp": tp or 0,
"magic": magic,
"comment": comment,
"deviation": 10,
"type_filling": 0 # 自动适配,不传也行
}
}
result = api_post("/order/send", data)["data"]
print(f"下单结果: retcode={result['retcode']}, order={result['order']}, comment={result['comment']}")
return result
# 市价买入 0.01 手 XAUUSD
result = send_order("XAUUSD", 0.01, 0, 4180.0, sl=4170.0, tp=4190.0)
```
**order_type 说明:**
| 值 | 含义 |
|------|------|
| 0 | 市价买入 |
| 1 | 市价卖出 |
| 2 | 限价买入 |
| 3 | 限价卖出 |
| 4 | 止损买入 |
| 5 | 止损卖出 |
---
## 10. 历史成交
```
GET /history/deals?date_from={from}&date_to={to}&symbol={symbol}
```
```python
deals = api("/history/deals", params={
"date_from": "2026-07-01",
"date_to": "2026-07-03",
"symbol": "XAUUSD"
})["data"]
for d in deals:
print(f"{d['ticket']} {d['time']} {d['symbol']} "
f"手数:{d['volume']} 价格:{d['price']} 盈亏:{d['profit']}")
```
---
## 11. 全局变量(MQL5 信号桥)
> 让 MQL5 指标/EA 把信号写出来,Python 通过 Bridge 读取。不用翻译 MQL5 代码。
说明:
- 这部分属于“MT5 指标/EA 信号导出”能力,不影响 Bridge 的账户、行情、历史、持仓、挂单、预检、下单等基础 API。
- 如果你只更新了远程 `Mt5Bridge.dll`,没有替换 `Alpha Trend.ex5`Bridge 仍然可以正常工作,`/gvar` 也仍会返回旧版指标写出的变量名。
- 只有在你需要“已收盘 K 线信号”和“带周期/参数作用域的新变量名”时,才需要重新编译并替换新版 `Alpha Trend.ex5`
#### 列出所有全局变量
```
GET /gvar
```
```python
gvars = api("/gvar")
print(gvars)
# {"data": [{"name": "AT_Trend_XAUUSD", "value": 1}, ...], "count": 5}
```
#### 读取指定变量
```
GET /gvar/{name}
```
```python
trend = api("/gvar/AT_Trend_XAUUSD")["value"]
print(f"趋势方向: {'多头' if trend == 1 else '空头'}")
```
#### 写入变量
```
POST /gvar/{name}
```
Body: `{"value": 75.5}`
```python
def set_gvar(name, value):
return api_post(f"/gvar/{name}", {"value": value})
set_gvar("MY_RSI", 75.5)
```
#### 删除变量
```
DELETE /gvar/{name}
```
---
## MQL5 指标 → Python 完整流程
### 第一步:改指标源码,输出已收盘 K 线信号
这一步是“升级指标导出行为”的可选步骤,不是 Bridge 基础 API 的必需步骤。
```mql5
// 在 OnCalculate 末尾加
int last_closed = rates_total - 2;
string key = StringFormat("MY_SIGNAL_%s_%s", _Symbol, EnumToString(_Period));
GlobalVariableSet(key, signal_value[last_closed]);
```
### 第二步:Python 读取信号
```python
def read_signal():
try:
return api(f"/gvar/MY_SIGNAL_XAUUSD_PERIOD_H1")["value"]
except:
return None
signal = read_signal()
print(f"指标信号: {signal}")
```
如果你仍在使用旧版 `Alpha Trend.ex5`,则读取方式应继续对应旧键名,例如:
```python
trend = api("/gvar/AT_Trend_XAUUSD")["value"]
buy_signal = api("/gvar/AT_Buy_XAUUSD")["value"]
```
### 第三步:根据信号做决策
```python
def on_tick():
signal = read_signal()
if signal != 1:
return # 没信号,不动
if not bridge.has_position("XAUUSD"):
bid, ask = bridge.tick("XAUUSD")
bridge.buy("XAUUSD", 0.01, ask, sl=ask - 50, tp=ask + 100)
print("指标发出买入信号,已开多")
```
## 完整策略模板
```python
import requests
import pandas as pd
import time
from datetime import datetime
BRIDGE = "http://61.164.252.86:13485"
KEY = "your-api-key"
class Mt5Bridge:
def __init__(self):
self.headers = {"X-API-Key": KEY}
def _get(self, path, params=None):
r = requests.get(f"{BRIDGE}{path}", params=params, headers=self.headers)
r.raise_for_status()
return r.json()
def _post(self, path, data):
r = requests.post(f"{BRIDGE}{path}", json=data, headers=self.headers)
r.raise_for_status()
return r.json()
def stream_ticks(self, symbol, on_tick):
"""订阅 tick 推送,on_tick 回调收到 dict,阻塞运行。需 pip install websocket-client"""
import websocket
url = f"{BRIDGE.replace('http', 'ws', 1)}/stream/ticks/{symbol}"
ws = websocket.WebSocketApp(
url,
header=[f"X-API-Key: {KEY}"],
on_message=lambda ws, msg: on_tick(eval(msg)),
on_error=lambda ws, err: print(f"ws error: {err}"),
)
ws.run_forever()
def stream_ticks_async(self, symbol, on_tick):
"""后台线程订阅 tick,不阻塞主线程"""
import threading
t = threading.Thread(target=self.stream_ticks, args=(symbol, on_tick), daemon=True)
t.start()
return t
# ── 行情 ──
def tick(self, symbol):
d = self._get(f"/symbols/{symbol}/tick")["data"][0]
return d["bid"], d["ask"]
def rates(self, symbol, timeframe, count):
return self._get("/rates/from-pos", params={
"symbol": symbol, "timeframe": f"TIMEFRAME_{timeframe}",
"start_pos": 0, "count": count
})["data"]
def to_df(self, symbol, timeframe, count):
df = pd.DataFrame(self.rates(symbol, timeframe, count))
df["time"] = pd.to_datetime(df["time"])
df.set_index("time", inplace=True)
return df
# ── 账户 ──
def account(self):
return self._get("/account")["data"][0]
# ── 持仓 ──
def positions(self, symbol=None):
return self._get("/positions", params={"symbol": symbol} if symbol else None)["data"]
def has_position(self, symbol):
return any(p["symbol"] == symbol for p in self.positions())
def close(self, ticket, volume=None):
body = {"ticket": ticket}
if volume: body["volume"] = volume
return self._post("/position/close", body)["data"]
def modify_position(self, ticket, sl=0, tp=0):
return self._post("/position/modify", {"ticket": ticket, "sl": sl, "tp": tp})["data"]
def close_by(self, ticket_a, ticket_b):
return self._post("/position/close-by", {"position": ticket_a, "position_by": ticket_b})["data"]
def close_batch(self, magic=None, symbol=None, deviation=None):
body = {k: v for k, v in {"magic": magic, "symbol": symbol, "deviation": deviation}.items() if v is not None}
return self._post("/positions/close-batch", body)
# ── 挂单 ──
def orders(self, symbol=None):
return self._get("/orders", params={"symbol": symbol} if symbol else None)["data"]
def cancel_order(self, ticket):
return self._post("/order/cancel", {"ticket": ticket})["data"]
def modify_order(self, ticket, price, sl=0, tp=0):
return self._post("/order/modify", {"ticket": ticket, "price": price, "sl": sl, "tp": tp})["data"]
# ── 下单 ──
def buy(self, symbol, volume, price, sl=0, tp=0, magic=0, comment=""):
return self._send(symbol, volume, 0, price, sl, tp, magic, comment)
def sell(self, symbol, volume, price, sl=0, tp=0, magic=0, comment=""):
return self._send(symbol, volume, 1, price, sl, tp, magic, comment)
def _send(self, symbol, volume, order_type, price, sl, tp, magic, comment):
return self._post("/order/send", {
"request": {
"action": 1, "symbol": symbol, "volume": volume,
"order_type": order_type, "price": price,
"sl": sl, "tp": tp, "magic": magic,
"comment": comment, "deviation": 10
}
})["data"]
def check(self, symbol, volume, order_type, price, sl=0, tp=0):
return self._post("/order/check", {
"action": 1, "symbol": symbol, "volume": volume,
"order_type": order_type, "price": price,
"sl": sl, "tp": tp, "magic": 0, "comment": "", "deviation": 10
})["data"]
# ══════════════════════════════════════════════
# 策略示例:均线金叉死叉
# ══════════════════════════════════════════════
class MAStrategy:
def __init__(self, bridge, symbol, fast=20, slow=60):
self.bridge = bridge
self.symbol = symbol
self.fast = fast
self.slow = slow
def signal(self):
"""计算信号:1=买入, -1=卖出, 0=观望"""
df = self.bridge.to_df(self.symbol, "H1", self.slow + 5)
df["ma_fast"] = df["close"].rolling(self.fast).mean()
df["ma_slow"] = df["close"].rolling(self.slow).mean()
# 使用最近两根已收盘 K 线
prev = df.iloc[-3]
curr = df.iloc[-2]
# 金叉
if prev["ma_fast"] <= prev["ma_slow"] and curr["ma_fast"] > curr["ma_slow"]:
return 1
# 死叉
if prev["ma_fast"] >= prev["ma_slow"] and curr["ma_fast"] < curr["ma_slow"]:
return -1
return 0
def run(self):
sig = self.signal()
bid, ask = self.bridge.tick(self.symbol)
acc = self.bridge.account()
print(f"[{datetime.now()}] {self.symbol} Bid:{bid} Ask:{ask} "
f"Balance:{acc['balance']} Equity:{acc['equity']} Signal:{sig}")
if sig == 1 and not self.bridge.has_position(self.symbol):
print(" → 金叉,开多")
self.bridge.buy(self.symbol, 0.01, ask, sl=ask - 50, tp=ask + 100)
elif sig == -1 and not self.bridge.has_position(self.symbol):
print(" → 死叉,开空")
self.bridge.sell(self.symbol, 0.01, bid, sl=bid + 50, tp=bid - 100)
# ══════════════════════════════════════════════
# 运行
# ══════════════════════════════════════════════
if __name__ == "__main__":
bridge = Mt5Bridge()
strategy = MAStrategy(bridge, "XAUUSD", fast=20, slow=60)
while True:
try:
strategy.run()
except Exception as e:
print(f"Error: {e}")
time.sleep(60) # 每分钟检查一次
```
---
## 浏览器快速验证
在浏览器地址栏直接输入:
```
http://61.164.252.86:13485/health?key=your-api-key
http://61.164.252.86:13485/account?key=your-api-key
http://61.164.252.86:13485/symbols/XAUUSD/tick?key=your-api-key
http://61.164.252.86:13485/rates/from-date?symbol=XAUUSD&timeframe=TIMEFRAME_H1&date_from=2026-07-01&date_to=2026-07-03&key=your-api-key
```
---
## PowerShell 快速测试
```powershell
$headers = @{ "X-API-Key" = "your-api-key" }
Invoke-RestMethod "http://61.164.252.86:13485/health" -Headers $headers
Invoke-RestMethod "http://61.164.252.86:13485/account" -Headers $headers
Invoke-RestMethod "http://61.164.252.86:13485/symbols/XAUUSD/tick" -Headers $headers
Invoke-RestMethod "http://61.164.252.86:13485/rates/from-date?symbol=XAUUSD&timeframe=TIMEFRAME_H1&date_from=2026-07-01&date_to=2026-07-03" -Headers $headers
```
---
## 常见问题
2026-07-09 06:12:37 +00:00
> **第一次调这个 bridge?先去读 [§「⚠️ 隐含约定与已知陷阱」](#-隐含约定与已知陷阱先读)** —— 7 个非显而易见的约定(date_to 排他、`entry` 字段语义反、`/order/send` 不返回 fill 价、`/account` 空 data 误判 等),不读这一节直接调接口几乎必踩。
2026-07-09 05:08:16 +08:00
### 返回 "Unauthorized"
API Key 错误或没带。检查 Header 中的 `X-API-Key` 或 URL 中的 `?key=`
### 返回 "对于该符号,不支持市场执行"
`order_type` 填错了,MT5 中有些品种不支持市价单,有些不支持挂单。先调用 `/order/check` 预检。
### 返回 "没有足够的资金"
保证金不足,减小手数或检查 `account.margin_free`
### 闭仓 P&L 永远是 0 / 跟 MT5 terminal 对不上
大概率是过滤了 `entry==0` 的 deal(以为 OUT),实际这个 bridge 是 `entry==1` 才是 OUT(关仓)。详见 §隐含约定 P2。
### 算出来的 P&L 跟 broker 对差 1.4× / 1.5×
多半是手动把 `positions.profit` 当 quote currency 处理。`positions.profit` 是 deposit currencyUSD),不是 quote currency。详见 §隐含约定 P4。
### 返回 "无法连接到远程服务器"
2026-07-09 06:12:37 +00:00
Bridge 未运行或网络不通,先检查 `/health`
### 业务字段返回默认值 → 怎么区分"无数据"和"真状态"
`/account` 在账户未加载完成时返回 `data:[]` + 全零字段(HTTP 200)。客户端若只看 `bool(response)``field == 0`,会把"无数据"误判成"零值正常状态"。详见 §隐含约定 P7。