加入对esp32客户端的支持

main
mula.liu 2026-10-08 00:07:59 +08:00
parent 654ab18a1f
commit 513b05b6ff
11 changed files with 376 additions and 44 deletions

BIN
.DS_Store vendored

Binary file not shown.

View File

@ -19,7 +19,7 @@ REDIS_MAX_CONNECTIONS=50
# Application Configuration
# ======================
# JWT Secret Key (IMPORTANT: Change this in production!)
JWT_SECRET_KEY=cosmo
JWT_SECRET_KEY=cosmo_4f8b9e2a7c1d0f5e6a8b9c0d1e2f3a4b5c6d7e8f9a0b1c2d3e4f5a6b7c8d9e0f
# CORS - Support both internal IP access and external domain
# Format: comma-separated list of origins

View File

@ -251,10 +251,34 @@ ls -lh /opt/cosmo/data/backups/
### 系统更新
```bash
# 拉取最新代码并重启
# 拉取最新代码、重建镜像、重启,并同步数据库结构
./scripts/deploy.sh --update
```
`--update` 在重启后会执行 `scripts/upgrade_db.py --apply`,把新版本模型里新增的
**表 / 字段 / 索引**补上(后端启动时的 `create_all()` 只建表,不会给已有表加列)。
这个脚本只做增量:已有字段、类型、约束和数据都不会被修改或删除。
想先看清楚会执行什么 DDL,可以单独跑一次 dry-run(不加 `--apply` 只打印不执行):
```bash
cd /opt/cosmo
docker-compose exec -T backend python scripts/upgrade_db.py # 预览
docker-compose exec -T backend python scripts/upgrade_db.py --apply # 执行
```
> **表结构之外的数据不在它的职责范围内**:
> - `system_settings` 的新增配置项由后端启动时自动补齐,无需干预;
> - `rocket_configs` 里的新增火箭型号属于业务数据,需要在管理端重新录入,
> 或者从开发库同步:
> ```bash
> # 开发机导出
> docker exec -e PGPASSWORD=xxx postgres17 pg_dump -U postgres -d cosmo_db -t rocket_configs > rockets.sql
> # 服务器导入
> docker-compose cp rockets.sql postgres:/tmp/rockets.sql
> docker-compose exec -T postgres psql -U postgres -d cosmo_db -f /tmp/rockets.sql
> ```
### 清理操作
```bash
@ -308,45 +332,64 @@ docker-compose restart frontend
编辑 `.env.production`:
```bash
DATABASE_PASSWORD=strong_random_password_here
JWT_SECRET_KEY=至少32位的随机字符串 # 必填,缺失时 docker-compose 会直接报错退出
```
`JWT_SECRET_KEY` 现在会显式注入 backend 容器(`${JWT_SECRET_KEY:?...}`)。
以前它只存在于 `backend/.env` 里,容器里读不到,于是回落到代码里的默认值——
所有部署共用同一个密钥,而且每次重启容器用户都会被踢下线。
缺失时 `docker-compose up` 会报 `JWT_SECRET_KEY is required, set it in .env` 并退出,属于预期行为。
> 改了密钥会让现有登录态全部失效,用户需要重新登录。
> 本地开发不受影响(`backend/.env` 直接生效,不走 compose)。
### 2. 配置 HTTPS(推荐)
使用 Let's Encrypt 免费证书:
Nginx 跑在 `cosmo_frontend` 容器里,宿主机上没有 Nginx,所以 **不能用 `certbot --nginx`**;
`nginx.conf` 里 `location ~ /\.` 会拒掉 `/.well-known/acme-challenge/`,webroot 模式也会 403。
用 standalone 签发:
```bash
# 安装 certbot
sudo apt install certbot python3-certbot-nginx
sudo apt install certbot
# 获取证书
sudo certbot --nginx -d your-domain.com
# 证书自动续期
sudo crontab -e
# 添加: 0 3 * * * certbot renew --quiet
# 80 端口被前端容器占着,先停掉再签
cd /opt/cosmo && docker-compose stop frontend
sudo certbot certonly --standalone -d your-domain.com
docker-compose start frontend
```
修改 `nginx/nginx.conf` 添加 SSL 配置:
证书在宿主机 `/etc/letsencrypt/`,`docker-compose.yml` 已经把该目录只读挂进前端容器,
并把 443 端口映射出去了。剩下只差取消 `nginx/nginx.conf` 里那三行注释:
```nginx
server {
listen 443 ssl http2;
server_name your-domain.com;
ssl_certificate /etc/letsencrypt/live/your-domain.com/fullchain.pem;
ssl_certificate_key /etc/letsencrypt/live/your-domain.com/privkey.pem;
# 其他配置...
}
# HTTP 重定向到 HTTPS
server {
listen 80;
server_name your-domain.com;
return 301 https://$server_name$request_uri;
}
server {
listen 80;
listen 443 ssl; # 取消注释
ssl_certificate /etc/letsencrypt/live/your-domain.com/fullchain.pem; # 换成你的域名
ssl_certificate_key /etc/letsencrypt/live/your-domain.com/privkey.pem;
server_name _;
# ... 后面的 location 原样不动,80 和 443 共用一套
```
```bash
cd /opt/cosmo && docker-compose exec frontend nginx -t && docker-compose restart frontend
```
证书自动续期(续期时同样要腾出 80 端口,并让容器重新读一次证书):
```bash
sudo crontab -e
# 0 3 * * * cd /opt/cosmo && docker-compose stop frontend && certbot renew --quiet; docker-compose start frontend
```
> **别加 HTTP→HTTPS 的 301。** ESP32 固件用的是裸 `HTTPClient`,不跟跳转:`/api` 一跳转,
> 终端就拿不到数据了。要么终端那边直接把地址填成 `https://your-domain.com/api`,
> 要么让 `/api/` 继续留在 80 上。
>
> 固件支持 https(`WiFiClientSecure`,不校验证书链,只要链路加密),
> 地址配置见 `terminal/firmware/README.md` 的「后端地址」一节。
### 3. 防火墙配置
```bash

View File

@ -13,6 +13,7 @@ from sqlalchemy import select, update
from pydantic import BaseModel
from typing import Optional, Dict, Any
from app.config import settings
from app.database import get_db
from app.models.db import Resource
from app.services.celestial_body_service import celestial_body_service
@ -22,6 +23,26 @@ logger = logging.getLogger(__name__)
router = APIRouter(prefix="/celestial/resources", tags=["celestial-resource"])
# icon 在界面上只有一个槽位(ResourceManager 只渲染 items[0]),换图必须是覆盖语义,
# 否则旧行既看不到也删不掉,resources 表会一直堆积同一张图的替代品。
# texture / model 是多值列表(每条可单独设 scale),不能套用覆盖。
SINGLE_VALUE_TYPES = ("icon",)
def remove_upload_file(relative_path: str) -> None:
"""Delete an uploaded file; a missing file is not an error.
file_path is stored relative to the upload root ("icon/x.png"), so it has to be
resolved against UPLOAD_DIR first - os.path.exists("icon/x.png") is always False.
"""
path = os.path.join(settings.upload_dir, relative_path)
if os.path.exists(path):
try:
os.remove(path)
logger.info(f"Deleted file: {path}")
except Exception as e:
logger.warning(f"Failed to delete file {path}: {e}")
# Pydantic models
class ResourceUpdate(BaseModel):
@ -96,6 +117,22 @@ async def upload_resource(
# Determine MIME type
mime_type = file.content_type
# Replace previous records of single-value types (files are removed after commit)
replaced_paths: list = []
if resource_type in SINGLE_VALUE_TYPES:
stale = await db.execute(
select(Resource).where(
Resource.body_id == body_id,
Resource.resource_type == resource_type,
)
)
for old in stale.scalars().all():
replaced_paths.append(old.file_path)
await db.delete(old)
# Flush the deletes first: the unit of work would insert before deleting
# and trip the (body_id, resource_type, file_path) unique index.
await db.flush()
# Create resource record
resource = await resource_service.create_resource(
{
@ -112,6 +149,10 @@ async def upload_resource(
await db.commit()
await db.refresh(resource)
for stale_path in replaced_paths:
if stale_path != relative_path: # same-second re-upload reuses the path
remove_upload_file(stale_path)
logger.info(f"Uploaded resource for {body.name} ({body.type}): {relative_path} ({file_size} bytes)")
return {
@ -178,12 +219,7 @@ async def delete_resource(
# Delete file if it exists
file_path = resource.file_path
if os.path.exists(file_path):
try:
os.remove(file_path)
logger.info(f"Deleted file: {file_path}")
except Exception as e:
logger.warning(f"Failed to delete file {file_path}: {e}")
remove_upload_file(file_path)
# Delete database record
deleted = await resource_service.delete_resource(resource_id, db)

View File

@ -61,23 +61,27 @@ services:
DATABASE_NAME: ${DATABASE_NAME:-cosmo_db}
DATABASE_USER: ${DATABASE_USER:-postgres}
DATABASE_PASSWORD: ${DATABASE_PASSWORD:-postgres}
DATABASE_POOL_SIZE: 20
DATABASE_MAX_OVERFLOW: 10
DATABASE_POOL_SIZE: ${DATABASE_POOL_SIZE:-20}
DATABASE_MAX_OVERFLOW: ${DATABASE_MAX_OVERFLOW:-10}
# Redis
REDIS_HOST: redis
REDIS_PORT: 6379
REDIS_DB: 0
REDIS_PASSWORD: ${REDIS_PASSWORD:-}
REDIS_MAX_CONNECTIONS: 50
REDIS_MAX_CONNECTIONS: ${REDIS_MAX_CONNECTIONS:-50}
# Application
APP_NAME: "Cosmo - Deep Space Explorer"
API_PREFIX: /api
# backend/.env is excluded by backend/.dockerignore, so this has to come from here.
# Without it the app silently signs every token with the hard-coded dev default,
# which lets anyone forge user / terminal tokens. Fail fast instead of guessing.
JWT_SECRET_KEY: ${JWT_SECRET_KEY:?JWT_SECRET_KEY is required, set it in .env}
CORS_ORIGINS: ${CORS_ORIGINS:-http://localhost}
# Cache
CACHE_TTL_DAYS: 3
CACHE_TTL_DAYS: ${CACHE_TTL_DAYS:-3}
# Upload
UPLOAD_DIR: /app/upload
MAX_UPLOAD_SIZE: 10485760
MAX_UPLOAD_SIZE: ${MAX_UPLOAD_SIZE:-10485760}
# Proxy (for accessing NASA JPL Horizons API)
HTTP_PROXY: ${HTTP_PROXY:-}
HTTPS_PROXY: ${HTTPS_PROXY:-}
@ -113,8 +117,11 @@ services:
restart: unless-stopped
volumes:
- ./nginx/nginx.conf:/etc/nginx/nginx.conf:ro
# certbot 在宿主机上签发的证书,nginx 容器里要能读到(启用 443 时需要)
- /etc/letsencrypt:/etc/letsencrypt:ro
ports:
- "80:80"
- "443:443"
depends_on:
- backend
networks:

View File

@ -40,8 +40,14 @@ http {
keepalive 32;
}
# One server block serves both plain HTTP and HTTPS, so enabling TLS is just
# uncommenting the three lines below (see DEPLOYMENT.md 配置 HTTPS).
server {
listen 80;
#listen 443 ssl; # 同一个 server 兼容 80/443
#ssl_certificate /etc/letsencrypt/live/your-domain.com/fullchain.pem;
#ssl_certificate_key /etc/letsencrypt/live/your-domain.com/privkey.pem;
# 证书目录已由 docker-compose 只读挂载进容器
server_name _;
# Root directory for static files

View File

@ -247,6 +247,13 @@ update_system() {
# Restart services
docker-compose up -d
# Schema sync: the backend only runs create_all() on boot, which creates missing
# tables but never adds columns to tables that already exist.
log "Upgrading database schema (additive only)..."
docker-compose exec -T backend mkdir -p /app/scripts
docker-compose cp scripts/upgrade_db.py backend:/app/scripts/upgrade_db.py
docker-compose exec -T backend python scripts/upgrade_db.py --apply
log "✓ System updated"
}

View File

@ -0,0 +1,170 @@
#!/usr/bin/env python3
"""
Schema upgrade script for an existing Cosmo database.
The backend runs `Base.metadata.create_all()` on every boot, which creates tables
added since the first deploy but never touches tables that already exist. Columns
added to the ORM models afterwards stay missing on the server. This script diffs the
ORM metadata against the live database and applies what is missing:
- missing tables (with their indexes and constraints, via create_all)
- missing columns on existing tables (ADD COLUMN IF NOT EXISTS)
- missing indexes on existing tables (CREATE INDEX IF NOT EXISTS)
It is additive only: existing columns, types, constraints and data are never changed
or dropped. Data (new rocket models, system settings, ...) is out of scope here.
Usage:
python scripts/upgrade_db.py # dry run, print the DDL
python scripts/upgrade_db.py --apply # execute it
"""
import asyncio
import logging
import os
import sys
from pathlib import Path
# Runs from the repo (<project>/scripts, app code in <project>/backend) and from
# inside the backend image (deploy.sh copies it to /app/scripts, app code in /app).
BACKEND_DIR = Path(__file__).resolve().parent.parent / "backend"
if not (BACKEND_DIR / "app").is_dir():
BACKEND_DIR = Path(__file__).resolve().parent.parent
sys.path.insert(0, str(BACKEND_DIR))
os.chdir(BACKEND_DIR)
from sqlalchemy import inspect, text
from sqlalchemy.schema import CreateIndex, CreateTable
from app.database import engine, Base
from app.config import settings
import app.models.db # noqa: F401 (register every model on Base.metadata)
logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(message)s")
logger = logging.getLogger(__name__)
def column_stmt(table_name: str, col, dialect, strict: bool) -> str:
"""ALTER TABLE ... ADD COLUMN for one missing column.
strict=False drops NOT NULL, used as fallback when the table already has rows
and the model gives no default Postgres could fill in.
"""
sql = (f'ALTER TABLE "{table_name}" ADD COLUMN IF NOT EXISTS "{col.name}" '
f'{col.type.compile(dialect=dialect)}')
if strict and not col.nullable:
sql += " NOT NULL"
default = col.server_default.arg if col.server_default is not None else None
if default is None and col.default is not None and not col.default.is_callable:
default = col.default.arg
if hasattr(default, "compile"): # func.now() and friends
sql += " DEFAULT CURRENT_TIMESTAMP"
elif isinstance(default, str):
sql += " DEFAULT '" + default.replace("'", "''") + "'"
elif isinstance(default, bool):
sql += " DEFAULT " + ("TRUE" if default else "FALSE")
elif default is not None:
sql += f" DEFAULT {default}"
return sql
def diff(conn):
"""Compare ORM metadata with the live schema; returns (changes, notes)."""
live = inspect(conn)
tables = live.get_table_names()
if not tables:
return [], ["空库:后端首次启动会自己建全部表,无需本脚本。"]
changes, notes = [], []
for name, table in Base.metadata.tables.items():
if name not in tables:
changes.append({"kind": "table", "label": name,
"sql": str(CreateTable(table).compile(dialect=conn.dialect))})
continue
existing = {c["name"] for c in live.get_columns(name)}
for col in table.columns:
if col.name in existing:
continue
changes.append({
"kind": "column", "label": f"{name}.{col.name}",
"sql": column_stmt(name, col, conn.dialect, strict=True),
"fallback": column_stmt(name, col, conn.dialect, strict=False),
})
pk_cols = {c.name for c in table.primary_key}
# Compare by covered columns, not only by name: the hand-written init_db.sql
# names indexes idx_*, the ORM declares ix_*, so a name-only check wants to
# add a duplicate index over identical columns on every existing database.
covered = set()
for item in live.get_indexes(name):
if item.get("column_names"):
covered.add((tuple(item["column_names"]), bool(item.get("unique"))))
for item in live.get_unique_constraints(name):
if item.get("column_names"):
covered.add((tuple(item["column_names"]), True))
live_names = {i["name"] for i in live.get_indexes(name)}
for index in table.indexes:
cols = tuple(c.name for c in index.columns)
if set(cols) <= pk_cols:
continue # the primary key already covers single-PK indexes
if index.name in live_names or (cols, bool(index.unique)) in covered:
continue # same columns already indexed (unique constraint counts)
sql = str(CreateIndex(index, if_not_exists=True).compile(dialect=conn.dialect))
changes.append({"kind": "index", "label": index.name, "sql": sql})
# Missing foreign keys on existing tables are reported, not applied: they can
# fail on orphan rows and silently lock out writes.
for name, table in Base.metadata.tables.items():
if name not in tables:
continue
live_fk_cols = {c for fk in live.get_foreign_keys(name) for c in fk["constrained_columns"]}
for fk in table.foreign_keys:
if fk.parent.name not in live_fk_cols:
notes.append(f"{name}: 缺少外键 {fk.parent.name} -> {fk.target_fullname}"
f"(脚本不加外键,确认数据干净后手工执行)")
return changes, notes
async def main(apply: bool):
logger.info(f"Database: {settings.database_url.split('@')[-1]}")
async with engine.connect() as conn:
changes, notes = await conn.run_sync(diff)
for change in changes:
logger.info(f"[{change['kind']}] {change['label']}")
logger.info(" " + " ".join(change["sql"].split()))
if not changes:
logger.info("✓ Schema is up to date, nothing to do")
for note in notes:
logger.warning(f"! {note}")
if not changes or not apply:
if changes:
logger.info("Dry run only. Re-run with --apply to execute.")
return
async with engine.begin() as conn:
for change in changes:
try:
# SAVEPOINT per statement: one failure must not abort the whole upgrade.
async with conn.begin_nested():
if change["kind"] == "table":
await conn.run_sync(lambda c, t=change["label"]: Base.metadata.create_all(
c, tables=[Base.metadata.tables[t]], checkfirst=True))
else:
await conn.execute(text(change["sql"]))
logger.info(f"✓ {change['kind']} {change['label']} applied")
except Exception as e:
if change["kind"] != "column":
raise
async with conn.begin_nested():
await conn.execute(text(change["fallback"]))
logger.warning(f"! {change['label']}: 已按 NULL 建列(原表有数据,NOT NULL 填不出默认值),"
f"请回填后再收紧 ({e})")
logger.info("✓ Schema upgrade finished")
if __name__ == "__main__":
asyncio.run(main("--apply" in sys.argv))

View File

@ -120,9 +120,10 @@ FQBN='esp32:esp32:esp32s3:PSRAM=opi,FlashSize=16M,FlashMode=dio,CDCOnBoot=defaul
- **版本门控**:`/api/terminal/follows` 每个天体多返回一个 `icon_ver`(= 图标资源行的 id,
也是图标端点的 `ETag`)。固件只在 `icon_ver` 变化时下载,稳态零流量;实测一个天体
拉 40+64 两张 ~130ms。
- **一个天体维护了多张图标**时后端取**最新**(`created_at desc, id desc`)。
⚠️ 管理端 `ResourceManager.tsx` 的预览和删除用的是 `items[0]`(`created_at` **最旧**那条),
重复上传时两边指向的不是同一行,换完图要对一下 `stat` 里的版本号。
- **换图是覆盖语义**:同一个天体同一种资源类型只保留一行,重复上传时后端删掉旧行
并删掉旧文件(`app/api/celestial_resource.py`)。终端只在 `icon_ver` 变化时重下,
管理端预览/删除的 `items[0]` 和后端取的那一行从此指向同一张图。
(只有 `icon` 是覆盖;`texture` / `model` 是多值列表,追加语义保持不变。)
- **固件缓存** `src/iconstore.{h,cpp}`:一个关注天体一个槽(上限 `MAX_BODIES=6`),
点阵 `heap_caps_malloc(MALLOC_CAP_SPIRAM|MALLOC_CAP_8BIT)` 惰性分配,每槽
3200+8192=11392B,最坏 6 槽约 68KB PSRAM;取消关注的槽在下次 sync 时释放(缓冲区留着复用)。
@ -156,6 +157,41 @@ curl --noproxy '*' -H "Authorization: Bearer $TOK" -o /tmp/ic_599_40.bin \
凭据、后端地址、token 都存在 NVS 命名空间 `cosmo`。
### 后端地址(换服务器 / 上 https 只看这一节)
地址存 NVS 键 `api`,**只有 NVS 里还没有这个键时**才用编译期的默认值:
```c
// cosmo_terminal.ino:26
#define DEFAULT_API "http://192.168.31.2:8000/api"
```
换正式地址有三种改法,前两种不用重新烧录:
| 改法 | 怎么做 | 适用 |
| --- | --- | --- |
| 串口 | `api https://your-domain.com/api`(写 NVS,立刻重新取数) | 设备在手边,最快 |
| 配网页 | 串口 `portal` 或首次开机 → `http://192.168.4.1` → 后端地址 | 没有串口线 |
| 重烧 | 改 `cosmo_terminal.ino:26` 的 `DEFAULT_API` 再编译烧录;**只有 NVS 里没有 `api` 键的设备**才会读到它 | 批量出厂 |
> 重烧固件**不清 NVS**:已经配过网/敲过 `api` 的设备,改了 `DEFAULT_API` 也还是旧地址。
> 要么串口 `api <新地址>` 覆盖一次,要么串口 `reset` 清空整个 NVS(连 WiFi 凭据和 token
> 一起清,设备会回到配网+配对流程),让它重新回落到 `DEFAULT_API`。
> 配网页现在预填**当前生效的地址**(以前预填 `DEFAULT_API`,重开配网页直接点保存
> 会把地址刷回编译期那个局域网 IP)。
`http://` 和 `https://` 都支持:`src/api.cpp:tls_make()` 按前缀分流,`https://` 走
`WiFiClientSecure`(`setHandshakeTimeout(10)`),其余走裸 `HTTPClient`。
- **不校验证书链**(`setInsecure()`):只要链路加密。代价是理论上可被中间人,
收益是换 CA / 证书续期都不用重烧固件。要严格校验就把 `setInsecure()` 换成
`setCACert(root_ca)`,代价是根证书编进固件、换 CA 必须重烧。
- TLS 握手瞬时要多占几十 KB 堆(`WiFiClientSecure` 的 mbedTLS 上下文)。这块板子是
OPI PSRAM、堆本来就宽,编译后 RAM 仍在 16% 左右,但这条只在编译层面验证过,
上正式 https 服务后建议串口 `stat` 看一眼剩余堆。
- **别把服务器配成 HTTP→HTTPS 的 301**:`HTTPClient` 不跟跳转,终端会直接拿不到数据。
要么地址直接填 `https://`,要么让 `/api/` 继续留在 80 上。
## 绑定与取数
```

View File

@ -93,7 +93,9 @@ static void build_identity()
/* ---------------- 配网 ---------------- */
static void start_provision()
{
portal_start(g_ap_ssid, DEFAULT_API);
/* 预填 NVS 里当前生效的地址:填 DEFAULT_API 的话,重开配网页直接点保存
会把后端地址刷回编译期那个局域网 IP */
portal_start(g_ap_ssid, g_ui.api);
g_ui.wifi_up = false; /* 配网时 STA 已断开,状态栏不能再显示 NET:ONLINE */
set_screen(SC_PROVISION);
Serial.printf("[portal] AP=%s -> http://192.168.4.1\n", g_ap_ssid);

View File

@ -1,10 +1,33 @@
#include <Arduino.h>
#include <HTTPClient.h>
#include <ArduinoJson.h>
#include <WiFiClientSecure.h>
#include <memory>
#include <string.h>
#include "api.h"
/* https 用 TLS,但不校验证书链(setInsecure):要的是链路加密,不是身份认证。
板子自带的根证书包一换签发机构(Let's Encrypt 换 CA)就得重烧固件,中间证书
轮换也会连不上。要严格校验,把根证书 PEM 编进固件再改成 setCACert(root_ca)。 */
static std::unique_ptr<WiFiClientSecure> tls_make(const char *url)
{
if (strncmp(url, "https://", 8) != 0) return nullptr;
auto tls = std::make_unique<WiFiClientSecure>();
tls->setInsecure();
tls->setHandshakeTimeout(10); /* 秒;TLS 握手比 HTTP 更容易卡住 */
return tls;
}
/* HTTPClient 只是借用这个 client,但 ~HTTPClient() 无条件 _client->stop(),而 end()
在连接已断开或 keep-alive 复用时不会把 _client 置空。所以 tls 必须在 HTTPClient
之前声明(局部变量逆序析构 = tls 比 http 活得久),否则析构时踩已释放的 client。 */
static void http_begin(HTTPClient &http, const char *url, const std::unique_ptr<WiFiClientSecure> &tls)
{
if (tls) http.begin(*tls, url);
else http.begin(url);
}
/* 中文标题被过滤空了就用 event_type 兜底:"close_approach" -> "CLOSE APPROACH" */
static void copy_upper(char *dst, size_t cap, const char *src)
{
@ -63,10 +86,11 @@ static void copy_title(char *dst, size_t cap, const char *src)
static bool http_json(const char *url, const char *body, const char *token,
JsonDocument &doc, int *status)
{
auto tls = tls_make(url); /* 声明在 http 之前,见 http_begin 注释 */
HTTPClient http;
http.setConnectTimeout(6000);
http.setTimeout(8000);
http.begin(url);
http_begin(http, url, tls);
http.addHeader("Content-Type", "application/json");
if (token && *token) http.addHeader("Authorization", String("Bearer ") + token);
int code = body ? http.POST(body) : http.GET();
@ -192,10 +216,11 @@ bool api_icon(const char *base, const char *token, const char *body_id, int size
void *dst, size_t cap, int *http_status)
{
String url = String(base) + "/terminal/bodies/" + body_id + "/icon?size=" + size;
auto tls = tls_make(url.c_str()); /* 声明在 http 之前,见 http_begin 注释 */
HTTPClient http;
http.setConnectTimeout(6000);
http.setTimeout(8000);
http.begin(url);
http_begin(http, url.c_str(), tls);
if (token && *token) http.addHeader("Authorization", String("Bearer ") + token);
int code = http.GET();
if (http_status) *http_status = code;