lingyuzeng/luci-app-openclaw
1
0
Fork 0
mirror of https://github.com/hotwa/luci-app-openclaw.git synced 2026-03-31 04:52:33 +00:00
Code Issues Packages Projects Releases 1 Wiki Activity

release: v1.0.4 — 适配 OpenClaw 2026.3.2

Browse Source
This commit is contained in:
10000ge10000
2026-03-05 15:20:31 +08:00
parent 649e73120f
commit 5f381eebd7
12 changed files with 315 additions and 53 deletions
Download Patch File Download Diff File Expand all files Collapse all files

174
root/usr/share/openclaw/oc-config.sh
View File

@@ -359,7 +359,8 @@ configure_model() {
echo -e " ${CYAN}9)${NC} xAI Grok (Grok-3/3-mini)"
echo -e " ${CYAN}10)${NC} Groq (Llama 4, Llama 3.3)"
echo -e " ${CYAN}11)${NC} 硅基流动 SiliconFlow"
echo -e " ${CYAN}12)${NC} 自定义 OpenAI 兼容 API"
echo -e " ${CYAN}12)${NC} Ollama (本地模型，无需 API Key)"
echo -e " ${CYAN}13)${NC} 自定义 OpenAI 兼容 API"
echo -e " ${CYAN}0)${NC} 返回"
echo ""
prompt_with_default "请选择" "1" choice
@@ -373,14 +374,6 @@ configure_model() {
echo -e " ${CYAN}预启用模型认证插件...${NC}"
enable_auth_plugins
echo ""
# 检查并安装 gemini-cli (官方向导的 Google Gemini OAuth 依赖)
if ! command -v gemini >/dev/null 2>&1; then
local npm_bin="${NODE_BASE}/bin/npm"
if [ -x "$npm_bin" ]; then
echo -e " ${CYAN}安装 Gemini CLI (官方向导 Google OAuth 依赖)...${NC}"
"$npm_bin" install -g @google/gemini-cli --prefix="$OC_GLOBAL" >/dev/null 2>&1 || true
fi
fi
(oc_cmd configure --section model) || echo -e " ${YELLOW}配置向导已退出${NC}"
echo ""
ask_restart
@@ -718,6 +711,122 @@ configure_model() {
fi
;;
12)
echo ""
echo -e " ${BOLD}🦙 Ollama 本地模型配置${NC}"
echo -e " ${YELLOW}Ollama 在本地或局域网运行大模型，无需 API Key${NC}"
echo -e " ${YELLOW}安装 Ollama: https://ollama.com${NC}"
echo ""
echo -e " ${CYAN}连接方式:${NC}"
echo -e " ${CYAN}a)${NC} 本机运行 (localhost:11434)"
echo -e " ${CYAN}b)${NC} 局域网其他设备"
echo ""
prompt_with_default "请选择" "a" ollama_mode
local ollama_url=""
case "$ollama_mode" in
b)
prompt_with_default "Ollama 地址 (如 192.168.1.100:11434)" "" ollama_host
if [ -n "$ollama_host" ]; then
# 补全协议前缀
case "$ollama_host" in
http://*|https://*) ollama_url="${ollama_host}" ;;
*) ollama_url="http://${ollama_host}" ;;
esac
# v2026.3.2: Ollama 使用原生 API，baseUrl 不带 /v1
ollama_url=$(echo "$ollama_url" | sed 's|/v1/*$||;s|/*$||')
fi
;;
*)
ollama_url="http://127.0.0.1:11434"
;;
esac
if [ -n "$ollama_url" ]; then
# 检测 Ollama 是否可达
echo ""
echo -e " ${CYAN}检测 Ollama 连通性...${NC}"
local ollama_base=$(echo "$ollama_url" | sed 's|/v1$||')
local ollama_check=$(curl -sf --connect-timeout 3 --max-time 5 "${ollama_base}/api/tags" 2>/dev/null || echo "")
if [ -n "$ollama_check" ]; then
echo -e " ${GREEN}✅ Ollama 已连接${NC}"
# 列出已安装的模型
local model_list=$("$NODE_BIN" -e "
try{
const d=JSON.parse(process.argv[1]);
(d.models||[]).forEach((m,i)=>console.log(' '+(i+1)+') '+m.name));
}catch(e){}
" "$ollama_check" 2>/dev/null)
if [ -n "$model_list" ]; then
echo -e " ${CYAN}已安装的模型:${NC}"
echo "$model_list"
echo -e " ${CYAN}m)${NC} 手动输入模型名"
echo ""
prompt_with_default "请选择模型" "1" ollama_sel
if [ "$ollama_sel" = "m" ]; then
prompt_with_default "请输入模型名称" "llama3.3" model_name
elif echo "$ollama_sel" | grep -qE '^[0-9]+$'; then
model_name=$("$NODE_BIN" -e "
try{
const d=JSON.parse(process.argv[1]);
const m=(d.models||[])[parseInt(process.argv[2])-1];
console.log(m?m.name:'');
}catch(e){console.log('');}
" "$ollama_check" "$ollama_sel" 2>/dev/null)
if [ -z "$model_name" ]; then
echo -e " ${YELLOW}无效选择，使用默认模型${NC}"
model_name="llama3.3"
fi
fi
else
echo -e " ${YELLOW}未检测到已安装模型，请先在 Ollama 中拉取模型:${NC}"
echo -e " ${CYAN} ollama pull llama3.3${NC}"
echo ""
prompt_with_default "请输入模型名称" "llama3.3" model_name
fi
else
echo -e " ${YELLOW}⚠️ 无法连接 Ollama (${ollama_base})${NC}"
echo -e " ${YELLOW} 请确认 Ollama 已启动并可访问${NC}"
echo -e " ${CYAN} 提示: 如果 Ollama 在其他设备上，需设置 OLLAMA_HOST=0.0.0.0${NC}"
echo ""
prompt_with_default "仍要继续配置? (y/n)" "n" force_continue
if ! confirm_yes "$force_continue"; then
return
fi
prompt_with_default "请输入模型名称" "llama3.3" model_name
fi
if [ -n "$model_name" ]; then
# Ollama 无需 API Key，使用占位符
auth_set_apikey ollama "ollama-local" "ollama:local"
# v2026.3.2: Ollama 使用原生 ollama API，不再走 OpenAI 兼容层
_RCP_PROV="ollama" _RCP_URL="$ollama_url" _RCP_KEY="ollama-local" _RCP_MID="$model_name" _RCP_MNAME="$model_name" "$NODE_BIN" -e "
const fs=require('fs');
let d={};
try{d=JSON.parse(fs.readFileSync('${CONFIG_FILE}','utf8'));}catch(e){}
if(!d.models)d.models={};
if(!d.models.providers)d.models.providers={};
d.models.mode='merge';
d.models.providers['ollama']={
baseUrl:process.env._RCP_URL,
apiKey:process.env._RCP_KEY,
api:'ollama',
models:[{
id:process.env._RCP_MID,
name:process.env._RCP_MNAME,
reasoning:false,
input:['text'],
cost:{input:0,output:0,cacheRead:0,cacheWrite:0},
contextWindow:128000,
maxTokens:32000
}]
};
fs.writeFileSync('${CONFIG_FILE}',JSON.stringify(d,null,2));
" 2>/dev/null
chown openclaw:openclaw "$CONFIG_FILE" 2>/dev/null || true
register_and_set_model "ollama/${model_name}"
echo -e " ${GREEN}✅ Ollama 已配置，活跃模型: ollama/${model_name}${NC}"
echo -e " ${CYAN} Ollama 地址: ${ollama_url}${NC}"
fi
fi
;;
13)
echo ""
echo -e " ${BOLD}自定义 OpenAI 兼容 API${NC}"
echo -e " ${YELLOW}支持任何兼容 OpenAI API 格式的服务商${NC}"
@@ -1108,6 +1217,34 @@ health_check() {
local gw_port=$(json_get gateway.port)
gw_port=${gw_port:-18789}
# ── v2026.3.2: 使用官方 config validate 验证配置 ──
if command -v openclaw >/dev/null 2>&1 || [ -n "$OC_ENTRY" ]; then
echo -e " ${CYAN}验证配置文件格式...${NC}"
local validate_out=""
validate_out=$(oc_cmd config validate --json 2>/dev/null) || true
if [ -n "$validate_out" ]; then
local has_errors=$("$NODE_BIN" -e "
try{const r=JSON.parse(process.argv[1]);
if(r.valid===true){console.log('OK');}
else if(r.errors&&r.errors.length>0){r.errors.forEach(e=>console.log('ERR:'+e.message));}
else{console.log('OK');}}catch(e){console.log('SKIP');}
" "$validate_out" 2>/dev/null)
if [ "$has_errors" = "OK" ]; then
echo -e " ${GREEN}✅ 配置文件格式有效${NC}"
elif [ "$has_errors" = "SKIP" ]; then
echo -e " ${YELLOW}⚠️ 无法解析验证结果，跳过${NC}"
else
echo -e " ${RED}❌ 配置文件存在错误:${NC}"
echo "$has_errors" | while IFS= read -r line; do
echo -e " ${YELLOW}• ${line#ERR:}${NC}"
done
fi
else
echo -e " ${YELLOW}⚠️ config validate 不可用，跳过格式验证${NC}"
fi
echo ""
fi
# ── 自动修复: 移除旧版错误写入的顶层 models.xxx 无效键 ──
if [ -f "$CONFIG_FILE" ]; then
local has_bad_models=$("$NODE_BIN" -e "
@@ -1152,6 +1289,19 @@ health_check() {
echo -e " ${RED}❌ HTTP 响应异常 (${http_code})${NC}"
fi
# v2026.3.2: 使用 gateway health --json 做深度健康检查 (HTTP /health 已被 SPA 接管)
local health_resp=$(oc_cmd gateway health --json 2>/dev/null)
if [ -n "$health_resp" ]; then
local health_ok=$("$NODE_BIN" -e "try{const h=JSON.parse(process.argv[1]);console.log(h.ok?'ok':'fail');}catch(e){console.log('parse_error');}" "$health_resp" 2>/dev/null)
if [ "$health_ok" = "ok" ]; then
echo -e " ${GREEN}✅ Gateway 健康检查正常${NC}"
elif [ "$health_ok" = "parse_error" ]; then
echo -e " ${YELLOW}⚠️ Gateway 健康检查响应无法解析${NC}"
else
echo -e " ${YELLOW}⚠️ Gateway 健康检查异常${NC}"
fi
fi
if [ -f "$CONFIG_FILE" ]; then
echo -e " ${GREEN}✅ 配置文件存在${NC}"
else
@@ -1296,9 +1446,9 @@ reset_to_defaults() {
local _node_bin
_node_bin=$(which node 2>/dev/null || echo "$NODE_BIN")
if command -v timeout >/dev/null 2>&1; then
timeout 10 sh -c "\"$_node_bin\" \"$OC_ENTRY\" onboard --non-interactive --accept-risk" >/dev/null 2>&1 || true
timeout 10 sh -c "\"$_node_bin\" \"$OC_ENTRY\" onboard --non-interactive --accept-risk --tools-profile coding" >/dev/null 2>&1 || true
else
"$_node_bin" "$OC_ENTRY" onboard --non-interactive --accept-risk >/dev/null 2>&1 &
"$_node_bin" "$OC_ENTRY" onboard --non-interactive --accept-risk --tools-profile coding >/dev/null 2>&1 &
local _ob_pid=$!
sleep 10
kill "$_ob_pid" 2>/dev/null || true
@@ -1318,6 +1468,8 @@ reset_to_defaults() {
json_set gateway.controlUi.dangerouslyDisableDeviceAuth true
json_set gateway.controlUi.dangerouslyAllowHostHeaderOriginFallback true
json_set gateway.tailscale.mode off
json_set acp.dispatch.enabled false
json_set tools.profile coding
# 同步 token 到 UCI
. /lib/functions.sh 2>/dev/null || true