8900 commited on
Update hf-sync-manager.mjs
Browse files- hf-sync-manager.mjs +128 -46
hf-sync-manager.mjs
CHANGED
|
@@ -6,27 +6,24 @@ import fs from "node:fs";
|
|
| 6 |
import path from "node:path";
|
| 7 |
import { execSync } from "node:child_process";
|
| 8 |
|
| 9 |
-
// ββ config ββββββββββββββββ
|
| 10 |
-
|
| 11 |
var HOME = process.env.OPENCLAW_HOME || process.env.HOME || "/home/user";
|
| 12 |
var OPENCLAW_DIR = path.join(HOME, ".openclaw");
|
| 13 |
var WORKSPACE = path.join(OPENCLAW_DIR, "workspace");
|
| 14 |
var HF_TOKEN = (process.env.HF_TOKEN || "").trim();
|
| 15 |
var DATASET_ID = (process.env.HF_DATASET_ID || "").trim();
|
| 16 |
var REPO_DIR = "/tmp/oc-dataset";
|
| 17 |
-
var INTERVAL = 30 * 60 * 1000;
|
| 18 |
-
|
| 19 |
-
var RUNNING_FLAG = "/tmp/hf-sync-running";
|
| 20 |
|
| 21 |
var EXCLUDE = ["sessions", "qmd", "canvas", ".git"];
|
| 22 |
var EXCLUDE_EXT = [".bak", ".tmp", ".log"];
|
| 23 |
|
| 24 |
-
// ββ
|
| 25 |
|
| 26 |
function log(msg) { console.log("[hf-sync] " + msg); }
|
| 27 |
|
| 28 |
function sleep(ms) {
|
| 29 |
-
return new Promise(function(
|
| 30 |
}
|
| 31 |
|
| 32 |
function repoUrl() {
|
|
@@ -43,14 +40,6 @@ env: Object.assign({}, process.env, { GIT_TERMINAL_PROMPT: "0" })
|
|
| 43 |
}).toString().trim();
|
| 44 |
}
|
| 45 |
|
| 46 |
-
function isValidJson(p) {
|
| 47 |
-
try {
|
| 48 |
-
if (!fs.existsSync(p)) return false;
|
| 49 |
-
JSON.parse(fs.readFileSync(p, "utf-8").trim());
|
| 50 |
-
return true;
|
| 51 |
-
} catch (e) { return false; }
|
| 52 |
-
}
|
| 53 |
-
|
| 54 |
function shouldExclude(name) {
|
| 55 |
var i;
|
| 56 |
for (i = 0; i < EXCLUDE.length; i++) {
|
|
@@ -62,20 +51,19 @@ if (name.indexOf(EXCLUDE_EXT[i]) >= 0) return true;
|
|
| 62 |
return false;
|
| 63 |
}
|
| 64 |
|
| 65 |
-
// ββ
|
| 66 |
|
| 67 |
function ensureRepo() {
|
| 68 |
if (fs.existsSync(path.join(REPO_DIR, ".git"))) return;
|
| 69 |
execSync("rm -rf " + REPO_DIR, { stdio: "pipe" });
|
| 70 |
try {
|
| 71 |
execSync("git clone --depth 1 " + repoUrl() + " " + REPO_DIR, {
|
| 72 |
-
stdio:
|
| 73 |
-
|
| 74 |
-
env: Object.assign({}, process.env, { GIT_TERMINAL_PROMPT: "0" })
|
| 75 |
});
|
| 76 |
log("Repo cloned");
|
| 77 |
} catch (e) {
|
| 78 |
-
log("Clone failed,
|
| 79 |
fs.mkdirSync(REPO_DIR, { recursive: true });
|
| 80 |
execSync("git init " + REPO_DIR, { stdio: "pipe" });
|
| 81 |
git("remote add origin " + repoUrl());
|
|
@@ -113,22 +101,19 @@ return 0;
|
|
| 113 |
}
|
| 114 |
}
|
| 115 |
|
| 116 |
-
// ββ
|
| 117 |
|
| 118 |
-
function
|
| 119 |
if (!fs.existsSync(srcDir)) return 0;
|
| 120 |
fs.mkdirSync(dstDir, { recursive: true });
|
| 121 |
-
|
| 122 |
-
var items = fs.readdirSync(srcDir);
|
| 123 |
var count = 0;
|
| 124 |
-
|
| 125 |
-
items.forEach(function(name) {
|
| 126 |
if (shouldExclude(name)) return;
|
| 127 |
-
var src
|
| 128 |
-
var dst
|
| 129 |
try {
|
| 130 |
if (fs.statSync(src).isDirectory()) {
|
| 131 |
-
count +=
|
| 132 |
} else {
|
| 133 |
fs.copyFileSync(src, dst);
|
| 134 |
count++;
|
|
@@ -137,25 +122,19 @@ count++;
|
|
| 137 |
log("Copy error [" + name + "]: " + e.message);
|
| 138 |
}
|
| 139 |
});
|
| 140 |
-
|
| 141 |
return count;
|
| 142 |
}
|
| 143 |
|
| 144 |
-
|
| 145 |
-
|
| 146 |
-
function mirrorFromRepo(srcDir, dstDir) {
|
| 147 |
if (!fs.existsSync(srcDir)) return 0;
|
| 148 |
fs.mkdirSync(dstDir, { recursive: true });
|
| 149 |
-
|
| 150 |
-
var items = fs.readdirSync(srcDir);
|
| 151 |
var count = 0;
|
| 152 |
-
|
| 153 |
-
|
| 154 |
-
var
|
| 155 |
-
var dst = path.join(dstDir, name);
|
| 156 |
try {
|
| 157 |
if (fs.statSync(src).isDirectory()) {
|
| 158 |
-
count +=
|
| 159 |
} else {
|
| 160 |
fs.copyFileSync(src, dst);
|
| 161 |
count++;
|
|
@@ -164,11 +143,10 @@ count++;
|
|
| 164 |
log("Restore error [" + name + "]: " + e.message);
|
| 165 |
}
|
| 166 |
});
|
| 167 |
-
|
| 168 |
return count;
|
| 169 |
}
|
| 170 |
|
| 171 |
-
// ββ
|
| 172 |
|
| 173 |
function seedWorkspaceIfNeeded() {
|
| 174 |
fs.mkdirSync(WORKSPACE, { recursive: true });
|
|
@@ -209,9 +187,9 @@ fs.writeFileSync(agents, [
|
|
| 209 |
"",
|
| 210 |
"## Memory rules",
|
| 211 |
"",
|
| 212 |
-
"- Write important facts to MEMORY.md when asked
|
| 213 |
"- Log daily context to memory/YYYY-MM-DD.md",
|
| 214 |
-
"- Do not ask for information
|
| 215 |
].join("\n") + "\n", "utf-8");
|
| 216 |
log("Seeded workspace/AGENTS.md");
|
| 217 |
}
|
|
@@ -228,6 +206,110 @@ log("Seeded workspace/MEMORY.md");
|
|
| 228 |
}
|
| 229 |
}
|
| 230 |
|
| 231 |
-
//
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 232 |
|
| 233 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
import path from "node:path";
|
| 7 |
import { execSync } from "node:child_process";
|
| 8 |
|
|
|
|
|
|
|
| 9 |
var HOME = process.env.OPENCLAW_HOME || process.env.HOME || "/home/user";
|
| 10 |
var OPENCLAW_DIR = path.join(HOME, ".openclaw");
|
| 11 |
var WORKSPACE = path.join(OPENCLAW_DIR, "workspace");
|
| 12 |
var HF_TOKEN = (process.env.HF_TOKEN || "").trim();
|
| 13 |
var DATASET_ID = (process.env.HF_DATASET_ID || "").trim();
|
| 14 |
var REPO_DIR = "/tmp/oc-dataset";
|
| 15 |
+
var INTERVAL = 30 * 60 * 1000;
|
| 16 |
+
var RUNNING_FLAG = "/tmp/.hf-sync-running";
|
|
|
|
| 17 |
|
| 18 |
var EXCLUDE = ["sessions", "qmd", "canvas", ".git"];
|
| 19 |
var EXCLUDE_EXT = [".bak", ".tmp", ".log"];
|
| 20 |
|
| 21 |
+
// βββββββββββββββββββββββββββββ
|
| 22 |
|
| 23 |
function log(msg) { console.log("[hf-sync] " + msg); }
|
| 24 |
|
| 25 |
function sleep(ms) {
|
| 26 |
+
return new Promise(function(resolve) { setTimeout(resolve, ms); });
|
| 27 |
}
|
| 28 |
|
| 29 |
function repoUrl() {
|
|
|
|
| 40 |
}).toString().trim();
|
| 41 |
}
|
| 42 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
function shouldExclude(name) {
|
| 44 |
var i;
|
| 45 |
for (i = 0; i < EXCLUDE.length; i++) {
|
|
|
|
| 51 |
return false;
|
| 52 |
}
|
| 53 |
|
| 54 |
+
// βββββββββββββββββββββββββββββ
|
| 55 |
|
| 56 |
function ensureRepo() {
|
| 57 |
if (fs.existsSync(path.join(REPO_DIR, ".git"))) return;
|
| 58 |
execSync("rm -rf " + REPO_DIR, { stdio: "pipe" });
|
| 59 |
try {
|
| 60 |
execSync("git clone --depth 1 " + repoUrl() + " " + REPO_DIR, {
|
| 61 |
+
stdio: "pipe", timeout: 120000,
|
| 62 |
+
env: Object.assign({}, process.env, { GIT_TERMINAL_PROMPT: "0" })
|
|
|
|
| 63 |
});
|
| 64 |
log("Repo cloned");
|
| 65 |
} catch (e) {
|
| 66 |
+
log("Clone failed, init empty: " + e.message);
|
| 67 |
fs.mkdirSync(REPO_DIR, { recursive: true });
|
| 68 |
execSync("git init " + REPO_DIR, { stdio: "pipe" });
|
| 69 |
git("remote add origin " + repoUrl());
|
|
|
|
| 101 |
}
|
| 102 |
}
|
| 103 |
|
| 104 |
+
// βββββββββββββββββββββββββββββ
|
| 105 |
|
| 106 |
+
function mirrorDirToRepo(srcDir, dstDir) {
|
| 107 |
if (!fs.existsSync(srcDir)) return 0;
|
| 108 |
fs.mkdirSync(dstDir, { recursive: true });
|
|
|
|
|
|
|
| 109 |
var count = 0;
|
| 110 |
+
fs.readdirSync(srcDir).forEach(function(name) {
|
|
|
|
| 111 |
if (shouldExclude(name)) return;
|
| 112 |
+
var src = path.join(srcDir, name);
|
| 113 |
+
var dst = path.join(dstDir, name);
|
| 114 |
try {
|
| 115 |
if (fs.statSync(src).isDirectory()) {
|
| 116 |
+
count += mirrorDirToRepo(src, dst);
|
| 117 |
} else {
|
| 118 |
fs.copyFileSync(src, dst);
|
| 119 |
count++;
|
|
|
|
| 122 |
log("Copy error [" + name + "]: " + e.message);
|
| 123 |
}
|
| 124 |
});
|
|
|
|
| 125 |
return count;
|
| 126 |
}
|
| 127 |
|
| 128 |
+
function mirrorDirFromRepo(srcDir, dstDir) {
|
|
|
|
|
|
|
| 129 |
if (!fs.existsSync(srcDir)) return 0;
|
| 130 |
fs.mkdirSync(dstDir, { recursive: true });
|
|
|
|
|
|
|
| 131 |
var count = 0;
|
| 132 |
+
fs.readdirSync(srcDir).forEach(function(name) {
|
| 133 |
+
var src = path.join(srcDir, name);
|
| 134 |
+
var dst = path.join(dstDir, name);
|
|
|
|
| 135 |
try {
|
| 136 |
if (fs.statSync(src).isDirectory()) {
|
| 137 |
+
count += mirrorDirFromRepo(src, dst);
|
| 138 |
} else {
|
| 139 |
fs.copyFileSync(src, dst);
|
| 140 |
count++;
|
|
|
|
| 143 |
log("Restore error [" + name + "]: " + e.message);
|
| 144 |
}
|
| 145 |
});
|
|
|
|
| 146 |
return count;
|
| 147 |
}
|
| 148 |
|
| 149 |
+
// βββββββββββββββββββββββββββββ
|
| 150 |
|
| 151 |
function seedWorkspaceIfNeeded() {
|
| 152 |
fs.mkdirSync(WORKSPACE, { recursive: true });
|
|
|
|
| 187 |
"",
|
| 188 |
"## Memory rules",
|
| 189 |
"",
|
| 190 |
+
"- Write important facts to MEMORY.md when asked",
|
| 191 |
"- Log daily context to memory/YYYY-MM-DD.md",
|
| 192 |
+
"- Do not ask for information already provided"
|
| 193 |
].join("\n") + "\n", "utf-8");
|
| 194 |
log("Seeded workspace/AGENTS.md");
|
| 195 |
}
|
|
|
|
| 206 |
}
|
| 207 |
}
|
| 208 |
|
| 209 |
+
// βββββββββββββββββββββββββββββ
|
| 210 |
+
|
| 211 |
+
function bootRestore() {
|
| 212 |
+
if (!HF_TOKEN || !DATASET_ID) {
|
| 213 |
+
log("No HF_TOKEN/HF_DATASET_ID - skipping Dataset restore");
|
| 214 |
+
seedWorkspaceIfNeeded();
|
| 215 |
+
return;
|
| 216 |
+
}
|
| 217 |
+
|
| 218 |
+
log("Boot restore from " + DATASET_ID + "β¦");
|
| 219 |
+
try {
|
| 220 |
+
ensureRepo();
|
| 221 |
+
pull();
|
| 222 |
+
} catch (e) {
|
| 223 |
+
log("Repo init failed (" + e.message + ") - starting fresh");
|
| 224 |
+
seedWorkspaceIfNeeded();
|
| 225 |
+
return;
|
| 226 |
+
}
|
| 227 |
+
|
| 228 |
+
var repoOC = path.join(REPO_DIR, "openclaw");
|
| 229 |
+
if (fs.existsSync(repoOC)) {
|
| 230 |
+
var n = mirrorDirFromRepo(repoOC, OPENCLAW_DIR);
|
| 231 |
+
log("Boot restore done: " + n + " file(s) -> " + OPENCLAW_DIR);
|
| 232 |
+
} else {
|
| 233 |
+
log("No data in Dataset yet - starting fresh");
|
| 234 |
+
}
|
| 235 |
+
|
| 236 |
+
seedWorkspaceIfNeeded();
|
| 237 |
+
log("Boot restore complete");
|
| 238 |
+
}
|
| 239 |
+
|
| 240 |
+
function runSync() {
|
| 241 |
+
if (!HF_TOKEN || !DATASET_ID) return;
|
| 242 |
+
|
| 243 |
+
var ts = new Date().toISOString().replace("T", " ").substring(0, 19);
|
| 244 |
+
log("Sync at " + ts + "β¦");
|
| 245 |
+
|
| 246 |
+
try {
|
| 247 |
+
ensureRepo();
|
| 248 |
+
pull();
|
| 249 |
+
|
| 250 |
+
var repoOC = path.join(REPO_DIR, "openclaw");
|
| 251 |
+
if (fs.existsSync(repoOC)) {
|
| 252 |
+
execSync("rm -rf " + repoOC, { stdio: "pipe" });
|
| 253 |
+
}
|
| 254 |
+
fs.mkdirSync(repoOC, { recursive: true });
|
| 255 |
+
|
| 256 |
+
var n = mirrorDirToRepo(OPENCLAW_DIR, repoOC);
|
| 257 |
+
log("Mirrored " + n + " file(s)");
|
| 258 |
+
|
| 259 |
+
var pushed = push();
|
| 260 |
+
if (pushed > 0) {
|
| 261 |
+
log("Sync done: " + pushed + " file(s) pushed");
|
| 262 |
+
} else {
|
| 263 |
+
log("Sync done: no changes");
|
| 264 |
+
}
|
| 265 |
+
|
| 266 |
+
} catch (e) {
|
| 267 |
+
log("Sync failed (retry in 30 min): " + e.message);
|
| 268 |
+
}
|
| 269 |
+
}
|
| 270 |
+
|
| 271 |
+
// βββββββββββββββββββββββββββββ
|
| 272 |
+
|
| 273 |
+
(async function() {
|
| 274 |
+
log("Startingβ¦");
|
| 275 |
+
log("OPENCLAW_DIR : " + OPENCLAW_DIR);
|
| 276 |
+
log("DATASET : " + (DATASET_ID || "NOT SET"));
|
| 277 |
|
| 278 |
+
fs.mkdirSync(OPENCLAW_DIR, { recursive: true });
|
| 279 |
+
fs.mkdirSync(WORKSPACE, { recursive: true });
|
| 280 |
+
|
| 281 |
+
try {
|
| 282 |
+
execSync("git config --global user.email openclaw@hf.space", { stdio: "pipe" });
|
| 283 |
+
execSync("git config --global user.name OpenClaw-Sync", { stdio: "pipe" });
|
| 284 |
+
execSync("git config --global http.postBuffer 52428800", { stdio: "pipe" });
|
| 285 |
+
execSync("git config --global pull.rebase false", { stdio: "pipe" });
|
| 286 |
+
} catch (e) { /* non-fatal */ }
|
| 287 |
+
|
| 288 |
+
var isGatewayRestart = fs.existsSync(RUNNING_FLAG);
|
| 289 |
+
|
| 290 |
+
if (isGatewayRestart) {
|
| 291 |
+
log("Gateway restart - skipping boot restore (data on disk)");
|
| 292 |
+
} else {
|
| 293 |
+
log("Container start - running full boot restore");
|
| 294 |
+
bootRestore();
|
| 295 |
+
}
|
| 296 |
+
|
| 297 |
+
fs.writeFileSync(RUNNING_FLAG, String(process.pid), "utf-8");
|
| 298 |
+
|
| 299 |
+
process.on("exit", function() { try { fs.unlinkSync(RUNNING_FLAG); } catch (e) { /* ok */ } });
|
| 300 |
+
process.on("SIGTERM", function() { try { fs.unlinkSync(RUNNING_FLAG); } catch (e) { /* ok */ } process.exit(0); });
|
| 301 |
+
process.on("SIGINT", function() { try { fs.unlinkSync(RUNNING_FLAG); } catch (e) { /* ok */ } process.exit(0); });
|
| 302 |
+
|
| 303 |
+
if (!HF_TOKEN || !DATASET_ID) {
|
| 304 |
+
log("Sync disabled: set HF_TOKEN and HF_DATASET_ID in Secrets");
|
| 305 |
+
return;
|
| 306 |
+
}
|
| 307 |
+
|
| 308 |
+
log("Sync loop starting (first run in 10 min)β¦");
|
| 309 |
+
await sleep(10 * 60 * 1000);
|
| 310 |
+
|
| 311 |
+
while (true) {
|
| 312 |
+
runSync();
|
| 313 |
+
await sleep(INTERVAL);
|
| 314 |
+
}
|
| 315 |
+
})().catch(function(e) { console.error("[hf-sync] Fatal: " + e.message); });
|