8900 commited on
Update hf-sync-manager.mjs
Browse files- hf-sync-manager.mjs +62 -156
hf-sync-manager.mjs
CHANGED
|
@@ -3,33 +3,34 @@
|
|
| 3 |
//
|
| 4 |
// TRACK 1 - Settings (openclaw.json):
|
| 5 |
// Boot: repo/settings/openclaw.json -> local disk
|
| 6 |
-
//
|
| 7 |
-
//
|
|
|
|
|
|
|
|
|
|
| 8 |
//
|
| 9 |
// TRACK 2 - Memory (conversations, history, context, agents, logs):
|
| 10 |
// Boot: repo/data/* -> local disk (full restore, all sessions)
|
| 11 |
-
// Upload: local disk -> repo/data/* every 3h
|
| 12 |
-
//
|
| 13 |
// ============================================================
|
| 14 |
|
| 15 |
import fs from "node:fs";
|
| 16 |
import path from "node:path";
|
| 17 |
import { execSync } from "node:child_process";
|
| 18 |
|
| 19 |
-
// -- config -----------------------------------
|
| 20 |
|
| 21 |
var HOME = process.env.OPENCLAW_HOME || process.env.HOME || "/home/user";
|
| 22 |
var OPENCLAW_DIR = path.join(HOME, ".openclaw");
|
| 23 |
var HF_TOKEN = (process.env.HF_TOKEN || "").trim();
|
| 24 |
var DATASET_ID = (process.env.HF_DATASET_ID || "").trim();
|
| 25 |
var REPO_DIR = "/tmp/hf-memory-sync";
|
| 26 |
-
var UPLOAD_INTERVAL = 3 * 60 * 60 * 1000;
|
| 27 |
|
| 28 |
-
|
| 29 |
-
var
|
| 30 |
-
var REPO_SETTINGS = "settings"; // repo subfolder
|
| 31 |
|
| 32 |
-
// Memory track: these dirs/files inside OPENCLAW_DIR
|
| 33 |
var MEMORY_TARGETS = [
|
| 34 |
"memory",
|
| 35 |
"history",
|
|
@@ -39,8 +40,8 @@ var MEMORY_TARGETS = [
|
|
| 39 |
"agents"
|
| 40 |
];
|
| 41 |
|
| 42 |
-
// Never touch these in either track
|
| 43 |
var NEVER_SYNC = [
|
|
|
|
| 44 |
"openclaw.json.bak",
|
| 45 |
".sync-state.json",
|
| 46 |
"settings.json",
|
|
@@ -52,7 +53,7 @@ var NEVER_SYNC = [
|
|
| 52 |
"canvas"
|
| 53 |
];
|
| 54 |
|
| 55 |
-
// -- helpers -----------------------------------
|
| 56 |
|
| 57 |
function log(msg) {
|
| 58 |
console.log("[hf-sync] " + msg);
|
|
@@ -64,7 +65,7 @@ function sleep(ms) {
|
|
| 64 |
|
| 65 |
function repoUrl() {
|
| 66 |
return "https://user:" + HF_TOKEN +
|
| 67 |
-
|
| 68 |
}
|
| 69 |
|
| 70 |
function git(args) {
|
|
@@ -77,16 +78,40 @@ function git(args) {
|
|
| 77 |
}
|
| 78 |
|
| 79 |
function isMemoryTarget(name) {
|
| 80 |
-
|
|
|
|
| 81 |
if (name === NEVER_SYNC[i]) return false;
|
| 82 |
}
|
| 83 |
-
for (
|
| 84 |
if (name === MEMORY_TARGETS[i]) return true;
|
| 85 |
}
|
| 86 |
return false;
|
| 87 |
}
|
| 88 |
|
| 89 |
-
//
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 90 |
|
| 91 |
function ensureRepo() {
|
| 92 |
if (fs.existsSync(path.join(REPO_DIR, ".git"))) return;
|
|
@@ -121,7 +146,6 @@ function pullLatest() {
|
|
| 121 |
}
|
| 122 |
}
|
| 123 |
|
| 124 |
-
// Single commit+push for both tracks
|
| 125 |
function commitAndPush(label) {
|
| 126 |
try {
|
| 127 |
git("add -A");
|
|
@@ -142,17 +166,6 @@ function commitAndPush(label) {
|
|
| 142 |
}
|
| 143 |
}
|
| 144 |
|
| 145 |
-
// -- copy helper ------------------------------
|
| 146 |
-
|
| 147 |
-
function copyItem(src, dst) {
|
| 148 |
-
if (fs.statSync(src).isDirectory()) {
|
| 149 |
-
execSync("cp -r " + src + " " + dst, { stdio: "pipe" });
|
| 150 |
-
} else {
|
| 151 |
-
fs.mkdirSync(path.dirname(dst), { recursive: true });
|
| 152 |
-
fs.copyFileSync(src, dst);
|
| 153 |
-
}
|
| 154 |
-
}
|
| 155 |
-
|
| 156 |
// ============================================================
|
| 157 |
// TRACK 1: Settings
|
| 158 |
// ============================================================
|
|
@@ -162,14 +175,25 @@ function settingsBootRestore() {
|
|
| 162 |
var localFile = path.join(OPENCLAW_DIR, SETTINGS_FILE);
|
| 163 |
|
| 164 |
if (!fs.existsSync(repoFile)) {
|
| 165 |
-
log("[settings] No settings in repo yet,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 166 |
return false;
|
| 167 |
}
|
| 168 |
|
| 169 |
try {
|
| 170 |
fs.mkdirSync(OPENCLAW_DIR, { recursive: true });
|
|
|
|
|
|
|
|
|
|
|
|
|
| 171 |
fs.copyFileSync(repoFile, localFile);
|
| 172 |
-
log("[settings] Boot restore done: openclaw.json <- repo");
|
| 173 |
return true;
|
| 174 |
} catch (e) {
|
| 175 |
log("[settings] Boot restore failed: " + e.message);
|
|
@@ -179,8 +203,10 @@ function settingsBootRestore() {
|
|
| 179 |
|
| 180 |
function settingsUpload() {
|
| 181 |
var localFile = path.join(OPENCLAW_DIR, SETTINGS_FILE);
|
| 182 |
-
|
| 183 |
-
|
|
|
|
|
|
|
| 184 |
return 0;
|
| 185 |
}
|
| 186 |
|
|
@@ -188,7 +214,9 @@ function settingsUpload() {
|
|
| 188 |
var repoFile = path.join(repoDir, SETTINGS_FILE);
|
| 189 |
|
| 190 |
fs.mkdirSync(repoDir, { recursive: true });
|
|
|
|
| 191 |
fs.copyFileSync(localFile, repoFile);
|
|
|
|
| 192 |
return 1;
|
| 193 |
}
|
| 194 |
|
|
@@ -198,10 +226,7 @@ function settingsUpload() {
|
|
| 198 |
|
| 199 |
function memoryBootRestore() {
|
| 200 |
var src = path.join(REPO_DIR, "data");
|
| 201 |
-
if (!fs.existsSync(src))
|
| 202 |
-
log("[memory] No memory data in repo yet");
|
| 203 |
-
return 0;
|
| 204 |
-
}
|
| 205 |
|
| 206 |
var items = fs.readdirSync(src);
|
| 207 |
var copied = 0;
|
|
@@ -249,123 +274,4 @@ function memoryUpload() {
|
|
| 249 |
function scanMemoryDisk() {
|
| 250 |
if (!fs.existsSync(OPENCLAW_DIR)) return [];
|
| 251 |
return fs.readdirSync(OPENCLAW_DIR).filter(isMemoryTarget);
|
| 252 |
-
}
|
| 253 |
-
|
| 254 |
-
// ============================================================
|
| 255 |
-
// Boot sequence (runs once at startup)
|
| 256 |
-
// ============================================================
|
| 257 |
-
|
| 258 |
-
function bootRestore() {
|
| 259 |
-
if (!HF_TOKEN || !DATASET_ID) {
|
| 260 |
-
log("HF_TOKEN or HF_DATASET_ID not set, skipping all restore");
|
| 261 |
-
return;
|
| 262 |
-
}
|
| 263 |
-
|
| 264 |
-
log("Boot restore starting from " + DATASET_ID + "...");
|
| 265 |
-
try {
|
| 266 |
-
ensureRepo();
|
| 267 |
-
pullLatest();
|
| 268 |
-
} catch (e) {
|
| 269 |
-
log("Repo init failed (non-fatal): " + e.message);
|
| 270 |
-
return;
|
| 271 |
-
}
|
| 272 |
-
|
| 273 |
-
settingsBootRestore();
|
| 274 |
-
var restored = memoryBootRestore();
|
| 275 |
-
if (restored > 0) {
|
| 276 |
-
log("[memory] Boot restore done: " + restored + " item(s) -> " + OPENCLAW_DIR);
|
| 277 |
-
} else {
|
| 278 |
-
log("[memory] Boot restore: no memory data in repository yet");
|
| 279 |
-
}
|
| 280 |
-
}
|
| 281 |
-
|
| 282 |
-
// ============================================================
|
| 283 |
-
// Upload cycle (runs every 3h, NEVER downloads)
|
| 284 |
-
// ============================================================
|
| 285 |
-
|
| 286 |
-
function runUploadCycle() {
|
| 287 |
-
if (!HF_TOKEN || !DATASET_ID) return;
|
| 288 |
-
|
| 289 |
-
var ts = new Date().toISOString().replace("T", " ").substring(0, 19);
|
| 290 |
-
log("Upload cycle starting at " + ts + "...");
|
| 291 |
-
|
| 292 |
-
try {
|
| 293 |
-
ensureRepo();
|
| 294 |
-
pullLatest(); // pull repo meta only, then we overwrite with our data
|
| 295 |
-
|
| 296 |
-
// Track 1: settings (overwrite single copy)
|
| 297 |
-
var sCopied = settingsUpload();
|
| 298 |
-
log("[settings] Copied " + sCopied + " file(s) to repo");
|
| 299 |
-
|
| 300 |
-
// Track 2: memory (upload from disk, never read back)
|
| 301 |
-
var targets = scanMemoryDisk();
|
| 302 |
-
if (targets.length > 0) {
|
| 303 |
-
var mCopied = memoryUpload();
|
| 304 |
-
log("[memory] Copied " + mCopied + " item(s) to repo: " + targets.join(", "));
|
| 305 |
-
} else {
|
| 306 |
-
log("[memory] No memory targets found on disk yet");
|
| 307 |
-
}
|
| 308 |
-
|
| 309 |
-
var changed = commitAndPush("sync");
|
| 310 |
-
if (changed > 0) {
|
| 311 |
-
log("Upload done: " + changed + " file(s) committed at " + ts);
|
| 312 |
-
} else {
|
| 313 |
-
log("Upload done: no changes since last sync");
|
| 314 |
-
}
|
| 315 |
-
} catch (e) {
|
| 316 |
-
log("Upload cycle failed (will retry in 3h): " + e.message);
|
| 317 |
-
}
|
| 318 |
-
}
|
| 319 |
-
|
| 320 |
-
// ============================================================
|
| 321 |
-
// Upload loop
|
| 322 |
-
// ============================================================
|
| 323 |
-
|
| 324 |
-
async function startUploadLoop() {
|
| 325 |
-
log("First upload scheduled in 10 min...");
|
| 326 |
-
await sleep(10 * 60 * 1000);
|
| 327 |
-
runUploadCycle();
|
| 328 |
-
|
| 329 |
-
while (true) {
|
| 330 |
-
await sleep(UPLOAD_INTERVAL);
|
| 331 |
-
runUploadCycle();
|
| 332 |
-
}
|
| 333 |
-
}
|
| 334 |
-
|
| 335 |
-
// ============================================================
|
| 336 |
-
// Main
|
| 337 |
-
// ============================================================
|
| 338 |
-
|
| 339 |
-
async function main() {
|
| 340 |
-
log("Starting...");
|
| 341 |
-
log("OPENCLAW_DIR : " + OPENCLAW_DIR);
|
| 342 |
-
log("HF_DATASET_ID : " + (DATASET_ID || "NOT SET"));
|
| 343 |
-
log("Settings target : " + SETTINGS_FILE + " (single copy, overwrite)");
|
| 344 |
-
log("Memory targets : " + MEMORY_TARGETS.join(", "));
|
| 345 |
-
log("Upload every : " + (UPLOAD_INTERVAL / 3600000) + "h");
|
| 346 |
-
|
| 347 |
-
fs.mkdirSync(OPENCLAW_DIR, { recursive: true });
|
| 348 |
-
|
| 349 |
-
try {
|
| 350 |
-
execSync("git config --global user.email openclaw@hf.space", { stdio: "pipe" });
|
| 351 |
-
execSync("git config --global user.name OpenClaw-Memory-Sync", { stdio: "pipe" });
|
| 352 |
-
execSync("git config --global http.postBuffer 52428800", { stdio: "pipe" });
|
| 353 |
-
execSync("git config --global pull.rebase false", { stdio: "pipe" });
|
| 354 |
-
} catch (e) { /* non-fatal */ }
|
| 355 |
-
|
| 356 |
-
bootRestore();
|
| 357 |
-
|
| 358 |
-
if (HF_TOKEN && DATASET_ID) {
|
| 359 |
-
startUploadLoop().catch(function(e) {
|
| 360 |
-
log("Upload loop crashed: " + e.message);
|
| 361 |
-
});
|
| 362 |
-
} else {
|
| 363 |
-
log("Upload loop disabled: set HF_TOKEN and HF_DATASET_ID in Space Secrets");
|
| 364 |
-
}
|
| 365 |
-
|
| 366 |
-
log("Ready.");
|
| 367 |
-
}
|
| 368 |
-
|
| 369 |
-
main().catch(function(e) {
|
| 370 |
-
console.error("[hf-sync] Fatal: " + e.message);
|
| 371 |
-
});
|
|
|
|
| 3 |
//
|
| 4 |
// TRACK 1 - Settings (openclaw.json):
|
| 5 |
// Boot: repo/settings/openclaw.json -> local disk
|
| 6 |
+
// ONLY if repo file passes JSON validation
|
| 7 |
+
// If invalid: keep local file, log warning
|
| 8 |
+
// Upload: local disk -> repo/settings/openclaw.json
|
| 9 |
+
// ONLY if local file passes JSON validation
|
| 10 |
+
// Single copy, always overwrite, never accumulate
|
| 11 |
//
|
| 12 |
// TRACK 2 - Memory (conversations, history, context, agents, logs):
|
| 13 |
// Boot: repo/data/* -> local disk (full restore, all sessions)
|
| 14 |
+
// Upload: local disk -> repo/data/* every 3h
|
| 15 |
+
// Upload only, NEVER download during runtime
|
| 16 |
// ============================================================
|
| 17 |
|
| 18 |
import fs from "node:fs";
|
| 19 |
import path from "node:path";
|
| 20 |
import { execSync } from "node:child_process";
|
| 21 |
|
| 22 |
+
// -- config ---------------------------------------------------
|
| 23 |
|
| 24 |
var HOME = process.env.OPENCLAW_HOME || process.env.HOME || "/home/user";
|
| 25 |
var OPENCLAW_DIR = path.join(HOME, ".openclaw");
|
| 26 |
var HF_TOKEN = (process.env.HF_TOKEN || "").trim();
|
| 27 |
var DATASET_ID = (process.env.HF_DATASET_ID || "").trim();
|
| 28 |
var REPO_DIR = "/tmp/hf-memory-sync";
|
| 29 |
+
var UPLOAD_INTERVAL = 3 * 60 * 60 * 1000;
|
| 30 |
|
| 31 |
+
var SETTINGS_FILE = "openclaw.json";
|
| 32 |
+
var REPO_SETTINGS = "settings";
|
|
|
|
| 33 |
|
|
|
|
| 34 |
var MEMORY_TARGETS = [
|
| 35 |
"memory",
|
| 36 |
"history",
|
|
|
|
| 40 |
"agents"
|
| 41 |
];
|
| 42 |
|
|
|
|
| 43 |
var NEVER_SYNC = [
|
| 44 |
+
"openclaw.json",
|
| 45 |
"openclaw.json.bak",
|
| 46 |
".sync-state.json",
|
| 47 |
"settings.json",
|
|
|
|
| 53 |
"canvas"
|
| 54 |
];
|
| 55 |
|
| 56 |
+
// -- helpers --------------------------------------------------
|
| 57 |
|
| 58 |
function log(msg) {
|
| 59 |
console.log("[hf-sync] " + msg);
|
|
|
|
| 65 |
|
| 66 |
function repoUrl() {
|
| 67 |
return "https://user:" + HF_TOKEN +
|
| 68 |
+
"@huggingface.co/datasets/" + DATASET_ID;
|
| 69 |
}
|
| 70 |
|
| 71 |
function git(args) {
|
|
|
|
| 78 |
}
|
| 79 |
|
| 80 |
function isMemoryTarget(name) {
|
| 81 |
+
var i;
|
| 82 |
+
for (i = 0; i < NEVER_SYNC.length; i++) {
|
| 83 |
if (name === NEVER_SYNC[i]) return false;
|
| 84 |
}
|
| 85 |
+
for (i = 0; i < MEMORY_TARGETS.length; i++) {
|
| 86 |
if (name === MEMORY_TARGETS[i]) return true;
|
| 87 |
}
|
| 88 |
return false;
|
| 89 |
}
|
| 90 |
|
| 91 |
+
// FIX: validate JSON before any config file operation
|
| 92 |
+
// Returns true if file exists and contains valid JSON
|
| 93 |
+
function isValidJson(filePath) {
|
| 94 |
+
try {
|
| 95 |
+
if (!fs.existsSync(filePath)) return false;
|
| 96 |
+
var content = fs.readFileSync(filePath, "utf-8").trim();
|
| 97 |
+
if (!content) return false;
|
| 98 |
+
JSON.parse(content);
|
| 99 |
+
return true;
|
| 100 |
+
} catch (e) {
|
| 101 |
+
return false;
|
| 102 |
+
}
|
| 103 |
+
}
|
| 104 |
+
|
| 105 |
+
function copyItem(src, dst) {
|
| 106 |
+
if (fs.statSync(src).isDirectory()) {
|
| 107 |
+
execSync("cp -r " + src + " " + dst, { stdio: "pipe" });
|
| 108 |
+
} else {
|
| 109 |
+
fs.mkdirSync(path.dirname(dst), { recursive: true });
|
| 110 |
+
fs.copyFileSync(src, dst);
|
| 111 |
+
}
|
| 112 |
+
}
|
| 113 |
+
|
| 114 |
+
// -- git repo setup ------------------------------------------
|
| 115 |
|
| 116 |
function ensureRepo() {
|
| 117 |
if (fs.existsSync(path.join(REPO_DIR, ".git"))) return;
|
|
|
|
| 146 |
}
|
| 147 |
}
|
| 148 |
|
|
|
|
| 149 |
function commitAndPush(label) {
|
| 150 |
try {
|
| 151 |
git("add -A");
|
|
|
|
| 166 |
}
|
| 167 |
}
|
| 168 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 169 |
// ============================================================
|
| 170 |
// TRACK 1: Settings
|
| 171 |
// ============================================================
|
|
|
|
| 175 |
var localFile = path.join(OPENCLAW_DIR, SETTINGS_FILE);
|
| 176 |
|
| 177 |
if (!fs.existsSync(repoFile)) {
|
| 178 |
+
log("[settings] No settings in repo yet, keeping local defaults");
|
| 179 |
+
return false;
|
| 180 |
+
}
|
| 181 |
+
|
| 182 |
+
// FIX: validate repo file before overwriting local config
|
| 183 |
+
if (!isValidJson(repoFile)) {
|
| 184 |
+
log("[settings] WARN: repo openclaw.json is invalid JSON, skipping restore to protect local config");
|
| 185 |
+
log("[settings] Corrupted repo file will be overwritten on next upload");
|
| 186 |
return false;
|
| 187 |
}
|
| 188 |
|
| 189 |
try {
|
| 190 |
fs.mkdirSync(OPENCLAW_DIR, { recursive: true });
|
| 191 |
+
// Backup local before overwriting
|
| 192 |
+
if (fs.existsSync(localFile)) {
|
| 193 |
+
fs.copyFileSync(localFile, localFile + ".pre-restore-bak");
|
| 194 |
+
}
|
| 195 |
fs.copyFileSync(repoFile, localFile);
|
| 196 |
+
log("[settings] Boot restore done: openclaw.json <- repo (validated OK)");
|
| 197 |
return true;
|
| 198 |
} catch (e) {
|
| 199 |
log("[settings] Boot restore failed: " + e.message);
|
|
|
|
| 203 |
|
| 204 |
function settingsUpload() {
|
| 205 |
var localFile = path.join(OPENCLAW_DIR, SETTINGS_FILE);
|
| 206 |
+
|
| 207 |
+
// FIX: validate local file before uploading to repo
|
| 208 |
+
if (!isValidJson(localFile)) {
|
| 209 |
+
log("[settings] WARN: local openclaw.json is invalid JSON, skipping upload");
|
| 210 |
return 0;
|
| 211 |
}
|
| 212 |
|
|
|
|
| 214 |
var repoFile = path.join(repoDir, SETTINGS_FILE);
|
| 215 |
|
| 216 |
fs.mkdirSync(repoDir, { recursive: true });
|
| 217 |
+
// Single copy, direct overwrite - no version accumulation
|
| 218 |
fs.copyFileSync(localFile, repoFile);
|
| 219 |
+
log("[settings] Uploaded openclaw.json -> repo (validated OK)");
|
| 220 |
return 1;
|
| 221 |
}
|
| 222 |
|
|
|
|
| 226 |
|
| 227 |
function memoryBootRestore() {
|
| 228 |
var src = path.join(REPO_DIR, "data");
|
| 229 |
+
if (!fs.existsSync(src)) return 0;
|
|
|
|
|
|
|
|
|
|
| 230 |
|
| 231 |
var items = fs.readdirSync(src);
|
| 232 |
var copied = 0;
|
|
|
|
| 274 |
function scanMemoryDisk() {
|
| 275 |
if (!fs.existsSync(OPENCLAW_DIR)) return [];
|
| 276 |
return fs.readdirSync(OPENCLAW_DIR).filter(isMemoryTarget);
|
| 277 |
+
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|