8900 commited on
Update hf-sync-manager.mjs
Browse files- hf-sync-manager.mjs +351 -72
hf-sync-manager.mjs
CHANGED
|
@@ -1,101 +1,380 @@
|
|
| 1 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
import path from "node:path";
|
| 3 |
import { execSync } from "node:child_process";
|
| 4 |
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
var
|
| 8 |
-
var
|
| 9 |
-
var
|
| 10 |
-
var
|
| 11 |
-
var
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
-
|
| 14 |
-
var botMemory = { conversations:[], memory:[], logs:[] };
|
| 15 |
|
| 16 |
-
function log(msg){ console.log("[hf-sync] "+msg); }
|
| 17 |
-
function repoUrl(){ return "https://user:"+HF_TOKEN+"@huggingface.co/datasets/"+DATASET_ID; }
|
| 18 |
-
function git(args,cwd){ return execSync("git "+args,{ cwd:cwd||REPO_DIR, stdio:"pipe", timeout:120000, env:Object.assign({},process.env,{GIT_TERMINAL_PROMPT:"0"}) }).toString().trim(); }
|
| 19 |
-
function sleep(ms){ return new Promise(resolve=>setTimeout(resolve,ms)); }
|
| 20 |
|
| 21 |
-
function
|
| 22 |
-
|
| 23 |
-
execSync("rm -rf "+REPO_DIR,{stdio:"pipe"});
|
| 24 |
-
try{ git("clone --depth 1 "+repoUrl()+" "+REPO_DIR,"/tmp"); } catch(e){ git("init",REPO_DIR); git("remote add origin "+repoUrl(),REPO_DIR); }
|
| 25 |
-
try{ git("config user.email openclaw@hf.space"); git("config user.name OpenClaw-Memory-Sync"); git("config pull.rebase false"); } catch(e){ }
|
| 26 |
}
|
| 27 |
|
| 28 |
-
function
|
| 29 |
-
|
| 30 |
-
|
| 31 |
}
|
| 32 |
|
| 33 |
-
function
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
git('commit -m "memory-sync '+ts+'"');
|
| 41 |
-
try{ git("push --quiet origin HEAD:main"); } catch(e){ git("push --quiet origin HEAD:master"); }
|
| 42 |
-
return count;
|
| 43 |
-
}catch(e){ log("Push failed: "+e.message); return 0; }
|
| 44 |
}
|
| 45 |
|
| 46 |
-
function
|
| 47 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
|
| 49 |
-
|
| 50 |
-
function getRepoDateDir(dateStr){ var dir=path.join(REPO_DIR,"dataset-memory",dateStr); fs.mkdirSync(dir,{recursive:true}); return dir; }
|
| 51 |
|
| 52 |
-
function
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
}
|
| 59 |
|
| 60 |
-
function
|
| 61 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
}
|
| 63 |
|
| 64 |
-
function
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 68 |
}
|
| 69 |
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 75 |
}
|
| 76 |
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
log("OPENCLAW_DIR : "+OPENCLAW_DIR);
|
| 80 |
-
log("HF_DATASET_ID: "+(DATASET_ID||"NOT SET"));
|
| 81 |
-
log("Upload interval: "+(UPLOAD_INTERVAL/3600000)+"h");
|
| 82 |
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 90 |
|
| 91 |
-
|
| 92 |
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 96 |
}
|
|
|
|
|
|
|
|
|
|
| 97 |
|
| 98 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 99 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 100 |
|
| 101 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
// ============================================================
|
| 2 |
+
// HF Sync Manager - Full Mirror Edition
|
| 3 |
+
//
|
| 4 |
+
// Mirrors ~/.openclaw/ to HF Dataset every 30 min.
|
| 5 |
+
//
|
| 6 |
+
// CRITICAL RULES:
|
| 7 |
+
// 1. NEVER restore openclaw.json from Dataset.
|
| 8 |
+
// openclaw.json is owned by setup-hf-config.mjs only.
|
| 9 |
+
// Restoring it causes an infinite restart loop:
|
| 10 |
+
// user saves -> gateway restarts -> we overwrite their
|
| 11 |
+
// settings with old Dataset copy -> OpenClaw restarts again.
|
| 12 |
+
//
|
| 13 |
+
// 2. RUNNING_FLAG is never deleted on process exit.
|
| 14 |
+
// It lives in /tmp which is only cleared on true container
|
| 15 |
+
// restart. Gateway restarts do NOT clear /tmp.
|
| 16 |
+
// This correctly distinguishes the two restart types.
|
| 17 |
+
//
|
| 18 |
+
// Excluded from sync:
|
| 19 |
+
// openclaw.json (owned by setup, never touch)
|
| 20 |
+
// openclaw.json.bak (backup files)
|
| 21 |
+
// agents/*/sessions/ (session transcripts, runtime only)
|
| 22 |
+
// agents/*/qmd/ (vector index, rebuilt on boot)
|
| 23 |
+
// canvas/ (UI runtime files)
|
| 24 |
+
// *.bak *.tmp *.log
|
| 25 |
+
// ============================================================
|
| 26 |
+
|
| 27 |
+
import fs from "node:fs";
|
| 28 |
import path from "node:path";
|
| 29 |
import { execSync } from "node:child_process";
|
| 30 |
|
| 31 |
+
// ββ config ββββββββββββββββ
|
| 32 |
+
|
| 33 |
+
var HOME = process.env.OPENCLAW_HOME || process.env.HOME || "/home/user";
|
| 34 |
+
var OPENCLAW_DIR = path.join(HOME, ".openclaw");
|
| 35 |
+
var WORKSPACE = path.join(OPENCLAW_DIR, "workspace");
|
| 36 |
+
var HF_TOKEN = (process.env.HF_TOKEN || "").trim();
|
| 37 |
+
var DATASET_ID = (process.env.HF_DATASET_ID || "").trim();
|
| 38 |
+
var REPO_DIR = "/tmp/oc-dataset";
|
| 39 |
+
var INTERVAL = 30 * 60 * 1000; // 30 minutes
|
| 40 |
+
|
| 41 |
+
// This flag file distinguishes container restart from gateway restart.
|
| 42 |
+
// Container restart: /tmp is wiped, flag is gone -> do full restore.
|
| 43 |
+
// Gateway restart: /tmp survives, flag exists -> skip restore.
|
| 44 |
+
// IMPORTANT: we never delete this flag on process exit.
|
| 45 |
+
var RUNNING_FLAG = "/tmp/.hf-sync-running";
|
| 46 |
+
|
| 47 |
+
// Files/dirs to EXCLUDE from sync in any direction
|
| 48 |
+
var EXCLUDE_NAMES = [
|
| 49 |
+
"openclaw.json", // NEVER touch - owned by setup-hf-config.mjs
|
| 50 |
+
"openclaw.json.bak", // backup files
|
| 51 |
+
"sessions", // session transcripts inside agents/*/sessions/
|
| 52 |
+
"qmd", // sqlite vector index inside agents/*/qmd/
|
| 53 |
+
"canvas", // UI runtime files
|
| 54 |
+
".git"
|
| 55 |
+
];
|
| 56 |
+
var EXCLUDE_EXT = [".bak", ".tmp", ".log"];
|
| 57 |
|
| 58 |
+
// ββ helpers ββββββββββββββββ
|
|
|
|
| 59 |
|
| 60 |
+
function log(msg) { console.log("[hf-sync] " + msg); }
|
|
|
|
|
|
|
|
|
|
| 61 |
|
| 62 |
+
function sleep(ms) {
|
| 63 |
+
return new Promise(function(resolve) { setTimeout(resolve, ms); });
|
|
|
|
|
|
|
|
|
|
| 64 |
}
|
| 65 |
|
| 66 |
+
function repoUrl() {
|
| 67 |
+
return "https://user:" + HF_TOKEN +
|
| 68 |
+
"@huggingface.co/datasets/" + DATASET_ID;
|
| 69 |
}
|
| 70 |
|
| 71 |
+
function git(args) {
|
| 72 |
+
return execSync("git " + args, {
|
| 73 |
+
cwd: REPO_DIR,
|
| 74 |
+
stdio: "pipe",
|
| 75 |
+
timeout: 120000,
|
| 76 |
+
env: Object.assign({}, process.env, { GIT_TERMINAL_PROMPT: "0" })
|
| 77 |
+
}).toString().trim();
|
|
|
|
|
|
|
|
|
|
|
|
|
| 78 |
}
|
| 79 |
|
| 80 |
+
function shouldExclude(name) {
|
| 81 |
+
var i;
|
| 82 |
+
for (i = 0; i < EXCLUDE_NAMES.length; i++) {
|
| 83 |
+
if (name === EXCLUDE_NAMES[i]) return true;
|
| 84 |
+
}
|
| 85 |
+
for (i = 0; i < EXCLUDE_EXT.length; i++) {
|
| 86 |
+
if (name.indexOf(EXCLUDE_EXT[i]) >= 0) return true;
|
| 87 |
+
}
|
| 88 |
+
return false;
|
| 89 |
+
}
|
| 90 |
|
| 91 |
+
// ββ git repo setup ββββββββββββββββββββ
|
|
|
|
| 92 |
|
| 93 |
+
function ensureRepo() {
|
| 94 |
+
if (fs.existsSync(path.join(REPO_DIR, ".git"))) return;
|
| 95 |
+
execSync("rm -rf " + REPO_DIR, { stdio: "pipe" });
|
| 96 |
+
try {
|
| 97 |
+
execSync("git clone --depth 1 " + repoUrl() + " " + REPO_DIR, {
|
| 98 |
+
stdio: "pipe", timeout: 120000,
|
| 99 |
+
env: Object.assign({}, process.env, { GIT_TERMINAL_PROMPT: "0" })
|
| 100 |
+
});
|
| 101 |
+
log("Repo cloned");
|
| 102 |
+
} catch (e) {
|
| 103 |
+
log("Clone failed, init empty: " + e.message);
|
| 104 |
+
fs.mkdirSync(REPO_DIR, { recursive: true });
|
| 105 |
+
execSync("git init " + REPO_DIR, { stdio: "pipe" });
|
| 106 |
+
git("remote add origin " + repoUrl());
|
| 107 |
+
}
|
| 108 |
+
try {
|
| 109 |
+
git("config user.email openclaw@hf.space");
|
| 110 |
+
git("config user.name OpenClaw-Sync");
|
| 111 |
+
git("config pull.rebase false");
|
| 112 |
+
} catch (e) { /* non-fatal */ }
|
| 113 |
}
|
| 114 |
|
| 115 |
+
function pull() {
|
| 116 |
+
try { git("fetch --quiet origin"); } catch (e) { /* ignore */ }
|
| 117 |
+
try { git("pull --quiet --no-rebase origin main"); }
|
| 118 |
+
catch (e) {
|
| 119 |
+
try { git("pull --quiet --no-rebase origin master"); }
|
| 120 |
+
catch (e2) { /* empty repo */ }
|
| 121 |
+
}
|
| 122 |
}
|
| 123 |
|
| 124 |
+
function push() {
|
| 125 |
+
try {
|
| 126 |
+
git("add -A");
|
| 127 |
+
var changed = git("diff --cached --name-only");
|
| 128 |
+
if (!changed) return 0;
|
| 129 |
+
var n = changed.split("\n").filter(Boolean).length;
|
| 130 |
+
var ts = new Date().toISOString().replace("T", " ").substring(0, 19);
|
| 131 |
+
git('commit -m "sync ' + ts + '"');
|
| 132 |
+
try { git("push --quiet origin HEAD:main"); }
|
| 133 |
+
catch (e) { git("push --quiet origin HEAD:master"); }
|
| 134 |
+
return n;
|
| 135 |
+
} catch (e) {
|
| 136 |
+
log("Push failed: " + e.message);
|
| 137 |
+
return 0;
|
| 138 |
+
}
|
| 139 |
}
|
| 140 |
|
| 141 |
+
// ββ directory mirror: local -> repo ββββββββ
|
| 142 |
+
|
| 143 |
+
function mirrorToRepo(srcDir, dstDir) {
|
| 144 |
+
if (!fs.existsSync(srcDir)) return 0;
|
| 145 |
+
fs.mkdirSync(dstDir, { recursive: true });
|
| 146 |
+
var count = 0;
|
| 147 |
+
fs.readdirSync(srcDir).forEach(function(name) {
|
| 148 |
+
if (shouldExclude(name)) return;
|
| 149 |
+
var src = path.join(srcDir, name);
|
| 150 |
+
var dst = path.join(dstDir, name);
|
| 151 |
+
try {
|
| 152 |
+
if (fs.statSync(src).isDirectory()) {
|
| 153 |
+
count += mirrorToRepo(src, dst);
|
| 154 |
+
} else {
|
| 155 |
+
fs.copyFileSync(src, dst);
|
| 156 |
+
count++;
|
| 157 |
+
}
|
| 158 |
+
} catch (e) {
|
| 159 |
+
log("Copy error [" + name + "]: " + e.message);
|
| 160 |
+
}
|
| 161 |
+
});
|
| 162 |
+
return count;
|
| 163 |
}
|
| 164 |
|
| 165 |
+
// ββ directory mirror: repo -> local ββββββββ
|
| 166 |
+
// Note: shouldExclude ensures openclaw.json is never restored
|
|
|
|
|
|
|
|
|
|
| 167 |
|
| 168 |
+
function mirrorFromRepo(srcDir, dstDir) {
|
| 169 |
+
if (!fs.existsSync(srcDir)) return 0;
|
| 170 |
+
fs.mkdirSync(dstDir, { recursive: true });
|
| 171 |
+
var count = 0;
|
| 172 |
+
fs.readdirSync(srcDir).forEach(function(name) {
|
| 173 |
+
if (shouldExclude(name)) return; // never restore openclaw.json
|
| 174 |
+
var src = path.join(srcDir, name);
|
| 175 |
+
var dst = path.join(dstDir, name);
|
| 176 |
+
try {
|
| 177 |
+
if (fs.statSync(src).isDirectory()) {
|
| 178 |
+
count += mirrorFromRepo(src, dst);
|
| 179 |
+
} else {
|
| 180 |
+
fs.copyFileSync(src, dst);
|
| 181 |
+
count++;
|
| 182 |
+
}
|
| 183 |
+
} catch (e) {
|
| 184 |
+
log("Restore error [" + name + "]: " + e.message);
|
| 185 |
+
}
|
| 186 |
+
});
|
| 187 |
+
return count;
|
| 188 |
+
}
|
| 189 |
+
|
| 190 |
+
// ββ seed default workspace files βββββββββββββ
|
| 191 |
+
|
| 192 |
+
function seedWorkspaceIfNeeded() {
|
| 193 |
+
fs.mkdirSync(WORKSPACE, { recursive: true });
|
| 194 |
+
fs.mkdirSync(path.join(WORKSPACE, "memory"), { recursive: true });
|
| 195 |
+
|
| 196 |
+
var soul = path.join(WORKSPACE, "SOUL.md");
|
| 197 |
+
if (!fs.existsSync(soul)) {
|
| 198 |
+
fs.writeFileSync(soul, [
|
| 199 |
+
"# Soul",
|
| 200 |
+
"",
|
| 201 |
+
"You are a helpful, warm, concise AI assistant.",
|
| 202 |
+
"",
|
| 203 |
+
"## Language",
|
| 204 |
+
"",
|
| 205 |
+
"Default language: Simplified Chinese.",
|
| 206 |
+
"Always reply in Chinese unless the user writes in another language first.",
|
| 207 |
+
"",
|
| 208 |
+
"## Tone",
|
| 209 |
+
"",
|
| 210 |
+
"- Natural and friendly, not overly formal",
|
| 211 |
+
"- Concise and to the point",
|
| 212 |
+
"- Ask one clarifying question at a time when needed"
|
| 213 |
+
].join("\n") + "\n", "utf-8");
|
| 214 |
+
log("Seeded workspace/SOUL.md");
|
| 215 |
+
}
|
| 216 |
+
|
| 217 |
+
var agents = path.join(WORKSPACE, "AGENTS.md");
|
| 218 |
+
if (!fs.existsSync(agents)) {
|
| 219 |
+
fs.writeFileSync(agents, [
|
| 220 |
+
"# Agent Instructions",
|
| 221 |
+
"",
|
| 222 |
+
"## Boot sequence",
|
| 223 |
+
"",
|
| 224 |
+
"1. Read SOUL.md - language and persona",
|
| 225 |
+
"2. Read USER.md if present - user profile",
|
| 226 |
+
"3. Read MEMORY.md - long-term facts and rules",
|
| 227 |
+
"4. Read today and yesterday memory/YYYY-MM-DD.md if present",
|
| 228 |
+
"",
|
| 229 |
+
"## Memory rules",
|
| 230 |
+
"",
|
| 231 |
+
"- Write important facts to MEMORY.md when asked",
|
| 232 |
+
"- Log daily context to memory/YYYY-MM-DD.md",
|
| 233 |
+
"- Do not ask for information already provided"
|
| 234 |
+
].join("\n") + "\n", "utf-8");
|
| 235 |
+
log("Seeded workspace/AGENTS.md");
|
| 236 |
+
}
|
| 237 |
+
|
| 238 |
+
var mem = path.join(WORKSPACE, "MEMORY.md");
|
| 239 |
+
if (!fs.existsSync(mem)) {
|
| 240 |
+
fs.writeFileSync(mem, [
|
| 241 |
+
"# Long-term Memory",
|
| 242 |
+
"",
|
| 243 |
+
"<!-- OpenClaw writes important facts here. -->",
|
| 244 |
+
"<!-- Loaded at the start of every session. -->"
|
| 245 |
+
].join("\n") + "\n", "utf-8");
|
| 246 |
+
log("Seeded workspace/MEMORY.md");
|
| 247 |
+
}
|
| 248 |
+
}
|
| 249 |
|
| 250 |
+
// ββ boot restore (container restart only) ββββββ
|
| 251 |
|
| 252 |
+
function bootRestore() {
|
| 253 |
+
if (!HF_TOKEN || !DATASET_ID) {
|
| 254 |
+
log("No HF_TOKEN/HF_DATASET_ID - skipping Dataset restore");
|
| 255 |
+
seedWorkspaceIfNeeded();
|
| 256 |
+
return;
|
| 257 |
+
}
|
| 258 |
+
|
| 259 |
+
log("Boot restore from " + DATASET_ID + "β¦");
|
| 260 |
+
try {
|
| 261 |
+
ensureRepo();
|
| 262 |
+
pull();
|
| 263 |
+
} catch (e) {
|
| 264 |
+
log("Repo init failed (" + e.message + ") - starting fresh");
|
| 265 |
+
seedWorkspaceIfNeeded();
|
| 266 |
+
return;
|
| 267 |
+
}
|
| 268 |
+
|
| 269 |
+
var repoOC = path.join(REPO_DIR, "openclaw");
|
| 270 |
+
if (fs.existsSync(repoOC)) {
|
| 271 |
+
// mirrorFromRepo skips openclaw.json automatically via shouldExclude
|
| 272 |
+
var n = mirrorFromRepo(repoOC, OPENCLAW_DIR);
|
| 273 |
+
log("Boot restore done: " + n + " file(s) restored (openclaw.json excluded)");
|
| 274 |
+
} else {
|
| 275 |
+
log("No data in Dataset yet - starting fresh");
|
| 276 |
+
}
|
| 277 |
+
|
| 278 |
+
seedWorkspaceIfNeeded();
|
| 279 |
+
log("Boot restore complete");
|
| 280 |
+
}
|
| 281 |
+
|
| 282 |
+
// ββ sync cycle (every 30 min) ββββββββββ
|
| 283 |
+
|
| 284 |
+
function runSync() {
|
| 285 |
+
if (!HF_TOKEN || !DATASET_ID) return;
|
| 286 |
+
|
| 287 |
+
var ts = new Date().toISOString().replace("T", " ").substring(0, 19);
|
| 288 |
+
log("Sync at " + ts + "β¦");
|
| 289 |
+
|
| 290 |
+
try {
|
| 291 |
+
ensureRepo();
|
| 292 |
+
pull();
|
| 293 |
+
|
| 294 |
+
var repoOC = path.join(REPO_DIR, "openclaw");
|
| 295 |
+
if (fs.existsSync(repoOC)) {
|
| 296 |
+
execSync("rm -rf " + repoOC, { stdio: "pipe" });
|
| 297 |
+
}
|
| 298 |
+
fs.mkdirSync(repoOC, { recursive: true });
|
| 299 |
+
|
| 300 |
+
// mirrorToRepo also skips openclaw.json via shouldExclude
|
| 301 |
+
// We explicitly copy openclaw.json separately only for backup
|
| 302 |
+
var n = mirrorToRepo(OPENCLAW_DIR, repoOC);
|
| 303 |
+
|
| 304 |
+
// Upload openclaw.json separately without excluding it from upload
|
| 305 |
+
// (we exclude it from RESTORE, not from upload)
|
| 306 |
+
var localCfg = path.join(OPENCLAW_DIR, "openclaw.json");
|
| 307 |
+
if (fs.existsSync(localCfg)) {
|
| 308 |
+
try {
|
| 309 |
+
var content = fs.readFileSync(localCfg, "utf-8").trim();
|
| 310 |
+
JSON.parse(content); // validate
|
| 311 |
+
fs.copyFileSync(localCfg, path.join(repoOC, "openclaw.json"));
|
| 312 |
+
n++;
|
| 313 |
+
log("openclaw.json uploaded (validated)");
|
| 314 |
+
} catch (e) {
|
| 315 |
+
log("openclaw.json skipped (invalid JSON): " + e.message);
|
| 316 |
}
|
| 317 |
+
}
|
| 318 |
+
|
| 319 |
+
log("Mirrored " + n + " file(s) to Dataset");
|
| 320 |
|
| 321 |
+
var pushed = push();
|
| 322 |
+
if (pushed > 0) {
|
| 323 |
+
log("Sync done: " + pushed + " file(s) pushed");
|
| 324 |
+
} else {
|
| 325 |
+
log("Sync done: no changes");
|
| 326 |
}
|
| 327 |
+
} catch (e) {
|
| 328 |
+
log("Sync failed (retry in 30 min): " + e.message);
|
| 329 |
+
}
|
| 330 |
+
}
|
| 331 |
+
|
| 332 |
+
// ββ main βββββββββββββββββββββββββ
|
| 333 |
+
|
| 334 |
+
(async function() {
|
| 335 |
+
log("Startingβ¦");
|
| 336 |
+
log("OPENCLAW_DIR : " + OPENCLAW_DIR);
|
| 337 |
+
log("DATASET : " + (DATASET_ID || "NOT SET"));
|
| 338 |
+
log("RUNNING_FLAG : " + RUNNING_FLAG);
|
| 339 |
+
|
| 340 |
+
fs.mkdirSync(OPENCLAW_DIR, { recursive: true });
|
| 341 |
+
fs.mkdirSync(WORKSPACE, { recursive: true });
|
| 342 |
+
|
| 343 |
+
try {
|
| 344 |
+
execSync("git config --global user.email openclaw@hf.space", { stdio: "pipe" });
|
| 345 |
+
execSync("git config --global user.name OpenClaw-Sync", { stdio: "pipe" });
|
| 346 |
+
execSync("git config --global http.postBuffer 52428800", { stdio: "pipe" });
|
| 347 |
+
execSync("git config --global pull.rebase false", { stdio: "pipe" });
|
| 348 |
+
} catch (e) { /* non-fatal */ }
|
| 349 |
+
|
| 350 |
+
// Detect restart type:
|
| 351 |
+
// /tmp survives gateway restart but is wiped on container restart.
|
| 352 |
+
// We NEVER delete this flag - only container restart clears it.
|
| 353 |
+
var isGatewayRestart = fs.existsSync(RUNNING_FLAG);
|
| 354 |
|
| 355 |
+
if (isGatewayRestart) {
|
| 356 |
+
log("Gateway restart detected - skipping boot restore");
|
| 357 |
+
log("(openclaw.json and workspace already on disk, nothing to restore)");
|
| 358 |
+
} else {
|
| 359 |
+
log("Container start detected - running full boot restore");
|
| 360 |
+
bootRestore();
|
| 361 |
+
}
|
| 362 |
+
|
| 363 |
+
// Write flag. Never delete it (not on SIGTERM, not on exit).
|
| 364 |
+
// Container restart is the only thing that clears /tmp.
|
| 365 |
+
fs.writeFileSync(RUNNING_FLAG, String(process.pid), "utf-8");
|
| 366 |
+
log("Running flag set: " + RUNNING_FLAG);
|
| 367 |
+
|
| 368 |
+
if (!HF_TOKEN || !DATASET_ID) {
|
| 369 |
+
log("Sync disabled: set HF_TOKEN and HF_DATASET_ID in Space Secrets");
|
| 370 |
+
return;
|
| 371 |
+
}
|
| 372 |
+
|
| 373 |
+
log("Sync loop starting (first run in 10 min)β¦");
|
| 374 |
+
await sleep(10 * 60 * 1000);
|
| 375 |
+
|
| 376 |
+
while (true) {
|
| 377 |
+
runSync();
|
| 378 |
+
await sleep(INTERVAL);
|
| 379 |
+
}
|
| 380 |
+
})().catch(function(e) { console.error("[hf-sync] Fatal: " + e.message); });
|