server.mjs
sha256:dc02e0ce2c9c3616e8ce241f8a700e1558caae53cb1af50daeffdb2685cf9c3d
feat(delegation): 7C-L1 live gate smoke + hosted bridge/gat…
Human
minor
⚠ breaking
31 days ago
| 1 | /** |
| 2 | * Knowtation Hub Bridge — Connect GitHub + Back up now + indexer + search for hosted product. |
| 3 | * Stores GitHub token per user; sync fetches notes + full proposals from canister and pushes to repo (snapshot JSON + markdown). |
| 4 | * Index/search: pull vault from canister, chunk → embed → sqlite-vec per user; search via POST /api/v1/search. |
| 5 | * On Netlify, tokens and vector DBs persist via Netlify Blobs (set by netlify/functions/bridge.mjs). |
| 6 | * Env: SESSION_SECRET, CANISTER_URL, HUB_BASE_URL; optional HUB_UI_ORIGIN, HUB_UI_PATH (default /hub), GITHUB_*, EMBEDDING_*, BRIDGE_PORT, DATA_DIR. |
| 7 | * Consolidation: CONSOLIDATION_LLM_API_KEY / OPENAI_API_KEY, CONSOLIDATION_LLM_MODEL; CONSOLIDATION_MEMORY_ENCRYPT=true omits raw event payloads from consolidation LLM prompts. |
| 8 | */ |
| 9 | |
| 10 | import fs from 'fs'; |
| 11 | import path from 'path'; |
| 12 | import os from 'os'; |
| 13 | import { fileURLToPath } from 'url'; |
| 14 | import crypto from 'crypto'; |
| 15 | import dotenv from 'dotenv'; |
| 16 | import express from 'express'; |
| 17 | import jwt from 'jsonwebtoken'; |
| 18 | import multer from 'multer'; |
| 19 | import AdmZip from 'adm-zip'; |
| 20 | import { parseCanisterProposalGetBody } from '../../lib/canister-proposal-response-parse.mjs'; |
| 21 | import { runImport } from '../../lib/import.mjs'; |
| 22 | import { IMPORT_SOURCE_TYPES } from '../../lib/import-source-types.mjs'; |
| 23 | import { commitImageToRepo, parseGitHubRepoUrl, validateImageExtension, validateMagicBytes } from '../../lib/github-commit-image.mjs'; |
| 24 | import { mergeProvenanceFrontmatter } from '../../lib/hub-provenance.mjs'; |
| 25 | import { createIndexTimer } from './index-timing.mjs'; |
| 26 | import { computeChunkContentHashTagged } from '../../lib/chunk-content-hash.mjs'; |
| 27 | import { |
| 28 | defaultBridgeEmbeddingModelForProvider, |
| 29 | resolveIndexerChunkOptions, |
| 30 | } from '../../lib/indexer-chunk-options.mjs'; |
| 31 | import { |
| 32 | runWithConcurrency, |
| 33 | parseEmbedConcurrency, |
| 34 | parseEmbedBatchSize, |
| 35 | } from '../../lib/parallel-embed-pool.mjs'; |
| 36 | import { partitionChunksForReindex } from '../../lib/index-partition.mjs'; |
| 37 | import { |
| 38 | estimateEmbedSeconds, |
| 39 | shouldUseBackgroundIndex, |
| 40 | parseSyncBudgetSeconds, |
| 41 | parseMaxSyncChunks, |
| 42 | } from '../../lib/bridge-index-preflight-estimate.mjs'; |
| 43 | import { |
| 44 | acquireJobLock, |
| 45 | releaseJobLock, |
| 46 | peekJobLock, |
| 47 | } from '../../lib/bridge-index-job-lock.mjs'; |
| 48 | import { |
| 49 | setLastIndexedAt, |
| 50 | getLastIndexedAt, |
| 51 | } from '../../lib/bridge-index-last-indexed.mjs'; |
| 52 | import { signInternalRequest } from '../../lib/bridge-internal-hmac.mjs'; |
| 53 | import { assertBackgroundKickoffOk } from '../../lib/bridge-index-kickoff-response.mjs'; |
| 54 | import { writeNote } from '../../lib/write.mjs'; |
| 55 | import { resolveVaultRelativePath, parseFrontmatterAndBody } from '../../lib/vault.mjs'; |
| 56 | import { |
| 57 | resolveEffectiveCanisterUser, |
| 58 | getScopeForUserVaultFromScopeMap, |
| 59 | resolveAllowedVaultIdsForHostedContext, |
| 60 | } from '../lib/hosted-workspace-resolve.mjs'; |
| 61 | import { applyScopeFilterToNotes, applyScopeFilterToProposals } from '../lib/scope-filter.mjs'; |
| 62 | import { actorMayApproveProposals } from '../lib/hub-evaluator-may-approve.mjs'; |
| 63 | import { |
| 64 | buildCalendarTimeline, |
| 65 | listSourceCalendarsForClient, |
| 66 | } from '../../lib/calendar/timeline.mjs'; |
| 67 | import { importIcsIntoVault } from '../../lib/calendar/event-store.mjs'; |
| 68 | import { patchSourceCalendar, parseSourceCalendarPatchBody } from '../../lib/calendar/source-calendar-patch.mjs'; |
| 69 | import { retrieveAgentCalendarContext } from '../../lib/calendar/agent-retrieval.mjs'; |
| 70 | import { materializeListFrontmatter } from '../gateway/note-facets.mjs'; |
| 71 | import { registerBridgeDelegationRoutes } from './delegation-routes.mjs'; |
| 72 | |
| 73 | // When Netlify bundles as CJS, import.meta.url is empty; avoid it in serverless so the app loads and routes register. |
| 74 | const inServerless = Boolean(process.env.AWS_LAMBDA_FUNCTION_NAME || process.env.NETLIFY); |
| 75 | let projectRoot; |
| 76 | if (inServerless) { |
| 77 | projectRoot = process.cwd(); |
| 78 | } else { |
| 79 | projectRoot = path.resolve(path.dirname(fileURLToPath(import.meta.url)), '..', '..'); |
| 80 | } |
| 81 | const __dirname = path.join(projectRoot, 'hub', 'bridge'); |
| 82 | const envPath = path.join(projectRoot, '.env'); |
| 83 | if (fs.existsSync(envPath)) dotenv.config({ path: envPath }); |
| 84 | |
| 85 | const PORT = parseInt(process.env.BRIDGE_PORT || process.env.PORT || '3341', 10); |
| 86 | const BASE_URL = (process.env.HUB_BASE_URL || `http://localhost:${PORT}`).replace(/\/$/, ''); |
| 87 | const CANISTER_URL = (process.env.CANISTER_URL || '').replace(/\/$/, ''); |
| 88 | const HUB_UI_ORIGIN = (process.env.HUB_UI_ORIGIN || BASE_URL).replace(/\/$/, ''); |
| 89 | // Path under HUB_UI_ORIGIN where the Hub app lives (e.g. /hub). Empty string = root. |
| 90 | const HUB_UI_PATH = (process.env.HUB_UI_PATH || '/hub').replace(/\/$/, ''); |
| 91 | const SESSION_SECRET = process.env.SESSION_SECRET || process.env.HUB_JWT_SECRET; |
| 92 | const CANISTER_AUTH_SECRET = process.env.CANISTER_AUTH_SECRET || ''; |
| 93 | const HOSTED_CONTEXT_FETCH_TIMEOUT_MS = (() => { |
| 94 | const n = parseInt(String(process.env.HOSTED_CONTEXT_FETCH_TIMEOUT_MS || ''), 10); |
| 95 | if (!Number.isFinite(n)) return 3000; |
| 96 | return Math.min(10_000, Math.max(250, n)); |
| 97 | })(); |
| 98 | const HOSTED_CONTEXT_CACHE_TTL_MS = 60_000; |
| 99 | const canisterVaultIdsCache = new Map(); |
| 100 | |
| 101 | function hostedContextAbortSignal() { |
| 102 | return typeof AbortSignal !== 'undefined' && typeof AbortSignal.timeout === 'function' |
| 103 | ? AbortSignal.timeout(HOSTED_CONTEXT_FETCH_TIMEOUT_MS) |
| 104 | : undefined; |
| 105 | } |
| 106 | |
| 107 | /** |
| 108 | * Base headers for all bridge→canister requests. |
| 109 | * Includes x-gateway-auth when CANISTER_AUTH_SECRET is configured so the |
| 110 | * canister's gatewayAuthorized() check (Phase 0) passes. |
| 111 | * Uses the same env var name as the gateway (CANISTER_AUTH_SECRET). |
| 112 | */ |
| 113 | function canisterHeaders(extra = {}) { |
| 114 | const h = { Accept: 'application/json', ...extra }; |
| 115 | if (CANISTER_AUTH_SECRET) h['x-gateway-auth'] = CANISTER_AUTH_SECRET; |
| 116 | return h; |
| 117 | } |
| 118 | // On Netlify Lambda /var/task/ is read-only; only /tmp is writable. |
| 119 | // Use /tmp/knowtation-bridge-data when serverless and DATA_DIR is not explicitly set. |
| 120 | const DATA_DIR = process.env.DATA_DIR |
| 121 | ? (path.isAbsolute(process.env.DATA_DIR) ? process.env.DATA_DIR : path.join(projectRoot, process.env.DATA_DIR)) |
| 122 | : (inServerless ? path.join(os.tmpdir(), 'knowtation-bridge-data') : path.join(projectRoot, 'data')); |
| 123 | const TOKENS_FILE = path.join(DATA_DIR, 'hub_github_tokens.json'); |
| 124 | const ROLES_FILE = path.join(DATA_DIR, 'hub_roles.json'); |
| 125 | const INVITES_FILE = path.join(DATA_DIR, 'hub_invites.json'); |
| 126 | const WORKSPACE_FILE = path.join(DATA_DIR, 'hub_workspace.json'); |
| 127 | const VAULT_ACCESS_FILE = path.join(DATA_DIR, 'hub_vault_access.json'); |
| 128 | const SCOPE_FILE = path.join(DATA_DIR, 'hub_scope.json'); |
| 129 | const EVALUATOR_MAY_APPROVE_FILE = path.join(DATA_DIR, 'hub_evaluator_may_approve.json'); |
| 130 | const VALID_ROLES = new Set(['admin', 'editor', 'viewer', 'evaluator']); |
| 131 | const INVITE_EXPIRY_MS = 7 * 24 * 60 * 60 * 1000; |
| 132 | |
| 133 | const adminUserIdsSet = new Set( |
| 134 | (process.env.HUB_ADMIN_USER_IDS || '') |
| 135 | .split(',') |
| 136 | .map((s) => s.trim()) |
| 137 | .filter(Boolean) |
| 138 | ); |
| 139 | |
| 140 | function sanitizeUserId(uid) { |
| 141 | return String(uid).replace(/[^a-zA-Z0-9_-]/g, '_').slice(0, 128) || 'default'; |
| 142 | } |
| 143 | |
| 144 | function sanitizeVaultId(vaultId) { |
| 145 | return String(vaultId || 'default').replace(/[^a-zA-Z0-9_-]/g, '_').slice(0, 64) || 'default'; |
| 146 | } |
| 147 | |
| 148 | let warnedOllamaLocalhostOnNetlify = false; |
| 149 | |
| 150 | /** Trim + default empty env so accidental whitespace does not break provider matching or Ollama URL. */ |
| 151 | function getBridgeEmbeddingConfig() { |
| 152 | const pEnv = process.env.EMBEDDING_PROVIDER; |
| 153 | const provider = ( |
| 154 | pEnv == null || String(pEnv).trim() === '' ? 'ollama' : String(pEnv).trim() |
| 155 | ).toLowerCase(); |
| 156 | const mEnv = process.env.EMBEDDING_MODEL; |
| 157 | const model = |
| 158 | mEnv == null || String(mEnv).trim() === '' |
| 159 | ? defaultBridgeEmbeddingModelForProvider(provider) |
| 160 | : String(mEnv).trim(); |
| 161 | const oEnv = process.env.OLLAMA_URL; |
| 162 | const ollama_url = |
| 163 | oEnv == null || String(oEnv).trim() === '' ? 'http://localhost:11434' : String(oEnv).trim(); |
| 164 | if (inServerless && provider === 'ollama' && !warnedOllamaLocalhostOnNetlify) { |
| 165 | warnedOllamaLocalhostOnNetlify = true; |
| 166 | const t = String(ollama_url).trim() || 'http://localhost:11434'; |
| 167 | try { |
| 168 | if (/^https?:\/\//i.test(t)) { |
| 169 | const u = new URL(t); |
| 170 | if (u.hostname === 'localhost' || u.hostname === '127.0.0.1') { |
| 171 | console.warn( |
| 172 | '[bridge] EMBEDDING_PROVIDER=ollama with localhost OLLAMA_URL cannot reach your machine from Netlify. ' + |
| 173 | 'Set EMBEDDING_PROVIDER=openai and OPENAI_API_KEY, or OLLAMA_URL to a public https:// Ollama API base.', |
| 174 | ); |
| 175 | } |
| 176 | } |
| 177 | } catch (_) { |
| 178 | /* embed path will throw a clearer error via normalizeOllamaEmbedBaseUrl */ |
| 179 | } |
| 180 | } |
| 181 | return { |
| 182 | provider, |
| 183 | model, |
| 184 | ollama_url, |
| 185 | }; |
| 186 | } |
| 187 | |
| 188 | /** |
| 189 | * Undici/fetch often throws TypeError with message "Invalid URL" only — map to actionable text for operators. |
| 190 | * @param {unknown} err |
| 191 | * @param {'index'|'search'|'embed'} kind |
| 192 | */ |
| 193 | function bridgeEmbedFailureMessage(err, kind) { |
| 194 | const raw = err && typeof err.message === 'string' ? err.message : String(err); |
| 195 | if (raw !== 'Invalid URL' && !raw.includes('Invalid URL')) return raw; |
| 196 | const c = getBridgeEmbeddingConfig(); |
| 197 | const hasOpenAiKey = Boolean( |
| 198 | process.env.OPENAI_API_KEY && String(process.env.OPENAI_API_KEY).trim(), |
| 199 | ); |
| 200 | const hasVoyageKey = Boolean(process.env.VOYAGE_API_KEY && String(process.env.VOYAGE_API_KEY).trim()); |
| 201 | return ( |
| 202 | `${raw} (${kind}). On Netlify, Invalid URL often means sqlite-vec was esbuild-bundled ` + |
| 203 | '(stack: getLoadablePath / input ".") — set [functions].external_node_modules for sqlite-vec and better-sqlite3 in netlify.toml. ' + |
| 204 | `Resolved EMBEDDING_PROVIDER="${c.provider}"; OPENAI_API_KEY ${hasOpenAiKey ? 'is set' : 'is missing'}; ` + |
| 205 | `VOYAGE_API_KEY ${hasVoyageKey ? 'is set' : 'is missing'}. ` + |
| 206 | 'If provider is ollama, OLLAMA_URL must be a full http(s) URL. Remove bad HTTP_PROXY/HTTPS_PROXY if set. ' + |
| 207 | 'See hub/bridge/README.md (semantic index/search).' |
| 208 | ); |
| 209 | } |
| 210 | |
| 211 | const DB_FILENAME = 'knowtation_vectors.db'; |
| 212 | |
| 213 | function getBridgeStoreConfig(uid, vectorsDirOverride) { |
| 214 | const vectorsDir = vectorsDirOverride ?? (() => { |
| 215 | const d = path.join(DATA_DIR, 'vectors', sanitizeUserId(uid)); |
| 216 | if (!fs.existsSync(d)) fs.mkdirSync(d, { recursive: true }); |
| 217 | return d; |
| 218 | })(); |
| 219 | return { |
| 220 | vector_store: 'sqlite-vec', |
| 221 | data_dir: vectorsDir, |
| 222 | embedding: getBridgeEmbeddingConfig(), |
| 223 | // Bridge owns the data lifecycle (downloads from blob → re-indexes → uploads to blob). |
| 224 | // A dimension change (e.g. OpenAI 1536 → DeepInfra 1024) can only resolve via a full |
| 225 | // re-embed of every vault in this DB. CLI keeps the throw so an accidental swap surfaces |
| 226 | // loudly. See `lib/vector-store-sqlite.mjs ensureCollection` migration logic. |
| 227 | allow_dimension_migration: true, |
| 228 | }; |
| 229 | } |
| 230 | |
| 231 | const isServerless = Boolean(process.env.AWS_LAMBDA_FUNCTION_NAME || process.env.NETLIFY); |
| 232 | |
| 233 | function ensureDataDir() { |
| 234 | if (!fs.existsSync(DATA_DIR)) fs.mkdirSync(DATA_DIR, { recursive: true }); |
| 235 | } |
| 236 | |
| 237 | const ALGO = 'aes-256-gcm'; |
| 238 | const IV_LEN = 16; |
| 239 | const TAG_LEN = 16; |
| 240 | const SALT_LEN = 16; |
| 241 | // Ciphertext format (v2): saltB64url.ivB64url.tagB64url.encB64url (4 parts) |
| 242 | // Legacy format (v1): ivB64url.tagB64url.encB64url (3 parts — decrypt will return null → graceful reconnect) |
| 243 | function encrypt(text, secret) { |
| 244 | const salt = crypto.randomBytes(SALT_LEN); |
| 245 | const key = crypto.scryptSync(secret, salt, 32); |
| 246 | const iv = crypto.randomBytes(IV_LEN); |
| 247 | const cipher = crypto.createCipheriv(ALGO, key, iv); |
| 248 | const enc = Buffer.concat([cipher.update(text, 'utf8'), cipher.final()]); |
| 249 | const tag = cipher.getAuthTag(); |
| 250 | return ( |
| 251 | salt.toString('base64url') + '.' + |
| 252 | iv.toString('base64url') + '.' + |
| 253 | tag.toString('base64url') + '.' + |
| 254 | enc.toString('base64url') |
| 255 | ); |
| 256 | } |
| 257 | function decrypt(encrypted, secret) { |
| 258 | const parts = encrypted.split('.'); |
| 259 | // v1 ciphertexts had 3 parts (hardcoded salt); treat as not-found so the |
| 260 | // caller falls through to "prompt reconnect" without crashing. |
| 261 | if (parts.length !== 4) return null; |
| 262 | const [saltB, ivB, tagB, encB] = parts; |
| 263 | if (!saltB || !ivB || !tagB || !encB) return null; |
| 264 | try { |
| 265 | const key = crypto.scryptSync(secret, Buffer.from(saltB, 'base64url'), 32); |
| 266 | const decipher = crypto.createDecipheriv(ALGO, key, Buffer.from(ivB, 'base64url')); |
| 267 | decipher.setAuthTag(Buffer.from(tagB, 'base64url')); |
| 268 | return decipher.update(Buffer.from(encB, 'base64url')) + decipher.final('utf8'); |
| 269 | } catch { |
| 270 | return null; |
| 271 | } |
| 272 | } |
| 273 | |
| 274 | function parseAndDecryptTokens(raw) { |
| 275 | if (!raw || typeof raw !== 'object' || Array.isArray(raw)) return {}; |
| 276 | const out = {}; |
| 277 | let decryptFailures = 0; |
| 278 | for (const [uid, v] of Object.entries(raw)) { |
| 279 | if (v && typeof v.token === 'string') { |
| 280 | const t = decrypt(v.token, SESSION_SECRET); |
| 281 | if (t) out[uid] = { token: t, repo: v.repo || null }; |
| 282 | else decryptFailures++; |
| 283 | } |
| 284 | } |
| 285 | if (decryptFailures > 0) { |
| 286 | console.warn( |
| 287 | '[bridge] loadTokens: decrypt failed for', |
| 288 | decryptFailures, |
| 289 | 'stored GitHub token(s). If SESSION_SECRET was rotated on the bridge, run Connect GitHub again to re-store the token.' |
| 290 | ); |
| 291 | } |
| 292 | return out; |
| 293 | } |
| 294 | |
| 295 | async function loadTokens(blobStore) { |
| 296 | if (!blobStore) { |
| 297 | ensureDataDir(); |
| 298 | if (!fs.existsSync(TOKENS_FILE)) return {}; |
| 299 | try { |
| 300 | const raw = JSON.parse(fs.readFileSync(TOKENS_FILE, 'utf8')); |
| 301 | return parseAndDecryptTokens(raw); |
| 302 | } catch (_) { |
| 303 | return {}; |
| 304 | } |
| 305 | } |
| 306 | try { |
| 307 | const rawStr = await blobStore.get('hub_github_tokens'); |
| 308 | if (!rawStr) return {}; |
| 309 | const raw = JSON.parse(rawStr); |
| 310 | return parseAndDecryptTokens(raw); |
| 311 | } catch (_) { |
| 312 | return {}; |
| 313 | } |
| 314 | } |
| 315 | |
| 316 | async function saveTokens(blobStore, tokens) { |
| 317 | const toWrite = {}; |
| 318 | for (const [uid, v] of Object.entries(tokens)) { |
| 319 | toWrite[uid] = { token: encrypt(v.token, SESSION_SECRET), repo: v.repo || null }; |
| 320 | } |
| 321 | const str = JSON.stringify(toWrite, null, 2); |
| 322 | if (!blobStore) { |
| 323 | ensureDataDir(); |
| 324 | fs.writeFileSync(TOKENS_FILE, str, 'utf8'); |
| 325 | return; |
| 326 | } |
| 327 | await blobStore.set('hub_github_tokens', str); |
| 328 | } |
| 329 | |
| 330 | // ——— Roles & invites (hosted parity: same contract as self-hosted hub/roles.mjs, hub/invites.mjs) ——— |
| 331 | async function loadRoles(blobStore) { |
| 332 | if (!blobStore) { |
| 333 | ensureDataDir(); |
| 334 | if (!fs.existsSync(ROLES_FILE)) return {}; |
| 335 | try { |
| 336 | const data = JSON.parse(fs.readFileSync(ROLES_FILE, 'utf8')); |
| 337 | const roles = data.roles != null ? data.roles : data; |
| 338 | return typeof roles === 'object' && roles !== null ? roles : {}; |
| 339 | } catch (_) { |
| 340 | return {}; |
| 341 | } |
| 342 | } |
| 343 | try { |
| 344 | const rawStr = await blobStore.get('hub_roles'); |
| 345 | if (!rawStr) return {}; |
| 346 | const data = JSON.parse(rawStr); |
| 347 | const roles = data.roles != null ? data.roles : data; |
| 348 | return typeof roles === 'object' && roles !== null ? roles : {}; |
| 349 | } catch (_) { |
| 350 | return {}; |
| 351 | } |
| 352 | } |
| 353 | |
| 354 | async function saveRoles(blobStore, roles) { |
| 355 | const obj = {}; |
| 356 | for (const [sub, role] of Object.entries(roles)) { |
| 357 | if (typeof sub === 'string' && sub.trim() && VALID_ROLES.has(role)) obj[sub.trim()] = role; |
| 358 | } |
| 359 | const str = JSON.stringify({ roles: obj }, null, 2); |
| 360 | if (!blobStore) { |
| 361 | ensureDataDir(); |
| 362 | fs.writeFileSync(ROLES_FILE, str, 'utf8'); |
| 363 | return; |
| 364 | } |
| 365 | await blobStore.set('hub_roles', str); |
| 366 | } |
| 367 | |
| 368 | function bridgeEnvEvaluatorMayApprove() { |
| 369 | return process.env.HUB_EVALUATOR_MAY_APPROVE === '1'; |
| 370 | } |
| 371 | |
| 372 | async function loadEvaluatorMayApproveMap(blobStore) { |
| 373 | if (!blobStore) { |
| 374 | ensureDataDir(); |
| 375 | if (!fs.existsSync(EVALUATOR_MAY_APPROVE_FILE)) return {}; |
| 376 | try { |
| 377 | const data = JSON.parse(fs.readFileSync(EVALUATOR_MAY_APPROVE_FILE, 'utf8')); |
| 378 | const m = data?.evaluator_may_approve != null ? data.evaluator_may_approve : data; |
| 379 | if (typeof m !== 'object' || m === null) return {}; |
| 380 | const out = {}; |
| 381 | for (const [k, v] of Object.entries(m)) { |
| 382 | if (typeof k === 'string' && k.trim()) out[k.trim()] = Boolean(v); |
| 383 | } |
| 384 | return out; |
| 385 | } catch (_) { |
| 386 | return {}; |
| 387 | } |
| 388 | } |
| 389 | try { |
| 390 | const rawStr = await blobStore.get('hub_evaluator_may_approve'); |
| 391 | if (!rawStr) return {}; |
| 392 | const data = JSON.parse(rawStr); |
| 393 | const m = data?.evaluator_may_approve != null ? data.evaluator_may_approve : data; |
| 394 | if (typeof m !== 'object' || m === null) return {}; |
| 395 | const out = {}; |
| 396 | for (const [k, v] of Object.entries(m)) { |
| 397 | if (typeof k === 'string' && k.trim()) out[k.trim()] = Boolean(v); |
| 398 | } |
| 399 | return out; |
| 400 | } catch (_) { |
| 401 | return {}; |
| 402 | } |
| 403 | } |
| 404 | |
| 405 | async function saveEvaluatorMayApproveMap(blobStore, map) { |
| 406 | const obj = {}; |
| 407 | for (const [k, v] of Object.entries(map)) { |
| 408 | if (typeof k === 'string' && k.trim()) obj[k.trim()] = Boolean(v); |
| 409 | } |
| 410 | const str = JSON.stringify({ evaluator_may_approve: obj }, null, 2); |
| 411 | if (!blobStore) { |
| 412 | ensureDataDir(); |
| 413 | fs.writeFileSync(EVALUATOR_MAY_APPROVE_FILE, str, 'utf8'); |
| 414 | return; |
| 415 | } |
| 416 | await blobStore.set('hub_evaluator_may_approve', str); |
| 417 | } |
| 418 | |
| 419 | /** Effective “may approve proposals” for Hub UI and gateway (admin always; evaluator from map + env). */ |
| 420 | function mayApproveProposalsForUser(uid, storedRoles, mayMap) { |
| 421 | const role = effectiveRole(uid, storedRoles); |
| 422 | return actorMayApproveProposals(uid, role, mayMap, bridgeEnvEvaluatorMayApprove()); |
| 423 | } |
| 424 | |
| 425 | async function loadInvites(blobStore) { |
| 426 | if (!blobStore) { |
| 427 | ensureDataDir(); |
| 428 | if (!fs.existsSync(INVITES_FILE)) return {}; |
| 429 | try { |
| 430 | const data = JSON.parse(fs.readFileSync(INVITES_FILE, 'utf8')); |
| 431 | const invites = data.invites && typeof data.invites === 'object' ? data.invites : {}; |
| 432 | return invites; |
| 433 | } catch (_) { |
| 434 | return {}; |
| 435 | } |
| 436 | } |
| 437 | try { |
| 438 | const rawStr = await blobStore.get('hub_invites'); |
| 439 | if (!rawStr) return {}; |
| 440 | const data = JSON.parse(rawStr); |
| 441 | const invites = data.invites && typeof data.invites === 'object' ? data.invites : {}; |
| 442 | return invites; |
| 443 | } catch (_) { |
| 444 | return {}; |
| 445 | } |
| 446 | } |
| 447 | |
| 448 | async function saveInvites(blobStore, invites) { |
| 449 | const obj = {}; |
| 450 | for (const [token, entry] of Object.entries(invites)) { |
| 451 | if (typeof token === 'string' && token && entry && typeof entry.role === 'string' && typeof entry.created_at === 'string') { |
| 452 | obj[token] = { role: entry.role, created_at: entry.created_at }; |
| 453 | } |
| 454 | } |
| 455 | const str = JSON.stringify({ invites: obj }, null, 2); |
| 456 | if (!blobStore) { |
| 457 | ensureDataDir(); |
| 458 | fs.writeFileSync(INVITES_FILE, str, 'utf8'); |
| 459 | return; |
| 460 | } |
| 461 | await blobStore.set('hub_invites', str); |
| 462 | } |
| 463 | |
| 464 | async function loadWorkspace(blobStore) { |
| 465 | if (!blobStore) { |
| 466 | ensureDataDir(); |
| 467 | if (!fs.existsSync(WORKSPACE_FILE)) return { owner_user_id: null }; |
| 468 | try { |
| 469 | const data = JSON.parse(fs.readFileSync(WORKSPACE_FILE, 'utf8')); |
| 470 | const id = data?.owner_user_id; |
| 471 | return { owner_user_id: typeof id === 'string' && id.trim() ? id.trim() : null }; |
| 472 | } catch (_) { |
| 473 | return { owner_user_id: null }; |
| 474 | } |
| 475 | } |
| 476 | try { |
| 477 | const rawStr = await blobStore.get('hub_workspace'); |
| 478 | if (!rawStr) return { owner_user_id: null }; |
| 479 | const data = JSON.parse(rawStr); |
| 480 | const id = data?.owner_user_id; |
| 481 | return { owner_user_id: typeof id === 'string' && id.trim() ? id.trim() : null }; |
| 482 | } catch (_) { |
| 483 | return { owner_user_id: null }; |
| 484 | } |
| 485 | } |
| 486 | |
| 487 | async function saveWorkspace(blobStore, ownerUserId) { |
| 488 | const payload = JSON.stringify( |
| 489 | { owner_user_id: ownerUserId && String(ownerUserId).trim() ? String(ownerUserId).trim() : null }, |
| 490 | null, |
| 491 | 2, |
| 492 | ); |
| 493 | if (!blobStore) { |
| 494 | ensureDataDir(); |
| 495 | fs.writeFileSync(WORKSPACE_FILE, payload, 'utf8'); |
| 496 | return; |
| 497 | } |
| 498 | await blobStore.set('hub_workspace', payload); |
| 499 | } |
| 500 | |
| 501 | async function loadVaultAccess(blobStore) { |
| 502 | if (!blobStore) { |
| 503 | ensureDataDir(); |
| 504 | if (!fs.existsSync(VAULT_ACCESS_FILE)) return {}; |
| 505 | try { |
| 506 | const data = JSON.parse(fs.readFileSync(VAULT_ACCESS_FILE, 'utf8')); |
| 507 | const out = {}; |
| 508 | if (data && typeof data === 'object') { |
| 509 | for (const [uid, arr] of Object.entries(data)) { |
| 510 | if (typeof uid === 'string' && uid.trim() && Array.isArray(arr)) { |
| 511 | out[uid.trim()] = arr.filter((v) => typeof v === 'string' && v.trim()).map((v) => v.trim()); |
| 512 | } |
| 513 | } |
| 514 | } |
| 515 | return out; |
| 516 | } catch (_) { |
| 517 | return {}; |
| 518 | } |
| 519 | } |
| 520 | try { |
| 521 | const rawStr = await blobStore.get('hub_vault_access'); |
| 522 | if (!rawStr) return {}; |
| 523 | const data = JSON.parse(rawStr); |
| 524 | const out = {}; |
| 525 | if (data && typeof data === 'object') { |
| 526 | for (const [uid, arr] of Object.entries(data)) { |
| 527 | if (typeof uid === 'string' && uid.trim() && Array.isArray(arr)) { |
| 528 | out[uid.trim()] = arr.filter((v) => typeof v === 'string' && v.trim()).map((v) => v.trim()); |
| 529 | } |
| 530 | } |
| 531 | } |
| 532 | return out; |
| 533 | } catch (_) { |
| 534 | return {}; |
| 535 | } |
| 536 | } |
| 537 | |
| 538 | async function saveVaultAccess(blobStore, access) { |
| 539 | const obj = {}; |
| 540 | for (const [uid, arr] of Object.entries(access || {})) { |
| 541 | if (typeof uid === 'string' && uid.trim() && Array.isArray(arr)) { |
| 542 | obj[uid.trim()] = arr.filter((v) => typeof v === 'string' && v.trim()).map((v) => v.trim()); |
| 543 | } |
| 544 | } |
| 545 | const str = JSON.stringify(obj, null, 2); |
| 546 | if (!blobStore) { |
| 547 | ensureDataDir(); |
| 548 | fs.writeFileSync(VAULT_ACCESS_FILE, str, 'utf8'); |
| 549 | return; |
| 550 | } |
| 551 | await blobStore.set('hub_vault_access', str); |
| 552 | } |
| 553 | |
| 554 | async function loadScope(blobStore) { |
| 555 | if (!blobStore) { |
| 556 | ensureDataDir(); |
| 557 | if (!fs.existsSync(SCOPE_FILE)) return {}; |
| 558 | try { |
| 559 | const data = JSON.parse(fs.readFileSync(SCOPE_FILE, 'utf8')); |
| 560 | return data && typeof data === 'object' ? data : {}; |
| 561 | } catch (_) { |
| 562 | return {}; |
| 563 | } |
| 564 | } |
| 565 | try { |
| 566 | const rawStr = await blobStore.get('hub_scope'); |
| 567 | if (!rawStr) return {}; |
| 568 | const data = JSON.parse(rawStr); |
| 569 | return data && typeof data === 'object' ? data : {}; |
| 570 | } catch (_) { |
| 571 | return {}; |
| 572 | } |
| 573 | } |
| 574 | |
| 575 | async function saveScope(blobStore, scope) { |
| 576 | const cleaned = {}; |
| 577 | for (const [uid, vaultMap] of Object.entries(scope || {})) { |
| 578 | if (typeof uid !== 'string' || !uid.trim() || !vaultMap || typeof vaultMap !== 'object') continue; |
| 579 | cleaned[uid.trim()] = {}; |
| 580 | for (const [vaultId, rules] of Object.entries(vaultMap)) { |
| 581 | if (typeof vaultId !== 'string' || !vaultId.trim() || !rules || typeof rules !== 'object') continue; |
| 582 | const projects = Array.isArray(rules.projects) |
| 583 | ? rules.projects.filter((p) => typeof p === 'string' && p.trim()).map((p) => p.trim()) |
| 584 | : []; |
| 585 | const folders = Array.isArray(rules.folders) |
| 586 | ? rules.folders.filter((f) => typeof f === 'string' && f.trim()).map((f) => f.trim()) |
| 587 | : []; |
| 588 | if (projects.length > 0 || folders.length > 0) { |
| 589 | cleaned[uid.trim()][vaultId.trim()] = { projects, folders }; |
| 590 | } |
| 591 | } |
| 592 | } |
| 593 | const str = JSON.stringify(cleaned, null, 2); |
| 594 | if (!blobStore) { |
| 595 | ensureDataDir(); |
| 596 | fs.writeFileSync(SCOPE_FILE, str, 'utf8'); |
| 597 | return; |
| 598 | } |
| 599 | await blobStore.set('hub_scope', str); |
| 600 | } |
| 601 | |
| 602 | /** Remove vault id from all hub_vault_access lists and hub_scope maps (hosted team). */ |
| 603 | async function stripHostedVaultFromAccessAndScope(blobStore, vaultId) { |
| 604 | const id = String(vaultId || '').trim(); |
| 605 | if (!id || id === 'default') return; |
| 606 | const access = await loadVaultAccess(blobStore); |
| 607 | const nextAccess = {}; |
| 608 | for (const [uid, arr] of Object.entries(access)) { |
| 609 | if (!Array.isArray(arr)) continue; |
| 610 | const filtered = arr.filter((x) => String(x).trim() !== id); |
| 611 | if (filtered.length > 0) nextAccess[uid] = filtered; |
| 612 | } |
| 613 | await saveVaultAccess(blobStore, nextAccess); |
| 614 | |
| 615 | const scope = await loadScope(blobStore); |
| 616 | const nextScope = {}; |
| 617 | for (const [uid, vmap] of Object.entries(scope)) { |
| 618 | if (!vmap || typeof vmap !== 'object') continue; |
| 619 | const inner = {}; |
| 620 | for (const [vid, rules] of Object.entries(vmap)) { |
| 621 | if (String(vid).trim() === id) continue; |
| 622 | inner[vid] = rules; |
| 623 | } |
| 624 | if (Object.keys(inner).length > 0) nextScope[uid] = inner; |
| 625 | } |
| 626 | await saveScope(blobStore, nextScope); |
| 627 | } |
| 628 | |
| 629 | /** Drop bridge vector store for (effective user, vault). */ |
| 630 | async function removeHostedVectorBlobForVault(blobStore, effectiveUid, vaultId) { |
| 631 | const safeUid = sanitizeUserId(effectiveUid); |
| 632 | const vid = sanitizeVaultId(vaultId); |
| 633 | const localDir = path.join(DATA_DIR, 'vectors', safeUid, vid); |
| 634 | if (!blobStore) { |
| 635 | if (fs.existsSync(localDir)) fs.rmSync(localDir, { recursive: true, force: true }); |
| 636 | return; |
| 637 | } |
| 638 | const key = 'vectors/' + safeUid + '/' + vid; |
| 639 | try { |
| 640 | if (typeof blobStore.delete === 'function') await blobStore.delete(key); |
| 641 | } catch (_) { |
| 642 | /* Netlify Blobs may omit delete; ignore */ |
| 643 | } |
| 644 | } |
| 645 | |
| 646 | /** @returns {Promise<string[]>} */ |
| 647 | async function fetchCanisterVaultIdsForUser(canisterUserId) { |
| 648 | if (!CANISTER_URL || !canisterUserId) return ['default']; |
| 649 | const cacheKey = String(canisterUserId); |
| 650 | const now = Date.now(); |
| 651 | const hit = canisterVaultIdsCache.get(cacheKey); |
| 652 | if (hit && hit.expires > now) return [...hit.ids]; |
| 653 | try { |
| 654 | const signal = hostedContextAbortSignal(); |
| 655 | const vRes = await fetch(CANISTER_URL + '/api/v1/vaults', { |
| 656 | method: 'GET', |
| 657 | headers: canisterHeaders({ 'X-User-Id': canisterUserId }), |
| 658 | ...(signal ? { signal } : {}), |
| 659 | }); |
| 660 | if (!vRes.ok) return ['default']; |
| 661 | const data = await vRes.json(); |
| 662 | const vaults = Array.isArray(data.vaults) ? data.vaults : []; |
| 663 | if (vaults.length === 0) return ['default']; |
| 664 | const ids = vaults.map((v) => String(v.id || 'default')).filter(Boolean); |
| 665 | canisterVaultIdsCache.set(cacheKey, { expires: now + HOSTED_CONTEXT_CACHE_TTL_MS, ids }); |
| 666 | return ids; |
| 667 | } catch (_) { |
| 668 | return ['default']; |
| 669 | } |
| 670 | } |
| 671 | |
| 672 | function explicitVaultAccessForUser(accessMap, actorUid) { |
| 673 | const raw = accessMap && typeof accessMap === 'object' ? accessMap[actorUid] : null; |
| 674 | if (!Array.isArray(raw) || raw.length === 0) return null; |
| 675 | const out = raw.map((x) => String(x).trim()).filter(Boolean); |
| 676 | return out.length > 0 ? out : null; |
| 677 | } |
| 678 | |
| 679 | /** |
| 680 | * @param {import('express').Request} req |
| 681 | * @param {string} actorUid |
| 682 | * @returns {Promise<{ ok: true, effectiveCanisterUid: string, actorUid: string, vaultId: string, scope: { projects: string[], folders: string[] } | null, allowedVaultIds: string[], delegating: boolean } | { ok: false, status: number, code: string, error: string }>} |
| 683 | */ |
| 684 | async function resolveHostedBridgeContext(req, actorUid) { |
| 685 | const vaultId = sanitizeVaultId(req.headers['x-vault-id']); |
| 686 | const workspace = await loadWorkspace(req.blobStore); |
| 687 | const roles = await loadRoles(req.blobStore); |
| 688 | const access = await loadVaultAccess(req.blobStore); |
| 689 | const scopeMap = await loadScope(req.blobStore); |
| 690 | const ownerId = workspace.owner_user_id; |
| 691 | const { effective, delegate } = resolveEffectiveCanisterUser({ |
| 692 | actorSub: actorUid, |
| 693 | workspaceOwnerId: ownerId, |
| 694 | storedRoles: roles, |
| 695 | adminUserIdsSet, |
| 696 | }); |
| 697 | const explicitVaultIds = explicitVaultAccessForUser(access, actorUid); |
| 698 | const canisterIds = |
| 699 | delegate && explicitVaultIds ? explicitVaultIds : await fetchCanisterVaultIdsForUser(effective); |
| 700 | const allowedVaultIds = resolveAllowedVaultIdsForHostedContext({ |
| 701 | delegate, |
| 702 | actorUid, |
| 703 | accessMap: access, |
| 704 | canisterIds, |
| 705 | }); |
| 706 | if (!allowedVaultIds.includes(vaultId)) { |
| 707 | return { |
| 708 | ok: false, |
| 709 | status: 403, |
| 710 | code: 'FORBIDDEN', |
| 711 | error: 'Access to this vault is not allowed.', |
| 712 | }; |
| 713 | } |
| 714 | let scope = getScopeForUserVaultFromScopeMap(scopeMap, actorUid, vaultId); |
| 715 | // Evaluators must see the full vault (per allowed_vault_ids) to review proposals in context; |
| 716 | // project/folder scope still applies to viewer/editor/admin delegating members. |
| 717 | const actorRole = effectiveRole(actorUid, roles); |
| 718 | if (actorRole === 'evaluator') { |
| 719 | scope = null; |
| 720 | } |
| 721 | return { |
| 722 | ok: true, |
| 723 | effectiveCanisterUid: effective, |
| 724 | actorUid, |
| 725 | vaultId, |
| 726 | scope, |
| 727 | allowedVaultIds, |
| 728 | delegating: delegate, |
| 729 | }; |
| 730 | } |
| 731 | |
| 732 | /** |
| 733 | * Hosted settings need the actor's vault allowlist without first proving access |
| 734 | * to a specific vault. This keeps Business-only delegated users from being |
| 735 | * denied while the UI is still deciding which vault to select. |
| 736 | * |
| 737 | * @param {import('express').Request} req |
| 738 | * @param {string} actorUid |
| 739 | */ |
| 740 | async function resolveHostedBridgeSettingsContext(req, actorUid) { |
| 741 | const workspace = await loadWorkspace(req.blobStore); |
| 742 | const roles = await loadRoles(req.blobStore); |
| 743 | const access = await loadVaultAccess(req.blobStore); |
| 744 | const ownerId = workspace.owner_user_id; |
| 745 | const { effective, delegate } = resolveEffectiveCanisterUser({ |
| 746 | actorSub: actorUid, |
| 747 | workspaceOwnerId: ownerId, |
| 748 | storedRoles: roles, |
| 749 | adminUserIdsSet, |
| 750 | }); |
| 751 | const explicitVaultIds = explicitVaultAccessForUser(access, actorUid); |
| 752 | const canisterIds = |
| 753 | delegate && explicitVaultIds ? explicitVaultIds : await fetchCanisterVaultIdsForUser(effective); |
| 754 | const allowedVaultIds = resolveAllowedVaultIdsForHostedContext({ |
| 755 | delegate, |
| 756 | actorUid, |
| 757 | accessMap: access, |
| 758 | canisterIds, |
| 759 | }); |
| 760 | return { |
| 761 | effectiveCanisterUid: effective, |
| 762 | actorUid, |
| 763 | allowedVaultIds, |
| 764 | delegating: delegate, |
| 765 | workspaceOwnerId: ownerId, |
| 766 | role: effectiveRole(actorUid, roles), |
| 767 | }; |
| 768 | } |
| 769 | |
| 770 | function effectiveRole(uid, storedRoles) { |
| 771 | if (!uid) return 'member'; |
| 772 | const stored = storedRoles && storedRoles[uid]; |
| 773 | if (stored && VALID_ROLES.has(stored)) return stored; |
| 774 | return adminUserIdsSet.has(uid) ? 'admin' : 'member'; |
| 775 | } |
| 776 | |
| 777 | /** Return a directory path that contains (or will contain) knowtation_vectors.db for this user and vault. Rehydrates from Blob if needed. Phase 15: keyed by (uid, vault_id). */ |
| 778 | async function getVectorsDirForUser(req, uid) { |
| 779 | const safeUid = sanitizeUserId(uid); |
| 780 | const vaultId = sanitizeVaultId(req.headers['x-vault-id']); |
| 781 | if (!req.blobStore) { |
| 782 | const d = path.join(DATA_DIR, 'vectors', safeUid, vaultId); |
| 783 | if (!fs.existsSync(d)) fs.mkdirSync(d, { recursive: true }); |
| 784 | return d; |
| 785 | } |
| 786 | const dir = path.join(os.tmpdir(), 'knowtation-bridge-vectors', safeUid, vaultId); |
| 787 | if (!fs.existsSync(dir)) fs.mkdirSync(dir, { recursive: true }); |
| 788 | const key = 'vectors/' + safeUid + '/' + vaultId; |
| 789 | try { |
| 790 | const data = await req.blobStore.get(key, { type: 'arrayBuffer' }); |
| 791 | if (data && data.byteLength > 0) { |
| 792 | fs.writeFileSync(path.join(dir, DB_FILENAME), Buffer.from(data)); |
| 793 | } |
| 794 | } catch (_) { |
| 795 | // No existing blob or read error; start fresh |
| 796 | } |
| 797 | return dir; |
| 798 | } |
| 799 | |
| 800 | /** Persist user's vector DB from disk to Blob (call after index). Phase 15: key includes vault_id. */ |
| 801 | async function persistVectorsToBlob(req, uid, vectorsDir) { |
| 802 | if (!req.blobStore) return; |
| 803 | const dbPath = path.join(vectorsDir, DB_FILENAME); |
| 804 | if (!fs.existsSync(dbPath)) return; |
| 805 | const vaultId = sanitizeVaultId(req.headers['x-vault-id']); |
| 806 | const key = 'vectors/' + sanitizeUserId(uid) + '/' + vaultId; |
| 807 | const buf = fs.readFileSync(dbPath); |
| 808 | const arrayBuffer = buf.buffer.slice(buf.byteOffset, buf.byteOffset + buf.byteLength); |
| 809 | await req.blobStore.set(key, arrayBuffer); |
| 810 | } |
| 811 | |
| 812 | function signState(payload) { |
| 813 | const payloadStr = JSON.stringify(payload); |
| 814 | const sig = crypto.createHmac('sha256', SESSION_SECRET).update(payloadStr).digest('hex'); |
| 815 | return Buffer.from(payloadStr).toString('base64url') + '.' + sig; |
| 816 | } |
| 817 | |
| 818 | function verifyState(stateStr, maxAgeMs = 600000) { |
| 819 | if (!stateStr || typeof stateStr !== 'string') return null; |
| 820 | const [b64, sig] = stateStr.split('.'); |
| 821 | if (!b64 || !sig) return null; |
| 822 | try { |
| 823 | const payload = JSON.parse(Buffer.from(b64, 'base64url').toString()); |
| 824 | const expected = crypto.createHmac('sha256', SESSION_SECRET).update(JSON.stringify(payload)).digest('hex'); |
| 825 | if (expected !== sig) return null; |
| 826 | if (Date.now() - (payload.ts || 0) > maxAgeMs) return null; |
| 827 | return payload; |
| 828 | } catch (_) { |
| 829 | return null; |
| 830 | } |
| 831 | } |
| 832 | |
| 833 | function userIdFromJwt(token) { |
| 834 | try { |
| 835 | const payload = jwt.verify(token, SESSION_SECRET); |
| 836 | return payload.sub ?? null; |
| 837 | } catch (_) { |
| 838 | return null; |
| 839 | } |
| 840 | } |
| 841 | |
| 842 | const app = express(); |
| 843 | app.use(express.json({ limit: '1mb' })); |
| 844 | |
| 845 | app.use((req, _res, next) => { |
| 846 | req.blobStore = globalThis.__netlify_blob_store || null; |
| 847 | next(); |
| 848 | }); |
| 849 | |
| 850 | // Background-function marker. `netlify/functions/bridge-index-background.mjs` |
| 851 | // validates an HMAC-signed inbound request, then sets |
| 852 | // `globalThis.__bridge_internal_request = { canisterUid, vaultId, jobId }` BEFORE |
| 853 | // invoking the same Express app via serverless-http. The index handler reads |
| 854 | // `req.bridgeInternalRequest` to decide whether to route (sync path) or skip |
| 855 | // routing and execute inline (background path). Globals are used because |
| 856 | // serverless-http does not let us inject per-request locals from the wrapper. |
| 857 | app.use((req, _res, next) => { |
| 858 | req.bridgeInternalRequest = globalThis.__bridge_internal_request || null; |
| 859 | next(); |
| 860 | }); |
| 861 | |
| 862 | app.use((_req, res, next) => { |
| 863 | res.set('Access-Control-Allow-Origin', process.env.HUB_CORS_ORIGIN || '*'); |
| 864 | res.set('Access-Control-Allow-Methods', 'GET, POST, DELETE, OPTIONS'); |
| 865 | res.set('Access-Control-Allow-Headers', 'Authorization, Content-Type, X-Vault-Id, X-User-Id'); |
| 866 | res.set('Access-Control-Allow-Credentials', 'true'); |
| 867 | next(); |
| 868 | }); |
| 869 | |
| 870 | // When Netlify rewrites /* to /.netlify/functions/bridge/:splat, Express sees the full path; strip prefix so routes match. |
| 871 | if (inServerless) { |
| 872 | const bridgePrefix = '/.netlify/functions/bridge'; |
| 873 | app.use((req, _res, next) => { |
| 874 | if (req.url.startsWith(bridgePrefix)) { |
| 875 | req.url = req.url.slice(bridgePrefix.length) || '/'; |
| 876 | } |
| 877 | next(); |
| 878 | }); |
| 879 | } |
| 880 | |
| 881 | // Public deploy probe (no auth): compare to Netlify **knowtation-bridge** Production commit vs gateway. |
| 882 | app.get('/api/v1/bridge-version', (_req, res) => { |
| 883 | res.json({ |
| 884 | service: 'knowtation-bridge', |
| 885 | commit: process.env.COMMIT_REF || process.env.VERCEL_GIT_COMMIT_SHA || null, |
| 886 | deploy_id: process.env.DEPLOY_ID || null, |
| 887 | context: process.env.CONTEXT || null, |
| 888 | netlify: Boolean(process.env.NETLIFY || process.env.AWS_LAMBDA_FUNCTION_NAME), |
| 889 | }); |
| 890 | }); |
| 891 | |
| 892 | // ——— Roles & invites (hosted parity) ——— |
| 893 | async function requireBridgeAuth(req, res, next) { |
| 894 | const auth = req.headers.authorization; |
| 895 | const token = auth && auth.startsWith('Bearer ') ? auth.slice(7) : null; |
| 896 | const uid = token ? userIdFromJwt(token) : null; |
| 897 | if (!uid) return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 898 | req.uid = uid; |
| 899 | next(); |
| 900 | } |
| 901 | |
| 902 | async function requireBridgeAdmin(req, res, next) { |
| 903 | const roles = await loadRoles(req.blobStore); |
| 904 | const role = effectiveRole(req.uid, roles); |
| 905 | if (role !== 'admin') return res.status(403).json({ error: 'Admin only', code: 'FORBIDDEN' }); |
| 906 | next(); |
| 907 | } |
| 908 | |
| 909 | /** Import / index parity: viewers cannot write; default role is member (treated like editor for hosted). */ |
| 910 | async function requireBridgeEditorOrAdmin(req, res, next) { |
| 911 | const roles = await loadRoles(req.blobStore); |
| 912 | const role = effectiveRole(req.uid, roles); |
| 913 | if (role === 'viewer') { |
| 914 | return res.status(403).json({ error: 'This action requires editor or admin.', code: 'FORBIDDEN' }); |
| 915 | } |
| 916 | next(); |
| 917 | } |
| 918 | |
| 919 | app.get('/api/v1/roles', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 920 | try { |
| 921 | const roles = await loadRoles(req.blobStore); |
| 922 | const evaluator_may_approve = await loadEvaluatorMayApproveMap(req.blobStore); |
| 923 | res.json({ roles, evaluator_may_approve }); |
| 924 | } catch (e) { |
| 925 | console.error('[bridge] GET /api/v1/roles', e?.message); |
| 926 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 927 | } |
| 928 | }); |
| 929 | |
| 930 | app.post('/api/v1/roles', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 931 | const { user_id: userId, role } = req.body || {}; |
| 932 | if (!userId || typeof userId !== 'string' || !userId.trim()) { |
| 933 | return res.status(400).json({ error: 'user_id required (e.g. github:12345)', code: 'BAD_REQUEST' }); |
| 934 | } |
| 935 | const r = (role || 'editor').toLowerCase(); |
| 936 | if (!VALID_ROLES.has(r)) { |
| 937 | return res.status(400).json({ error: 'role must be admin, editor, viewer, or evaluator', code: 'BAD_REQUEST' }); |
| 938 | } |
| 939 | try { |
| 940 | const roles = await loadRoles(req.blobStore); |
| 941 | const uidKey = userId.trim(); |
| 942 | roles[uidKey] = r; |
| 943 | await saveRoles(req.blobStore, roles); |
| 944 | const mayMap = await loadEvaluatorMayApproveMap(req.blobStore); |
| 945 | if (r === 'evaluator' && req.body && Object.prototype.hasOwnProperty.call(req.body, 'evaluator_may_approve')) { |
| 946 | mayMap[uidKey] = Boolean(req.body.evaluator_may_approve); |
| 947 | await saveEvaluatorMayApproveMap(req.blobStore, mayMap); |
| 948 | } else if (r !== 'evaluator' && Object.prototype.hasOwnProperty.call(mayMap, uidKey)) { |
| 949 | delete mayMap[uidKey]; |
| 950 | await saveEvaluatorMayApproveMap(req.blobStore, mayMap); |
| 951 | } |
| 952 | res.json({ ok: true }); |
| 953 | } catch (e) { |
| 954 | console.error('[bridge] POST /api/v1/roles', e?.message); |
| 955 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 956 | } |
| 957 | }); |
| 958 | |
| 959 | app.post('/api/v1/roles/evaluator-may-approve', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 960 | const { user_id: userId, evaluator_may_approve: flag } = req.body || {}; |
| 961 | if (!userId || typeof userId !== 'string' || !userId.trim()) { |
| 962 | return res.status(400).json({ error: 'user_id required', code: 'BAD_REQUEST' }); |
| 963 | } |
| 964 | if (typeof flag !== 'boolean') { |
| 965 | return res.status(400).json({ error: 'evaluator_may_approve must be boolean', code: 'BAD_REQUEST' }); |
| 966 | } |
| 967 | const uidKey = userId.trim(); |
| 968 | try { |
| 969 | const roles = await loadRoles(req.blobStore); |
| 970 | if (effectiveRole(uidKey, roles) !== 'evaluator') { |
| 971 | return res.status(400).json({ error: 'User must have evaluator role', code: 'BAD_REQUEST' }); |
| 972 | } |
| 973 | const mayMap = await loadEvaluatorMayApproveMap(req.blobStore); |
| 974 | mayMap[uidKey] = flag; |
| 975 | await saveEvaluatorMayApproveMap(req.blobStore, mayMap); |
| 976 | res.json({ ok: true }); |
| 977 | } catch (e) { |
| 978 | console.error('[bridge] POST /api/v1/roles/evaluator-may-approve', e?.message); |
| 979 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 980 | } |
| 981 | }); |
| 982 | |
| 983 | app.get('/api/v1/invites', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 984 | try { |
| 985 | const invitesMap = await loadInvites(req.blobStore); |
| 986 | const now = Date.now(); |
| 987 | const list = []; |
| 988 | for (const [token, entry] of Object.entries(invitesMap)) { |
| 989 | const created = new Date(entry.created_at).getTime(); |
| 990 | const expires_at = new Date(created + INVITE_EXPIRY_MS).toISOString(); |
| 991 | if (now - created <= INVITE_EXPIRY_MS) { |
| 992 | list.push({ token, role: entry.role, created_at: entry.created_at, expires_at }); |
| 993 | } |
| 994 | } |
| 995 | list.sort((a, b) => new Date(b.created_at) - new Date(a.created_at)); |
| 996 | res.json({ invites: list }); |
| 997 | } catch (e) { |
| 998 | console.error('[bridge] GET /api/v1/invites', e?.message); |
| 999 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1000 | } |
| 1001 | }); |
| 1002 | |
| 1003 | app.post('/api/v1/invites', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 1004 | const role = (req.body?.role || 'editor').toLowerCase(); |
| 1005 | if (!['viewer', 'editor', 'admin', 'evaluator'].includes(role)) { |
| 1006 | return res.status(400).json({ error: 'role must be viewer, editor, admin, or evaluator', code: 'BAD_REQUEST' }); |
| 1007 | } |
| 1008 | try { |
| 1009 | const token = crypto.randomBytes(24).toString('base64url'); |
| 1010 | const created_at = new Date().toISOString(); |
| 1011 | const expires_at = new Date(Date.now() + INVITE_EXPIRY_MS).toISOString(); |
| 1012 | const invites = await loadInvites(req.blobStore); |
| 1013 | invites[token] = { role, created_at }; |
| 1014 | await saveInvites(req.blobStore, invites); |
| 1015 | const base = (HUB_UI_ORIGIN + (HUB_UI_PATH || '/hub') + '/').replace(/(\/)+$/, '/'); |
| 1016 | const invite_url = base + '?invite=' + encodeURIComponent(token); |
| 1017 | res.status(201).json({ invite_url, token, role, created_at, expires_at }); |
| 1018 | } catch (e) { |
| 1019 | console.error('[bridge] POST /api/v1/invites', e?.message); |
| 1020 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1021 | } |
| 1022 | }); |
| 1023 | |
| 1024 | app.delete('/api/v1/invites/:token', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 1025 | const token = req.params.token; |
| 1026 | if (!token) return res.status(400).json({ error: 'token required', code: 'BAD_REQUEST' }); |
| 1027 | try { |
| 1028 | const invites = await loadInvites(req.blobStore); |
| 1029 | const had = token in invites; |
| 1030 | delete invites[token]; |
| 1031 | await saveInvites(req.blobStore, invites); |
| 1032 | res.json({ ok: true, removed: had }); |
| 1033 | } catch (e) { |
| 1034 | console.error('[bridge] DELETE /api/v1/invites/:token', e?.message); |
| 1035 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1036 | } |
| 1037 | }); |
| 1038 | |
| 1039 | app.post('/api/v1/invites/consume', requireBridgeAuth, async (req, res) => { |
| 1040 | const token = req.body?.token; |
| 1041 | if (!token || typeof token !== 'string' || !token.trim()) { |
| 1042 | return res.status(400).json({ error: 'token required', code: 'BAD_REQUEST' }); |
| 1043 | } |
| 1044 | const uid = req.uid; |
| 1045 | try { |
| 1046 | const invites = await loadInvites(req.blobStore); |
| 1047 | const entry = invites[token]; |
| 1048 | if (!entry) { |
| 1049 | return res.status(404).json({ error: 'Invite not found or already used', code: 'NOT_FOUND' }); |
| 1050 | } |
| 1051 | const created = new Date(entry.created_at).getTime(); |
| 1052 | if (Date.now() - created > INVITE_EXPIRY_MS) { |
| 1053 | delete invites[token]; |
| 1054 | await saveInvites(req.blobStore, invites); |
| 1055 | return res.status(410).json({ error: 'Invite expired', code: 'EXPIRED' }); |
| 1056 | } |
| 1057 | const roles = await loadRoles(req.blobStore); |
| 1058 | roles[uid] = entry.role; |
| 1059 | await saveRoles(req.blobStore, roles); |
| 1060 | delete invites[token]; |
| 1061 | await saveInvites(req.blobStore, invites); |
| 1062 | res.json({ ok: true, role: entry.role }); |
| 1063 | } catch (e) { |
| 1064 | console.error('[bridge] POST /api/v1/invites/consume', e?.message); |
| 1065 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1066 | } |
| 1067 | }); |
| 1068 | |
| 1069 | // For gateway GET /api/v1/settings: return role from bridge store so invited users get correct role |
| 1070 | app.get('/api/v1/role', requireBridgeAuth, async (req, res) => { |
| 1071 | try { |
| 1072 | const roles = await loadRoles(req.blobStore); |
| 1073 | const mayMap = await loadEvaluatorMayApproveMap(req.blobStore); |
| 1074 | const role = effectiveRole(req.uid, roles); |
| 1075 | const may_approve_proposals = mayApproveProposalsForUser(req.uid, roles, mayMap); |
| 1076 | res.json({ role, may_approve_proposals }); |
| 1077 | } catch (e) { |
| 1078 | console.error('[bridge] GET /api/v1/role', e?.message); |
| 1079 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1080 | } |
| 1081 | }); |
| 1082 | |
| 1083 | app.get('/api/v1/workspace', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 1084 | try { |
| 1085 | const w = await loadWorkspace(req.blobStore); |
| 1086 | res.json({ owner_user_id: w.owner_user_id }); |
| 1087 | } catch (e) { |
| 1088 | console.error('[bridge] GET /api/v1/workspace', e?.message); |
| 1089 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1090 | } |
| 1091 | }); |
| 1092 | |
| 1093 | app.post('/api/v1/workspace', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 1094 | const raw = req.body?.owner_user_id; |
| 1095 | const owner_user_id = |
| 1096 | raw === null || raw === undefined || raw === '' |
| 1097 | ? null |
| 1098 | : typeof raw === 'string' && raw.trim() |
| 1099 | ? raw.trim() |
| 1100 | : null; |
| 1101 | try { |
| 1102 | await saveWorkspace(req.blobStore, owner_user_id); |
| 1103 | const w = await loadWorkspace(req.blobStore); |
| 1104 | res.json({ ok: true, owner_user_id: w.owner_user_id }); |
| 1105 | } catch (e) { |
| 1106 | console.error('[bridge] POST /api/v1/workspace', e?.message); |
| 1107 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1108 | } |
| 1109 | }); |
| 1110 | |
| 1111 | app.get('/api/v1/vault-access', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 1112 | try { |
| 1113 | const access = await loadVaultAccess(req.blobStore); |
| 1114 | res.json({ access }); |
| 1115 | } catch (e) { |
| 1116 | console.error('[bridge] GET /api/v1/vault-access', e?.message); |
| 1117 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1118 | } |
| 1119 | }); |
| 1120 | |
| 1121 | app.post('/api/v1/vault-access', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 1122 | const access = req.body?.access; |
| 1123 | if (!access || typeof access !== 'object') { |
| 1124 | return res.status(400).json({ error: 'access object required', code: 'BAD_REQUEST' }); |
| 1125 | } |
| 1126 | try { |
| 1127 | await saveVaultAccess(req.blobStore, access); |
| 1128 | const out = await loadVaultAccess(req.blobStore); |
| 1129 | res.json({ ok: true, access: out }); |
| 1130 | } catch (e) { |
| 1131 | console.error('[bridge] POST /api/v1/vault-access', e?.message); |
| 1132 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1133 | } |
| 1134 | }); |
| 1135 | |
| 1136 | app.get('/api/v1/scope', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 1137 | try { |
| 1138 | const scope = await loadScope(req.blobStore); |
| 1139 | res.json({ scope }); |
| 1140 | } catch (e) { |
| 1141 | console.error('[bridge] GET /api/v1/scope', e?.message); |
| 1142 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1143 | } |
| 1144 | }); |
| 1145 | |
| 1146 | app.post('/api/v1/scope', requireBridgeAuth, requireBridgeAdmin, async (req, res) => { |
| 1147 | const scope = req.body?.scope; |
| 1148 | if (!scope || typeof scope !== 'object') { |
| 1149 | return res.status(400).json({ error: 'scope object required', code: 'BAD_REQUEST' }); |
| 1150 | } |
| 1151 | try { |
| 1152 | await saveScope(req.blobStore, scope); |
| 1153 | const out = await loadScope(req.blobStore); |
| 1154 | res.json({ ok: true, scope: out }); |
| 1155 | } catch (e) { |
| 1156 | console.error('[bridge] POST /api/v1/scope', e?.message); |
| 1157 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1158 | } |
| 1159 | }); |
| 1160 | |
| 1161 | app.get('/api/v1/hosted-context', requireBridgeAuth, async (req, res) => { |
| 1162 | try { |
| 1163 | const actor = req.uid; |
| 1164 | const workspace = await loadWorkspace(req.blobStore); |
| 1165 | const roles = await loadRoles(req.blobStore); |
| 1166 | const ctx = await resolveHostedBridgeContext(req, actor); |
| 1167 | if (!ctx.ok) { |
| 1168 | return res.status(ctx.status).json({ error: ctx.error, code: ctx.code }); |
| 1169 | } |
| 1170 | const role = effectiveRole(actor, roles); |
| 1171 | const mayMap = await loadEvaluatorMayApproveMap(req.blobStore); |
| 1172 | const may_approve_proposals = mayApproveProposalsForUser(actor, roles, mayMap); |
| 1173 | res.json({ |
| 1174 | actor_sub: actor, |
| 1175 | workspace_owner_id: workspace.owner_user_id, |
| 1176 | effective_canister_user_id: ctx.effectiveCanisterUid, |
| 1177 | delegating: ctx.delegating, |
| 1178 | allowed_vault_ids: ctx.allowedVaultIds, |
| 1179 | scope: ctx.scope, |
| 1180 | role, |
| 1181 | may_approve_proposals, |
| 1182 | }); |
| 1183 | } catch (e) { |
| 1184 | console.error('[bridge] GET /api/v1/hosted-context', e?.message); |
| 1185 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1186 | } |
| 1187 | }); |
| 1188 | |
| 1189 | app.get('/api/v1/hosted-context/settings', requireBridgeAuth, async (req, res) => { |
| 1190 | try { |
| 1191 | const actor = req.uid; |
| 1192 | const ctx = await resolveHostedBridgeSettingsContext(req, actor); |
| 1193 | const mayMap = await loadEvaluatorMayApproveMap(req.blobStore); |
| 1194 | const may_approve_proposals = mayApproveProposalsForUser(actor, { [actor]: ctx.role }, mayMap); |
| 1195 | res.json({ |
| 1196 | actor_sub: actor, |
| 1197 | workspace_owner_id: ctx.workspaceOwnerId, |
| 1198 | effective_canister_user_id: ctx.effectiveCanisterUid, |
| 1199 | delegating: ctx.delegating, |
| 1200 | allowed_vault_ids: ctx.allowedVaultIds, |
| 1201 | scope: null, |
| 1202 | role: ctx.role, |
| 1203 | may_approve_proposals, |
| 1204 | }); |
| 1205 | } catch (e) { |
| 1206 | console.error('[bridge] GET /api/v1/hosted-context/settings', e?.message); |
| 1207 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1208 | } |
| 1209 | }); |
| 1210 | |
| 1211 | // ——— Connect GitHub ——— |
| 1212 | if (process.env.GITHUB_CLIENT_ID && process.env.GITHUB_CLIENT_SECRET) { |
| 1213 | app.get('/auth/github-connect', (req, res) => { |
| 1214 | const token = req.query.token || (req.headers.authorization && req.headers.authorization.startsWith('Bearer ') && req.headers.authorization.slice(7)); |
| 1215 | const uid = token ? userIdFromJwt(token) : null; |
| 1216 | if (!uid) { |
| 1217 | return res.redirect(HUB_UI_ORIGIN + HUB_UI_PATH + '/?github_connect_error=not_authenticated'); |
| 1218 | } |
| 1219 | const state = signState({ uid, ts: Date.now() }); |
| 1220 | const redirectUri = BASE_URL + '/auth/callback/github-connect'; |
| 1221 | const url = 'https://github.com/login/oauth/authorize?client_id=' + encodeURIComponent(process.env.GITHUB_CLIENT_ID) |
| 1222 | + '&redirect_uri=' + encodeURIComponent(redirectUri) |
| 1223 | + '&scope=repo' |
| 1224 | + '&state=' + encodeURIComponent(state); |
| 1225 | res.redirect(url); |
| 1226 | }); |
| 1227 | |
| 1228 | app.get('/auth/callback/github-connect', async (req, res) => { |
| 1229 | const { code, state } = req.query || {}; |
| 1230 | const hubBase = HUB_UI_ORIGIN + HUB_UI_PATH + '/'; |
| 1231 | console.log('[bridge] callback: hubBase=%s (ORIGIN=%s PATH=%s)', hubBase, HUB_UI_ORIGIN, HUB_UI_PATH); |
| 1232 | const payload = verifyState(state); |
| 1233 | if (!payload) { |
| 1234 | const url = hubBase + '?github_connect_error=error_state'; |
| 1235 | console.log('[bridge] redirect (error_state): %s', url); |
| 1236 | return res.redirect(302, url); |
| 1237 | } |
| 1238 | if (!code) { |
| 1239 | const url = hubBase + '?github_connect_error=error_code'; |
| 1240 | console.log('[bridge] redirect (error_code): %s', url); |
| 1241 | return res.redirect(302, url); |
| 1242 | } |
| 1243 | const uid = payload.uid; |
| 1244 | const redirectUri = BASE_URL + '/auth/callback/github-connect'; |
| 1245 | const tokenRes = await fetch('https://github.com/login/oauth/access_token', { |
| 1246 | method: 'POST', |
| 1247 | headers: { Accept: 'application/json', 'Content-Type': 'application/json' }, |
| 1248 | body: JSON.stringify({ |
| 1249 | client_id: process.env.GITHUB_CLIENT_ID, |
| 1250 | client_secret: process.env.GITHUB_CLIENT_SECRET, |
| 1251 | code, |
| 1252 | redirect_uri: redirectUri, |
| 1253 | }), |
| 1254 | }); |
| 1255 | const data = await tokenRes.json(); |
| 1256 | if (!data.access_token) { |
| 1257 | const url = hubBase + '?github_connect_error=error_token'; |
| 1258 | console.log('[bridge] redirect (error_token): %s', url); |
| 1259 | return res.redirect(302, url); |
| 1260 | } |
| 1261 | const tokensByUser = await loadTokens(req.blobStore); |
| 1262 | tokensByUser[uid] = { token: data.access_token, repo: tokensByUser[uid]?.repo || null }; |
| 1263 | try { |
| 1264 | await saveTokens(req.blobStore, tokensByUser); |
| 1265 | } catch (e) { |
| 1266 | console.error('[bridge] saveTokens after GitHub OAuth failed:', e?.message || e); |
| 1267 | const url = hubBase + '?github_connect_error=blob_storage'; |
| 1268 | return res.redirect(302, url); |
| 1269 | } |
| 1270 | const redirectTo = hubBase + '?github_connected=1'; |
| 1271 | console.log('[bridge] redirect after connect: HUB_UI_ORIGIN=%s HUB_UI_PATH=%s redirectTo=%s', HUB_UI_ORIGIN, HUB_UI_PATH, redirectTo); |
| 1272 | res.redirect(302, redirectTo); |
| 1273 | }); |
| 1274 | } |
| 1275 | |
| 1276 | // ——— Delete vault (canister + team access/scope + vector blob) ——— |
| 1277 | app.delete('/api/v1/vaults/:vaultId', requireBridgeAuth, requireBridgeEditorOrAdmin, async (req, res) => { |
| 1278 | if (!CANISTER_URL) { |
| 1279 | return res.status(503).json({ error: 'CANISTER_URL not configured', code: 'NOT_AVAILABLE' }); |
| 1280 | } |
| 1281 | const vaultId = sanitizeVaultId(req.params.vaultId); |
| 1282 | if (!req.params.vaultId || String(req.params.vaultId).trim() === '' || vaultId === 'default') { |
| 1283 | return res.status(400).json({ error: 'Cannot delete the default vault', code: 'BAD_REQUEST' }); |
| 1284 | } |
| 1285 | |
| 1286 | const prevVaultHeader = req.headers['x-vault-id']; |
| 1287 | req.headers['x-vault-id'] = vaultId; |
| 1288 | const hctx = await resolveHostedBridgeContext(req, req.uid); |
| 1289 | req.headers['x-vault-id'] = prevVaultHeader; |
| 1290 | |
| 1291 | if (!hctx.ok) { |
| 1292 | return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 1293 | } |
| 1294 | |
| 1295 | const workspace = await loadWorkspace(req.blobStore); |
| 1296 | const owner = workspace.owner_user_id && String(workspace.owner_user_id).trim(); |
| 1297 | if (owner && req.uid !== owner) { |
| 1298 | return res.status(403).json({ |
| 1299 | error: 'Only the workspace owner can delete vaults.', |
| 1300 | code: 'FORBIDDEN', |
| 1301 | }); |
| 1302 | } |
| 1303 | |
| 1304 | let canRes; |
| 1305 | try { |
| 1306 | canRes = await fetch(`${CANISTER_URL}/api/v1/vaults/${encodeURIComponent(vaultId)}`, { |
| 1307 | method: 'DELETE', |
| 1308 | headers: canisterHeaders({ 'X-User-Id': hctx.effectiveCanisterUid }), |
| 1309 | }); |
| 1310 | } catch (e) { |
| 1311 | console.error('[bridge] DELETE vault canister fetch', e?.message); |
| 1312 | return res.status(502).json({ error: 'Could not reach canister', code: 'BAD_GATEWAY' }); |
| 1313 | } |
| 1314 | |
| 1315 | const text = await canRes.text(); |
| 1316 | if (!canRes.ok) { |
| 1317 | let errMsg = text; |
| 1318 | try { |
| 1319 | const j = JSON.parse(text); |
| 1320 | if (j && j.error) errMsg = j.error; |
| 1321 | } catch (_) {} |
| 1322 | return res.status(canRes.status >= 400 ? canRes.status : 502).json({ |
| 1323 | error: errMsg || 'Canister error', |
| 1324 | code: 'UPSTREAM_ERROR', |
| 1325 | }); |
| 1326 | } |
| 1327 | |
| 1328 | await stripHostedVaultFromAccessAndScope(req.blobStore, vaultId); |
| 1329 | await removeHostedVectorBlobForVault(req.blobStore, hctx.effectiveCanisterUid, vaultId); |
| 1330 | |
| 1331 | try { |
| 1332 | const data = text ? JSON.parse(text) : {}; |
| 1333 | res.json({ ok: true, ...data }); |
| 1334 | } catch (_) { |
| 1335 | res.json({ ok: true, deleted_vault_id: vaultId }); |
| 1336 | } |
| 1337 | }); |
| 1338 | |
| 1339 | /** |
| 1340 | * Full proposal documents for GitHub backup (list + GET each id). Same scope as notes. |
| 1341 | * @param {string} canisterUrl |
| 1342 | * @param {string} canisterUid |
| 1343 | * @param {string} vaultId |
| 1344 | * @param {{ projects?: string[], folders?: string[] } | null | undefined} scope |
| 1345 | */ |
| 1346 | async function fetchFullProposalsForGithubBackup(canisterUrl, canisterUid, vaultId, scope) { |
| 1347 | const base = String(canisterUrl || '').replace(/\/$/, ''); |
| 1348 | const headers = canisterHeaders({ 'X-User-Id': canisterUid, 'X-Vault-Id': vaultId }); |
| 1349 | const listRes = await fetch(`${base}/api/v1/proposals`, { method: 'GET', headers }); |
| 1350 | if (!listRes.ok) { |
| 1351 | const err = new Error(`Canister proposals list ${listRes.status}`); |
| 1352 | err.status = 502; |
| 1353 | throw err; |
| 1354 | } |
| 1355 | let listJson; |
| 1356 | try { |
| 1357 | listJson = await listRes.json(); |
| 1358 | } catch { |
| 1359 | const err = new Error('Invalid canister proposals list JSON'); |
| 1360 | err.status = 502; |
| 1361 | throw err; |
| 1362 | } |
| 1363 | const stubs = Array.isArray(listJson.proposals) ? listJson.proposals : []; |
| 1364 | const full = []; |
| 1365 | for (const stub of stubs) { |
| 1366 | const id = stub && stub.proposal_id ? String(stub.proposal_id) : ''; |
| 1367 | if (!id) continue; |
| 1368 | const oneRes = await fetch(`${base}/api/v1/proposals/${encodeURIComponent(id)}`, { |
| 1369 | method: 'GET', |
| 1370 | headers, |
| 1371 | }); |
| 1372 | if (!oneRes.ok) { |
| 1373 | const err = new Error(`Canister proposal ${id} ${oneRes.status}`); |
| 1374 | err.status = 502; |
| 1375 | throw err; |
| 1376 | } |
| 1377 | const text = await oneRes.text(); |
| 1378 | const body = parseCanisterProposalGetBody(id, text, stub); |
| 1379 | if (body._knowtation_backup_json_unparseable) { |
| 1380 | console.error('[bridge] vault/sync proposal JSON parse failed', { id, preview: text.slice(0, 300) }); |
| 1381 | } |
| 1382 | full.push(body); |
| 1383 | } |
| 1384 | return applyScopeFilterToProposals(full, scope); |
| 1385 | } |
| 1386 | |
| 1387 | // ——— Back up now: fetch vault from canister, push to GitHub ——— |
| 1388 | app.post('/api/v1/vault/sync', requireBridgeAuth, requireBridgeEditorOrAdmin, async (req, res) => { |
| 1389 | const uid = req.uid; |
| 1390 | |
| 1391 | const hctx = await resolveHostedBridgeContext(req, uid); |
| 1392 | if (!hctx.ok) { |
| 1393 | return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 1394 | } |
| 1395 | const canisterUid = hctx.effectiveCanisterUid; |
| 1396 | |
| 1397 | const tokensByUser = await loadTokens(req.blobStore); |
| 1398 | const conn = tokensByUser[uid]; |
| 1399 | const repo = req.body?.repo || conn?.repo; |
| 1400 | if (!conn?.token) { |
| 1401 | return res.status(400).json({ error: 'GitHub not connected', code: 'GITHUB_NOT_CONNECTED' }); |
| 1402 | } |
| 1403 | if (!repo || typeof repo !== 'string') { |
| 1404 | return res.status(400).json({ error: 'Repo required', code: 'REPO_REQUIRED', hint: 'Send { "repo": "owner/name" } or set repo after connecting GitHub.' }); |
| 1405 | } |
| 1406 | |
| 1407 | const [owner, name] = repo.split('/').filter(Boolean); |
| 1408 | if (!owner || !name) { |
| 1409 | return res.status(400).json({ error: 'Invalid repo format', code: 'BAD_REQUEST' }); |
| 1410 | } |
| 1411 | |
| 1412 | const vaultId = sanitizeVaultId(req.headers['x-vault-id']); |
| 1413 | |
| 1414 | // Fetch vault from canister (export) |
| 1415 | let exportRes; |
| 1416 | try { |
| 1417 | exportRes = await fetch(CANISTER_URL + '/api/v1/export', { |
| 1418 | method: 'GET', |
| 1419 | headers: canisterHeaders({ 'X-User-Id': canisterUid, 'X-Vault-Id': vaultId }), |
| 1420 | }); |
| 1421 | } catch (e) { |
| 1422 | return res.status(502).json({ error: 'Could not reach canister', code: 'BAD_GATEWAY' }); |
| 1423 | } |
| 1424 | if (!exportRes.ok) { |
| 1425 | return res.status(502).json({ error: 'Canister error', code: 'BAD_GATEWAY', status: exportRes.status }); |
| 1426 | } |
| 1427 | let vault; |
| 1428 | try { |
| 1429 | vault = await exportRes.json(); |
| 1430 | } catch (_) { |
| 1431 | return res.status(502).json({ error: 'Invalid canister response', code: 'BAD_GATEWAY' }); |
| 1432 | } |
| 1433 | let notes = vault.notes || []; |
| 1434 | if (hctx.scope) { |
| 1435 | notes = applyScopeFilterToNotes(notes, hctx.scope); |
| 1436 | } |
| 1437 | |
| 1438 | let proposals = []; |
| 1439 | try { |
| 1440 | proposals = await fetchFullProposalsForGithubBackup(CANISTER_URL, canisterUid, vaultId, hctx.scope); |
| 1441 | } catch (e) { |
| 1442 | console.error('[bridge] vault/sync proposals fetch', e?.message); |
| 1443 | return res.status(e.status || 502).json({ |
| 1444 | error: e.message || 'Could not fetch proposals for backup', |
| 1445 | code: 'BAD_GATEWAY', |
| 1446 | }); |
| 1447 | } |
| 1448 | |
| 1449 | // Store repo for next time |
| 1450 | if (req.body?.repo && (!conn.repo || conn.repo !== repo)) { |
| 1451 | tokensByUser[uid] = { ...conn, repo }; |
| 1452 | await saveTokens(req.blobStore, tokensByUser); |
| 1453 | } |
| 1454 | |
| 1455 | // Push to GitHub: get default branch, create blobs, create tree, commit, push |
| 1456 | const ghToken = conn.token; |
| 1457 | const ghApi = 'https://api.github.com'; |
| 1458 | // GitHub requires a non-empty User-Agent; some serverless runtimes send none → 403 "Administrative rules". |
| 1459 | const ghHeaders = { |
| 1460 | Authorization: 'token ' + ghToken, |
| 1461 | Accept: 'application/vnd.github.v3+json', |
| 1462 | 'Content-Type': 'application/json', |
| 1463 | 'User-Agent': 'KnowtationHub-Bridge/1.0 (+https://knowtation.store)', |
| 1464 | }; |
| 1465 | const headsRefEnc = (branch) => encodeURIComponent(`heads/${String(branch || 'main').trim()}`); |
| 1466 | |
| 1467 | let defaultBranch; |
| 1468 | try { |
| 1469 | const repoRes = await fetch(`${ghApi}/repos/${owner}/${name}`, { headers: ghHeaders }); |
| 1470 | if (!repoRes.ok) { |
| 1471 | if (repoRes.status === 404) { |
| 1472 | return res.status(400).json({ error: 'Repo not found or no access', code: 'REPO_NOT_FOUND' }); |
| 1473 | } |
| 1474 | throw new Error('GitHub API ' + repoRes.status); |
| 1475 | } |
| 1476 | const repoData = await repoRes.json(); |
| 1477 | defaultBranch = String(repoData.default_branch || 'main').trim() || 'main'; |
| 1478 | } catch (e) { |
| 1479 | return res.status(502).json({ error: 'GitHub API error', code: 'BAD_GATEWAY' }); |
| 1480 | } |
| 1481 | |
| 1482 | // GET single ref: documented as /git/ref/{ref} with ref = heads/<branch> (URL-encoded). Avoids edge cases with /git/refs/... on some hosts. |
| 1483 | const refRes = await fetch(`${ghApi}/repos/${owner}/${name}/git/ref/${headsRefEnc(defaultBranch)}`, { headers: ghHeaders }); |
| 1484 | let baseSha = null; |
| 1485 | let baseTreeSha = null; |
| 1486 | if (refRes.ok) { |
| 1487 | const refData = await refRes.json(); |
| 1488 | baseSha = refData.object?.sha; |
| 1489 | if (!baseSha) { |
| 1490 | return res.status(502).json({ error: 'Invalid ref response', code: 'BAD_GATEWAY' }); |
| 1491 | } |
| 1492 | const baseTreeRes = await fetch(`${ghApi}/repos/${owner}/${name}/git/commits/${baseSha}`, { headers: ghHeaders }); |
| 1493 | if (!baseTreeRes.ok) { |
| 1494 | return res.status(502).json({ error: 'Could not get base commit', code: 'BAD_GATEWAY' }); |
| 1495 | } |
| 1496 | const baseCommit = await baseTreeRes.json(); |
| 1497 | baseTreeSha = baseCommit.tree?.sha; |
| 1498 | } else if (refRes.status === 404) { |
| 1499 | // Repo exists on GitHub but has no commits yet (Quick setup / empty repo) — no refs/heads/* yet. |
| 1500 | baseSha = null; |
| 1501 | baseTreeSha = null; |
| 1502 | } else { |
| 1503 | const refErrBody = await refRes.text(); |
| 1504 | console.warn('[bridge] GitHub GET ref failed', { owner, name, branch: defaultBranch, status: refRes.status, body: refErrBody.slice(0, 500) }); |
| 1505 | if (refRes.status === 403 || refRes.status === 401) { |
| 1506 | return res.status(502).json({ |
| 1507 | error: |
| 1508 | 'GitHub denied access when reading the branch (often missing User-Agent or expired token). Use Settings → Connect GitHub again.', |
| 1509 | code: 'BAD_GATEWAY', |
| 1510 | }); |
| 1511 | } |
| 1512 | return res.status(502).json({ |
| 1513 | error: 'Could not read branch on GitHub. If the repo is new with no commits, try Back up again after redeploying the bridge; otherwise check bridge logs.', |
| 1514 | code: 'BAD_GATEWAY', |
| 1515 | }); |
| 1516 | } |
| 1517 | |
| 1518 | const tree = []; |
| 1519 | for (const note of notes) { |
| 1520 | const path = note.path || 'note.md'; |
| 1521 | const content = (note.frontmatter && note.frontmatter !== '{}' ? '---\n' + note.frontmatter + '\n---\n\n' : '') + (note.body || ''); |
| 1522 | const blobRes = await fetch(`${ghApi}/repos/${owner}/${name}/git/blobs`, { |
| 1523 | method: 'POST', |
| 1524 | headers: ghHeaders, |
| 1525 | body: JSON.stringify({ content: Buffer.from(content, 'utf8').toString('base64'), encoding: 'base64' }), |
| 1526 | }); |
| 1527 | if (!blobRes.ok) { |
| 1528 | return res.status(502).json({ error: 'GitHub blob failed', code: 'BAD_GATEWAY' }); |
| 1529 | } |
| 1530 | const blob = await blobRes.json(); |
| 1531 | tree.push({ path, mode: '100644', type: 'blob', sha: blob.sha }); |
| 1532 | } |
| 1533 | |
| 1534 | const snapshotObj = { |
| 1535 | format_version: 1, |
| 1536 | kind: 'knowtation-hosted-backup', |
| 1537 | exported_at: new Date().toISOString(), |
| 1538 | vault_id: vaultId, |
| 1539 | proposals, |
| 1540 | }; |
| 1541 | const snapshotJson = JSON.stringify(snapshotObj); |
| 1542 | const snapBlobRes = await fetch(`${ghApi}/repos/${owner}/${name}/git/blobs`, { |
| 1543 | method: 'POST', |
| 1544 | headers: ghHeaders, |
| 1545 | body: JSON.stringify({ |
| 1546 | content: Buffer.from(snapshotJson, 'utf8').toString('base64'), |
| 1547 | encoding: 'base64', |
| 1548 | }), |
| 1549 | }); |
| 1550 | if (!snapBlobRes.ok) { |
| 1551 | return res.status(502).json({ error: 'GitHub blob failed (snapshot)', code: 'BAD_GATEWAY' }); |
| 1552 | } |
| 1553 | const snapBlob = await snapBlobRes.json(); |
| 1554 | tree.push({ |
| 1555 | path: '.knowtation/backup/v1/snapshot.json', |
| 1556 | mode: '100644', |
| 1557 | type: 'blob', |
| 1558 | sha: snapBlob.sha, |
| 1559 | }); |
| 1560 | |
| 1561 | const isInitialCommit = !baseSha; |
| 1562 | if (isInitialCommit && notes.length === 0 && proposals.length === 0) { |
| 1563 | const placeholder = |
| 1564 | '# Knowtation vault backup\n\n' |
| 1565 | + 'This folder is written by **Back up now** on hosted Knowtation.\n\n' |
| 1566 | + '- **Markdown files** elsewhere in this repo are your vault **notes**.\n' |
| 1567 | + '- **`.knowtation/backup/v1/snapshot.json`** holds full **proposal** records (status, review, enrich metadata, bodies).\n\n' |
| 1568 | + 'Your vault had no notes or proposals yet. Add content in the Hub and run **Back up now** again.\n'; |
| 1569 | const blobRes = await fetch(`${ghApi}/repos/${owner}/${name}/git/blobs`, { |
| 1570 | method: 'POST', |
| 1571 | headers: ghHeaders, |
| 1572 | body: JSON.stringify({ |
| 1573 | content: Buffer.from(placeholder, 'utf8').toString('base64'), |
| 1574 | encoding: 'base64', |
| 1575 | }), |
| 1576 | }); |
| 1577 | if (!blobRes.ok) { |
| 1578 | return res.status(502).json({ error: 'GitHub blob failed', code: 'BAD_GATEWAY' }); |
| 1579 | } |
| 1580 | const blob = await blobRes.json(); |
| 1581 | tree.push({ path: '.knowtation/README.md', mode: '100644', type: 'blob', sha: blob.sha }); |
| 1582 | } |
| 1583 | |
| 1584 | const treePayload = baseTreeSha ? { base_tree: baseTreeSha, tree } : { tree }; |
| 1585 | const treeRes = await fetch(`${ghApi}/repos/${owner}/${name}/git/trees`, { |
| 1586 | method: 'POST', |
| 1587 | headers: ghHeaders, |
| 1588 | body: JSON.stringify(treePayload), |
| 1589 | }); |
| 1590 | if (!treeRes.ok) { |
| 1591 | return res.status(502).json({ error: 'GitHub tree failed', code: 'BAD_GATEWAY' }); |
| 1592 | } |
| 1593 | const newTree = await treeRes.json(); |
| 1594 | |
| 1595 | const commitRes = await fetch(`${ghApi}/repos/${owner}/${name}/git/commits`, { |
| 1596 | method: 'POST', |
| 1597 | headers: ghHeaders, |
| 1598 | body: JSON.stringify({ |
| 1599 | message: 'Knowtation Hub backup ' + new Date().toISOString(), |
| 1600 | tree: newTree.sha, |
| 1601 | parents: baseSha ? [baseSha] : [], |
| 1602 | }), |
| 1603 | }); |
| 1604 | if (!commitRes.ok) { |
| 1605 | return res.status(502).json({ error: 'GitHub commit failed', code: 'BAD_GATEWAY' }); |
| 1606 | } |
| 1607 | const newCommit = await commitRes.json(); |
| 1608 | |
| 1609 | let refUpdateRes; |
| 1610 | if (baseSha) { |
| 1611 | refUpdateRes = await fetch(`${ghApi}/repos/${owner}/${name}/git/refs/${headsRefEnc(defaultBranch)}`, { |
| 1612 | method: 'PATCH', |
| 1613 | headers: ghHeaders, |
| 1614 | body: JSON.stringify({ sha: newCommit.sha, force: false }), |
| 1615 | }); |
| 1616 | } else { |
| 1617 | refUpdateRes = await fetch(`${ghApi}/repos/${owner}/${name}/git/refs`, { |
| 1618 | method: 'POST', |
| 1619 | headers: ghHeaders, |
| 1620 | body: JSON.stringify({ ref: `refs/heads/${defaultBranch}`, sha: newCommit.sha }), |
| 1621 | }); |
| 1622 | } |
| 1623 | if (!refUpdateRes.ok) { |
| 1624 | return res.status(502).json({ error: 'GitHub push failed', code: 'BAD_GATEWAY' }); |
| 1625 | } |
| 1626 | |
| 1627 | res.json({ |
| 1628 | ok: true, |
| 1629 | message: 'Synced', |
| 1630 | notesCount: notes.length, |
| 1631 | proposalsCount: proposals.length, |
| 1632 | }); |
| 1633 | }); |
| 1634 | |
| 1635 | // Optional: GET status for Settings (connected + repo) |
| 1636 | app.get('/api/v1/vault/github-status', async (req, res) => { |
| 1637 | const auth = req.headers.authorization; |
| 1638 | const token = auth && auth.startsWith('Bearer ') ? auth.slice(7) : null; |
| 1639 | const uid = token ? userIdFromJwt(token) : null; |
| 1640 | if (!uid) { |
| 1641 | return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 1642 | } |
| 1643 | const tokensByUser = await loadTokens(req.blobStore); |
| 1644 | const conn = tokensByUser[uid]; |
| 1645 | res.json({ |
| 1646 | github_connected: Boolean(conn?.token), |
| 1647 | repo: conn?.repo || null, |
| 1648 | }); |
| 1649 | }); |
| 1650 | |
| 1651 | // Internal: GET GitHub connection (token + repo) for the gateway to use for image upload. |
| 1652 | // Server-to-server only — never exposed to the browser. Auth required. |
| 1653 | app.get('/api/v1/vault/github-token', async (req, res) => { |
| 1654 | const auth = req.headers.authorization; |
| 1655 | const token = auth && auth.startsWith('Bearer ') ? auth.slice(7) : null; |
| 1656 | const uid = token ? userIdFromJwt(token) : null; |
| 1657 | if (!uid) { |
| 1658 | return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 1659 | } |
| 1660 | try { |
| 1661 | const tokensByUser = await loadTokens(req.blobStore); |
| 1662 | const conn = tokensByUser[uid]; |
| 1663 | if (!conn?.token) { |
| 1664 | return res.status(400).json({ error: 'GitHub not connected', code: 'GITHUB_NOT_CONNECTED' }); |
| 1665 | } |
| 1666 | res.json({ token: conn.token, repo: conn.repo || null }); |
| 1667 | } catch (e) { |
| 1668 | res.status(500).json({ error: e.message || 'Internal error', code: 'INTERNAL_ERROR' }); |
| 1669 | } |
| 1670 | }); |
| 1671 | |
| 1672 | /** Max notes per canister POST /api/v1/notes/batch (must match hub/icp NOTES_BATCH cap). */ |
| 1673 | const CANISTER_NOTES_BATCH_MAX = 100; |
| 1674 | |
| 1675 | /** |
| 1676 | * @param {string} canisterUid |
| 1677 | * @param {string} actorUid |
| 1678 | * @param {string} vaultId |
| 1679 | * @param {{ path: string, body: string, frontmatter?: Record<string, unknown> }[]} notes |
| 1680 | */ |
| 1681 | async function postNotesBatchToCanister(canisterUid, actorUid, vaultId, notes) { |
| 1682 | if (!notes.length) return; |
| 1683 | for (let offset = 0; offset < notes.length; offset += CANISTER_NOTES_BATCH_MAX) { |
| 1684 | const chunk = notes.slice(offset, offset + CANISTER_NOTES_BATCH_MAX); |
| 1685 | const r = await fetch(CANISTER_URL + '/api/v1/notes/batch', { |
| 1686 | method: 'POST', |
| 1687 | headers: canisterHeaders({ |
| 1688 | 'Content-Type': 'application/json', |
| 1689 | 'X-User-Id': canisterUid, |
| 1690 | 'X-Actor-Id': actorUid, |
| 1691 | 'X-Vault-Id': vaultId, |
| 1692 | }), |
| 1693 | body: JSON.stringify({ notes: chunk }), |
| 1694 | }); |
| 1695 | const text = await r.text(); |
| 1696 | if (!r.ok) { |
| 1697 | throw new Error(`Canister batch note write failed (${r.status}): ${text.slice(0, 800)}`); |
| 1698 | } |
| 1699 | } |
| 1700 | } |
| 1701 | |
| 1702 | /** |
| 1703 | * Sanitize a user-supplied filename before writing it to disk. |
| 1704 | * - Strips all directory components (path traversal prevention). |
| 1705 | * - Removes every character that is not alphanumeric, a dot, hyphen, or underscore. |
| 1706 | * - Truncates to 200 chars so filesystem limits are never approached. |
| 1707 | * - Falls back to 'upload' when the result would be empty. |
| 1708 | */ |
| 1709 | function sanitizeUploadFilename(rawName) { |
| 1710 | const base = path.basename(rawName || ''); |
| 1711 | const safe = base.replace(/[^a-zA-Z0-9._-]/g, '_').slice(0, 200); |
| 1712 | return safe || 'upload'; |
| 1713 | } |
| 1714 | |
| 1715 | const importTempDirMiddleware = (req, _res, next) => { |
| 1716 | req._importTempDir = fs.mkdtempSync(path.join(os.tmpdir(), 'knowtation-bridge-import-')); |
| 1717 | next(); |
| 1718 | }; |
| 1719 | const bridgeImportUpload = multer({ |
| 1720 | storage: multer.diskStorage({ |
| 1721 | destination: (req, _file, cb) => cb(null, req._importTempDir), |
| 1722 | filename: (_req, file, cb) => cb(null, sanitizeUploadFilename(file.originalname)), |
| 1723 | }), |
| 1724 | limits: { fileSize: 100 * 1024 * 1024 }, |
| 1725 | }).single('file'); |
| 1726 | |
| 1727 | // ——— Phase 18: GitHub image upload + image proxy ——— |
| 1728 | |
| 1729 | const bridgeImageUpload = multer({ |
| 1730 | storage: multer.memoryStorage(), |
| 1731 | limits: { fileSize: 25 * 1024 * 1024 }, |
| 1732 | }).single('image'); |
| 1733 | |
| 1734 | app.post( |
| 1735 | /^\/api\/v1\/notes\/(.+)\/upload-image$/, |
| 1736 | requireBridgeAuth, |
| 1737 | requireBridgeEditorOrAdmin, |
| 1738 | bridgeImageUpload, |
| 1739 | async (req, res) => { |
| 1740 | try { |
| 1741 | if (!req.file) return res.status(400).json({ error: 'image file required', code: 'BAD_REQUEST' }); |
| 1742 | |
| 1743 | const originalName = req.file.originalname || 'image.jpg'; |
| 1744 | try { validateImageExtension(originalName); } catch (e) { |
| 1745 | return res.status(400).json({ error: e.message, code: 'BAD_REQUEST' }); |
| 1746 | } |
| 1747 | if (!(req.file.mimetype || '').toLowerCase().startsWith('image/')) { |
| 1748 | return res.status(400).json({ error: 'File content-type must be image/*', code: 'BAD_REQUEST' }); |
| 1749 | } |
| 1750 | const ext = originalName.split('.').pop().toLowerCase(); |
| 1751 | try { validateMagicBytes(req.file.buffer, ext); } catch (e) { |
| 1752 | return res.status(400).json({ error: e.message, code: 'BAD_REQUEST' }); |
| 1753 | } |
| 1754 | |
| 1755 | const tokensByUser = await loadTokens(req.blobStore); |
| 1756 | const conn = tokensByUser[req.uid]; |
| 1757 | if (!conn?.token) { |
| 1758 | return res.status(400).json({ error: 'GitHub not connected. Go to Settings → Backup → Connect GitHub.', code: 'GITHUB_NOT_CONNECTED' }); |
| 1759 | } |
| 1760 | if (!conn.repo) { |
| 1761 | return res.status(400).json({ error: 'GitHub repo not set. Back up once first to set the remote.', code: 'GITHUB_NOT_CONFIGURED' }); |
| 1762 | } |
| 1763 | |
| 1764 | const now = new Date(); |
| 1765 | const yearMonth = `${now.getFullYear()}-${String(now.getMonth() + 1).padStart(2, '0')}`; |
| 1766 | const safeName = originalName.replace(/[^a-zA-Z0-9._-]/g, '_').slice(0, 128); |
| 1767 | const uniqueName = `${Date.now()}-${safeName}`; |
| 1768 | const repoFilePath = `media/images/${yearMonth}/${uniqueName}`; |
| 1769 | |
| 1770 | const result = await commitImageToRepo({ |
| 1771 | accessToken: conn.token, |
| 1772 | repoUrl: conn.repo, |
| 1773 | filePath: repoFilePath, |
| 1774 | fileBuffer: req.file.buffer, |
| 1775 | commitMessage: `Add image: ${safeName}`, |
| 1776 | }); |
| 1777 | |
| 1778 | res.json({ |
| 1779 | url: result.url, |
| 1780 | inserted_markdown: ``, |
| 1781 | sha: result.sha, |
| 1782 | repo_path: repoFilePath, |
| 1783 | repo_private: result.isPrivate === true, |
| 1784 | }); |
| 1785 | } catch (e) { |
| 1786 | console.error('[bridge] upload-image error:', e?.message); |
| 1787 | const msg = e.message || String(e); |
| 1788 | const clientErr = /not found|not connected|lacks permission|lacks repo|Reconnect|scope|remote/i.test(msg); |
| 1789 | res.status(clientErr ? 400 : 500).json({ error: msg, code: clientErr ? 'BAD_REQUEST' : 'RUNTIME_ERROR' }); |
| 1790 | } |
| 1791 | } |
| 1792 | ); |
| 1793 | |
| 1794 | // Image proxy: serve raw.githubusercontent.com images via the stored GitHub token. |
| 1795 | // Accepts JWT via ?token= query param (browsers cannot send headers for <img> tags). |
| 1796 | const BRIDGE_IMAGE_PROXY_SIZE_LIMIT = 10 * 1024 * 1024; |
| 1797 | app.get('/api/v1/vault/image-proxy', async (req, res) => { |
| 1798 | const auth = req.headers.authorization; |
| 1799 | const headerToken = auth && auth.startsWith('Bearer ') ? auth.slice(7) : null; |
| 1800 | const queryToken = typeof req.query.token === 'string' ? req.query.token : null; |
| 1801 | const uid = (headerToken || queryToken) ? userIdFromJwt(headerToken || queryToken) : null; |
| 1802 | if (!uid) return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 1803 | |
| 1804 | const rawUrl = typeof req.query.url === 'string' ? req.query.url : ''; |
| 1805 | if (!/^https:\/\/raw\.githubusercontent\.com\/[^/]+\/[^/]+\/.+$/i.test(rawUrl)) { |
| 1806 | return res.status(400).json({ error: 'url must be a raw.githubusercontent.com path', code: 'BAD_REQUEST' }); |
| 1807 | } |
| 1808 | |
| 1809 | let accessToken = ''; |
| 1810 | try { |
| 1811 | const tokensByUser = await loadTokens(req.blobStore); |
| 1812 | const conn = tokensByUser[uid]; |
| 1813 | if (conn?.token) accessToken = conn.token; |
| 1814 | } catch (_) {} |
| 1815 | |
| 1816 | const fetchHeaders = { 'User-Agent': 'Knowtation-Hub/1.0' }; |
| 1817 | if (accessToken) fetchHeaders.Authorization = `token ${accessToken}`; |
| 1818 | |
| 1819 | let upstream; |
| 1820 | try { |
| 1821 | upstream = await fetch(rawUrl, { headers: fetchHeaders }); |
| 1822 | } catch (e) { |
| 1823 | return res.status(502).json({ error: 'Failed to fetch image from GitHub', code: 'UPSTREAM_ERROR' }); |
| 1824 | } |
| 1825 | if (!upstream.ok) { |
| 1826 | return res.status(upstream.status).json({ error: 'Image not found on GitHub', code: 'UPSTREAM_ERROR' }); |
| 1827 | } |
| 1828 | const ct = upstream.headers.get('content-type') || ''; |
| 1829 | if (!ct.startsWith('image/')) { |
| 1830 | return res.status(400).json({ error: 'URL does not point to an image', code: 'BAD_REQUEST' }); |
| 1831 | } |
| 1832 | const buf = Buffer.from(await upstream.arrayBuffer()); |
| 1833 | if (buf.byteLength > BRIDGE_IMAGE_PROXY_SIZE_LIMIT) { |
| 1834 | return res.status(400).json({ error: 'Image too large (max 10 MB)', code: 'BAD_REQUEST' }); |
| 1835 | } |
| 1836 | res.setHeader('Content-Type', ct); |
| 1837 | res.setHeader('Content-Length', buf.byteLength); |
| 1838 | res.setHeader('Cache-Control', 'private, max-age=3600'); |
| 1839 | res.setHeader('X-Content-Type-Options', 'nosniff'); |
| 1840 | res.send(buf); |
| 1841 | }); |
| 1842 | |
| 1843 | app.post( |
| 1844 | '/api/v1/import', |
| 1845 | requireBridgeAuth, |
| 1846 | requireBridgeEditorOrAdmin, |
| 1847 | importTempDirMiddleware, |
| 1848 | bridgeImportUpload, |
| 1849 | async (req, res) => { |
| 1850 | const tempDir = req._importTempDir; |
| 1851 | try { |
| 1852 | if (!CANISTER_URL) { |
| 1853 | return res.status(503).json({ error: 'Canister not configured', code: 'SERVICE_UNAVAILABLE' }); |
| 1854 | } |
| 1855 | const sourceType = req.body && req.body.source_type ? String(req.body.source_type).trim() : ''; |
| 1856 | if (!IMPORT_SOURCE_TYPES.includes(sourceType)) { |
| 1857 | return res.status(400).json({ |
| 1858 | error: `source_type must be one of: ${IMPORT_SOURCE_TYPES.join(', ')}`, |
| 1859 | code: 'BAD_REQUEST', |
| 1860 | }); |
| 1861 | } |
| 1862 | const sheetId = req.body && req.body.spreadsheet_id ? String(req.body.spreadsheet_id).trim() : ''; |
| 1863 | const sheetsRange = req.body && req.body.sheets_range ? String(req.body.sheets_range).trim() : undefined; |
| 1864 | if (sourceType === 'google-sheets') { |
| 1865 | if (!sheetId) { |
| 1866 | return res |
| 1867 | .status(400) |
| 1868 | .json({ error: 'google-sheets: spreadsheet_id is required in the multipart body', code: 'BAD_REQUEST' }); |
| 1869 | } |
| 1870 | if (req.file) { |
| 1871 | return res |
| 1872 | .status(400) |
| 1873 | .json({ error: 'google-sheets: do not send a file; use spreadsheet_id only', code: 'BAD_REQUEST' }); |
| 1874 | } |
| 1875 | } else if (!req.file) { |
| 1876 | return res.status(400).json({ error: 'file required', code: 'BAD_REQUEST' }); |
| 1877 | } |
| 1878 | const project = req.body && req.body.project ? String(req.body.project).trim() : undefined; |
| 1879 | const outputDir = req.body && req.body.output_dir ? String(req.body.output_dir).trim() : undefined; |
| 1880 | const tagsRaw = req.body && req.body.tags ? String(req.body.tags) : ''; |
| 1881 | const tags = tagsRaw ? tagsRaw.split(',').map((s) => s.trim()).filter(Boolean) : []; |
| 1882 | let inputPath = sourceType === 'google-sheets' ? sheetId : req.file.path; |
| 1883 | if (sourceType !== 'google-sheets' && req.file && req.file.originalname && req.file.originalname.toLowerCase().endsWith('.zip')) { |
| 1884 | const extractDir = path.join(tempDir, 'extracted'); |
| 1885 | fs.mkdirSync(extractDir, { recursive: true }); |
| 1886 | const zip = new AdmZip(req.file.path); |
| 1887 | // Zip-slip protection: every entry must resolve inside extractDir |
| 1888 | const extractDirResolved = path.resolve(extractDir) + path.sep; |
| 1889 | for (const entry of zip.getEntries()) { |
| 1890 | const entryResolved = path.resolve(extractDir, entry.entryName); |
| 1891 | if (entryResolved !== path.resolve(extractDir) && !entryResolved.startsWith(extractDirResolved)) { |
| 1892 | return res.status(400).json({ error: 'Invalid zip entry: path traversal detected', code: 'BAD_REQUEST' }); |
| 1893 | } |
| 1894 | } |
| 1895 | zip.extractAllTo(extractDir, true); |
| 1896 | inputPath = extractDir; |
| 1897 | } |
| 1898 | const hctx = await resolveHostedBridgeContext(req, req.uid); |
| 1899 | if (!hctx.ok) { |
| 1900 | return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 1901 | } |
| 1902 | const vaultPath = path.join(tempDir, 'vault-work'); |
| 1903 | fs.mkdirSync(vaultPath, { recursive: true }); |
| 1904 | const result = await runImport(sourceType, inputPath, { |
| 1905 | project, |
| 1906 | outputDir, |
| 1907 | tags, |
| 1908 | vaultPath, |
| 1909 | ...(sheetsRange ? { sheetsRange } : {}), |
| 1910 | }); |
| 1911 | const importStamp = mergeProvenanceFrontmatter({}, { |
| 1912 | sub: hctx.actorUid, |
| 1913 | kind: 'import', |
| 1914 | }); |
| 1915 | /** @type {{ path: string, body: string, frontmatter: Record<string, unknown> }[]} */ |
| 1916 | const notesForCanister = []; |
| 1917 | for (const item of result.imported || []) { |
| 1918 | if (item.path && typeof item.path === 'string') { |
| 1919 | try { |
| 1920 | writeNote(vaultPath, item.path, { frontmatter: importStamp }); |
| 1921 | const safe = resolveVaultRelativePath(vaultPath, item.path); |
| 1922 | const fullPath = path.join(vaultPath, safe); |
| 1923 | const markdownFull = fs.readFileSync(fullPath, 'utf8'); |
| 1924 | const parsed = parseFrontmatterAndBody(markdownFull); |
| 1925 | const fm = |
| 1926 | parsed.frontmatter && typeof parsed.frontmatter === 'object' && !Array.isArray(parsed.frontmatter) |
| 1927 | ? /** @type {Record<string, unknown>} */ ({ ...parsed.frontmatter }) |
| 1928 | : {}; |
| 1929 | notesForCanister.push({ |
| 1930 | path: safe.replace(/\\/g, '/'), |
| 1931 | body: parsed.body || '', |
| 1932 | frontmatter: fm, |
| 1933 | }); |
| 1934 | } catch (e) { |
| 1935 | console.error('[bridge] import prepare note for canister failed for', item.path, e?.message || e); |
| 1936 | return res.status(502).json({ |
| 1937 | error: e.message || 'Canister write failed', |
| 1938 | code: 'BAD_GATEWAY', |
| 1939 | }); |
| 1940 | } |
| 1941 | } |
| 1942 | } |
| 1943 | try { |
| 1944 | await postNotesBatchToCanister( |
| 1945 | hctx.effectiveCanisterUid, |
| 1946 | hctx.actorUid, |
| 1947 | hctx.vaultId, |
| 1948 | notesForCanister, |
| 1949 | ); |
| 1950 | } catch (e) { |
| 1951 | console.error('[bridge] import canister batch write failed', e?.message || e); |
| 1952 | return res.status(502).json({ |
| 1953 | error: e.message || 'Canister write failed', |
| 1954 | code: 'BAD_GATEWAY', |
| 1955 | }); |
| 1956 | } |
| 1957 | return res.json({ imported: result.imported, count: result.count }); |
| 1958 | } catch (e) { |
| 1959 | const msg = e.message || String(e); |
| 1960 | const clientError = |
| 1961 | /OPENAI_API_KEY|required for transcription|Unsupported format|file not found|not found:|Transcription failed|413|Payload Too Large|25MB|Whisper accepts|Only https|blocked|private IP|timed out|exceeds \d+ bytes|Invalid URL|URL is required|Extract mode requires|Could not extract|DNS resolution failed|Too many redirects|non-https/i.test( |
| 1962 | msg, |
| 1963 | ); |
| 1964 | res.status(clientError ? 400 : 500).json({ |
| 1965 | error: msg, |
| 1966 | code: clientError ? 'BAD_REQUEST' : 'RUNTIME_ERROR', |
| 1967 | }); |
| 1968 | } finally { |
| 1969 | if (tempDir && fs.existsSync(tempDir)) { |
| 1970 | try { |
| 1971 | fs.rmSync(tempDir, { recursive: true, force: true }); |
| 1972 | } catch (_) {} |
| 1973 | } |
| 1974 | } |
| 1975 | }, |
| 1976 | ); |
| 1977 | |
| 1978 | /** |
| 1979 | * @param {unknown} raw |
| 1980 | * @returns {'auto' | 'bookmark' | 'extract'} |
| 1981 | */ |
| 1982 | function normalizeBridgeImportUrlMode(raw) { |
| 1983 | const s = typeof raw === 'string' ? raw.trim().toLowerCase() : ''; |
| 1984 | if (s === 'bookmark' || s === 'extract' || s === 'auto') return s; |
| 1985 | return 'auto'; |
| 1986 | } |
| 1987 | |
| 1988 | /** |
| 1989 | * @param {unknown} body |
| 1990 | * @returns {string[]} |
| 1991 | */ |
| 1992 | function tagsFromBridgeImportUrlBody(body) { |
| 1993 | const t = body && body.tags; |
| 1994 | if (Array.isArray(t)) return t.map((x) => String(x).trim()).filter(Boolean); |
| 1995 | if (typeof t === 'string') return t.split(',').map((s) => s.trim()).filter(Boolean); |
| 1996 | return []; |
| 1997 | } |
| 1998 | |
| 1999 | app.post('/api/v1/import-url', requireBridgeAuth, requireBridgeEditorOrAdmin, async (req, res) => { |
| 2000 | const tempDir = fs.mkdtempSync(path.join(os.tmpdir(), 'knowtation-bridge-import-url-')); |
| 2001 | try { |
| 2002 | if (!CANISTER_URL) { |
| 2003 | return res.status(503).json({ error: 'Canister not configured', code: 'SERVICE_UNAVAILABLE' }); |
| 2004 | } |
| 2005 | const body = req.body && typeof req.body === 'object' ? req.body : {}; |
| 2006 | const urlStr = typeof body.url === 'string' ? body.url.trim() : ''; |
| 2007 | if (!urlStr) return res.status(400).json({ error: 'url required', code: 'BAD_REQUEST' }); |
| 2008 | const urlMode = normalizeBridgeImportUrlMode(body.mode); |
| 2009 | const project = body.project != null && String(body.project).trim() !== '' ? String(body.project).trim() : undefined; |
| 2010 | const outputDir = |
| 2011 | body.output_dir != null && String(body.output_dir).trim() !== '' ? String(body.output_dir).trim() : undefined; |
| 2012 | const tags = tagsFromBridgeImportUrlBody(body); |
| 2013 | |
| 2014 | const hctx = await resolveHostedBridgeContext(req, req.uid); |
| 2015 | if (!hctx.ok) { |
| 2016 | return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 2017 | } |
| 2018 | const vaultPath = path.join(tempDir, 'vault-work'); |
| 2019 | fs.mkdirSync(vaultPath, { recursive: true }); |
| 2020 | const result = await runImport('url', urlStr, { project, outputDir, tags, vaultPath, urlMode }); |
| 2021 | const importStamp = mergeProvenanceFrontmatter({}, { |
| 2022 | sub: hctx.actorUid, |
| 2023 | kind: 'import', |
| 2024 | }); |
| 2025 | /** @type {{ path: string, body: string, frontmatter: Record<string, unknown> }[]} */ |
| 2026 | const notesForCanister = []; |
| 2027 | for (const item of result.imported || []) { |
| 2028 | if (item.path && typeof item.path === 'string') { |
| 2029 | try { |
| 2030 | writeNote(vaultPath, item.path, { frontmatter: importStamp }); |
| 2031 | const safe = resolveVaultRelativePath(vaultPath, item.path); |
| 2032 | const fullPath = path.join(vaultPath, safe); |
| 2033 | const markdownFull = fs.readFileSync(fullPath, 'utf8'); |
| 2034 | const parsed = parseFrontmatterAndBody(markdownFull); |
| 2035 | const fm = |
| 2036 | parsed.frontmatter && typeof parsed.frontmatter === 'object' && !Array.isArray(parsed.frontmatter) |
| 2037 | ? /** @type {Record<string, unknown>} */ ({ ...parsed.frontmatter }) |
| 2038 | : {}; |
| 2039 | notesForCanister.push({ |
| 2040 | path: safe.replace(/\\/g, '/'), |
| 2041 | body: parsed.body || '', |
| 2042 | frontmatter: fm, |
| 2043 | }); |
| 2044 | } catch (e) { |
| 2045 | console.error('[bridge] import-url prepare note for canister failed for', item.path, e?.message || e); |
| 2046 | return res.status(502).json({ |
| 2047 | error: e.message || 'Canister write failed', |
| 2048 | code: 'BAD_GATEWAY', |
| 2049 | }); |
| 2050 | } |
| 2051 | } |
| 2052 | } |
| 2053 | try { |
| 2054 | await postNotesBatchToCanister(hctx.effectiveCanisterUid, hctx.actorUid, hctx.vaultId, notesForCanister); |
| 2055 | } catch (e) { |
| 2056 | console.error('[bridge] import-url canister batch write failed', e?.message || e); |
| 2057 | return res.status(502).json({ |
| 2058 | error: e.message || 'Canister write failed', |
| 2059 | code: 'BAD_GATEWAY', |
| 2060 | }); |
| 2061 | } |
| 2062 | return res.json({ imported: result.imported, count: result.count }); |
| 2063 | } catch (e) { |
| 2064 | const msg = e.message || String(e); |
| 2065 | const clientError = |
| 2066 | /OPENAI_API_KEY|required for transcription|Unsupported format|file not found|not found:|Transcription failed|413|Payload Too Large|25MB|Whisper accepts|Only https|blocked|private IP|timed out|exceeds \d+ bytes|Invalid URL|URL is required|Extract mode requires|Could not extract|DNS resolution failed|Too many redirects|non-https/i.test( |
| 2067 | msg, |
| 2068 | ); |
| 2069 | res.status(clientError ? 400 : 500).json({ |
| 2070 | error: msg, |
| 2071 | code: clientError ? 'BAD_REQUEST' : 'RUNTIME_ERROR', |
| 2072 | }); |
| 2073 | } finally { |
| 2074 | if (tempDir && fs.existsSync(tempDir)) { |
| 2075 | try { |
| 2076 | fs.rmSync(tempDir, { recursive: true, force: true }); |
| 2077 | } catch (_) {} |
| 2078 | } |
| 2079 | } |
| 2080 | }); |
| 2081 | |
| 2082 | // ——— Index + Search (hosted: indexer runs in bridge, canister does not run Node) ——— |
| 2083 | // BATCH_EMBED + INDEXER_EMBED_CONCURRENCY together drive how much wall time the index |
| 2084 | // step takes against Netlify's 60 s sync-function cap. With DeepInfra (BAAI/bge-large-en-v1.5) |
| 2085 | // per-batch latencies trending 2.5 – 8.5 s, the previous serial loop at BATCH=10 ran ~65 – 80 s |
| 2086 | // for a 251-chunk vault and got killed. Defaults below (BATCH=50, CONCURRENCY=5) bring that |
| 2087 | // same vault under ~10 – 15 s when a full re-embed is needed; the content-hash cache below |
| 2088 | // makes subsequent re-indexes a few seconds regardless of vault size. |
| 2089 | const BATCH_EMBED_DEFAULT = parseEmbedBatchSize(process.env.INDEXER_EMBED_BATCH_SIZE); |
| 2090 | const EMBED_CONCURRENCY_DEFAULT = parseEmbedConcurrency(process.env.INDEXER_EMBED_CONCURRENCY); |
| 2091 | const BATCH_UPSERT = 50; |
| 2092 | const SYNC_BUDGET_SECONDS = parseSyncBudgetSeconds(process.env.INDEXER_SYNC_BUDGET_SECONDS); |
| 2093 | const MAX_SYNC_CHUNKS = parseMaxSyncChunks(process.env.INDEXER_MAX_SYNC_CHUNKS); |
| 2094 | |
| 2095 | /** |
| 2096 | * Kick off the `bridge-index-background` Netlify Function. Used by the |
| 2097 | * synchronous `POST /api/v1/index` handler when the preflight estimate exceeds |
| 2098 | * the sync budget (`SYNC_BUDGET_SECONDS`) or the chunk-count safety net |
| 2099 | * (`MAX_SYNC_CHUNKS`). The background function returns 202 instantly and runs |
| 2100 | * up to 15 min in a separate Lambda; this fetch only waits for that 202 so the |
| 2101 | * sync handler can return its own 202 to the browser without blocking on the |
| 2102 | * actual embed work. |
| 2103 | * |
| 2104 | * Two layers of auth on the inbound side (see `lib/bridge-internal-hmac.mjs`): |
| 2105 | * 1. The user JWT is forwarded verbatim so the background function still |
| 2106 | * runs `requireBridgeAuth` and the user must be a real authenticated user. |
| 2107 | * 2. An HMAC signature over (canisterUid, vaultId, jobId, ts) signed with |
| 2108 | * `SESSION_SECRET` proves the request originated from this sync handler |
| 2109 | * (the background URL is publicly addressable on Netlify). |
| 2110 | */ |
| 2111 | async function kickOffBackgroundIndex(req, jobId, canisterUid, vaultId) { |
| 2112 | if (!SESSION_SECRET) { |
| 2113 | throw new Error( |
| 2114 | 'kickOffBackgroundIndex: SESSION_SECRET is not set; cannot sign internal request', |
| 2115 | ); |
| 2116 | } |
| 2117 | const ts = Date.now(); |
| 2118 | const sig = signInternalRequest(SESSION_SECRET, { canisterUid, vaultId, jobId, ts }); |
| 2119 | const protocol = |
| 2120 | req.protocol || |
| 2121 | (req.headers['x-forwarded-proto'] && String(req.headers['x-forwarded-proto']).split(',')[0]) || |
| 2122 | 'https'; |
| 2123 | const host = (req.get && req.get('host')) || req.headers.host; |
| 2124 | if (!host) { |
| 2125 | throw new Error('kickOffBackgroundIndex: cannot determine host header for background URL'); |
| 2126 | } |
| 2127 | const url = `${protocol}://${host}/.netlify/functions/bridge-index-background`; |
| 2128 | const auth = req.headers.authorization || ''; |
| 2129 | // Background functions on Netlify return 202 within ~50–100 ms regardless of |
| 2130 | // what the function body does; we only await that 202 so the sync handler |
| 2131 | // can immediately return its own 202 to the browser. |
| 2132 | // |
| 2133 | // CRITICAL (May 2026 hotfix): we MUST inspect `response.status`. fetch() |
| 2134 | // resolves successfully on 4xx/5xx responses (it only throws on network |
| 2135 | // errors), so without this check a non-202 response (function not deployed, |
| 2136 | // wrong host header, HMAC misconfiguration, future routing bug, etc.) would |
| 2137 | // be silently treated as success — the sync handler would return |
| 2138 | // `202 status:"background"` to the browser while no work runs in the |
| 2139 | // background. The job lock would then sit for its full 16-min TTL blocking |
| 2140 | // any retry. See `lib/bridge-index-kickoff-response.mjs` for full context |
| 2141 | // and the failure-mode test matrix in |
| 2142 | // `test/bridge-index-kickoff-response.test.mjs`. |
| 2143 | const response = await fetch(url, { |
| 2144 | method: 'POST', |
| 2145 | headers: { |
| 2146 | 'content-type': 'application/json', |
| 2147 | authorization: auth, |
| 2148 | 'x-vault-id': vaultId, |
| 2149 | 'x-bridge-internal-uid': canisterUid, |
| 2150 | 'x-bridge-internal-vault-id': vaultId, |
| 2151 | 'x-bridge-internal-job-id': jobId, |
| 2152 | 'x-bridge-internal-ts': String(ts), |
| 2153 | 'x-bridge-internal-sig': sig, |
| 2154 | }, |
| 2155 | body: '{}', |
| 2156 | }); |
| 2157 | let body = ''; |
| 2158 | try { |
| 2159 | body = await response.text(); |
| 2160 | } catch (_) { |
| 2161 | // body read failure is non-fatal here — the helper accepts undefined body |
| 2162 | // and the status code alone is sufficient to detect the failure mode. |
| 2163 | } |
| 2164 | assertBackgroundKickoffOk(response, body); |
| 2165 | } |
| 2166 | |
| 2167 | /** |
| 2168 | * Read-only snapshot of "is the index for this vault currently being rebuilt |
| 2169 | * in the background, and when did it last finish successfully?". The Hub UI |
| 2170 | * polls this on page load to render `Last indexed: 2 minutes ago` next to the |
| 2171 | * Re-index button and to disable the button while a background job is live. |
| 2172 | * |
| 2173 | * Same auth + scope as `POST /api/v1/index`: the user must be authenticated |
| 2174 | * AND have the vault in their effective hosted-bridge context. |
| 2175 | */ |
| 2176 | app.get('/api/v1/index/status', requireBridgeAuth, async (req, res) => { |
| 2177 | const hctx = await resolveHostedBridgeContext(req, req.uid); |
| 2178 | if (!hctx.ok) return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 2179 | const canisterUid = hctx.effectiveCanisterUid; |
| 2180 | const vaultId = sanitizeVaultId(req.headers['x-vault-id']); |
| 2181 | const lastIndexed = req.blobStore |
| 2182 | ? await getLastIndexedAt(req.blobStore, { canisterUid, vaultId }) |
| 2183 | : null; |
| 2184 | const jobLock = req.blobStore |
| 2185 | ? await peekJobLock(req.blobStore, { canisterUid, vaultId }) |
| 2186 | : null; |
| 2187 | const inProgress = |
| 2188 | jobLock != null && |
| 2189 | Number.isFinite(jobLock.expiresAt) && |
| 2190 | jobLock.expiresAt > Date.now(); |
| 2191 | res.json({ |
| 2192 | lastIndexed, |
| 2193 | inProgress, |
| 2194 | job: inProgress ? jobLock : null, |
| 2195 | }); |
| 2196 | }); |
| 2197 | |
| 2198 | app.post('/api/v1/index', requireBridgeAuth, requireBridgeEditorOrAdmin, async (req, res) => { |
| 2199 | const uid = req.uid; |
| 2200 | const earlyVaultId = sanitizeVaultId(req.headers['x-vault-id']); |
| 2201 | const timer = createIndexTimer({ vaultId: earlyVaultId, canisterUid: null }); |
| 2202 | const hctx = await resolveHostedBridgeContext(req, uid); |
| 2203 | timer.step('resolve_context', { ok: hctx.ok }); |
| 2204 | if (!hctx.ok) { |
| 2205 | timer.finish({ ok: false, phase: 'resolve_context', status: hctx.status, code: hctx.code }); |
| 2206 | return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 2207 | } |
| 2208 | const canisterUid = hctx.effectiveCanisterUid; |
| 2209 | const vaultId = sanitizeVaultId(req.headers['x-vault-id']); |
| 2210 | let exportRes; |
| 2211 | try { |
| 2212 | exportRes = await fetch(CANISTER_URL + '/api/v1/export', { |
| 2213 | method: 'GET', |
| 2214 | headers: canisterHeaders({ 'X-User-Id': canisterUid, 'X-Vault-Id': vaultId }), |
| 2215 | }); |
| 2216 | } catch (e) { |
| 2217 | timer.finish({ ok: false, phase: 'canister_export_fetch', error: e?.message || String(e) }); |
| 2218 | return res.status(502).json({ error: 'Could not reach canister', code: 'BAD_GATEWAY' }); |
| 2219 | } |
| 2220 | if (!exportRes.ok) { |
| 2221 | timer.finish({ ok: false, phase: 'canister_export_status', status: exportRes.status }); |
| 2222 | return res.status(502).json({ error: 'Canister export failed', code: 'BAD_GATEWAY', status: exportRes.status }); |
| 2223 | } |
| 2224 | let vault; |
| 2225 | try { |
| 2226 | vault = await exportRes.json(); |
| 2227 | } catch (_) { |
| 2228 | timer.finish({ ok: false, phase: 'canister_export_parse' }); |
| 2229 | return res.status(502).json({ error: 'Invalid canister response', code: 'BAD_GATEWAY' }); |
| 2230 | } |
| 2231 | let notes = vault.notes || []; |
| 2232 | timer.step('canister_export', { note_count: notes.length, scoped: Boolean(hctx.scope) }); |
| 2233 | if (hctx.scope) { |
| 2234 | notes = applyScopeFilterToNotes(notes, hctx.scope); |
| 2235 | timer.step('scope_filter', { note_count_after: notes.length }); |
| 2236 | } |
| 2237 | try { |
| 2238 | if (!globalThis.__knowtation_bridge_embed_logged) { |
| 2239 | globalThis.__knowtation_bridge_embed_logged = true; |
| 2240 | const c = getBridgeEmbeddingConfig(); |
| 2241 | const hasOpenAiKey = Boolean( |
| 2242 | process.env.OPENAI_API_KEY && String(process.env.OPENAI_API_KEY).trim(), |
| 2243 | ); |
| 2244 | console.log( |
| 2245 | '[bridge] embedding (no secrets):', |
| 2246 | JSON.stringify({ |
| 2247 | provider: c.provider, |
| 2248 | model: c.model, |
| 2249 | ollama_url_set: Boolean(process.env.OLLAMA_URL && String(process.env.OLLAMA_URL).trim()), |
| 2250 | openai_key_set: hasOpenAiKey, |
| 2251 | }), |
| 2252 | ); |
| 2253 | } |
| 2254 | const { chunkNote } = await import('../../lib/chunk.mjs'); |
| 2255 | const { embedWithUsage, embeddingDimension } = await import('../../lib/embedding.mjs'); |
| 2256 | const { createVectorStore } = await import('../../lib/vector-store.mjs'); |
| 2257 | timer.step('import_modules'); |
| 2258 | |
| 2259 | const vectorsDir = await getVectorsDirForUser(req, canisterUid); |
| 2260 | const storeConfig = getBridgeStoreConfig(canisterUid, vectorsDir); |
| 2261 | timer.step('get_vectors_dir'); |
| 2262 | const chunkOpts = resolveIndexerChunkOptions(process.env, storeConfig.embedding); |
| 2263 | const allChunks = []; |
| 2264 | for (const n of notes) { |
| 2265 | const note = { |
| 2266 | body: n.body || '', |
| 2267 | path: n.path || 'note.md', |
| 2268 | project: undefined, |
| 2269 | tags: [], |
| 2270 | date: undefined, |
| 2271 | }; |
| 2272 | const chunks = chunkNote(note, chunkOpts); |
| 2273 | for (const c of chunks) allChunks.push(c); |
| 2274 | } |
| 2275 | // Tag every chunk with a versioned content hash + the namespaced store id so the |
| 2276 | // sqlite-vec backend's `getChunkHashes(vaultId)` lookup keys line up. The hash prefix |
| 2277 | // includes the active provider+model so a same-dimension model swap (e.g. BGE-large 1024 |
| 2278 | // → BGE-m3 1024) automatically invalidates the cache instead of silently keeping stale |
| 2279 | // vectors. See `lib/chunk-content-hash.mjs:computeChunkContentHashTagged`. |
| 2280 | const embeddingConfigForHash = storeConfig.embedding; |
| 2281 | const chunksWithHash = allChunks.map((chunk) => ({ |
| 2282 | chunk, |
| 2283 | storeId: `${vaultId}::${chunk.id}`, |
| 2284 | contentHash: computeChunkContentHashTagged(chunk, embeddingConfigForHash), |
| 2285 | })); |
| 2286 | timer.step('chunk_notes', { chunk_count: allChunks.length }); |
| 2287 | |
| 2288 | const dim = embeddingDimension(storeConfig.embedding); |
| 2289 | const store = await createVectorStore(storeConfig); |
| 2290 | await store.ensureCollection(dim); |
| 2291 | timer.step('ensure_collection', { dim }); |
| 2292 | |
| 2293 | // Empty vault: drop everything for this vault and persist (covers note-deletion case). |
| 2294 | if (chunksWithHash.length === 0) { |
| 2295 | let vectors_deleted = 0; |
| 2296 | if (typeof store.deleteByVaultId === 'function') { |
| 2297 | vectors_deleted = await store.deleteByVaultId(vaultId); |
| 2298 | } |
| 2299 | timer.step('delete_old_vectors_empty', { vectors_deleted }); |
| 2300 | await persistVectorsToBlob(req, canisterUid, vectorsDir); |
| 2301 | timer.step('persist_vectors_empty'); |
| 2302 | // Sidecar update so the Hub UI's "Last indexed" line stays correct even after |
| 2303 | // an all-notes-deleted re-index (notes=0 is a legitimate steady state). |
| 2304 | if (req.blobStore) { |
| 2305 | try { |
| 2306 | await setLastIndexedAt(req.blobStore, { |
| 2307 | canisterUid, |
| 2308 | vaultId, |
| 2309 | actorUid: sanitizeUserId(uid), |
| 2310 | notesProcessed: notes.length, |
| 2311 | chunksIndexed: 0, |
| 2312 | chunksEmbedded: 0, |
| 2313 | chunksSkippedCached: 0, |
| 2314 | vectorsDeleted: vectors_deleted, |
| 2315 | embeddingInputTokens: 0, |
| 2316 | durationMs: timer.totalMs(), |
| 2317 | mode: req.bridgeInternalRequest != null ? 'background' : 'sync', |
| 2318 | provider: storeConfig.embedding?.provider || null, |
| 2319 | model: storeConfig.embedding?.model || null, |
| 2320 | }); |
| 2321 | } catch (sidecarErr) { |
| 2322 | // Sidecar write failure must not fail the index — UI just falls back to "never indexed". |
| 2323 | console.warn('[bridge] setLastIndexedAt failed (empty path):', sidecarErr?.message || sidecarErr); |
| 2324 | } |
| 2325 | } |
| 2326 | // If this is a background-mode invocation, release the lock so subsequent |
| 2327 | // re-indexes are not falsely blocked. Use expectedJobId so a stale background |
| 2328 | // function (whose lock has since been overwritten) cannot clobber a newer one. |
| 2329 | if (req.bridgeInternalRequest != null && req.blobStore) { |
| 2330 | try { |
| 2331 | await releaseJobLock(req.blobStore, { |
| 2332 | canisterUid, |
| 2333 | vaultId, |
| 2334 | expectedJobId: req.bridgeInternalRequest.jobId, |
| 2335 | }); |
| 2336 | } catch (lockErr) { |
| 2337 | console.warn('[bridge] releaseJobLock failed (empty path):', lockErr?.message || lockErr); |
| 2338 | } |
| 2339 | } |
| 2340 | console.log( |
| 2341 | '[bridge] index', |
| 2342 | JSON.stringify({ |
| 2343 | vault_id: vaultId, |
| 2344 | canister_uid: sanitizeUserId(canisterUid), |
| 2345 | notes_processed: notes.length, |
| 2346 | chunks_indexed: 0, |
| 2347 | vectors_deleted, |
| 2348 | chunks_skipped_cached: 0, |
| 2349 | }), |
| 2350 | ); |
| 2351 | timer.finish({ |
| 2352 | ok: true, |
| 2353 | notes_processed: notes.length, |
| 2354 | chunks_indexed: 0, |
| 2355 | vectors_deleted, |
| 2356 | chunks_skipped_cached: 0, |
| 2357 | }); |
| 2358 | return res.json({ |
| 2359 | ok: true, |
| 2360 | notesProcessed: notes.length, |
| 2361 | chunksIndexed: 0, |
| 2362 | embedding_input_tokens: 0, |
| 2363 | vectors_deleted, |
| 2364 | chunksSkippedCached: 0, |
| 2365 | }); |
| 2366 | } |
| 2367 | |
| 2368 | // Content-hash cache lookup. If the store doesn't expose getChunkHashes (older |
| 2369 | // backend or test mock), treat every chunk as cache miss — correct, just slower. |
| 2370 | let existingHashes = new Map(); |
| 2371 | if (typeof store.getChunkHashes === 'function') { |
| 2372 | try { |
| 2373 | existingHashes = await store.getChunkHashes(vaultId); |
| 2374 | } catch (e) { |
| 2375 | console.warn( |
| 2376 | '[bridge] getChunkHashes failed; falling back to full re-embed for this vault:', |
| 2377 | e?.message || e, |
| 2378 | ); |
| 2379 | existingHashes = new Map(); |
| 2380 | } |
| 2381 | } |
| 2382 | const partitioned = partitionChunksForReindex(chunksWithHash, existingHashes); |
| 2383 | const toEmbed = partitioned.toEmbed; |
| 2384 | const chunks_skipped_cached = partitioned.skippedCachedCount; |
| 2385 | timer.step('cache_lookup', { |
| 2386 | cache_size: existingHashes.size, |
| 2387 | chunks_total: chunksWithHash.length, |
| 2388 | chunks_skipped_cached, |
| 2389 | chunks_to_embed: toEmbed.length, |
| 2390 | orphan_count: partitioned.orphanIds.length, |
| 2391 | }); |
| 2392 | |
| 2393 | // —— Auto-routing: sync vs background —— |
| 2394 | // The bridge runs as a Netlify synchronous function (60 s platform max). After the |
| 2395 | // OpenAI(1536)→DeepInfra(1024) switch, a 251-chunk full re-embed costs ~10–15 s |
| 2396 | // and a 1 500-chunk re-embed pushes past 30 s. To keep the snappy UX for the 99 % |
| 2397 | // case (small delta or cache-hit) AND eliminate timeout risk for the 1 % case |
| 2398 | // (first-time index, dim migration, big import), we estimate the embed wall-clock |
| 2399 | // here and either (a) continue inline OR (b) hand the work to the |
| 2400 | // `bridge-index-background` Netlify Function (15-min cap). |
| 2401 | // |
| 2402 | // The background path itself re-enters this same handler via |
| 2403 | // `req.bridgeInternalRequest` (set by the wrapper after HMAC verification); when |
| 2404 | // that's truthy we SKIP the routing decision and execute inline regardless of size. |
| 2405 | const isInternalBackgroundRequest = req.bridgeInternalRequest != null; |
| 2406 | const embeddingConfig = storeConfig.embedding; |
| 2407 | const BATCH_EMBED = BATCH_EMBED_DEFAULT; |
| 2408 | const EMBED_CONCURRENCY = EMBED_CONCURRENCY_DEFAULT; |
| 2409 | if (!isInternalBackgroundRequest && toEmbed.length > 0) { |
| 2410 | const estimatedSeconds = estimateEmbedSeconds({ |
| 2411 | chunksToEmbed: toEmbed.length, |
| 2412 | batchSize: BATCH_EMBED, |
| 2413 | concurrency: EMBED_CONCURRENCY, |
| 2414 | }); |
| 2415 | // `isFirstIndex` covers BOTH a true first-time index (no prior cache rows) AND |
| 2416 | // the post-dim-migration state where `ensureCollection` just dropped + recreated |
| 2417 | // the table (so `getChunkHashes` returned empty). Both require a full re-embed |
| 2418 | // and both should route to background regardless of estimate. |
| 2419 | const isFirstIndex = existingHashes.size === 0; |
| 2420 | const decision = shouldUseBackgroundIndex({ |
| 2421 | chunksToEmbed: toEmbed.length, |
| 2422 | estimatedSeconds, |
| 2423 | syncBudgetSeconds: SYNC_BUDGET_SECONDS, |
| 2424 | maxSyncChunks: MAX_SYNC_CHUNKS, |
| 2425 | isFirstIndex, |
| 2426 | }); |
| 2427 | timer.step('routing_decision', { |
| 2428 | chunks_to_embed: toEmbed.length, |
| 2429 | estimated_seconds: estimatedSeconds, |
| 2430 | is_first_index: isFirstIndex, |
| 2431 | sync_budget_seconds: SYNC_BUDGET_SECONDS, |
| 2432 | max_sync_chunks: MAX_SYNC_CHUNKS, |
| 2433 | decision: decision.shouldUseBackground ? 'background' : 'sync', |
| 2434 | reason: decision.reason, |
| 2435 | }); |
| 2436 | if (decision.shouldUseBackground) { |
| 2437 | if (!req.blobStore) { |
| 2438 | // No Blob store available (local self-host without Netlify Blobs): we cannot |
| 2439 | // safely run the background path because lock + sidecar persistence would be |
| 2440 | // lost. Fall through to sync — local self-host is single-tenant and operators |
| 2441 | // can tolerate a longer wait. |
| 2442 | timer.step('routing_fallback_no_blobstore'); |
| 2443 | } else { |
| 2444 | const lockResult = await acquireJobLock(req.blobStore, { |
| 2445 | canisterUid, |
| 2446 | vaultId, |
| 2447 | actorUid: sanitizeUserId(uid), |
| 2448 | chunksToEmbed: toEmbed.length, |
| 2449 | estimatedSeconds, |
| 2450 | reason: decision.reason, |
| 2451 | }); |
| 2452 | if (!lockResult.acquired) { |
| 2453 | timer.finish({ |
| 2454 | ok: true, |
| 2455 | phase: 'background_already_running', |
| 2456 | existing_job_id: lockResult.existing?.jobId || null, |
| 2457 | }); |
| 2458 | return res.status(409).json({ |
| 2459 | status: 'already_running', |
| 2460 | message: |
| 2461 | 'A background re-index is already running for this vault. Refresh in a minute.', |
| 2462 | jobId: lockResult.existing?.jobId || null, |
| 2463 | startedAt: lockResult.existing?.startedAt || null, |
| 2464 | }); |
| 2465 | } |
| 2466 | try { |
| 2467 | await kickOffBackgroundIndex(req, lockResult.jobId, canisterUid, vaultId); |
| 2468 | } catch (kickoffErr) { |
| 2469 | // Kickoff failed (network blip, missing SESSION_SECRET, etc.) — release the |
| 2470 | // lock so the user can retry, and surface the error. |
| 2471 | await releaseJobLock(req.blobStore, { |
| 2472 | canisterUid, |
| 2473 | vaultId, |
| 2474 | expectedJobId: lockResult.jobId, |
| 2475 | }); |
| 2476 | timer.finish({ |
| 2477 | ok: false, |
| 2478 | phase: 'background_kickoff', |
| 2479 | error: kickoffErr?.message || String(kickoffErr), |
| 2480 | }); |
| 2481 | return res.status(502).json({ |
| 2482 | error: 'Could not start background re-index', |
| 2483 | code: 'BACKGROUND_KICKOFF_FAILED', |
| 2484 | message: kickoffErr?.message || String(kickoffErr), |
| 2485 | }); |
| 2486 | } |
| 2487 | timer.finish({ |
| 2488 | ok: true, |
| 2489 | phase: 'background_started', |
| 2490 | job_id: lockResult.jobId, |
| 2491 | chunks_to_embed: toEmbed.length, |
| 2492 | estimated_seconds: estimatedSeconds, |
| 2493 | reason: decision.reason, |
| 2494 | }); |
| 2495 | return res.status(202).json({ |
| 2496 | status: 'background', |
| 2497 | jobId: lockResult.jobId, |
| 2498 | message: |
| 2499 | 'Large re-index started in the background. Refresh in 1–2 minutes — search will use the new vectors as soon as the job finishes.', |
| 2500 | estimatedSeconds, |
| 2501 | chunksToEmbed: toEmbed.length, |
| 2502 | reason: decision.reason, |
| 2503 | }); |
| 2504 | } |
| 2505 | } |
| 2506 | } |
| 2507 | // —— end auto-routing —— |
| 2508 | |
| 2509 | const embedBatches = []; |
| 2510 | for (let i = 0; i < toEmbed.length; i += BATCH_EMBED) { |
| 2511 | embedBatches.push(toEmbed.slice(i, i + BATCH_EMBED)); |
| 2512 | } |
| 2513 | let embedding_input_tokens = 0; |
| 2514 | let embed_total_ms = 0; |
| 2515 | let embed_max_batch_ms = 0; |
| 2516 | let embed_min_batch_ms = embedBatches.length > 0 ? Number.POSITIVE_INFINITY : 0; |
| 2517 | const embedBatchCount = embedBatches.length; |
| 2518 | // Result vectors keyed by toEmbed index (preserves order so the upsert step can |
| 2519 | // zip vectors[i] back to toEmbed[i].chunk without depending on completion order). |
| 2520 | const embedResults = await runWithConcurrency( |
| 2521 | embedBatches.map((batch, batchIndex) => async () => { |
| 2522 | const texts = batch.map((item) => item.chunk.text); |
| 2523 | const { vectors: batchVectors, embedding_input_tokens: batchTok } = await embedWithUsage( |
| 2524 | texts, |
| 2525 | embeddingConfig, |
| 2526 | { voyageInputType: 'document' }, |
| 2527 | ); |
| 2528 | return { batchIndex, batchVectors, batchTok }; |
| 2529 | }), |
| 2530 | { |
| 2531 | concurrency: EMBED_CONCURRENCY, |
| 2532 | onSettled: ({ index, ok, ms, error }) => { |
| 2533 | if (!ok) { |
| 2534 | timer.step('embed_batch_error', { |
| 2535 | batch_index: index, |
| 2536 | embed_ms: ms, |
| 2537 | error: error?.message || String(error), |
| 2538 | }); |
| 2539 | return; |
| 2540 | } |
| 2541 | embed_total_ms += ms; |
| 2542 | if (ms > embed_max_batch_ms) embed_max_batch_ms = ms; |
| 2543 | if (ms < embed_min_batch_ms) embed_min_batch_ms = ms; |
| 2544 | timer.step('embed_batch', { |
| 2545 | batch_index: index, |
| 2546 | batch_size: embedBatches[index].length, |
| 2547 | embed_ms: ms, |
| 2548 | }); |
| 2549 | }, |
| 2550 | }, |
| 2551 | ); |
| 2552 | const vectorsByEmbedIndex = new Array(toEmbed.length); |
| 2553 | for (const { batchIndex, batchVectors, batchTok } of embedResults) { |
| 2554 | embedding_input_tokens += batchTok; |
| 2555 | const start = batchIndex * BATCH_EMBED; |
| 2556 | const batch = embedBatches[batchIndex]; |
| 2557 | for (let j = 0; j < batch.length; j++) { |
| 2558 | vectorsByEmbedIndex[start + j] = batchVectors[j] || []; |
| 2559 | } |
| 2560 | } |
| 2561 | timer.step('embed_total', { |
| 2562 | batches: embedBatchCount, |
| 2563 | embed_total_ms, |
| 2564 | embed_avg_batch_ms: embedBatchCount > 0 ? Math.round(embed_total_ms / embedBatchCount) : 0, |
| 2565 | embed_min_batch_ms: embed_min_batch_ms === Number.POSITIVE_INFINITY ? 0 : embed_min_batch_ms, |
| 2566 | embed_max_batch_ms, |
| 2567 | embedding_input_tokens, |
| 2568 | concurrency: EMBED_CONCURRENCY, |
| 2569 | batch_size: BATCH_EMBED, |
| 2570 | provider: embeddingConfig?.provider || null, |
| 2571 | model: embeddingConfig?.model || null, |
| 2572 | }); |
| 2573 | |
| 2574 | // Orphans = chunk_ids in the store but not in the current export (deleted/renamed notes). |
| 2575 | let vectors_deleted = 0; |
| 2576 | if (partitioned.orphanIds.length > 0 && typeof store.deleteByChunkIds === 'function') { |
| 2577 | vectors_deleted = await store.deleteByChunkIds(partitioned.orphanIds); |
| 2578 | } else if ( |
| 2579 | partitioned.orphanIds.length === 0 && |
| 2580 | existingHashes.size === 0 && |
| 2581 | typeof store.deleteByVaultId === 'function' |
| 2582 | ) { |
| 2583 | // First run for this vault (no prior cache rows) — clear any leftover rows that |
| 2584 | // lacked content_hash but might still match the vault, so search cannot return paths |
| 2585 | // that no longer exist in the export. |
| 2586 | vectors_deleted = await store.deleteByVaultId(vaultId); |
| 2587 | } |
| 2588 | timer.step('delete_old_vectors', { |
| 2589 | vectors_deleted, |
| 2590 | orphan_count: partitioned.orphanIds.length, |
| 2591 | }); |
| 2592 | |
| 2593 | let upsert_total_ms = 0; |
| 2594 | const upsertBatchCount = Math.ceil(toEmbed.length / BATCH_UPSERT); |
| 2595 | for (let i = 0; i < toEmbed.length; i += BATCH_UPSERT) { |
| 2596 | const slice = toEmbed.slice(i, i + BATCH_UPSERT); |
| 2597 | const points = slice.map((item, j) => ({ |
| 2598 | id: item.storeId, |
| 2599 | vector: vectorsByEmbedIndex[i + j] || [], |
| 2600 | text: item.chunk.text, |
| 2601 | path: item.chunk.path, |
| 2602 | vault_id: vaultId, |
| 2603 | project: item.chunk.project, |
| 2604 | tags: item.chunk.tags, |
| 2605 | date: item.chunk.date, |
| 2606 | causal_chain_id: item.chunk.causal_chain_id, |
| 2607 | entity: item.chunk.entity, |
| 2608 | episode_id: item.chunk.episode_id, |
| 2609 | content_hash: item.contentHash, |
| 2610 | })); |
| 2611 | const upsertStart = Date.now(); |
| 2612 | await store.upsert(points); |
| 2613 | upsert_total_ms += Date.now() - upsertStart; |
| 2614 | } |
| 2615 | timer.step('upsert_total', { |
| 2616 | batches: upsertBatchCount, |
| 2617 | upsert_total_ms, |
| 2618 | upsert_avg_batch_ms: upsertBatchCount > 0 ? Math.round(upsert_total_ms / upsertBatchCount) : 0, |
| 2619 | points_upserted: toEmbed.length, |
| 2620 | }); |
| 2621 | await persistVectorsToBlob(req, canisterUid, vectorsDir); |
| 2622 | timer.step('persist_vectors'); |
| 2623 | // Sidecar update so the Hub UI's "Last indexed" line is correct after BOTH |
| 2624 | // the synchronous and the background path. The same record format is read |
| 2625 | // by `GET /api/v1/index/status` and rendered next to the Re-index button. |
| 2626 | if (req.blobStore) { |
| 2627 | try { |
| 2628 | await setLastIndexedAt(req.blobStore, { |
| 2629 | canisterUid, |
| 2630 | vaultId, |
| 2631 | actorUid: sanitizeUserId(uid), |
| 2632 | notesProcessed: notes.length, |
| 2633 | chunksIndexed: allChunks.length, |
| 2634 | chunksEmbedded: toEmbed.length, |
| 2635 | chunksSkippedCached: chunks_skipped_cached, |
| 2636 | vectorsDeleted: vectors_deleted, |
| 2637 | embeddingInputTokens: embedding_input_tokens, |
| 2638 | durationMs: timer.totalMs(), |
| 2639 | mode: req.bridgeInternalRequest != null ? 'background' : 'sync', |
| 2640 | provider: embeddingConfig?.provider || null, |
| 2641 | model: embeddingConfig?.model || null, |
| 2642 | }); |
| 2643 | } catch (sidecarErr) { |
| 2644 | // Sidecar write failure must not fail the index — UI just falls back to "never indexed". |
| 2645 | console.warn('[bridge] setLastIndexedAt failed:', sidecarErr?.message || sidecarErr); |
| 2646 | } |
| 2647 | } |
| 2648 | // Background path: release the job lock so a future re-index is not falsely blocked. |
| 2649 | // `expectedJobId` ensures we only release OUR lock — if a stale background job |
| 2650 | // finishes after a fresh background job has already acquired a new lock (rare, |
| 2651 | // but possible if the first job exceeded the lock TTL), we leave the new lock alone. |
| 2652 | if (req.bridgeInternalRequest != null && req.blobStore) { |
| 2653 | try { |
| 2654 | await releaseJobLock(req.blobStore, { |
| 2655 | canisterUid, |
| 2656 | vaultId, |
| 2657 | expectedJobId: req.bridgeInternalRequest.jobId, |
| 2658 | }); |
| 2659 | } catch (lockErr) { |
| 2660 | console.warn('[bridge] releaseJobLock failed:', lockErr?.message || lockErr); |
| 2661 | } |
| 2662 | } |
| 2663 | console.log( |
| 2664 | '[bridge] index', |
| 2665 | JSON.stringify({ |
| 2666 | vault_id: vaultId, |
| 2667 | canister_uid: sanitizeUserId(canisterUid), |
| 2668 | notes_processed: notes.length, |
| 2669 | chunks_indexed: allChunks.length, |
| 2670 | chunks_skipped_cached, |
| 2671 | chunks_embedded: toEmbed.length, |
| 2672 | vectors_deleted, |
| 2673 | mode: req.bridgeInternalRequest != null ? 'background' : 'sync', |
| 2674 | }), |
| 2675 | ); |
| 2676 | const indexResult = { |
| 2677 | ok: true, |
| 2678 | notesProcessed: notes.length, |
| 2679 | chunksIndexed: allChunks.length, |
| 2680 | chunksSkippedCached: chunks_skipped_cached, |
| 2681 | chunksEmbedded: toEmbed.length, |
| 2682 | embedding_input_tokens, |
| 2683 | vectors_deleted, |
| 2684 | }; |
| 2685 | timer.finish({ |
| 2686 | ok: true, |
| 2687 | notes_processed: notes.length, |
| 2688 | chunks_indexed: allChunks.length, |
| 2689 | chunks_skipped_cached, |
| 2690 | chunks_embedded: toEmbed.length, |
| 2691 | vectors_deleted, |
| 2692 | embedding_input_tokens, |
| 2693 | mode: req.bridgeInternalRequest != null ? 'background' : 'sync', |
| 2694 | }); |
| 2695 | res.json(indexResult); |
| 2696 | fireBridgeCaptureEvent( |
| 2697 | 'index', |
| 2698 | { |
| 2699 | note_count: notes.length, |
| 2700 | chunk_count: allChunks.length, |
| 2701 | chunks_skipped_cached, |
| 2702 | chunks_embedded: toEmbed.length, |
| 2703 | vectors_deleted, |
| 2704 | }, |
| 2705 | sanitizeUserId(uid), |
| 2706 | vaultId, |
| 2707 | ); |
| 2708 | return; |
| 2709 | } catch (e) { |
| 2710 | console.error('Bridge index error:', e); |
| 2711 | timer.finish({ ok: false, phase: 'catch', error: e?.message || String(e) }); |
| 2712 | // Background path: release the lock on error so the operator can retry without |
| 2713 | // waiting for the 16-min TTL. We do this defensively regardless of whether the |
| 2714 | // error happened before or after the lock was acquired. |
| 2715 | if (req.bridgeInternalRequest != null && req.blobStore) { |
| 2716 | try { |
| 2717 | await releaseJobLock(req.blobStore, { |
| 2718 | canisterUid: req.bridgeInternalRequest.canisterUid, |
| 2719 | vaultId: req.bridgeInternalRequest.vaultId, |
| 2720 | expectedJobId: req.bridgeInternalRequest.jobId, |
| 2721 | }); |
| 2722 | } catch (lockErr) { |
| 2723 | console.warn('[bridge] releaseJobLock failed (catch path):', lockErr?.message || lockErr); |
| 2724 | } |
| 2725 | } |
| 2726 | return res.status(500).json({ |
| 2727 | error: 'Index failed', |
| 2728 | code: 'INTERNAL_ERROR', |
| 2729 | message: bridgeEmbedFailureMessage(e, 'index'), |
| 2730 | }); |
| 2731 | } |
| 2732 | }); |
| 2733 | |
| 2734 | function truncateSnippet(text, maxChars = 300) { |
| 2735 | if (text == null || typeof text !== 'string') return ''; |
| 2736 | const t = text.trim(); |
| 2737 | if (t.length <= maxChars) return t; |
| 2738 | const slice = t.slice(0, maxChars); |
| 2739 | const lastSpace = slice.lastIndexOf(' '); |
| 2740 | return (lastSpace > maxChars / 2 ? slice.slice(0, lastSpace) : slice) + '…'; |
| 2741 | } |
| 2742 | |
| 2743 | /** |
| 2744 | * Batch document embeddings for hosted MCP `cluster` (and similar callers). |
| 2745 | * Auth + vault access mirror `POST /api/v1/search`: JWT in `Authorization`, `X-Vault-Id`, |
| 2746 | * `resolveHostedBridgeContext` (effective canister user + allowed vault ids + optional scope). |
| 2747 | * Embedding model/env match `POST /api/v1/index` via `getBridgeStoreConfig` + `embedWithUsage` with `voyageInputType: "document"`. |
| 2748 | */ |
| 2749 | const HOSTED_EMBED_MAX_TEXTS = 200; |
| 2750 | const HOSTED_EMBED_MAX_CHARS_PER_TEXT = 1200; |
| 2751 | |
| 2752 | app.post('/api/v1/embed', async (req, res) => { |
| 2753 | const auth = req.headers.authorization; |
| 2754 | const token = auth && auth.startsWith('Bearer ') ? auth.slice(7) : null; |
| 2755 | const uid = token ? userIdFromJwt(token) : null; |
| 2756 | if (!uid) { |
| 2757 | return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 2758 | } |
| 2759 | const hctx = await resolveHostedBridgeContext(req, uid); |
| 2760 | if (!hctx.ok) { |
| 2761 | return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 2762 | } |
| 2763 | const canisterUid = hctx.effectiveCanisterUid; |
| 2764 | const rawTexts = req.body?.texts; |
| 2765 | if (!Array.isArray(rawTexts)) { |
| 2766 | return res.status(400).json({ error: 'texts array required', code: 'BAD_REQUEST' }); |
| 2767 | } |
| 2768 | const texts = rawTexts |
| 2769 | .slice(0, HOSTED_EMBED_MAX_TEXTS) |
| 2770 | .map((t) => String(t ?? '').slice(0, HOSTED_EMBED_MAX_CHARS_PER_TEXT)); |
| 2771 | if (texts.length === 0) { |
| 2772 | return res.status(400).json({ error: 'texts must be non-empty', code: 'BAD_REQUEST' }); |
| 2773 | } |
| 2774 | try { |
| 2775 | const { embedWithUsage } = await import('../../lib/embedding.mjs'); |
| 2776 | const vectorsDir = await getVectorsDirForUser(req, canisterUid); |
| 2777 | const storeConfig = getBridgeStoreConfig(canisterUid, vectorsDir); |
| 2778 | const embeddingConfig = storeConfig.embedding; |
| 2779 | let embedding_input_tokens = 0; |
| 2780 | const vectors = []; |
| 2781 | for (let i = 0; i < texts.length; i += BATCH_EMBED) { |
| 2782 | const batch = texts.slice(i, i + BATCH_EMBED); |
| 2783 | const { vectors: batchVectors, embedding_input_tokens: batchTok } = await embedWithUsage( |
| 2784 | batch, |
| 2785 | embeddingConfig, |
| 2786 | { voyageInputType: 'document' }, |
| 2787 | ); |
| 2788 | embedding_input_tokens += batchTok; |
| 2789 | for (const v of batchVectors) { |
| 2790 | vectors.push(v); |
| 2791 | } |
| 2792 | } |
| 2793 | return res.json({ |
| 2794 | vectors, |
| 2795 | embedding_input_tokens, |
| 2796 | texts_count: texts.length, |
| 2797 | }); |
| 2798 | } catch (e) { |
| 2799 | console.error('Bridge embed batch error:', e); |
| 2800 | return res.status(500).json({ |
| 2801 | error: 'Embed failed', |
| 2802 | code: 'INTERNAL_ERROR', |
| 2803 | message: bridgeEmbedFailureMessage(e, 'embed'), |
| 2804 | }); |
| 2805 | } |
| 2806 | }); |
| 2807 | |
| 2808 | app.post('/api/v1/search', async (req, res) => { |
| 2809 | const auth = req.headers.authorization; |
| 2810 | const token = auth && auth.startsWith('Bearer ') ? auth.slice(7) : null; |
| 2811 | const uid = token ? userIdFromJwt(token) : null; |
| 2812 | if (!uid) { |
| 2813 | return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 2814 | } |
| 2815 | const hctx = await resolveHostedBridgeContext(req, uid); |
| 2816 | if (!hctx.ok) { |
| 2817 | return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 2818 | } |
| 2819 | const canisterUid = hctx.effectiveCanisterUid; |
| 2820 | const query = req.body?.query; |
| 2821 | // Auto-capture after successful response — fire-and-forget, does not affect latency. |
| 2822 | const _captureVaultId = sanitizeVaultId(req.headers['x-vault-id']); |
| 2823 | const _captureMode = req.body?.mode === 'keyword' ? 'keyword' : 'semantic'; |
| 2824 | res.on('finish', () => { |
| 2825 | if (res.statusCode >= 200 && res.statusCode < 300 && query) { |
| 2826 | fireBridgeCaptureEvent('search', { query, mode: _captureMode }, sanitizeUserId(uid), _captureVaultId); |
| 2827 | } |
| 2828 | }); |
| 2829 | if (!query || typeof query !== 'string') { |
| 2830 | return res.status(400).json({ error: 'query required', code: 'BAD_REQUEST' }); |
| 2831 | } |
| 2832 | const limit = Math.max(1, Math.min(parseInt(req.body?.limit, 10) || 20, 100)); |
| 2833 | const snippetChars = parseInt(req.body?.snippetChars, 10) || 300; |
| 2834 | try { |
| 2835 | const mode = req.body?.mode === 'keyword' ? 'keyword' : 'semantic'; |
| 2836 | const bridgeVaultId = sanitizeVaultId(req.headers['x-vault-id']); |
| 2837 | |
| 2838 | if (mode === 'keyword') { |
| 2839 | let exportRes; |
| 2840 | try { |
| 2841 | exportRes = await fetch(CANISTER_URL + '/api/v1/export', { |
| 2842 | method: 'GET', |
| 2843 | headers: canisterHeaders({ 'X-User-Id': canisterUid, 'X-Vault-Id': bridgeVaultId }), |
| 2844 | }); |
| 2845 | } catch (_e) { |
| 2846 | return res.status(502).json({ error: 'Could not reach canister', code: 'BAD_GATEWAY' }); |
| 2847 | } |
| 2848 | if (!exportRes.ok) { |
| 2849 | return res.status(502).json({ error: 'Canister export failed', code: 'BAD_GATEWAY', status: exportRes.status }); |
| 2850 | } |
| 2851 | let vault; |
| 2852 | try { |
| 2853 | vault = await exportRes.json(); |
| 2854 | } catch (_e) { |
| 2855 | return res.status(502).json({ error: 'Invalid canister response', code: 'BAD_GATEWAY' }); |
| 2856 | } |
| 2857 | let rawNotes = vault.notes || []; |
| 2858 | if (hctx.scope) { |
| 2859 | rawNotes = applyScopeFilterToNotes(rawNotes, hctx.scope); |
| 2860 | } |
| 2861 | const { noteRecordFromExportPayload, keywordSearchNotesArray } = await import('../../lib/keyword-search.mjs'); |
| 2862 | const { filterNotesByListOptions } = await import('../../lib/list-notes.mjs'); |
| 2863 | let shaped = rawNotes.map((n) => noteRecordFromExportPayload(n)); |
| 2864 | shaped = filterNotesByListOptions(shaped, { |
| 2865 | folder: req.body?.folder, |
| 2866 | project: req.body?.project, |
| 2867 | tag: req.body?.tag, |
| 2868 | since: req.body?.since, |
| 2869 | until: req.body?.until, |
| 2870 | chain: req.body?.chain, |
| 2871 | entity: req.body?.entity, |
| 2872 | episode: req.body?.episode, |
| 2873 | content_scope: req.body?.content_scope, |
| 2874 | }); |
| 2875 | const fields = |
| 2876 | req.body?.fields === 'path' || req.body?.fields === 'full' ? req.body.fields : 'path+snippet'; |
| 2877 | const out = keywordSearchNotesArray(shaped, query, { |
| 2878 | limit, |
| 2879 | order: req.body?.order, |
| 2880 | fields, |
| 2881 | snippetChars, |
| 2882 | match: req.body?.match === 'all_terms' ? 'all_terms' : 'phrase', |
| 2883 | countOnly: req.body?.count_only === true || req.body?.countOnly === true, |
| 2884 | }); |
| 2885 | if (out.results && hctx.scope) { |
| 2886 | return res.json({ ...out, results: applyScopeFilterToNotes(out.results, hctx.scope) }); |
| 2887 | } |
| 2888 | return res.json(out); |
| 2889 | } |
| 2890 | |
| 2891 | const { embed } = await import('../../lib/embedding.mjs'); |
| 2892 | const { createVectorStore } = await import('../../lib/vector-store.mjs'); |
| 2893 | const { filterHitsByContentScope, resolveSearchFolderForContentScope } = await import('../../lib/approval-log.mjs'); |
| 2894 | const { MAX_VECTOR_KNN } = await import('../../lib/vector-knn-limit.mjs'); |
| 2895 | |
| 2896 | const vectorsDir = await getVectorsDirForUser(req, canisterUid); |
| 2897 | const storeConfig = getBridgeStoreConfig(canisterUid, vectorsDir); |
| 2898 | const store = await createVectorStore(storeConfig); |
| 2899 | const [queryVector] = await embed([query], storeConfig.embedding, { voyageInputType: 'query' }); |
| 2900 | if (!queryVector) { |
| 2901 | return res.status(500).json({ error: 'Embedding failed', code: 'INTERNAL_ERROR' }); |
| 2902 | } |
| 2903 | const cs = req.body?.content_scope || 'all'; |
| 2904 | const userFolder = req.body?.folder; |
| 2905 | const resolved = resolveSearchFolderForContentScope(cs, userFolder); |
| 2906 | if (resolved.impossible) { |
| 2907 | return res.json({ results: [], query, mode: 'semantic' }); |
| 2908 | } |
| 2909 | let searchLimit = limit; |
| 2910 | if (resolved.wideNotesFetch) { |
| 2911 | searchLimit = Math.min(10000, Math.max(limit * 120, 2500)); |
| 2912 | } else if (cs !== 'all') { |
| 2913 | searchLimit = Math.min(10000, Math.max(limit * 40, 800)); |
| 2914 | } |
| 2915 | searchLimit = Math.min(searchLimit, MAX_VECTOR_KNN); |
| 2916 | const hits = await store.search(queryVector, { |
| 2917 | limit: searchLimit, |
| 2918 | vault_id: bridgeVaultId, |
| 2919 | project: req.body?.project, |
| 2920 | tag: req.body?.tag, |
| 2921 | folder: resolved.folder, |
| 2922 | since: req.body?.since, |
| 2923 | until: req.body?.until, |
| 2924 | order: req.body?.order, |
| 2925 | chain: req.body?.chain, |
| 2926 | entity: req.body?.entity, |
| 2927 | episode: req.body?.episode, |
| 2928 | }); |
| 2929 | let scopedHits = filterHitsByContentScope(hits || [], cs); |
| 2930 | scopedHits = scopedHits.slice(0, limit); |
| 2931 | let results = scopedHits.map((h) => ({ |
| 2932 | path: h.path, |
| 2933 | score: h.score, |
| 2934 | ...(typeof h.vec_distance === 'number' && Number.isFinite(h.vec_distance) |
| 2935 | ? { vec_distance: h.vec_distance } |
| 2936 | : {}), |
| 2937 | project: h.project ?? null, |
| 2938 | tags: h.tags ?? [], |
| 2939 | snippet: truncateSnippet(h.text, snippetChars), |
| 2940 | })); |
| 2941 | if (hctx.scope) { |
| 2942 | results = applyScopeFilterToNotes(results, hctx.scope); |
| 2943 | } |
| 2944 | return res.json({ results, query, mode: 'semantic' }); |
| 2945 | } catch (e) { |
| 2946 | console.error('Bridge search error:', e); |
| 2947 | return res.status(500).json({ |
| 2948 | error: 'Search failed', |
| 2949 | code: 'INTERNAL_ERROR', |
| 2950 | message: bridgeEmbedFailureMessage(e, 'search'), |
| 2951 | }); |
| 2952 | } |
| 2953 | }); |
| 2954 | |
| 2955 | app.use((err, req, res, _next) => { |
| 2956 | if (res.headersSent) return; |
| 2957 | console.error('[bridge] unhandled error:', err?.stack || err?.message || err); |
| 2958 | let status = 500; |
| 2959 | if (err instanceof multer.MulterError) { |
| 2960 | if (err.code === 'LIMIT_FILE_SIZE') status = 413; |
| 2961 | else status = 400; |
| 2962 | } else if (typeof err.status === 'number' && err.status >= 400 && err.status < 600) { |
| 2963 | status = err.status; |
| 2964 | } else if (typeof err.statusCode === 'number' && err.statusCode >= 400 && err.statusCode < 600) { |
| 2965 | status = err.statusCode; |
| 2966 | } |
| 2967 | res.status(status).json({ |
| 2968 | error: err.message || 'Internal error', |
| 2969 | code: err.code || 'INTERNAL_ERROR', |
| 2970 | }); |
| 2971 | }); |
| 2972 | |
| 2973 | // ——— Memory endpoints (Phase 8) ——— |
| 2974 | |
| 2975 | /** |
| 2976 | * Fire-and-forget memory event capture for hosted bridge endpoints. |
| 2977 | * Uses Netlify Blobs when available (hosted), falls back to file-based for self-hosted. |
| 2978 | * Never throws, never delays the response. |
| 2979 | */ |
| 2980 | function fireBridgeCaptureEvent(type, data, uid, vaultId) { |
| 2981 | (async () => { |
| 2982 | try { |
| 2983 | if (globalThis.__netlify_blob_store) { |
| 2984 | // Hosted: append to Netlify Blobs for durability across Lambda invocations. |
| 2985 | const { createMemoryEvent, MEMORY_EVENT_TYPES } = await import('../../lib/memory-event.mjs'); |
| 2986 | if (!MEMORY_EVENT_TYPES.includes(type)) return; |
| 2987 | const event = createMemoryEvent(type, data, { vaultId: vaultId || 'default' }); |
| 2988 | await blobsAppendMemoryEvent(uid, vaultId, event); |
| 2989 | } else { |
| 2990 | // Self-hosted: file-based. |
| 2991 | const { FileMemoryProvider } = await import('../../lib/memory-provider-file.mjs'); |
| 2992 | const { MemoryManager } = await import('../../lib/memory.mjs'); |
| 2993 | const mm = new MemoryManager(new FileMemoryProvider(bridgeMemoryDir(uid, vaultId || 'default'))); |
| 2994 | if (mm.shouldCapture(type)) mm.store(type, data); |
| 2995 | } |
| 2996 | } catch (_) {} |
| 2997 | })(); |
| 2998 | } |
| 2999 | |
| 3000 | // ——— Calendar (hosted parity — step 12): event store on bridge DATA_DIR + notes from canister ——— |
| 3001 | |
| 3002 | /** |
| 3003 | * Fetches canister note metadata for calendar timeline merge. Returns an empty array when |
| 3004 | * the canister is unreachable so the events layer can still succeed. |
| 3005 | * |
| 3006 | * @param {string} canisterUid |
| 3007 | * @param {string} actorUid |
| 3008 | * @param {string} vaultId |
| 3009 | * @returns {Promise<Array<{ path: string, frontmatter: object, date?: string|null, updated?: string|null, project?: string|null, tags?: string[] }>>} |
| 3010 | */ |
| 3011 | async function fetchCanisterNoteRecordsForTimeline(canisterUid, actorUid, vaultId) { |
| 3012 | if (!CANISTER_URL) return []; |
| 3013 | try { |
| 3014 | const upstream = await fetch(`${CANISTER_URL}/api/v1/notes?limit=10000&offset=0`, { |
| 3015 | headers: canisterHeaders({ |
| 3016 | 'x-user-id': canisterUid, |
| 3017 | 'x-actor-id': actorUid, |
| 3018 | 'x-vault-id': vaultId, |
| 3019 | }), |
| 3020 | }); |
| 3021 | if (!upstream.ok) return []; |
| 3022 | const data = await upstream.json(); |
| 3023 | if (!Array.isArray(data.notes)) return []; |
| 3024 | return data.notes |
| 3025 | .map((note) => { |
| 3026 | const path = typeof note.path === 'string' ? note.path.trim() : ''; |
| 3027 | if (!path) return null; |
| 3028 | const fm = materializeListFrontmatter(note.frontmatter ?? {}); |
| 3029 | return { |
| 3030 | path, |
| 3031 | frontmatter: note.frontmatter ?? {}, |
| 3032 | date: typeof fm.date === 'string' ? fm.date : null, |
| 3033 | updated: typeof note.updated === 'string' ? note.updated : null, |
| 3034 | project: typeof fm.project === 'string' ? fm.project : null, |
| 3035 | tags: Array.isArray(note.tags) ? note.tags.map(String) : [], |
| 3036 | }; |
| 3037 | }) |
| 3038 | .filter(Boolean); |
| 3039 | } catch (_) { |
| 3040 | return []; |
| 3041 | } |
| 3042 | } |
| 3043 | |
| 3044 | app.get('/api/v1/calendar/timeline', requireBridgeAuth, async (req, res) => { |
| 3045 | const from = typeof req.query.from === 'string' ? req.query.from.trim() : ''; |
| 3046 | const to = typeof req.query.to === 'string' ? req.query.to.trim() : ''; |
| 3047 | if (!from || !to) { |
| 3048 | return res.status(400).json({ error: '`from` and `to` are required', code: 'BAD_REQUEST' }); |
| 3049 | } |
| 3050 | const hctx = await resolveHostedBridgeContext(req, req.uid); |
| 3051 | if (!hctx.ok) return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 3052 | try { |
| 3053 | const noteRecords = await fetchCanisterNoteRecordsForTimeline( |
| 3054 | hctx.effectiveCanisterUid, |
| 3055 | hctx.actorUid, |
| 3056 | hctx.vaultId, |
| 3057 | ); |
| 3058 | const payload = buildCalendarTimeline({ |
| 3059 | dataDir: DATA_DIR, |
| 3060 | vaultId: hctx.vaultId, |
| 3061 | noteRecords, |
| 3062 | from, |
| 3063 | to, |
| 3064 | layers: req.query.layers, |
| 3065 | sourceCalendarIds: req.query.source_calendar_ids, |
| 3066 | scope: hctx.scope, |
| 3067 | }); |
| 3068 | return res.json(payload); |
| 3069 | } catch (e) { |
| 3070 | const message = e?.message ? String(e.message) : 'Invalid timeline request'; |
| 3071 | if ( |
| 3072 | message.includes('Unsupported timeline layer') |
| 3073 | || message.includes('Invalid') |
| 3074 | || message.includes('required') |
| 3075 | || message.includes('before') |
| 3076 | ) { |
| 3077 | return res.status(400).json({ error: message, code: 'BAD_REQUEST' }); |
| 3078 | } |
| 3079 | return res.status(500).json({ error: message, code: 'RUNTIME_ERROR' }); |
| 3080 | } |
| 3081 | }); |
| 3082 | |
| 3083 | app.get('/api/v1/calendar/agent-context', requireBridgeAuth, async (req, res) => { |
| 3084 | const from = typeof req.query.from === 'string' ? req.query.from.trim() : ''; |
| 3085 | const to = typeof req.query.to === 'string' ? req.query.to.trim() : ''; |
| 3086 | if (!from || !to) { |
| 3087 | return res.status(400).json({ error: '`from` and `to` are required', code: 'BAD_REQUEST' }); |
| 3088 | } |
| 3089 | const hctx = await resolveHostedBridgeContext(req, req.uid); |
| 3090 | if (!hctx.ok) return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 3091 | try { |
| 3092 | const payload = retrieveAgentCalendarContext(DATA_DIR, hctx.vaultId, { |
| 3093 | from, |
| 3094 | to, |
| 3095 | agentContextTier: req.query.agent_context_tier, |
| 3096 | sourceCalendarIds: req.query.source_calendar_ids, |
| 3097 | }); |
| 3098 | return res.json(payload); |
| 3099 | } catch (e) { |
| 3100 | const message = e?.message ? String(e.message) : 'Invalid agent context request'; |
| 3101 | if ( |
| 3102 | message.includes('agent_context_tier') |
| 3103 | || message.includes('Invalid') |
| 3104 | || message.includes('required') |
| 3105 | || message.includes('before') |
| 3106 | ) { |
| 3107 | return res.status(400).json({ error: message, code: 'BAD_REQUEST' }); |
| 3108 | } |
| 3109 | return res.status(500).json({ error: message, code: 'RUNTIME_ERROR' }); |
| 3110 | } |
| 3111 | }); |
| 3112 | |
| 3113 | app.get('/api/v1/calendar/source-calendars', requireBridgeAuth, async (req, res) => { |
| 3114 | const hctx = await resolveHostedBridgeSettingsContext(req, req.uid); |
| 3115 | const vaultId = sanitizeVaultId(req.headers['x-vault-id']); |
| 3116 | if (!hctx.allowedVaultIds.includes(vaultId)) { |
| 3117 | return res.status(403).json({ error: 'Access to this vault is not allowed.', code: 'FORBIDDEN' }); |
| 3118 | } |
| 3119 | try { |
| 3120 | return res.json({ |
| 3121 | schema: 'knowtation.source_calendars/v0', |
| 3122 | vault_id: vaultId, |
| 3123 | source_calendars: listSourceCalendarsForClient(DATA_DIR, vaultId), |
| 3124 | }); |
| 3125 | } catch (e) { |
| 3126 | return res.status(500).json({ error: e.message, code: 'RUNTIME_ERROR' }); |
| 3127 | } |
| 3128 | }); |
| 3129 | |
| 3130 | app.patch('/api/v1/calendar/source-calendars/:id', requireBridgeAuth, requireBridgeEditorOrAdmin, async (req, res) => { |
| 3131 | const sourceCalendarId = typeof req.params.id === 'string' ? decodeURIComponent(req.params.id).trim() : ''; |
| 3132 | if (!sourceCalendarId) { |
| 3133 | return res.status(400).json({ error: 'source calendar id is required', code: 'BAD_REQUEST' }); |
| 3134 | } |
| 3135 | const hctx = await resolveHostedBridgeContext(req, req.uid); |
| 3136 | if (!hctx.ok) return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 3137 | try { |
| 3138 | const patch = parseSourceCalendarPatchBody(req.body); |
| 3139 | const result = patchSourceCalendar(DATA_DIR, hctx.vaultId, sourceCalendarId, patch); |
| 3140 | return res.json({ |
| 3141 | schema: 'knowtation.source_calendar_patch/v0', |
| 3142 | vault_id: hctx.vaultId, |
| 3143 | policy_agent_context_tier_max_cap: result.policy_agent_context_tier_max_cap, |
| 3144 | source_calendar: result.source_calendar, |
| 3145 | }); |
| 3146 | } catch (e) { |
| 3147 | const message = e?.message ? String(e.message) : 'Patch failed'; |
| 3148 | if (e?.code === 'POLICY_CAP_EXCEEDED') { |
| 3149 | return res.status(403).json({ error: message, code: 'POLICY_CAP_EXCEEDED' }); |
| 3150 | } |
| 3151 | if (message.includes('not found')) { |
| 3152 | return res.status(404).json({ error: message, code: 'NOT_FOUND' }); |
| 3153 | } |
| 3154 | if ( |
| 3155 | message.includes('must be') |
| 3156 | || message.includes('required') |
| 3157 | || message.includes('exceeds policy') |
| 3158 | ) { |
| 3159 | return res.status(400).json({ error: message, code: 'BAD_REQUEST' }); |
| 3160 | } |
| 3161 | return res.status(500).json({ error: message, code: 'RUNTIME_ERROR' }); |
| 3162 | } |
| 3163 | }); |
| 3164 | |
| 3165 | app.post('/api/v1/calendar/events/import', requireBridgeAuth, requireBridgeEditorOrAdmin, async (req, res) => { |
| 3166 | const hctx = await resolveHostedBridgeContext(req, req.uid); |
| 3167 | if (!hctx.ok) return res.status(hctx.status).json({ error: hctx.error, code: hctx.code }); |
| 3168 | const body = req.body && typeof req.body === 'object' ? req.body : {}; |
| 3169 | const icsText = typeof body.ics_text === 'string' ? body.ics_text : ''; |
| 3170 | if (!icsText.trim()) { |
| 3171 | return res.status(400).json({ error: 'ics_text (string) is required', code: 'BAD_REQUEST' }); |
| 3172 | } |
| 3173 | try { |
| 3174 | const result = importIcsIntoVault(DATA_DIR, hctx.vaultId, { |
| 3175 | icsText, |
| 3176 | displayName: typeof body.display_name === 'string' ? body.display_name : undefined, |
| 3177 | sourceCalendarId: typeof body.source_calendar_id === 'string' ? body.source_calendar_id : undefined, |
| 3178 | connectorId: typeof body.connector_id === 'string' ? body.connector_id : undefined, |
| 3179 | defaultTimezone: typeof body.default_timezone === 'string' ? body.default_timezone : undefined, |
| 3180 | }); |
| 3181 | return res.status(200).json({ |
| 3182 | schema: 'knowtation.calendar_import/v0', |
| 3183 | vault_id: hctx.vaultId, |
| 3184 | ...result, |
| 3185 | }); |
| 3186 | } catch (e) { |
| 3187 | const message = e?.message ? String(e.message) : 'Import failed'; |
| 3188 | if ( |
| 3189 | message.includes('not found') |
| 3190 | || message.includes('required') |
| 3191 | || message.includes('exceeds') |
| 3192 | || message.includes('ICS') |
| 3193 | ) { |
| 3194 | return res.status(400).json({ error: message, code: 'BAD_REQUEST' }); |
| 3195 | } |
| 3196 | return res.status(500).json({ error: message, code: 'RUNTIME_ERROR' }); |
| 3197 | } |
| 3198 | }); |
| 3199 | |
| 3200 | // Agent delegation (hosted parity — 7C-L1): same handler family as self-hosted hub/server.mjs. |
| 3201 | registerBridgeDelegationRoutes(app, { |
| 3202 | dataDir: DATA_DIR, |
| 3203 | requireBridgeAuth, |
| 3204 | resolveHostedBridgeContext, |
| 3205 | }); |
| 3206 | |
| 3207 | function bridgeMemoryAuth(req) { |
| 3208 | const auth = req.headers.authorization; |
| 3209 | const token = auth && auth.startsWith('Bearer ') ? auth.slice(7) : null; |
| 3210 | const uid = token ? userIdFromJwt(token) : null; |
| 3211 | const vaultId = sanitizeVaultId(req.headers['x-vault-id'] || req.query.vault_id); |
| 3212 | const scope = req.query.scope === 'global' ? 'global' : 'vault'; |
| 3213 | return { uid: uid ? sanitizeUserId(uid) : null, vaultId, scope }; |
| 3214 | } |
| 3215 | |
| 3216 | function bridgeMemoryDir(uid, vaultId, scope) { |
| 3217 | if (scope === 'global') { |
| 3218 | return path.join(DATA_DIR, 'memory', uid, '_global'); |
| 3219 | } |
| 3220 | return path.join(DATA_DIR, 'memory', uid, vaultId); |
| 3221 | } |
| 3222 | |
| 3223 | app.get('/api/v1/memory/:key', async (req, res) => { |
| 3224 | const { uid, vaultId } = bridgeMemoryAuth(req); |
| 3225 | if (!uid) return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 3226 | try { |
| 3227 | const { FileMemoryProvider } = await import('../../lib/memory-provider-file.mjs'); |
| 3228 | const { MemoryManager } = await import('../../lib/memory.mjs'); |
| 3229 | const provider = new FileMemoryProvider(bridgeMemoryDir(uid, vaultId)); |
| 3230 | const mm = new MemoryManager(provider); |
| 3231 | const event = mm.getLatest(req.params.key); |
| 3232 | if (!event) return res.json({ key: req.params.key, value: null, updated_at: null }); |
| 3233 | res.json({ key: req.params.key, value: event.data, updated_at: event.ts, id: event.id }); |
| 3234 | } catch (e) { |
| 3235 | res.status(500).json({ error: e.message, code: 'RUNTIME_ERROR' }); |
| 3236 | } |
| 3237 | }); |
| 3238 | |
| 3239 | app.post('/api/v1/memory/store', requireBridgeAuth, requireBridgeEditorOrAdmin, express.json(), async (req, res) => { |
| 3240 | const { uid, vaultId } = bridgeMemoryAuth(req); |
| 3241 | try { |
| 3242 | const { FileMemoryProvider } = await import('../../lib/memory-provider-file.mjs'); |
| 3243 | const { MemoryManager } = await import('../../lib/memory.mjs'); |
| 3244 | const provider = new FileMemoryProvider(bridgeMemoryDir(uid, vaultId)); |
| 3245 | const mm = new MemoryManager(provider); |
| 3246 | const { key, value, ttl } = req.body || {}; |
| 3247 | if (!key || !value) return res.status(400).json({ error: 'key and value required', code: 'BAD_REQUEST' }); |
| 3248 | const data = typeof value === 'object' ? { key, ...value } : { key, text: String(value) }; |
| 3249 | const result = mm.store('user', data, { vaultId, ttl }); |
| 3250 | res.json(result); |
| 3251 | } catch (e) { |
| 3252 | res.status(500).json({ error: e.message, code: 'RUNTIME_ERROR' }); |
| 3253 | } |
| 3254 | }); |
| 3255 | |
| 3256 | app.get('/api/v1/memory', async (req, res) => { |
| 3257 | const { uid, vaultId } = bridgeMemoryAuth(req); |
| 3258 | if (!uid) return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 3259 | try { |
| 3260 | let events; |
| 3261 | if (globalThis.__netlify_blob_store) { |
| 3262 | // Hosted: read from Blobs. |
| 3263 | events = await blobsGetMemoryEvents(uid, vaultId); |
| 3264 | if (req.query.type) events = events.filter((e) => e.type === req.query.type); |
| 3265 | if (req.query.since) events = events.filter((e) => e.ts >= req.query.since); |
| 3266 | if (req.query.until) events = events.filter((e) => e.ts <= req.query.until); |
| 3267 | events.sort((a, b) => (b.ts > a.ts ? 1 : b.ts < a.ts ? -1 : 0)); |
| 3268 | events = events.slice(0, Math.min(parseInt(req.query.limit) || 20, 100)); |
| 3269 | } else { |
| 3270 | const { FileMemoryProvider } = await import('../../lib/memory-provider-file.mjs'); |
| 3271 | const { MemoryManager } = await import('../../lib/memory.mjs'); |
| 3272 | const mm = new MemoryManager(new FileMemoryProvider(bridgeMemoryDir(uid, vaultId))); |
| 3273 | events = mm.list({ |
| 3274 | type: req.query.type || undefined, |
| 3275 | since: req.query.since || undefined, |
| 3276 | until: req.query.until || undefined, |
| 3277 | limit: Math.min(parseInt(req.query.limit) || 20, 100), |
| 3278 | }); |
| 3279 | } |
| 3280 | res.json({ events, count: events.length }); |
| 3281 | } catch (e) { |
| 3282 | res.status(500).json({ error: e.message, code: 'RUNTIME_ERROR' }); |
| 3283 | } |
| 3284 | }); |
| 3285 | |
| 3286 | app.post('/api/v1/memory/search', express.json(), async (req, res) => { |
| 3287 | const { uid, vaultId } = bridgeMemoryAuth(req); |
| 3288 | if (!uid) return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 3289 | res.json({ results: [], count: 0, note: 'Hosted memory search requires vector provider (future).' }); |
| 3290 | }); |
| 3291 | |
| 3292 | app.delete('/api/v1/memory/clear', requireBridgeAuth, requireBridgeEditorOrAdmin, async (req, res) => { |
| 3293 | const { uid, vaultId } = bridgeMemoryAuth(req); |
| 3294 | try { |
| 3295 | const { FileMemoryProvider } = await import('../../lib/memory-provider-file.mjs'); |
| 3296 | const { MemoryManager } = await import('../../lib/memory.mjs'); |
| 3297 | const provider = new FileMemoryProvider(bridgeMemoryDir(uid, vaultId)); |
| 3298 | const mm = new MemoryManager(provider); |
| 3299 | const result = mm.clear({ |
| 3300 | type: req.query.type || undefined, |
| 3301 | before: req.query.before || undefined, |
| 3302 | }); |
| 3303 | res.json(result); |
| 3304 | } catch (e) { |
| 3305 | res.status(500).json({ error: e.message, code: 'RUNTIME_ERROR' }); |
| 3306 | } |
| 3307 | }); |
| 3308 | |
| 3309 | app.get('/api/v1/memory-stats', async (req, res) => { |
| 3310 | const { uid, vaultId } = bridgeMemoryAuth(req); |
| 3311 | if (!uid) return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 3312 | try { |
| 3313 | const { FileMemoryProvider } = await import('../../lib/memory-provider-file.mjs'); |
| 3314 | const { MemoryManager } = await import('../../lib/memory.mjs'); |
| 3315 | const provider = new FileMemoryProvider(bridgeMemoryDir(uid, vaultId)); |
| 3316 | const mm = new MemoryManager(provider); |
| 3317 | res.json(mm.stats()); |
| 3318 | } catch (e) { |
| 3319 | res.status(500).json({ error: e.message, code: 'RUNTIME_ERROR' }); |
| 3320 | } |
| 3321 | }); |
| 3322 | |
| 3323 | // ——— Hosted Consolidation (Phase 10 / Stream 1) ——— |
| 3324 | |
| 3325 | // ——— Blobs-backed memory helpers (hosted path) ——— |
| 3326 | |
| 3327 | /** Netlify Blobs key for a user's raw memory events. */ |
| 3328 | function memoryBlobKey(uid, vaultId) { |
| 3329 | return `memory/${uid}/${vaultId || 'default'}/events`; |
| 3330 | } |
| 3331 | |
| 3332 | /** Load memory events from Netlify Blobs (hosted) or return [] if unavailable. */ |
| 3333 | async function blobsGetMemoryEvents(uid, vaultId) { |
| 3334 | const store = globalThis.__netlify_blob_store; |
| 3335 | if (!store) return []; |
| 3336 | try { |
| 3337 | const raw = await store.get(memoryBlobKey(uid, vaultId), { type: 'text' }); |
| 3338 | if (!raw) return []; |
| 3339 | return JSON.parse(raw) || []; |
| 3340 | } catch (_) { return []; } |
| 3341 | } |
| 3342 | |
| 3343 | /** Persist memory events to Netlify Blobs, capped at 500 events. */ |
| 3344 | async function blobsSetMemoryEvents(uid, vaultId, events) { |
| 3345 | const store = globalThis.__netlify_blob_store; |
| 3346 | if (!store) return; |
| 3347 | try { |
| 3348 | await store.set(memoryBlobKey(uid, vaultId), JSON.stringify(events.slice(-500))); |
| 3349 | } catch (_) {} |
| 3350 | } |
| 3351 | |
| 3352 | /** Append a single event to Blobs memory store (read-modify-write). */ |
| 3353 | async function blobsAppendMemoryEvent(uid, vaultId, event) { |
| 3354 | const events = await blobsGetMemoryEvents(uid, vaultId); |
| 3355 | events.push(event); |
| 3356 | await blobsSetMemoryEvents(uid, vaultId, events); |
| 3357 | } |
| 3358 | |
| 3359 | // ——— Consolidation cost tracking ——— |
| 3360 | |
| 3361 | function utcDateString() { |
| 3362 | return new Date().toISOString().slice(0, 10); |
| 3363 | } |
| 3364 | |
| 3365 | function utcMonthString() { |
| 3366 | return new Date().toISOString().slice(0, 7); |
| 3367 | } |
| 3368 | |
| 3369 | /** Blobs key for per-user consolidation cost record. */ |
| 3370 | function consolidationCostBlobKey(uid) { |
| 3371 | return `memory/${uid}/consolidation-cost`; |
| 3372 | } |
| 3373 | |
| 3374 | /** Load consolidation cost record — Blobs on hosted, file on self-hosted. */ |
| 3375 | async function loadConsolidationCost(uid) { |
| 3376 | const store = globalThis.__netlify_blob_store; |
| 3377 | if (store) { |
| 3378 | try { |
| 3379 | const raw = await store.get(consolidationCostBlobKey(uid), { type: 'text' }); |
| 3380 | if (!raw) return {}; |
| 3381 | return JSON.parse(raw) || {}; |
| 3382 | } catch (_) { return {}; } |
| 3383 | } |
| 3384 | const filePath = path.join(DATA_DIR, 'consolidation', uid + '_cost.json'); |
| 3385 | try { |
| 3386 | const raw = JSON.parse(fs.readFileSync(filePath, 'utf8')); |
| 3387 | return raw && typeof raw === 'object' ? raw : {}; |
| 3388 | } catch (_) { return {}; } |
| 3389 | } |
| 3390 | |
| 3391 | /** Persist consolidation cost record — Blobs on hosted, file on self-hosted. */ |
| 3392 | async function saveConsolidationCost(uid, data) { |
| 3393 | const store = globalThis.__netlify_blob_store; |
| 3394 | if (store) { |
| 3395 | try { |
| 3396 | await store.set(consolidationCostBlobKey(uid), JSON.stringify(data)); |
| 3397 | } catch (_) {} |
| 3398 | return; |
| 3399 | } |
| 3400 | const filePath = path.join(DATA_DIR, 'consolidation', uid + '_cost.json'); |
| 3401 | const dir = path.dirname(filePath); |
| 3402 | if (!fs.existsSync(dir)) fs.mkdirSync(dir, { recursive: true }); |
| 3403 | fs.writeFileSync(filePath, JSON.stringify(data, null, 2), 'utf8'); |
| 3404 | } |
| 3405 | |
| 3406 | async function recordConsolidationPass(uid, costUsd) { |
| 3407 | const rec = await loadConsolidationCost(uid); |
| 3408 | const today = utcDateString(); |
| 3409 | const month = utcMonthString(); |
| 3410 | const prevCostDate = rec.cost_date; |
| 3411 | const prevMonth = rec.pass_month; |
| 3412 | return { |
| 3413 | last_pass: new Date().toISOString(), |
| 3414 | cost_today_usd: prevCostDate === today ? Number((rec.cost_today_usd || 0) + costUsd) : costUsd, |
| 3415 | cost_date: today, |
| 3416 | cost_cap_usd: process.env.CONSOLIDATION_COST_CAP_USD ? parseFloat(process.env.CONSOLIDATION_COST_CAP_USD) : null, |
| 3417 | pass_count_month: prevMonth === month ? (rec.pass_count_month || 0) + 1 : 1, |
| 3418 | pass_month: month, |
| 3419 | }; |
| 3420 | } |
| 3421 | |
| 3422 | /** |
| 3423 | * POST /api/v1/memory/consolidate |
| 3424 | * Body: { dry_run?, passes?, lookback_hours?, max_events_per_pass?, max_topics_per_pass?, llm?: { max_tokens? } } |
| 3425 | * Response: { topics, total_events, verify, discover, cost_usd, pass_id } |
| 3426 | */ |
| 3427 | app.post('/api/v1/memory/consolidate', requireBridgeAuth, requireBridgeEditorOrAdmin, express.json(), async (req, res) => { |
| 3428 | const { uid, vaultId } = bridgeMemoryAuth(req); |
| 3429 | |
| 3430 | const llmApiKey = process.env.CONSOLIDATION_LLM_API_KEY || process.env.OPENAI_API_KEY; |
| 3431 | if (!llmApiKey) { |
| 3432 | return res.status(503).json({ |
| 3433 | error: 'No LLM API key configured for hosted consolidation (CONSOLIDATION_LLM_API_KEY or OPENAI_API_KEY).', |
| 3434 | code: 'LLM_NOT_CONFIGURED', |
| 3435 | }); |
| 3436 | } |
| 3437 | |
| 3438 | const mergedBody = |
| 3439 | req.body && typeof req.body === 'object' ? { ...req.body } : {}; |
| 3440 | |
| 3441 | const { dry_run, passes } = mergedBody; |
| 3442 | |
| 3443 | // 30-minute server-side cooldown on real (non-dry-run) passes to prevent runaway costs. |
| 3444 | // Automated scheduler runs respect their own configured interval; this guards manual triggers. |
| 3445 | if (!dry_run) { |
| 3446 | try { |
| 3447 | const costRec = await loadConsolidationCost(uid); |
| 3448 | const lastPassAt = costRec?.last_pass; |
| 3449 | if (lastPassAt) { |
| 3450 | const elapsedMs = Date.now() - new Date(lastPassAt).getTime(); |
| 3451 | const cooldownMs = 30 * 60 * 1000; |
| 3452 | if (elapsedMs < cooldownMs) { |
| 3453 | const waitMin = Math.ceil((cooldownMs - elapsedMs) / 60_000); |
| 3454 | return res.status(429).json({ |
| 3455 | error: `Consolidation available again in ${waitMin} minute${waitMin === 1 ? '' : 's'}.`, |
| 3456 | code: 'RATE_LIMITED', |
| 3457 | retry_after_minutes: waitMin, |
| 3458 | }); |
| 3459 | } |
| 3460 | } |
| 3461 | } catch (_) { |
| 3462 | // If cost record can't be read, allow the pass through — don't block on a read error. |
| 3463 | } |
| 3464 | } |
| 3465 | |
| 3466 | try { |
| 3467 | const { FileMemoryProvider } = await import('../../lib/memory-provider-file.mjs'); |
| 3468 | const { MemoryManager } = await import('../../lib/memory.mjs'); |
| 3469 | const { consolidateMemory } = await import('../../lib/memory-consolidate.mjs'); |
| 3470 | const { computeCallCost } = await import('../../lib/daemon-cost.mjs'); |
| 3471 | const { createMemoryEvent } = await import('../../lib/memory-event.mjs'); |
| 3472 | |
| 3473 | // Hosted (Blobs): load events into a temp FileMemoryProvider so consolidateMemory |
| 3474 | // can read and write to it, then sync remaining events back to Blobs. |
| 3475 | // Self-hosted: use the normal file-based memory directory. |
| 3476 | let mm; |
| 3477 | let tempDir = null; |
| 3478 | const isHostedBlobs = Boolean(globalThis.__netlify_blob_store); |
| 3479 | |
| 3480 | if (isHostedBlobs) { |
| 3481 | const rawEvents = await blobsGetMemoryEvents(uid, vaultId); |
| 3482 | tempDir = fs.mkdtempSync(path.join(os.tmpdir(), 'knowtation-mm-')); |
| 3483 | if (rawEvents.length > 0) { |
| 3484 | fs.writeFileSync( |
| 3485 | path.join(tempDir, 'events.jsonl'), |
| 3486 | rawEvents.map((e) => JSON.stringify(e)).join('\n') + '\n', |
| 3487 | 'utf8', |
| 3488 | ); |
| 3489 | } |
| 3490 | mm = new MemoryManager(new FileMemoryProvider(tempDir)); |
| 3491 | } else { |
| 3492 | mm = new MemoryManager(new FileMemoryProvider(bridgeMemoryDir(uid, vaultId))); |
| 3493 | } |
| 3494 | |
| 3495 | const maxTok = |
| 3496 | mergedBody.llm && typeof mergedBody.llm === 'object' && mergedBody.llm.max_tokens != null |
| 3497 | ? Math.floor(Number(mergedBody.llm.max_tokens)) |
| 3498 | : 1024; |
| 3499 | const lbH = |
| 3500 | mergedBody.lookback_hours != null && Number.isFinite(Number(mergedBody.lookback_hours)) |
| 3501 | ? Number(mergedBody.lookback_hours) |
| 3502 | : 24; |
| 3503 | const maxEv = |
| 3504 | mergedBody.max_events_per_pass != null && Number.isFinite(Number(mergedBody.max_events_per_pass)) |
| 3505 | ? Number(mergedBody.max_events_per_pass) |
| 3506 | : 200; |
| 3507 | const maxTop = |
| 3508 | mergedBody.max_topics_per_pass != null && Number.isFinite(Number(mergedBody.max_topics_per_pass)) |
| 3509 | ? Number(mergedBody.max_topics_per_pass) |
| 3510 | : 10; |
| 3511 | |
| 3512 | const consolidationConfig = { |
| 3513 | data_dir: isHostedBlobs ? os.tmpdir() : DATA_DIR, |
| 3514 | llm: { |
| 3515 | provider: 'openai', |
| 3516 | api_key: llmApiKey, |
| 3517 | model: process.env.CONSOLIDATION_LLM_MODEL || 'gpt-4o-mini', |
| 3518 | }, |
| 3519 | daemon: { |
| 3520 | lookback_hours: lbH, |
| 3521 | max_events_per_pass: maxEv, |
| 3522 | max_topics_per_pass: maxTop, |
| 3523 | llm: { max_tokens: Number.isFinite(maxTok) ? maxTok : 1024 }, |
| 3524 | }, |
| 3525 | memory: { |
| 3526 | provider: 'file', |
| 3527 | encrypt: process.env.CONSOLIDATION_MEMORY_ENCRYPT === 'true', |
| 3528 | }, |
| 3529 | }; |
| 3530 | |
| 3531 | // Track LLM call cost via a wrapping llmFn. |
| 3532 | let totalCostUsd = 0; |
| 3533 | const { completeChat } = await import('../../lib/llm-complete.mjs'); |
| 3534 | const trackingLlmFn = async (cfg, callOpts) => { |
| 3535 | const rawResponse = await completeChat(consolidationConfig, callOpts); |
| 3536 | totalCostUsd += computeCallCost(callOpts, rawResponse); |
| 3537 | return rawResponse; |
| 3538 | }; |
| 3539 | |
| 3540 | const result = await consolidateMemory(consolidationConfig, { |
| 3541 | mm, |
| 3542 | dryRun: Boolean(dry_run), |
| 3543 | passes: passes ?? undefined, |
| 3544 | llmFn: dry_run ? undefined : trackingLlmFn, |
| 3545 | }); |
| 3546 | |
| 3547 | const pass_id = 'cpass_' + Date.now().toString(36) + '_' + Math.random().toString(36).slice(2, 6); |
| 3548 | |
| 3549 | if (!dry_run) { |
| 3550 | const updated = await recordConsolidationPass(uid, totalCostUsd); |
| 3551 | await saveConsolidationCost(uid, updated); |
| 3552 | |
| 3553 | // Store pass-level summary event. |
| 3554 | const passEvent = createMemoryEvent('consolidation_pass', { |
| 3555 | topics_count: Array.isArray(result.topics) ? result.topics.length : (result.topics ?? 0), |
| 3556 | total_events: result.total_events, |
| 3557 | cost_usd: totalCostUsd, |
| 3558 | pass_id, |
| 3559 | verify: result.verify ?? null, |
| 3560 | discover: result.discover ?? null, |
| 3561 | }); |
| 3562 | |
| 3563 | if (isHostedBlobs) { |
| 3564 | // Sync remaining events (post-consolidation) + pass summary back to Blobs. |
| 3565 | const remaining = mm.list({ limit: 500 }); |
| 3566 | await blobsSetMemoryEvents(uid, vaultId, [...remaining, passEvent]); |
| 3567 | } else { |
| 3568 | mm.store('consolidation_pass', passEvent.data); |
| 3569 | } |
| 3570 | } |
| 3571 | |
| 3572 | // Clean up temp dir used for hosted path. |
| 3573 | if (tempDir) { |
| 3574 | try { fs.rmSync(tempDir, { recursive: true, force: true }); } catch (_) {} |
| 3575 | } |
| 3576 | |
| 3577 | return res.json({ |
| 3578 | topics: result.topics, |
| 3579 | total_events: result.total_events, |
| 3580 | verify: result.verify ?? null, |
| 3581 | discover: result.discover ?? null, |
| 3582 | cost_usd: totalCostUsd, |
| 3583 | pass_id, |
| 3584 | dry_run: result.dry_run, |
| 3585 | }); |
| 3586 | } catch (e) { |
| 3587 | console.error('[bridge] POST /api/v1/memory/consolidate', e?.message); |
| 3588 | res.status(500).json({ error: e.message || 'Consolidation failed', code: 'RUNTIME_ERROR' }); |
| 3589 | } |
| 3590 | }); |
| 3591 | |
| 3592 | /** |
| 3593 | * GET /api/v1/memory/consolidate/status |
| 3594 | * Response: { last_pass, cost_today_usd, cost_cap_usd, pass_count_month } |
| 3595 | */ |
| 3596 | app.get('/api/v1/memory/consolidate/status', async (req, res) => { |
| 3597 | const { uid } = bridgeMemoryAuth(req); |
| 3598 | if (!uid) return res.status(401).json({ error: 'Unauthorized', code: 'UNAUTHORIZED' }); |
| 3599 | |
| 3600 | try { |
| 3601 | const rec = await loadConsolidationCost(uid); |
| 3602 | const today = utcDateString(); |
| 3603 | const month = utcMonthString(); |
| 3604 | const passCountMonth = rec.pass_month === month ? (rec.pass_count_month || 0) : 0; |
| 3605 | |
| 3606 | // Cooldown: minutes until the next manual consolidation is available. |
| 3607 | const lastPass = rec.last_pass ?? null; |
| 3608 | let cooldownMinutes = 0; |
| 3609 | if (lastPass) { |
| 3610 | const elapsedMs = Date.now() - new Date(lastPass).getTime(); |
| 3611 | const remaining = 30 * 60 * 1000 - elapsedMs; |
| 3612 | cooldownMinutes = remaining > 0 ? Math.ceil(remaining / 60_000) : 0; |
| 3613 | } |
| 3614 | |
| 3615 | return res.json({ |
| 3616 | last_pass: lastPass, |
| 3617 | pass_count_month: passCountMonth, |
| 3618 | cooldown_minutes: cooldownMinutes, |
| 3619 | // Legacy cost fields kept for backward compat |
| 3620 | cost_today_usd: rec.cost_date === today ? (rec.cost_today_usd || 0) : 0, |
| 3621 | cost_cap_usd: process.env.CONSOLIDATION_COST_CAP_USD |
| 3622 | ? parseFloat(process.env.CONSOLIDATION_COST_CAP_USD) |
| 3623 | : null, |
| 3624 | }); |
| 3625 | } catch (e) { |
| 3626 | console.error('[bridge] GET /api/v1/memory/consolidate/status', e?.message); |
| 3627 | res.status(500).json({ error: e.message || 'Internal error', code: 'RUNTIME_ERROR' }); |
| 3628 | } |
| 3629 | }); |
| 3630 | |
| 3631 | if (!isServerless) { |
| 3632 | if (!CANISTER_URL || !SESSION_SECRET) { |
| 3633 | console.error('Bridge: CANISTER_URL and SESSION_SECRET (or HUB_JWT_SECRET) are required.'); |
| 3634 | console.error(' Add them to the repo root .env (bridge loads ../../.env) or export in your shell.'); |
| 3635 | console.error(' Template: hub/bridge/.env.example'); |
| 3636 | process.exit(1); |
| 3637 | } |
| 3638 | app.listen(PORT, () => { |
| 3639 | console.log('Knowtation Hub Bridge listening on http://localhost:' + PORT); |
| 3640 | console.log(' Canister: ' + CANISTER_URL); |
| 3641 | console.log(' GitHub connect: ' + (process.env.GITHUB_CLIENT_ID ? 'enabled' : 'not configured')); |
| 3642 | console.log(' Index/Search: ' + (process.env.EMBEDDING_PROVIDER || 'ollama') + ' (run POST /api/v1/index to index)'); |
| 3643 | }); |
| 3644 | } |
| 3645 | |
| 3646 | export { app }; |
File History
1 commit
sha256:dc02e0ce2c9c3616e8ce241f8a700e1558caae53cb1af50daeffdb2685cf9c3d
feat(delegation): 7C-L1 live gate smoke + hosted bridge/gat…
Human
minor
⚠
31 days ago