@arbidocs/cli 0.3.47 → 0.3.49

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@arbidocs/cli",
-  "version": "0.3.47",
+  "version": "0.3.49",
   "description": "CLI tool for interacting with ARBI — login, manage workspaces, upload documents, query the RAG assistant",
   "main": "dist/index.js",
   "bin": {
@@ -21,9 +21,9 @@
     "preuninstall": "node scripts/preuninstall.js"
   },
   "dependencies": {
-    "@arbidocs/sdk": "0.3.47",
-    "@arbidocs/client": "0.3.47",
-    "@arbidocs/tui": "0.3.47",
+    "@arbidocs/sdk": "0.3.49",
+    "@arbidocs/client": "0.3.49",
+    "@arbidocs/tui": "0.3.49",
     "@inquirer/prompts": "^8.2.0",
     "chalk": "^5.6.2",
     "commander": "^13.1.0"

package/scripts/time-workspace-open.mjs

@@ -0,0 +1,245 @@
+#!/usr/bin/env node
+/**
+ * time-workspace-open.mjs
+ *
+ * Times the POST /v1/workspace/{id}/open call with a detailed breakdown:
+ *
+ *   - setup       — auth + crypto (workspace key unseal/seal). Excluded from "open" time.
+ *   - ttfb        — POST request start → first byte.  ≈ network RTT + backend processing
+ *                   (DB query, decrypt, serialize JSON, begin streaming).
+ *   - download    — first byte → last byte.  ≈ wire transfer time for the payload.
+ *   - parse       — JSON.parse on the raw body.
+ *   - total       — ttfb + download + parse.
+ *
+ *   Also reports:
+ *   - bytes       — raw response body size
+ *   - docs        — documents[] count (if present)
+ *   - per-doc     — bytes / docs
+ *
+ * Usage:
+ *   node packages/arbi-cli/scripts/time-workspace-open.mjs              # uses selected workspace
+ *   node packages/arbi-cli/scripts/time-workspace-open.mjs wrk-820e4c6d
+ *   node packages/arbi-cli/scripts/time-workspace-open.mjs wrk-820e4c6d --runs=3
+ *
+ * Requires:
+ *   - `arbi login` has been run with a valid cached session
+ *   - the SDK is built (it's an internal workspace dep so fresh-from-source works too)
+ */
+
+// Stub IndexedDB + structuredClone before anything else — @arbidocs/client's login
+// flow calls saveSession() which writes to IndexedDB; in Node we use fake-indexeddb.
+import 'fake-indexeddb/auto'
+
+import { performance } from 'node:perf_hooks'
+import { readFileSync } from 'node:fs'
+import { homedir } from 'node:os'
+import path from 'node:path'
+
+// ─── arg parsing ─────────────────────────────────────────────────────────
+const rawArgs = process.argv.slice(2)
+const positional = []
+const flags = {}
+for (const raw of rawArgs) {
+  if (raw.startsWith('--')) {
+    const m = raw.match(/^--([^=]+)(?:=(.*))?$/)
+    if (m) flags[m[1]] = m[2] ?? 'true'
+  } else {
+    positional.push(raw)
+  }
+}
+const RUNS = Number(flags.runs ?? 1)
+const WORKSPACE_ARG = positional[0]
+
+// ─── load CLI config/credentials from ~/.arbi ────────────────────────────
+const arbiDir = path.join(homedir(), '.arbi')
+const config = JSON.parse(readFileSync(path.join(arbiDir, 'config.json'), 'utf8'))
+const credentials = JSON.parse(readFileSync(path.join(arbiDir, 'credentials.json'), 'utf8'))
+
+const workspaceId = WORKSPACE_ARG ?? config.selectedWorkspaceId
+if (!workspaceId) {
+  console.error('No workspace ID given and none selected.')
+  process.exit(1)
+}
+
+console.log(`server:    ${config.baseUrl}`)
+console.log(`workspace: ${workspaceId}`)
+console.log(`runs:      ${RUNS}`)
+console.log('')
+
+// ─── import SDK + client ─────────────────────────────────────────────────
+const { createArbiClient, base64ToBytes } = await import('@arbidocs/client')
+const { generateEncryptedWorkspaceKey } = await import('@arbidocs/sdk')
+
+// ─── per-run timing ──────────────────────────────────────────────────────
+async function timeOneRun() {
+  const t0 = performance.now()
+
+  // ── setup: fresh arbi client + login to get a session ──────────────────
+  // Minimal in-memory sessionStorage shim so @arbidocs/client doesn't fall back
+  // to IndexedDB (which doesn't exist in Node). Only used by the auto-relogin
+  // middleware — we never actually trigger it since we log in fresh here.
+  const memorySessionStorage = {
+    async getSession() {
+      return {
+        signingPrivateKey: base64ToBytes(credentials.signingPrivateKeyBase64),
+        serverSessionKey: credentials.serverSessionKeyBase64
+          ? base64ToBytes(credentials.serverSessionKeyBase64)
+          : undefined,
+      }
+    },
+    async saveSession() {},
+    async clearSession() {},
+  }
+
+  const arbi = createArbiClient({
+    baseUrl: config.baseUrl,
+    deploymentDomain: config.deploymentDomain,
+    credentials: 'omit',
+    sessionStorage: memorySessionStorage,
+  })
+  await arbi.crypto.initSodium()
+
+  const signingPrivateKey = base64ToBytes(credentials.signingPrivateKeyBase64)
+  const loginResult = await arbi.auth.loginWithKey({
+    email: credentials.email,
+    signingPrivateKey,
+    ssoToken: credentials.ssoToken,
+  })
+
+  // Fetch workspace list to get the wrapped key
+  const { data: workspaces, error: wsError } = await arbi.fetch.GET('/v1/user/workspaces')
+  if (wsError || !workspaces) throw new Error('Failed to fetch workspaces')
+  const ws = workspaces.find((w) => w.external_id === workspaceId)
+  if (!ws) throw new Error(`workspace ${workspaceId} not found`)
+  if (!ws.wrapped_key) throw new Error(`workspace ${workspaceId} has no wrapped_key`)
+
+  const encryptedWorkspaceKey = await generateEncryptedWorkspaceKey(
+    arbi,
+    ws.wrapped_key,
+    loginResult.serverSessionKey,
+    credentials.signingPrivateKeyBase64
+  )
+
+  const setupMs = performance.now() - t0
+  const docCountHint = (ws.shared_document_count ?? 0) + (ws.private_document_count ?? 0)
+
+  // ── the actual /open POST with streaming fetch for TTFB measurement ────
+  const qs = flags['no-documents'] ? '?include_documents=false' : ''
+  const url = `${config.baseUrl}/v1/workspace/${workspaceId}/open${qs}`
+  const requestBody = JSON.stringify({ workspace_key: encryptedWorkspaceKey })
+
+  const tPostStart = performance.now()
+  const res = await fetch(url, {
+    method: 'POST',
+    headers: {
+      'content-type': 'application/json',
+      authorization: `Bearer ${loginResult.accessToken}`,
+    },
+    body: requestBody,
+  })
+  const tHeaders = performance.now() // headers received ≈ TTFB
+
+  if (!res.ok) {
+    const text = await res.text()
+    throw new Error(`POST /open failed ${res.status}: ${text.slice(0, 500)}`)
+  }
+
+  // Stream the body to measure download time separately from TTFB
+  const reader = res.body.getReader()
+  const chunks = []
+  let bytes = 0
+  // eslint-disable-next-line no-constant-condition
+  while (true) {
+    const { done, value } = await reader.read()
+    if (done) break
+    chunks.push(value)
+    bytes += value.byteLength
+  }
+  const tDownload = performance.now()
+
+  // Assemble + parse
+  const body = new Uint8Array(bytes)
+  let offset = 0
+  for (const chunk of chunks) {
+    body.set(chunk, offset)
+    offset += chunk.byteLength
+  }
+  const text = new TextDecoder().decode(body)
+  const json = JSON.parse(text)
+  const tParse = performance.now()
+
+  const ttfbMs = tHeaders - tPostStart
+  const downloadMs = tDownload - tHeaders
+  const parseMs = tParse - tDownload
+  const totalMs = tParse - tPostStart
+
+  const docs = Array.isArray(json.documents) ? json.documents.length : 0
+  const conversations = Array.isArray(json.conversations) ? json.conversations.length : 0
+  const tags = Array.isArray(json.tags) ? json.tags.length : 0
+
+  return {
+    setupMs,
+    ttfbMs,
+    downloadMs,
+    parseMs,
+    totalMs,
+    bytes,
+    docs,
+    conversations,
+    tags,
+    docCountHint,
+  }
+}
+
+function fmt(ms) {
+  return `${ms.toFixed(1).padStart(8)} ms`
+}
+function fmtBytes(n) {
+  if (n < 1024) return `${n} B`
+  if (n < 1024 * 1024) return `${(n / 1024).toFixed(1)} KB`
+  return `${(n / 1024 / 1024).toFixed(2)} MB`
+}
+
+// ─── run it ──────────────────────────────────────────────────────────────
+const results = []
+for (let i = 1; i <= RUNS; i++) {
+  process.stdout.write(`run ${i}/${RUNS} ... `)
+  try {
+    const r = await timeOneRun()
+    results.push(r)
+    console.log(
+      `total=${r.totalMs.toFixed(0)}ms ttfb=${r.ttfbMs.toFixed(0)}ms dl=${r.downloadMs.toFixed(0)}ms parse=${r.parseMs.toFixed(0)}ms size=${fmtBytes(r.bytes)} docs=${r.docs}`
+    )
+  } catch (err) {
+    console.log(`FAILED: ${err.message}`)
+    process.exit(1)
+  }
+}
+
+// ─── summary ─────────────────────────────────────────────────────────────
+console.log('')
+console.log('─── summary ──────────────────────────────────────────')
+const avg = (key) => results.reduce((s, r) => s + r[key], 0) / results.length
+const last = results[results.length - 1]
+
+console.log(`setup      ${fmt(avg('setupMs'))}  (auth + key prep, excluded from /open total)`)
+console.log(`ttfb       ${fmt(avg('ttfbMs'))}  (network RTT + backend: DB + decrypt + serialize)`)
+console.log(`download   ${fmt(avg('downloadMs'))}  (wire transfer of JSON body)`)
+console.log(`parse      ${fmt(avg('parseMs'))}  (JSON.parse on client)`)
+console.log(`─────────────────────────────────────`)
+console.log(`total      ${fmt(avg('totalMs'))}  (ttfb + download + parse)`)
+console.log('')
+console.log(`payload    ${fmtBytes(last.bytes)}  (${last.bytes.toLocaleString()} bytes)`)
+console.log(`documents  ${last.docs}`)
+console.log(`conversat. ${last.conversations}`)
+console.log(`tags       ${last.tags}`)
+if (last.docs > 0) {
+  const perDoc = last.bytes / last.docs
+  console.log(`per-doc    ${fmtBytes(perDoc)}  ${perDoc.toFixed(0)} B/doc`)
+  console.log('')
+  console.log('─── extrapolation ────────────────────────────────────')
+  for (const n of [500, 2000, 10000, 40000]) {
+    const projBytes = perDoc * n
+    console.log(`  ${String(n).padStart(6)} docs → ${fmtBytes(projBytes).padStart(10)}`)
+  }
+}