@askdb/studio 0.2.0-beta.2 → 0.2.0-beta.21

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/dist/server.js CHANGED
@@ -1,18 +1,26 @@
1
- import { existsSync, readFileSync, rmSync } from "node:fs";
1
+ import { randomUUID } from "node:crypto";
2
+ import { existsSync, readFileSync, rmSync, writeFileSync } from "node:fs";
2
3
  import { createServer } from "node:http";
3
4
  import { extname, relative, join, resolve } from "node:path";
4
5
  import { fileURLToPath } from "node:url";
5
6
  import { generateText as defaultGenerateText } from "ai";
6
7
  import { getAskDbRuntimeConfig } from "@askdb/config";
7
- import { ask, askDbAiKeyMissingMessage, createAskDbEmbeddingModel, createAskDbLanguageModelFromEnv, isBuiltInDialectId, loadSchema, resolveAskDbAiConfig, resolveAskDbEmbeddingConfig, suggestEnrichment, } from "@askdb/core";
8
- import { buildSchemaIndex, chunkContentHash, chunkSchema, createRetriever, loadChunkerSourcesFromDir, } from "@askdb/rag";
9
- import { createAiSdkEmbedder } from "@askdb/rag/embedders/ai-sdk";
10
- import { createFileStore } from "@askdb/rag/stores/file";
11
- import { buildDefaultTableBody, buildFrontmatter, buildSuggestionContext, buildSuggestionTarget, buildTableDraft, loadWorkspace, replaceH2Section, replaceTableDescription, saveTable, } from "@askdb/enrich";
8
+ import { aiKeyMissingMessage, createAiRegistry, resolveAiConfig, resolveEmbeddingConfig, } from "@askdb/ai";
9
+ import { azureProvider } from "@askdb/ai-azure";
10
+ import { googleProvider } from "@askdb/ai-google";
11
+ import { openaiProvider } from "@askdb/ai-openai";
12
+ import { ask, isBuiltInDialectId, loadSchema, suggestEnrichment, tenantScopeSchema, writeTenantPolicyMarkdown, tenantPolicyFrontmatterSchema, } from "@askdb/core";
13
+ import { buildSchemaIndex, chunkContentHash, chunkSchema, createAiSdkEmbedder, createFileStore, createMemoryStore, createPgvectorStore, createRetriever, loadChunkerSourcesFromDir, } from "@askdb/rag";
14
+ import { buildDefaultTableBody, buildFrontmatter, buildSuggestionContext, buildSuggestionTarget, buildTableDraft, loadWorkspace, replaceH2Section, replaceTableDescription, saveConcepts, saveTable, } from "@askdb/enrich";
15
+ const ai = createAiRegistry([openaiProvider, azureProvider, googleProvider]);
12
16
  const CLIENT_DIR = fileURLToPath(new URL("./client/", import.meta.url));
13
17
  const STUDIO_RAG_MOCK_DIMENSIONS = 64;
14
18
  const STUDIO_RAG_MOCK_EMBEDDER_ID = `studio:mock-lexical-${STUDIO_RAG_MOCK_DIMENSIONS}`;
15
19
  const DEFAULT_EMBEDDING_MODEL = "text-embedding-3-small";
20
+ let studioPgvectorStoreFactoryForTests;
21
+ export function setStudioPgvectorStoreFactoryForTests(factory) {
22
+ studioPgvectorStoreFactoryForTests = factory;
23
+ }
16
24
  export function createStudioServer(options) {
17
25
  const schemaDir = resolve(options.schema);
18
26
  if (!existsSync(schemaDir)) {
@@ -41,6 +49,22 @@ export function createStudioServer(options) {
41
49
  saveDraft(state, tableId, draft);
42
50
  return writeJson(res, 200, serializeWorkspace(state.workspace));
43
51
  }
52
+ if (req.method === "POST" && url.pathname === "/api/concepts") {
53
+ const body = await readJson(req);
54
+ const concepts = parseConceptsBody(body);
55
+ saveConceptsDraft(state, concepts);
56
+ return writeJson(res, 200, serializeWorkspace(state.workspace));
57
+ }
58
+ if (req.method === "POST" && url.pathname === "/api/tenant-policy") {
59
+ const body = await readJson(req);
60
+ const { frontmatter, bodyText } = parseTenantPolicyBody(body);
61
+ saveTenantPolicy(state, frontmatter, bodyText);
62
+ return writeJson(res, 200, serializeWorkspace(state.workspace));
63
+ }
64
+ if (req.method === "POST" && url.pathname === "/api/suggest-tenant-policy") {
65
+ const draft = await suggestTenantPolicyDraft(state);
66
+ return writeJson(res, 200, draft);
67
+ }
44
68
  if (req.method === "POST" && url.pathname === "/api/suggest") {
45
69
  const body = await readJson(req);
46
70
  const source = parseSuggestSource(body);
@@ -48,22 +72,41 @@ export function createStudioServer(options) {
48
72
  return writeJson(res, 200, { candidates });
49
73
  }
50
74
  if (req.method === "GET" && url.pathname === "/api/rag/status") {
51
- return writeJson(res, 200, getRagStatus(state.schemaDir));
75
+ return writeJson(res, 200, await getRagStatus(state));
52
76
  }
53
77
  if (req.method === "POST" && url.pathname === "/api/rag/index") {
54
- const result = await indexRag(state.schemaDir);
78
+ const result = await indexRag(state);
55
79
  return writeJson(res, 200, result);
56
80
  }
57
81
  if (req.method === "POST" && url.pathname === "/api/rag/query") {
58
82
  const body = await readJson(req);
59
83
  const query = parseRagQuery(body);
60
- const result = await queryRag(state.schemaDir, query);
84
+ const result = await queryRag(state, query);
61
85
  return writeJson(res, 200, result);
62
86
  }
63
87
  if (req.method === "POST" && url.pathname === "/api/ask") {
64
88
  const body = await readJson(req);
65
89
  const options = parseAskBody(body);
66
- const result = await askSampleQuestion(state.schemaDir, options);
90
+ const result = await askSampleQuestion(state, options);
91
+ return writeJson(res, 200, result);
92
+ }
93
+ if (req.method === "GET" && url.pathname === "/api/history") {
94
+ return writeJson(res, 200, readPlaygroundHistory(state.schemaDir));
95
+ }
96
+ if (req.method === "POST" && url.pathname === "/api/history") {
97
+ const body = await readJson(req);
98
+ const entry = parsePlaygroundHistoryEntry(body);
99
+ appendPlaygroundHistory(state.schemaDir, entry);
100
+ return writeJson(res, 200, { ok: true });
101
+ }
102
+ if (req.method === "DELETE" && url.pathname.startsWith("/api/history/")) {
103
+ const id = decodeURIComponent(url.pathname.slice("/api/history/".length));
104
+ const result = deletePlaygroundHistoryEntry(state.schemaDir, id);
105
+ return writeJson(res, 200, result);
106
+ }
107
+ if (req.method === "POST" && url.pathname === "/api/execute") {
108
+ const body = await readJson(req);
109
+ const result = await executeQuery(body);
67
110
  return writeJson(res, 200, result);
68
111
  }
69
112
  if (req.method === "GET" && !url.pathname.startsWith("/api/")) {
@@ -83,9 +126,7 @@ export function serializeWorkspace(workspace) {
83
126
  const rt = getAskDbRuntimeConfig();
84
127
  const aiConfig = (() => {
85
128
  try {
86
- return resolveAskDbAiConfig(rt.ai.aiEnv, {
87
- modelEnvVar: rt.ai.studioModel ? "ASKDB_STUDIO_MODEL" : undefined,
88
- });
129
+ return resolveAiConfig(rt.ai.aiEnv);
89
130
  }
90
131
  catch {
91
132
  // A misconfigured AI env (e.g. azure without resourceName) shouldn't crash the workspace
@@ -113,8 +154,33 @@ export function serializeWorkspace(workspace) {
113
154
  };
114
155
  }),
115
156
  concepts: workspace.concepts?.frontmatter.concepts ?? [],
157
+ tenantPolicy: (() => {
158
+ const schema = loadSchema(workspace.schemaDir);
159
+ return schema.tenantPolicy ?? null;
160
+ })(),
116
161
  };
117
162
  }
163
+ function saveConceptsDraft(state, concepts) {
164
+ saveConcepts(state.workspace, { concepts });
165
+ state.workspace = loadWorkspace(state.schemaDir);
166
+ }
167
+ function saveTenantPolicy(state, frontmatter, body) {
168
+ const filePath = join(state.schemaDir, "tenant-policy.md");
169
+ const md = writeTenantPolicyMarkdown(frontmatter, body);
170
+ writeFileSync(filePath, md, "utf8");
171
+ state.workspace = loadWorkspace(state.schemaDir);
172
+ }
173
+ function parseTenantPolicyBody(body) {
174
+ if (!isRecord(body) || !isRecord(body.frontmatter)) {
175
+ throw new StudioHttpError(400, "`frontmatter` object is required.");
176
+ }
177
+ const parsed = tenantPolicyFrontmatterSchema.safeParse(body.frontmatter);
178
+ if (!parsed.success) {
179
+ throw new StudioHttpError(400, `Invalid tenant policy: ${parsed.error.message}`);
180
+ }
181
+ const bodyText = typeof body.body === "string" ? body.body : "";
182
+ return { frontmatter: parsed.data, bodyText };
183
+ }
118
184
  function saveDraft(state, tableId, draft) {
119
185
  const table = state.workspace.tables.find((candidate) => candidate.physical.id === tableId);
120
186
  if (!table)
@@ -134,38 +200,118 @@ function saveDraft(state, tableId, draft) {
134
200
  }
135
201
  async function suggestForSource(workspace, source) {
136
202
  const rt = getAskDbRuntimeConfig();
137
- const modelEnvVar = rt.ai.studioModel
138
- ? "ASKDB_STUDIO_MODEL"
139
- : rt.ai.tuiModel
140
- ? "ASKDB_TUI_MODEL"
141
- : undefined;
142
- const model = await createAskDbLanguageModelFromEnv(rt.ai.aiEnv, {
143
- modelEnvVar,
144
- });
203
+ const model = await ai.createLanguageModelFromEnv(rt.ai.aiEnv);
145
204
  if (!model) {
146
- throw new StudioHttpError(400, askDbAiKeyMissingMessage("AI enrichment suggestions"));
205
+ throw new StudioHttpError(400, aiKeyMissingMessage("AI enrichment suggestions"));
147
206
  }
148
207
  const candidates = await suggestEnrichment(buildSuggestionTarget(workspace, source), buildSuggestionContext(workspace, source.tableId), model);
149
208
  return candidates.map((candidate) => ({ text: candidate.text }));
150
209
  }
151
- async function askSampleQuestion(schemaDir, options) {
210
+ async function suggestTenantPolicyDraft(state) {
211
+ const rt = getAskDbRuntimeConfig();
212
+ const model = await ai.createLanguageModelFromEnv(rt.ai.aiEnv);
213
+ if (!model) {
214
+ throw new StudioHttpError(400, aiKeyMissingMessage("AI tenant policy suggestion"));
215
+ }
216
+ // Build a compact DDL-like representation of the schema for the prompt
217
+ const workspace = state.workspace;
218
+ const schemaId = workspace.physical.schemaId;
219
+ const schemaDdl = workspace.tables
220
+ .map((t) => {
221
+ const cols = t.physical.columns
222
+ .map((c) => {
223
+ const parts = [c.name, c.type];
224
+ if (c.primaryKey)
225
+ parts.push("PRIMARY KEY");
226
+ if (!c.nullable)
227
+ parts.push("NOT NULL");
228
+ return ` ${parts.join(" ")}`;
229
+ })
230
+ .join(",\n");
231
+ const rels = (t.physical.relationships ?? [])
232
+ .map((r) => ` -- FK: ${r.from} -> ${r.to}`)
233
+ .join("\n");
234
+ return `CREATE TABLE ${t.physical.schema}.${t.physical.name} (\n${cols}\n);\n${rels}`.trim();
235
+ })
236
+ .join("\n\n");
237
+ const tableIds = workspace.tables.map((t) => t.physical.id);
238
+ const systemPrompt = `You are an expert database architect specializing in multi-tenant SaaS systems.
239
+ You analyze database schemas and produce a tenant-policy configuration in a specific JSON format.
240
+
241
+ The tenant policy identifies:
242
+ 1. **roots** — Tables whose rows represent tenants (organizations, companies, accounts, etc).
243
+ Each root has: id (table ID like "table:public.orgs"), tenantIdColumn (column ref like "table:public.orgs#id"), label (human-readable name).
244
+ If a root has a parent root (e.g. sub_org belongs to org), include parent: { root: "<parent-root-id>", foreignKey: "<fk-column-ref>" }.
245
+ 2. **hierarchy** — Edges showing parent→child relationships between roots or tenant-scoped tables.
246
+ Each edge: { parent: "<table-id>", child: "<table-id>", foreignKey: "<child-fk-column-ref>" }.
247
+ 3. **scopedTables** — Tables that belong to a tenant via a direct column or a JOIN chain.
248
+ Each entry: { id: "<table-id>", scopeThrough: [{ root: "<root-id>", column: "<column-ref>" }] } for direct FK,
249
+ or { id: "<table-id>", scopeThrough: [{ root: "<root-id>", join: [{ from: "<col-ref>", to: "<col-ref>" }] }] } for indirect.
250
+ 4. **polymorphicTables** — Tables with a type/id polymorphic association pattern.
251
+ Each entry: { id: "<table-id>", typeColumn: "<col-ref>", idColumn: "<col-ref>", mapping: { "<type_value>": "<target-table-id>" } }.
252
+ 5. **globalTables** — Tables shared across all tenants (lookup tables, reference data). Array of table IDs.
253
+ 6. **enforcement** — "strict" (reject queries on unknown tables) or "warn" (allow with warnings).
254
+
255
+ Column references use the format "table:<schema>.<table>#<column>".
256
+ Table IDs use the format "table:<schema>.<table>".
257
+
258
+ Available table IDs: ${JSON.stringify(tableIds)}
259
+
260
+ Return a JSON object with two keys:
261
+ - "frontmatter": the TenantPolicyFrontmatter object (schemaId, enforcement, roots, hierarchy?, scopedTables?, polymorphicTables?, globalTables?)
262
+ - "body": a markdown body string with sections "## Hierarchy", "## Scope rules", "## Sensitive interactions" explaining the policy in plain English.
263
+
264
+ Return ONLY valid JSON, no markdown fences or explanations.`;
265
+ const userPrompt = `Analyze this database schema and draft a complete multi-tenant policy.
266
+
267
+ Schema ID: ${schemaId}
268
+
269
+ ${schemaDdl}
270
+
271
+ Look for:
272
+ - Tables that represent organizations, companies, accounts, tenants, or workspaces — these are roots.
273
+ - Foreign key columns like org_id, company_id, tenant_id, account_id that scope data to a tenant.
274
+ - Tables with no tenant FK that are clearly shared/global (lookup tables, enums, config).
275
+ - Polymorphic association patterns (type_column + id_column pointing to multiple parent tables).
276
+ - Multi-level hierarchies (org → team → user, company → department → employee).
277
+
278
+ If the schema has no clear multi-tenant pattern, still make your best guess at what would be the tenant root.`;
279
+ const result = await defaultGenerateText({
280
+ model,
281
+ system: systemPrompt,
282
+ prompt: userPrompt,
283
+ });
284
+ let parsed;
285
+ try {
286
+ parsed = JSON.parse(result.text);
287
+ }
288
+ catch {
289
+ throw new StudioHttpError(500, `AI returned invalid JSON for tenant policy suggestion. Raw text: ${result.text.slice(0, 500)}`);
290
+ }
291
+ // Validate frontmatter shape
292
+ const validation = tenantPolicyFrontmatterSchema.safeParse(parsed.frontmatter);
293
+ if (!validation.success) {
294
+ throw new StudioHttpError(500, `AI-generated tenant policy frontmatter is invalid: ${validation.error.issues.map((i) => i.message).join("; ")}`);
295
+ }
296
+ parsed.frontmatter = validation.data;
297
+ return parsed;
298
+ }
299
+ async function askSampleQuestion(state, options) {
152
300
  const rt = getAskDbRuntimeConfig();
153
301
  const mockSql = rt.dev.mockSql;
154
- const aiConfig = mockSql ? undefined : resolveAskDbAiConfig(rt.ai.aiEnv);
302
+ const aiConfig = mockSql ? undefined : resolveAiConfig(rt.ai.aiEnv);
155
303
  if (!mockSql && !aiConfig) {
156
- throw new StudioHttpError(400, `${askDbAiKeyMissingMessage("Sample NL-to-SQL generation")} Set ASKDB_MOCK_SQL to bypass the live model.`);
304
+ throw new StudioHttpError(400, `${aiKeyMissingMessage("Sample NL-to-SQL generation")} Set ASKDB_MOCK_SQL to bypass the live model.`);
157
305
  }
158
- const schema = loadSchema(schemaDir);
306
+ const schema = loadSchema(state.schemaDir);
159
307
  const retrievedChunks = [];
160
308
  const usage = createRequestUsageCollector();
161
309
  const ragIndex = options.useRag
162
- ? await createCurrentStudioRagIndex(schemaDir, "using RAG for sample generation", usage)
310
+ ? await createCurrentStudioRagIndex(state, "using RAG for sample generation", usage)
163
311
  : undefined;
164
312
  const model = mockSql
165
313
  ? undefined
166
- : (await createAskDbLanguageModelFromEnv(rt.ai.aiEnv, {
167
- modelEnvVar: rt.ai.studioModel ? "ASKDB_STUDIO_MODEL" : undefined,
168
- }));
314
+ : (await ai.createLanguageModelFromEnv(rt.ai.aiEnv));
169
315
  const schemaProvider = "provider" in schema && typeof schema.provider === "string"
170
316
  ? schema.provider
171
317
  : undefined;
@@ -177,6 +323,8 @@ async function askSampleQuestion(schemaDir, options) {
177
323
  model,
178
324
  dialect,
179
325
  explain: true,
326
+ ...(options.tenantScope ? { tenantScope: options.tenantScope } : {}),
327
+ ...(options.tenantSqlMode ? { tenantSqlMode: options.tenantSqlMode } : {}),
180
328
  ...(ragIndex
181
329
  ? {
182
330
  retriever: async (params) => {
@@ -185,7 +333,7 @@ async function askSampleQuestion(schemaDir, options) {
185
333
  results = await ragIndex.retriever(params);
186
334
  }
187
335
  catch (error) {
188
- throw formatStudioRagEmbeddingError(error, ragIndex.config);
336
+ throw formatStudioRagOperationError(error, ragIndex.config);
189
337
  }
190
338
  retrievedChunks.splice(0, retrievedChunks.length, ...results);
191
339
  return results;
@@ -202,6 +350,8 @@ async function askSampleQuestion(schemaDir, options) {
202
350
  : {
203
351
  generateText: createTrackedGenerateText(usage),
204
352
  },
353
+ }).finally(async () => {
354
+ await ragIndex?.dispose();
205
355
  });
206
356
  return {
207
357
  sql: result.sql,
@@ -211,90 +361,115 @@ async function askSampleQuestion(schemaDir, options) {
211
361
  enabled: options.useRag,
212
362
  chunks: options.useRag ? retrievedChunks.map(serializeRagResult) : [],
213
363
  },
364
+ tenant: options.tenantScope
365
+ ? {
366
+ enabled: true,
367
+ sqlMode: options.tenantSqlMode ?? "sql-only",
368
+ bindings: result.tenantBindings ?? [],
369
+ params: result.tenantParams ?? [],
370
+ }
371
+ : null,
214
372
  usage: usage.toDto(),
215
373
  };
216
374
  }
217
- function getRagStatus(schemaDir) {
375
+ async function getRagStatus(state) {
218
376
  const config = resolveStudioRagEmbedderConfig();
219
- const sources = loadChunkerSourcesFromDir(schemaDir);
377
+ const sources = loadChunkerSourcesFromDir(state.schemaDir);
220
378
  const chunkResult = chunkSchema(sources);
221
- const lockPath = join(schemaDir, "schema.lock.json");
222
- const embeddingsJsonPath = join(schemaDir, "schema.embeddings.json");
223
- const embeddingsBinPath = join(schemaDir, "schema.embeddings.bin");
379
+ const lockPath = join(state.schemaDir, "schema.lock.json");
224
380
  const lock = readOptionalJson(lockPath);
225
- const embeddings = readOptionalJson(embeddingsJsonPath);
226
381
  const currentHashes = Object.fromEntries(chunkResult.chunks.map((chunk) => [chunk.id, chunkContentHash(chunk.text)]));
227
382
  const lockHashes = lock?.hashes ?? {};
228
383
  const hashIds = Object.keys(currentHashes);
229
- const stale = !lock ||
230
- !existsSync(embeddingsJsonPath) ||
231
- !existsSync(embeddingsBinPath) ||
232
- lock.embedderId !== config.embedderId ||
233
- embeddings?.dimensions !== config.dimensions ||
234
- Object.keys(lockHashes).length !== hashIds.length ||
235
- hashIds.some((id) => lockHashes[id] !== currentHashes[id]);
236
- return {
237
- schemaId: sources.schema.schemaId,
238
- embedder: {
239
- kind: config.kind,
240
- label: config.label,
241
- configured: config.configured,
242
- expectedId: config.embedderId,
243
- indexedId: lock?.embedderId ?? null,
244
- provider: config.kind === "ai-sdk" ? config.provider : null,
245
- model: config.kind === "ai-sdk" ? config.model : null,
246
- baseUrl: config.kind === "ai-sdk" ? config.baseUrl ?? null : null,
247
- },
248
- embedderId: lock?.embedderId ?? config.embedderId,
249
- expectedEmbedderId: config.embedderId,
250
- hasIndex: Boolean(lock && existsSync(embeddingsJsonPath) && existsSync(embeddingsBinPath)),
251
- stale,
252
- updatedAt: lock?.updatedAt ?? null,
253
- chunksTotal: chunkResult.chunks.length,
254
- chunksIndexed: Array.isArray(embeddings?.records) ? embeddings.records.length : 0,
255
- dimensions: typeof embeddings?.dimensions === "number" ? embeddings.dimensions : config.dimensions,
256
- expectedDimensions: config.dimensions,
257
- sensitiveExcluded: chunkResult.stats.sensitiveExcluded,
258
- sensitiveIncluded: chunkResult.stats.sensitiveIncluded,
259
- files: {
260
- lock: existsSync(lockPath),
261
- embeddingsJson: existsSync(embeddingsJsonPath),
262
- embeddingsBin: existsSync(embeddingsBinPath),
263
- },
264
- };
384
+ const store = await openStudioRagStore(state, config.dimensions);
385
+ try {
386
+ const chunksIndexed = await countStudioRagStoreChunks(store, sources.schema.schemaId);
387
+ const stale = !lock ||
388
+ lock.embedderId !== config.embedderId ||
389
+ Object.keys(lockHashes).length !== hashIds.length ||
390
+ hashIds.some((id) => lockHashes[id] !== currentHashes[id]) ||
391
+ chunksIndexed !== hashIds.length;
392
+ const fileArtifacts = store.kind === "file" && store.basePath
393
+ ? {
394
+ lock: existsSync(lockPath),
395
+ embeddingsJson: existsSync(`${store.basePath}.embeddings.json`),
396
+ embeddingsBin: existsSync(`${store.basePath}.embeddings.bin`),
397
+ }
398
+ : {
399
+ lock: existsSync(lockPath),
400
+ embeddingsJson: false,
401
+ embeddingsBin: false,
402
+ };
403
+ return {
404
+ schemaId: sources.schema.schemaId,
405
+ store: {
406
+ kind: store.kind,
407
+ basePath: store.basePath ?? null,
408
+ table: store.table ?? null,
409
+ indexStrategy: store.indexStrategy ?? null,
410
+ },
411
+ embedder: {
412
+ kind: config.kind,
413
+ label: config.label,
414
+ configured: config.configured,
415
+ expectedId: config.embedderId,
416
+ indexedId: lock?.embedderId ?? null,
417
+ provider: config.kind === "ai-sdk" ? config.provider : null,
418
+ model: config.kind === "ai-sdk" ? config.model : null,
419
+ baseUrl: config.kind === "ai-sdk" ? config.baseUrl ?? null : null,
420
+ },
421
+ embedderId: lock?.embedderId ?? config.embedderId,
422
+ expectedEmbedderId: config.embedderId,
423
+ hasIndex: Boolean(lock && chunksIndexed > 0),
424
+ stale,
425
+ updatedAt: lock?.updatedAt ?? null,
426
+ chunksTotal: chunkResult.chunks.length,
427
+ chunksIndexed,
428
+ dimensions: config.dimensions,
429
+ expectedDimensions: config.dimensions,
430
+ sensitiveExcluded: chunkResult.stats.sensitiveExcluded,
431
+ sensitiveIncluded: chunkResult.stats.sensitiveIncluded,
432
+ files: fileArtifacts,
433
+ };
434
+ }
435
+ finally {
436
+ await store.dispose();
437
+ }
265
438
  }
266
- async function indexRag(schemaDir) {
439
+ async function indexRag(state) {
267
440
  const config = resolveStudioRagEmbedderConfig();
268
441
  if (!config.configured) {
269
442
  throw new StudioHttpError(400, studioRagAiSdkKeyMissingMessage());
270
443
  }
271
444
  const usage = createRequestUsageCollector();
272
- clearIncompatibleRagStore(schemaDir, config);
273
- const sources = loadChunkerSourcesFromDir(schemaDir);
274
- const store = createFileStore({ basePath: join(schemaDir, "schema") });
445
+ clearIncompatibleRagStore(state, config);
446
+ const sources = loadChunkerSourcesFromDir(state.schemaDir);
447
+ const store = await openStudioRagStore(state, config.dimensions);
275
448
  let result;
276
449
  try {
277
450
  result = await buildSchemaIndex({
278
451
  schema: sources,
279
452
  embedder: await createStudioRagEmbedder(config, usage),
280
- store,
453
+ store: store.store,
281
454
  embedderId: config.embedderId,
282
- lockFilePath: join(schemaDir, "schema.lock.json"),
455
+ lockFilePath: join(state.schemaDir, "schema.lock.json"),
283
456
  });
284
457
  }
285
458
  catch (error) {
286
- throw formatStudioRagEmbeddingError(error, config);
459
+ throw formatStudioRagOperationError(error, config);
460
+ }
461
+ finally {
462
+ await store.dispose();
287
463
  }
288
- store.flush();
289
464
  return {
290
- status: getRagStatus(schemaDir),
465
+ status: await getRagStatus(state),
291
466
  stats: result.stats,
292
467
  usage: usage.toDto(),
293
468
  };
294
469
  }
295
- async function queryRag(schemaDir, query) {
470
+ async function queryRag(state, query) {
296
471
  const usage = createRequestUsageCollector();
297
- const ragIndex = await createCurrentStudioRagIndex(schemaDir, "querying chunks", usage);
472
+ const ragIndex = await createCurrentStudioRagIndex(state, "querying chunks", usage);
298
473
  let results;
299
474
  try {
300
475
  results = await ragIndex.retriever({
@@ -307,7 +482,10 @@ async function queryRag(schemaDir, query) {
307
482
  });
308
483
  }
309
484
  catch (error) {
310
- throw formatStudioRagEmbeddingError(error, ragIndex.config);
485
+ throw formatStudioRagOperationError(error, ragIndex.config);
486
+ }
487
+ finally {
488
+ await ragIndex.dispose();
311
489
  }
312
490
  return {
313
491
  question: query.question,
@@ -316,12 +494,12 @@ async function queryRag(schemaDir, query) {
316
494
  usage: usage.toDto(),
317
495
  };
318
496
  }
319
- async function createCurrentStudioRagIndex(schemaDir, action, usage) {
497
+ async function createCurrentStudioRagIndex(state, action, usage) {
320
498
  const config = resolveStudioRagEmbedderConfig();
321
499
  if (!config.configured) {
322
500
  throw new StudioHttpError(400, studioRagAiSdkKeyMissingMessage());
323
501
  }
324
- const status = getRagStatus(schemaDir);
502
+ const status = (await getRagStatus(state));
325
503
  if (!status.hasIndex) {
326
504
  throw new StudioHttpError(400, `Build the RAG index before ${action}.`);
327
505
  }
@@ -331,6 +509,7 @@ async function createCurrentStudioRagIndex(schemaDir, action, usage) {
331
509
  if (!status.schemaId || typeof status.chunksTotal !== "number") {
332
510
  throw new StudioHttpError(500, "Studio RAG status is missing schema metadata.");
333
511
  }
512
+ const store = await openStudioRagStore(state, config.dimensions);
334
513
  return {
335
514
  config,
336
515
  status: {
@@ -339,10 +518,85 @@ async function createCurrentStudioRagIndex(schemaDir, action, usage) {
339
518
  },
340
519
  retriever: createRetriever({
341
520
  embedder: await createStudioRagEmbedder(config, usage),
342
- store: createFileStore({ basePath: join(schemaDir, "schema") }),
521
+ store: store.store,
343
522
  }),
523
+ dispose: async () => {
524
+ await store.dispose();
525
+ },
344
526
  };
345
527
  }
528
+ function resolveStudioRagStoreConfig(state) {
529
+ const rt = getAskDbRuntimeConfig();
530
+ const kind = rt.structured.rag.store;
531
+ if (kind === "memory")
532
+ return { kind };
533
+ if (kind === "file") {
534
+ const basePath = rt.structured.rag.storeConfig.file?.basePath?.trim();
535
+ return { kind, basePath: basePath ? resolve(basePath) : join(state.schemaDir, "schema") };
536
+ }
537
+ const connectionString = pickFlat(rt.flat, "ASKDB_PGVECTOR_URL");
538
+ return {
539
+ kind,
540
+ connectionString,
541
+ table: rt.structured.rag.storeConfig.pgvector?.table?.trim() || undefined,
542
+ indexStrategy: pickFlat(rt.flat, "ASKDB_PGVECTOR_INDEX_STRATEGY"),
543
+ };
544
+ }
545
+ async function openStudioRagStore(state, dimensions) {
546
+ const config = resolveStudioRagStoreConfig(state);
547
+ if (config.kind === "memory") {
548
+ state.ragMemoryStore ??= createMemoryStore();
549
+ return {
550
+ kind: "memory",
551
+ store: state.ragMemoryStore,
552
+ dispose: async () => { },
553
+ };
554
+ }
555
+ if (config.kind === "file") {
556
+ const store = createFileStore({ basePath: config.basePath });
557
+ return {
558
+ kind: "file",
559
+ store,
560
+ basePath: config.basePath,
561
+ dispose: async () => {
562
+ store.flush();
563
+ },
564
+ };
565
+ }
566
+ if (!config.connectionString) {
567
+ throw new StudioHttpError(400, 'Studio pgvector RAG requires `ASKDB_PGVECTOR_URL` via `askdb.config.ts`.');
568
+ }
569
+ const pgvectorFactory = studioPgvectorStoreFactoryForTests ?? createPgvectorStore;
570
+ const store = pgvectorFactory({
571
+ connectionString: config.connectionString,
572
+ table: config.table,
573
+ dimensions,
574
+ ...(config.indexStrategy ? { indexStrategy: config.indexStrategy } : {}),
575
+ });
576
+ await store.ensureSchema();
577
+ return {
578
+ kind: "pgvector",
579
+ store,
580
+ table: config.table,
581
+ indexStrategy: config.indexStrategy,
582
+ dispose: async () => {
583
+ await store.close();
584
+ },
585
+ };
586
+ }
587
+ async function countStudioRagStoreChunks(store, schemaId) {
588
+ if (typeof store.store.count === "function") {
589
+ return store.store.count({ schemaId });
590
+ }
591
+ if (typeof store.store.size === "function") {
592
+ return store.store.size();
593
+ }
594
+ return 0;
595
+ }
596
+ function pickFlat(flat, key) {
597
+ const value = flat[key];
598
+ return typeof value === "string" && value.trim() !== "" ? value.trim() : undefined;
599
+ }
346
600
  function pickEnv(env, key) {
347
601
  const v = env[key];
348
602
  return typeof v === "string" && v.trim() !== "" ? v.trim() : undefined;
@@ -350,7 +604,7 @@ function pickEnv(env, key) {
350
604
  function resolveStudioRagEmbedderConfig() {
351
605
  const rt = getAskDbRuntimeConfig();
352
606
  const base = rt.ai.aiEnv;
353
- const explicitKind = pickEnv(base, "ASKDB_RAG_EMBEDDER") ?? pickEnv(base, "ASKDB_STUDIO_RAG_EMBEDDER");
607
+ const explicitKind = pickEnv(base, "ASKDB_RAG_EMBEDDER");
354
608
  const kind = explicitKind?.toLowerCase();
355
609
  if (kind === "mock") {
356
610
  return {
@@ -365,8 +619,8 @@ function resolveStudioRagEmbedderConfig() {
365
619
  throw new StudioHttpError(400, `Unsupported Studio RAG embedder: ${kind}`);
366
620
  }
367
621
  const env = buildStudioRagEmbeddingEnv(kind, base);
368
- const aiConfig = resolveAskDbEmbeddingConfig(env, {
369
- modelEnvVar: "ASKDB_STUDIO_RAG_EMBEDDER_MODEL",
622
+ const aiConfig = resolveEmbeddingConfig(env, {
623
+ modelEnvVar: "ASKDB_RAG_EMBEDDER_MODEL",
370
624
  modelDefault: DEFAULT_EMBEDDING_MODEL,
371
625
  });
372
626
  if (!aiConfig && kind === undefined) {
@@ -380,8 +634,7 @@ function resolveStudioRagEmbedderConfig() {
380
634
  }
381
635
  const provider = aiConfig?.provider ?? fallbackStudioRagProvider(kind, base);
382
636
  const model = aiConfig?.model ?? DEFAULT_EMBEDDING_MODEL;
383
- const dimensionOverride = readPositiveIntegerEnv(pickEnv(base, "ASKDB_RAG_EMBEDDER_DIMENSIONS") ??
384
- pickEnv(base, "ASKDB_STUDIO_RAG_EMBEDDER_DIMENSIONS"));
637
+ const dimensionOverride = readPositiveIntegerEnv(pickEnv(base, "ASKDB_RAG_EMBEDDER_DIMENSIONS"));
385
638
  const dimensions = dimensionOverride ?? defaultEmbeddingDimensions(model);
386
639
  return {
387
640
  kind: "ai-sdk",
@@ -397,14 +650,12 @@ function resolveStudioRagEmbedderConfig() {
397
650
  };
398
651
  }
399
652
  function buildStudioRagEmbeddingEnv(kind, base) {
400
- const apiKeyOverride = pickEnv(base, "ASKDB_RAG_EMBEDDER_API_KEY") ?? pickEnv(base, "ASKDB_STUDIO_RAG_EMBEDDER_API_KEY");
401
- const modelOverride = pickEnv(base, "ASKDB_RAG_EMBEDDER_MODEL") ?? pickEnv(base, "ASKDB_STUDIO_RAG_EMBEDDER_MODEL");
402
- const baseUrlOverride = pickEnv(base, "ASKDB_RAG_EMBEDDER_BASE_URL") ?? pickEnv(base, "ASKDB_STUDIO_RAG_EMBEDDER_BASE_URL");
653
+ const apiKeyOverride = pickEnv(base, "ASKDB_RAG_EMBEDDER_API_KEY");
654
+ const baseUrlOverride = pickEnv(base, "ASKDB_RAG_EMBEDDER_BASE_URL");
403
655
  return {
404
656
  ...base,
405
657
  ...(kind === "openai" ? { ASKDB_AI_PROVIDER: "openai" } : {}),
406
658
  ...(apiKeyOverride ? { ASKDB_AI_API_KEY: apiKeyOverride } : {}),
407
- ...(modelOverride ? { ASKDB_STUDIO_RAG_EMBEDDER_MODEL: modelOverride } : {}),
408
659
  ...(baseUrlOverride ? { ASKDB_AI_BASE_URL: baseUrlOverride } : {}),
409
660
  };
410
661
  }
@@ -426,7 +677,7 @@ async function createStudioRagEmbedder(config, usage) {
426
677
  if (!config.aiConfig) {
427
678
  throw new StudioHttpError(400, studioRagAiSdkKeyMissingMessage());
428
679
  }
429
- const model = await createAskDbEmbeddingModel(config.aiConfig, {
680
+ const model = await ai.createEmbeddingModel(config.aiConfig, {
430
681
  dimensions: config.requestDimensions,
431
682
  });
432
683
  return createAiSdkEmbedder({
@@ -441,11 +692,16 @@ async function createStudioRagEmbedder(config, usage) {
441
692
  },
442
693
  });
443
694
  }
444
- function formatStudioRagEmbeddingError(error, config) {
695
+ function formatStudioRagOperationError(error, config) {
696
+ if (error instanceof StudioHttpError)
697
+ return error;
445
698
  if (config.kind === "mock") {
446
699
  return new StudioHttpError(500, error instanceof Error ? error.message : String(error));
447
700
  }
448
701
  const apiError = findApiCallError(error);
702
+ if (!apiError) {
703
+ return new StudioHttpError(500, error instanceof Error ? error.message : String(error));
704
+ }
449
705
  const parts = [
450
706
  `Studio RAG embedding request failed for provider ${config.provider}, model ${config.model}.`,
451
707
  ];
@@ -488,15 +744,18 @@ function truncateForMessage(value) {
488
744
  return undefined;
489
745
  return trimmed.length > 500 ? `${trimmed.slice(0, 500)}...` : trimmed;
490
746
  }
491
- function clearIncompatibleRagStore(schemaDir, config) {
492
- const embeddingsJsonPath = join(schemaDir, "schema.embeddings.json");
747
+ function clearIncompatibleRagStore(state, config) {
748
+ const store = resolveStudioRagStoreConfig(state);
749
+ if (store.kind !== "file")
750
+ return;
751
+ const embeddingsJsonPath = `${store.basePath}.embeddings.json`;
493
752
  const embeddings = readOptionalJson(embeddingsJsonPath);
494
753
  if (embeddings?.dimensions === undefined || embeddings.dimensions === config.dimensions)
495
754
  return;
496
755
  for (const path of [
497
- join(schemaDir, "schema.embeddings.json"),
498
- join(schemaDir, "schema.embeddings.bin"),
499
- join(schemaDir, "schema.lock.json"),
756
+ `${store.basePath}.embeddings.json`,
757
+ `${store.basePath}.embeddings.bin`,
758
+ join(state.schemaDir, "schema.lock.json"),
500
759
  ]) {
501
760
  rmSync(path, { force: true });
502
761
  }
@@ -546,6 +805,116 @@ function stableTokenHash(token) {
546
805
  }
547
806
  return hash >>> 0;
548
807
  }
808
+ // ---------------------------------------------------------------------------
809
+ // Playground history helpers
810
+ // ---------------------------------------------------------------------------
811
+ const HISTORY_MAX_STORED = 200;
812
+ const HISTORY_MAX_RETURNED = 50;
813
+ function playgroundHistoryPath(schemaDir) {
814
+ return join(schemaDir, "playground-history.json");
815
+ }
816
+ function readPlaygroundHistory(schemaDir) {
817
+ const raw = readOptionalJson(playgroundHistoryPath(schemaDir));
818
+ const entries = Array.isArray(raw) ? raw : [];
819
+ return { entries: entries.slice(0, HISTORY_MAX_RETURNED) };
820
+ }
821
+ function appendPlaygroundHistory(schemaDir, entry) {
822
+ const histPath = playgroundHistoryPath(schemaDir);
823
+ const raw = readOptionalJson(histPath);
824
+ const existing = Array.isArray(raw)
825
+ ? raw
826
+ : [];
827
+ const newEntry = {
828
+ ...entry,
829
+ id: randomUUID(),
830
+ timestamp: new Date().toISOString(),
831
+ };
832
+ const updated = [newEntry, ...existing].slice(0, HISTORY_MAX_STORED);
833
+ writeFileSync(histPath, JSON.stringify(updated, null, 2), "utf8");
834
+ }
835
+ function deletePlaygroundHistoryEntry(schemaDir, id) {
836
+ const histPath = playgroundHistoryPath(schemaDir);
837
+ const raw = readOptionalJson(histPath);
838
+ const existing = Array.isArray(raw)
839
+ ? raw
840
+ : [];
841
+ const filtered = existing.filter((e) => e.id !== id);
842
+ if (filtered.length === existing.length) {
843
+ return { ok: false, error: "not found" };
844
+ }
845
+ writeFileSync(histPath, JSON.stringify(filtered, null, 2), "utf8");
846
+ return { ok: true };
847
+ }
848
+ function parsePlaygroundHistoryEntry(body) {
849
+ if (!isRecord(body)) {
850
+ throw new StudioHttpError(400, "Request body must be a JSON object.");
851
+ }
852
+ if (typeof body.question !== "string" || body.question.trim() === "") {
853
+ throw new StudioHttpError(400, "`question` is required.");
854
+ }
855
+ if (body.mode !== "full" && body.mode !== "rag") {
856
+ throw new StudioHttpError(400, "`mode` must be `full` or `rag`.");
857
+ }
858
+ if (typeof body.sqlMode !== "string") {
859
+ throw new StudioHttpError(400, "`sqlMode` is required.");
860
+ }
861
+ if (typeof body.sql !== "string") {
862
+ throw new StudioHttpError(400, "`sql` is required.");
863
+ }
864
+ return body;
865
+ }
866
+ // ---------------------------------------------------------------------------
867
+ // Execute endpoint helper
868
+ // ---------------------------------------------------------------------------
869
+ async function executeQuery(body) {
870
+ const databaseUrl = getAskDbRuntimeConfig().studio.execute.databaseUrl;
871
+ if (!databaseUrl) {
872
+ return { ok: false, error: "ASKDB_STUDIO_DATABASE_URL is not configured" };
873
+ }
874
+ if (!isRecord(body) || typeof body.sql !== "string" || body.sql.trim() === "") {
875
+ return { ok: false, error: "`sql` is required." };
876
+ }
877
+ const sql = body.sql;
878
+ const params = Array.isArray(body.params) ? body.params : [];
879
+ let pgMod;
880
+ try {
881
+ const mod = await import("pg");
882
+ pgMod = (mod.default ?? mod);
883
+ }
884
+ catch {
885
+ return { ok: false, error: "The `pg` package is required for query execution. Install it with `pnpm add pg`." };
886
+ }
887
+ const client = new pgMod.Client({ connectionString: databaseUrl });
888
+ await client.connect();
889
+ const startMs = Date.now();
890
+ try {
891
+ await client.query("BEGIN READ ONLY");
892
+ const result = await client.query({ text: sql, values: params });
893
+ await client.query("COMMIT");
894
+ const durationMs = Date.now() - startMs;
895
+ const columns = result.fields.map((f) => f.name);
896
+ const rows = result.rows.map((r) => columns.map((c) => r[c]));
897
+ const truncated = rows.length > 500;
898
+ return {
899
+ ok: true,
900
+ columns,
901
+ rows: rows.slice(0, 500),
902
+ rowCount: result.rowCount ?? rows.length,
903
+ durationMs,
904
+ truncated,
905
+ };
906
+ }
907
+ catch (err) {
908
+ await client.query("ROLLBACK").catch(() => { });
909
+ return {
910
+ ok: false,
911
+ error: err instanceof Error ? err.message : String(err),
912
+ };
913
+ }
914
+ finally {
915
+ await client.end();
916
+ }
917
+ }
549
918
  function readOptionalJson(path) {
550
919
  if (!existsSync(path))
551
920
  return undefined;
@@ -556,6 +925,17 @@ function readOptionalJson(path) {
556
925
  return undefined;
557
926
  }
558
927
  }
928
+ function parseConceptsBody(body) {
929
+ if (!isRecord(body) || !Array.isArray(body.concepts)) {
930
+ throw new StudioHttpError(400, "`concepts` array is required.");
931
+ }
932
+ for (const concept of body.concepts) {
933
+ if (!isRecord(concept) || typeof concept.id !== "string" || typeof concept.label !== "string") {
934
+ throw new StudioHttpError(400, "Each concept requires a string `id` and `label`.");
935
+ }
936
+ }
937
+ return body.concepts;
938
+ }
559
939
  function parseTableDraftBody(body) {
560
940
  if (!isRecord(body) || !isRecord(body.draft)) {
561
941
  throw new StudioHttpError(400, "`draft` is required.");
@@ -604,9 +984,26 @@ function parseAskBody(body) {
604
984
  if (mode !== "full" && mode !== "rag") {
605
985
  throw new StudioHttpError(400, "`mode` must be `full` or `rag`.");
606
986
  }
987
+ let tenantScope;
988
+ if (body.tenantScope !== undefined && body.tenantScope !== null) {
989
+ const parsed = tenantScopeSchema.safeParse(body.tenantScope);
990
+ if (!parsed.success) {
991
+ throw new StudioHttpError(400, `Invalid tenantScope: ${parsed.error.message}`);
992
+ }
993
+ tenantScope = parsed.data;
994
+ }
995
+ let tenantSqlMode;
996
+ if (typeof body.tenantSqlMode === "string") {
997
+ if (body.tenantSqlMode !== "sql-only" && body.tenantSqlMode !== "sql-params") {
998
+ throw new StudioHttpError(400, "`tenantSqlMode` must be `sql-only` or `sql-params`.");
999
+ }
1000
+ tenantSqlMode = body.tenantSqlMode;
1001
+ }
607
1002
  return {
608
1003
  question: body.question.trim(),
609
1004
  useRag: mode === "rag",
1005
+ tenantScope,
1006
+ tenantSqlMode,
610
1007
  };
611
1008
  }
612
1009
  function parseRagQuery(body) {