ktx/packages/cli/test/context/scan/local-enrichment.test.ts

872 lines
28 KiB
TypeScript
Raw Permalink Normal View History

2026-05-10 23:12:26 +02:00
import Database from 'better-sqlite3';
import { describe, expect, it, vi } from 'vitest';
test: split cli tests from source tree (#216) * feat(cli): define full warehouse dialect contract * test(cli): keep dialect edge tests focused * fix(cli): stabilize dialect contract foundation * refactor(connectors): own read-only query preparation * refactor(connectors): resolve dialects through registry * refactor(connectors): keep concrete dialect classes internal * chore(workspace): enforce dialect import boundary * refactor(cli): resolve relationship dialect at scan boundary * refactor(cli): use dialect display parsing for entity details * refactor(cli): use dialect display parsing for warehouse catalog * refactor(cli): use dialect SQL in relationship workflows * test(cli): verify solid dialect scan workflow closure * test: split cli tests from source tree * refactor(cli): standardize BigQuery scope listing * feat(sqlite): implement connector scope listing * test(connectors): cover required table listing * feat(cli): add warehouse driver registry * refactor(setup): route scope discovery through driver registry * refactor(cli): route local query execution through driver registry * refactor(historic-sql): route dialect support through driver registry * refactor(cli): test warehouse connections through driver registry * fix(cli): close driver registry type export gaps * Improve setup daemon diagnostics * refactor(setup): centralize rail-prefixed diagnostics + query-history fallback Extract errorMessage, writePrefixedLines, and flushPrefixedBufferedCommandOutput into clack.ts so the setup wizard, managed daemons, and embedding/agent steps share one rail-formatted writer. setup-databases.ts also adds a "disable query history and retry" option when the schema-context build fails and query history is the likely culprit, surfaced via a new failed-query-history-unavailable status. * fix(cli): carry catalog through the picker so BigQuery/Snowflake/SQL Server scope filters match The setup picker's KtxTableListEntry was a 2-level { schema, name }, so qualifiedTableId always wrote db.name into enabled_tables. When BigQuery, Snowflake, or SQL Server later ran fast ingest, their introspect step filtered the scope set with scopedTableNames(scope, { catalog: projectId|database, db }) — catalog was non-null on the introspect side but null in the scope refs, so every entry was rejected, the live-database adapter staged zero table files, and detect() failed with 'Adapter "live-database" did not recognize fetched source output'. Align the picker boundary with the canonical 3-level KtxTableRef: - Add catalog: string | null to KtxTableListEntry. - BigQuery/Snowflake/SQL Server listTables populate catalog from the resolved projectId / database; Postgres/MySQL/ClickHouse/SQLite set null. - qualifiedTableId emits catalog.schema.name when catalog is non-null (resolveEnabledTables already accepts the 3-part shape) and schemasFromEnabledTables now goes through parseDottedTableEntry so it recovers the schema correctly from both 2-part and 3-part entries. - Export parseDottedTableEntry from enabled-tables.ts (@internal) for picker reuse. Update listTables expectations in all seven connector tests and the setup / picker test fixtures. Add a picker regression test that covers the catalog-bearing round-trip (save + refine). * fix(cli): allow debug telemetry under opt-out env
2026-05-26 08:49:05 +02:00
import { buildDefaultKtxProjectConfig } from '../../../src/context/project/config.js';
2026-05-10 23:12:26 +02:00
import type {
2026-05-10 23:51:24 +02:00
KtxScanEnrichmentCompletedStage,
KtxScanEnrichmentFailedStage,
KtxScanEnrichmentStageLookup,
KtxScanEnrichmentStateStore,
test: split cli tests from source tree (#216) * feat(cli): define full warehouse dialect contract * test(cli): keep dialect edge tests focused * fix(cli): stabilize dialect contract foundation * refactor(connectors): own read-only query preparation * refactor(connectors): resolve dialects through registry * refactor(connectors): keep concrete dialect classes internal * chore(workspace): enforce dialect import boundary * refactor(cli): resolve relationship dialect at scan boundary * refactor(cli): use dialect display parsing for entity details * refactor(cli): use dialect display parsing for warehouse catalog * refactor(cli): use dialect SQL in relationship workflows * test(cli): verify solid dialect scan workflow closure * test: split cli tests from source tree * refactor(cli): standardize BigQuery scope listing * feat(sqlite): implement connector scope listing * test(connectors): cover required table listing * feat(cli): add warehouse driver registry * refactor(setup): route scope discovery through driver registry * refactor(cli): route local query execution through driver registry * refactor(historic-sql): route dialect support through driver registry * refactor(cli): test warehouse connections through driver registry * fix(cli): close driver registry type export gaps * Improve setup daemon diagnostics * refactor(setup): centralize rail-prefixed diagnostics + query-history fallback Extract errorMessage, writePrefixedLines, and flushPrefixedBufferedCommandOutput into clack.ts so the setup wizard, managed daemons, and embedding/agent steps share one rail-formatted writer. setup-databases.ts also adds a "disable query history and retry" option when the schema-context build fails and query history is the likely culprit, surfaced via a new failed-query-history-unavailable status. * fix(cli): carry catalog through the picker so BigQuery/Snowflake/SQL Server scope filters match The setup picker's KtxTableListEntry was a 2-level { schema, name }, so qualifiedTableId always wrote db.name into enabled_tables. When BigQuery, Snowflake, or SQL Server later ran fast ingest, their introspect step filtered the scope set with scopedTableNames(scope, { catalog: projectId|database, db }) — catalog was non-null on the introspect side but null in the scope refs, so every entry was rejected, the live-database adapter staged zero table files, and detect() failed with 'Adapter "live-database" did not recognize fetched source output'. Align the picker boundary with the canonical 3-level KtxTableRef: - Add catalog: string | null to KtxTableListEntry. - BigQuery/Snowflake/SQL Server listTables populate catalog from the resolved projectId / database; Postgres/MySQL/ClickHouse/SQLite set null. - qualifiedTableId emits catalog.schema.name when catalog is non-null (resolveEnabledTables already accepts the 3-part shape) and schemasFromEnabledTables now goes through parseDottedTableEntry so it recovers the schema correctly from both 2-part and 3-part entries. - Export parseDottedTableEntry from enabled-tables.ts (@internal) for picker reuse. Update listTables expectations in all seven connector tests and the setup / picker test fixtures. Add a picker regression test that covers the catalog-bearing round-trip (save + refine). * fix(cli): allow debug telemetry under opt-out env
2026-05-26 08:49:05 +02:00
} from '../../../src/context/scan/enrichment-state.js';
2026-05-10 23:12:26 +02:00
import {
createDeterministicLocalScanEnrichmentProviders,
runLocalScanEnrichment,
2026-05-10 23:51:24 +02:00
snapshotToKtxEnrichedSchema,
test: split cli tests from source tree (#216) * feat(cli): define full warehouse dialect contract * test(cli): keep dialect edge tests focused * fix(cli): stabilize dialect contract foundation * refactor(connectors): own read-only query preparation * refactor(connectors): resolve dialects through registry * refactor(connectors): keep concrete dialect classes internal * chore(workspace): enforce dialect import boundary * refactor(cli): resolve relationship dialect at scan boundary * refactor(cli): use dialect display parsing for entity details * refactor(cli): use dialect display parsing for warehouse catalog * refactor(cli): use dialect SQL in relationship workflows * test(cli): verify solid dialect scan workflow closure * test: split cli tests from source tree * refactor(cli): standardize BigQuery scope listing * feat(sqlite): implement connector scope listing * test(connectors): cover required table listing * feat(cli): add warehouse driver registry * refactor(setup): route scope discovery through driver registry * refactor(cli): route local query execution through driver registry * refactor(historic-sql): route dialect support through driver registry * refactor(cli): test warehouse connections through driver registry * fix(cli): close driver registry type export gaps * Improve setup daemon diagnostics * refactor(setup): centralize rail-prefixed diagnostics + query-history fallback Extract errorMessage, writePrefixedLines, and flushPrefixedBufferedCommandOutput into clack.ts so the setup wizard, managed daemons, and embedding/agent steps share one rail-formatted writer. setup-databases.ts also adds a "disable query history and retry" option when the schema-context build fails and query history is the likely culprit, surfaced via a new failed-query-history-unavailable status. * fix(cli): carry catalog through the picker so BigQuery/Snowflake/SQL Server scope filters match The setup picker's KtxTableListEntry was a 2-level { schema, name }, so qualifiedTableId always wrote db.name into enabled_tables. When BigQuery, Snowflake, or SQL Server later ran fast ingest, their introspect step filtered the scope set with scopedTableNames(scope, { catalog: projectId|database, db }) — catalog was non-null on the introspect side but null in the scope refs, so every entry was rejected, the live-database adapter staged zero table files, and detect() failed with 'Adapter "live-database" did not recognize fetched source output'. Align the picker boundary with the canonical 3-level KtxTableRef: - Add catalog: string | null to KtxTableListEntry. - BigQuery/Snowflake/SQL Server listTables populate catalog from the resolved projectId / database; Postgres/MySQL/ClickHouse/SQLite set null. - qualifiedTableId emits catalog.schema.name when catalog is non-null (resolveEnabledTables already accepts the 3-part shape) and schemasFromEnabledTables now goes through parseDottedTableEntry so it recovers the schema correctly from both 2-part and 3-part entries. - Export parseDottedTableEntry from enabled-tables.ts (@internal) for picker reuse. Update listTables expectations in all seven connector tests and the setup / picker test fixtures. Add a picker regression test that covers the catalog-bearing round-trip (save + refine). * fix(cli): allow debug telemetry under opt-out env
2026-05-26 08:49:05 +02:00
} from '../../../src/context/scan/local-enrichment.js';
2026-05-10 23:12:26 +02:00
import {
2026-05-10 23:51:24 +02:00
createKtxConnectorCapabilities,
type KtxQueryResult,
type KtxReadOnlyQueryInput,
type KtxEmbeddingPort,
2026-05-10 23:51:24 +02:00
type KtxScanConnector,
type KtxScanContext,
type KtxSchemaSnapshot,
test: split cli tests from source tree (#216) * feat(cli): define full warehouse dialect contract * test(cli): keep dialect edge tests focused * fix(cli): stabilize dialect contract foundation * refactor(connectors): own read-only query preparation * refactor(connectors): resolve dialects through registry * refactor(connectors): keep concrete dialect classes internal * chore(workspace): enforce dialect import boundary * refactor(cli): resolve relationship dialect at scan boundary * refactor(cli): use dialect display parsing for entity details * refactor(cli): use dialect display parsing for warehouse catalog * refactor(cli): use dialect SQL in relationship workflows * test(cli): verify solid dialect scan workflow closure * test: split cli tests from source tree * refactor(cli): standardize BigQuery scope listing * feat(sqlite): implement connector scope listing * test(connectors): cover required table listing * feat(cli): add warehouse driver registry * refactor(setup): route scope discovery through driver registry * refactor(cli): route local query execution through driver registry * refactor(historic-sql): route dialect support through driver registry * refactor(cli): test warehouse connections through driver registry * fix(cli): close driver registry type export gaps * Improve setup daemon diagnostics * refactor(setup): centralize rail-prefixed diagnostics + query-history fallback Extract errorMessage, writePrefixedLines, and flushPrefixedBufferedCommandOutput into clack.ts so the setup wizard, managed daemons, and embedding/agent steps share one rail-formatted writer. setup-databases.ts also adds a "disable query history and retry" option when the schema-context build fails and query history is the likely culprit, surfaced via a new failed-query-history-unavailable status. * fix(cli): carry catalog through the picker so BigQuery/Snowflake/SQL Server scope filters match The setup picker's KtxTableListEntry was a 2-level { schema, name }, so qualifiedTableId always wrote db.name into enabled_tables. When BigQuery, Snowflake, or SQL Server later ran fast ingest, their introspect step filtered the scope set with scopedTableNames(scope, { catalog: projectId|database, db }) — catalog was non-null on the introspect side but null in the scope refs, so every entry was rejected, the live-database adapter staged zero table files, and detect() failed with 'Adapter "live-database" did not recognize fetched source output'. Align the picker boundary with the canonical 3-level KtxTableRef: - Add catalog: string | null to KtxTableListEntry. - BigQuery/Snowflake/SQL Server listTables populate catalog from the resolved projectId / database; Postgres/MySQL/ClickHouse/SQLite set null. - qualifiedTableId emits catalog.schema.name when catalog is non-null (resolveEnabledTables already accepts the 3-part shape) and schemasFromEnabledTables now goes through parseDottedTableEntry so it recovers the schema correctly from both 2-part and 3-part entries. - Export parseDottedTableEntry from enabled-tables.ts (@internal) for picker reuse. Update listTables expectations in all seven connector tests and the setup / picker test fixtures. Add a picker regression test that covers the catalog-bearing round-trip (save + refine). * fix(cli): allow debug telemetry under opt-out env
2026-05-26 08:49:05 +02:00
} from '../../../src/context/scan/types.js';
2026-05-10 23:12:26 +02:00
function fakeScanEmbedding(options: { dimensions: number; maxBatchSize?: number }): KtxEmbeddingPort {
return {
dimensions: options.dimensions,
maxBatchSize: options.maxBatchSize ?? 64,
async embedBatch(texts) {
return texts.map((_, textIndex) =>
Array.from({ length: options.dimensions }, (__, dimensionIndex) => textIndex + dimensionIndex),
);
},
};
}
2026-05-10 23:51:24 +02:00
const snapshot: KtxSchemaSnapshot = {
2026-05-10 23:12:26 +02:00
connectionId: 'warehouse',
driver: 'postgres',
extractedAt: '2026-04-29T12:00:00.000Z',
scope: { schemas: ['public'] },
metadata: {},
tables: [
{
catalog: null,
db: 'public',
name: 'customers',
kind: 'table',
comment: 'Customer accounts',
estimatedRows: 2,
foreignKeys: [],
columns: [
{
name: 'id',
nativeType: 'integer',
normalizedType: 'integer',
dimensionType: 'number',
nullable: false,
primaryKey: true,
comment: 'Customer id',
},
],
},
{
catalog: null,
db: 'public',
name: 'orders',
kind: 'table',
comment: 'Customer orders',
estimatedRows: 3,
foreignKeys: [],
columns: [
{
name: 'id',
nativeType: 'integer',
normalizedType: 'integer',
dimensionType: 'number',
nullable: false,
primaryKey: true,
comment: 'Order id',
},
{
name: 'customer_id',
nativeType: 'integer',
normalizedType: 'integer',
dimensionType: 'number',
nullable: false,
primaryKey: false,
comment: 'Customer id',
},
],
},
],
};
2026-05-10 23:51:24 +02:00
function connector(): KtxScanConnector {
2026-05-10 23:12:26 +02:00
return {
id: 'test:warehouse',
driver: 'postgres',
2026-05-10 23:51:24 +02:00
capabilities: createKtxConnectorCapabilities({
2026-05-10 23:12:26 +02:00
tableSampling: true,
columnSampling: true,
readOnlySql: true,
columnStats: true,
}),
introspect: vi.fn(async () => snapshot),
test: split cli tests from source tree (#216) * feat(cli): define full warehouse dialect contract * test(cli): keep dialect edge tests focused * fix(cli): stabilize dialect contract foundation * refactor(connectors): own read-only query preparation * refactor(connectors): resolve dialects through registry * refactor(connectors): keep concrete dialect classes internal * chore(workspace): enforce dialect import boundary * refactor(cli): resolve relationship dialect at scan boundary * refactor(cli): use dialect display parsing for entity details * refactor(cli): use dialect display parsing for warehouse catalog * refactor(cli): use dialect SQL in relationship workflows * test(cli): verify solid dialect scan workflow closure * test: split cli tests from source tree * refactor(cli): standardize BigQuery scope listing * feat(sqlite): implement connector scope listing * test(connectors): cover required table listing * feat(cli): add warehouse driver registry * refactor(setup): route scope discovery through driver registry * refactor(cli): route local query execution through driver registry * refactor(historic-sql): route dialect support through driver registry * refactor(cli): test warehouse connections through driver registry * fix(cli): close driver registry type export gaps * Improve setup daemon diagnostics * refactor(setup): centralize rail-prefixed diagnostics + query-history fallback Extract errorMessage, writePrefixedLines, and flushPrefixedBufferedCommandOutput into clack.ts so the setup wizard, managed daemons, and embedding/agent steps share one rail-formatted writer. setup-databases.ts also adds a "disable query history and retry" option when the schema-context build fails and query history is the likely culprit, surfaced via a new failed-query-history-unavailable status. * fix(cli): carry catalog through the picker so BigQuery/Snowflake/SQL Server scope filters match The setup picker's KtxTableListEntry was a 2-level { schema, name }, so qualifiedTableId always wrote db.name into enabled_tables. When BigQuery, Snowflake, or SQL Server later ran fast ingest, their introspect step filtered the scope set with scopedTableNames(scope, { catalog: projectId|database, db }) — catalog was non-null on the introspect side but null in the scope refs, so every entry was rejected, the live-database adapter staged zero table files, and detect() failed with 'Adapter "live-database" did not recognize fetched source output'. Align the picker boundary with the canonical 3-level KtxTableRef: - Add catalog: string | null to KtxTableListEntry. - BigQuery/Snowflake/SQL Server listTables populate catalog from the resolved projectId / database; Postgres/MySQL/ClickHouse/SQLite set null. - qualifiedTableId emits catalog.schema.name when catalog is non-null (resolveEnabledTables already accepts the 3-part shape) and schemasFromEnabledTables now goes through parseDottedTableEntry so it recovers the schema correctly from both 2-part and 3-part entries. - Export parseDottedTableEntry from enabled-tables.ts (@internal) for picker reuse. Update listTables expectations in all seven connector tests and the setup / picker test fixtures. Add a picker regression test that covers the catalog-bearing round-trip (save + refine). * fix(cli): allow debug telemetry under opt-out env
2026-05-26 08:49:05 +02:00
listSchemas: vi.fn(async () => []),
listTables: vi.fn(async () => []),
2026-05-10 23:12:26 +02:00
sampleTable: vi.fn(async () => ({
headers: ['id', 'customer_id'],
rows: [[1, 10]],
totalRows: 1,
})),
sampleColumn: vi.fn(async () => ({
values: ['10', '11'],
nullCount: 0,
distinctCount: 2,
})),
};
}
class InMemorySqliteExecutor {
readonly db = new Database(':memory:');
2026-05-10 23:51:24 +02:00
executeReadOnly(input: KtxReadOnlyQueryInput, _ctx: KtxScanContext): Promise<KtxQueryResult> {
2026-05-10 23:12:26 +02:00
const rows = this.db.prepare(input.sql).all() as Record<string, unknown>[];
const headers = Object.keys(rows[0] ?? {});
return Promise.resolve({
headers,
rows: rows.map((row) => headers.map((header) => row[header])),
totalRows: rows.length,
rowCount: rows.length,
});
}
close(): void {
this.db.close();
}
}
2026-05-10 23:51:24 +02:00
function noDeclaredRelationshipSnapshot(): KtxSchemaSnapshot {
2026-05-10 23:12:26 +02:00
return {
connectionId: 'warehouse',
driver: 'sqlite',
extractedAt: '2026-05-07T00:00:00.000Z',
scope: {},
metadata: {},
tables: [
{
catalog: null,
db: null,
name: 'accounts',
kind: 'table',
comment: null,
estimatedRows: 2,
foreignKeys: [],
columns: [
{
name: 'id',
nativeType: 'INTEGER',
normalizedType: 'integer',
dimensionType: 'number',
nullable: false,
primaryKey: false,
comment: null,
},
],
},
{
catalog: null,
db: null,
name: 'orders',
kind: 'table',
comment: null,
estimatedRows: 3,
foreignKeys: [],
columns: [
{
name: 'id',
nativeType: 'INTEGER',
normalizedType: 'integer',
dimensionType: 'number',
nullable: false,
primaryKey: false,
comment: null,
},
{
name: 'account_id',
nativeType: 'INTEGER',
normalizedType: 'integer',
dimensionType: 'number',
nullable: false,
primaryKey: false,
comment: null,
},
],
},
],
};
}
2026-05-10 23:51:24 +02:00
function memoryEnrichmentStateStore(): KtxScanEnrichmentStateStore {
const records = new Map<string, KtxScanEnrichmentCompletedStage | KtxScanEnrichmentFailedStage>();
const key = (input: Pick<KtxScanEnrichmentStageLookup, 'runId' | 'stage'>) => `${input.runId}:${input.stage}`;
2026-05-10 23:12:26 +02:00
return {
2026-05-10 23:51:24 +02:00
async findCompletedStage<TOutput>(input: KtxScanEnrichmentStageLookup) {
2026-05-10 23:12:26 +02:00
const record = records.get(key(input));
if (!record || record.status !== 'completed' || record.inputHash !== input.inputHash) {
return null;
}
2026-05-10 23:51:24 +02:00
return record as KtxScanEnrichmentCompletedStage<TOutput>;
2026-05-10 23:12:26 +02:00
},
async saveCompletedStage(input) {
records.set(key(input), {
...input,
status: 'completed',
errorMessage: null,
});
},
async saveFailedStage(input) {
records.set(key(input), {
...input,
status: 'failed',
output: null,
});
},
async listRunStages(runId) {
return [...records.values()].filter((record) => record.runId === runId);
},
};
}
describe('local scan enrichment', () => {
it('maps a scan snapshot into relationship detector schema', () => {
2026-05-10 23:51:24 +02:00
const schema = snapshotToKtxEnrichedSchema(snapshot);
2026-05-10 23:12:26 +02:00
expect(schema.connectionId).toBe('warehouse');
expect(schema.tables).toHaveLength(2);
expect(schema.tables[1]?.columns.map((column) => column.name)).toEqual(['id', 'customer_id']);
expect(schema.tables[1]?.columns[1]).toMatchObject({
id: 'public.orders.customer_id',
tableId: 'public.orders',
primaryKey: false,
sampleValues: null,
embedding: null,
});
});
it('maps snapshot foreign keys into formal schema relationships', () => {
const source = noDeclaredRelationshipSnapshot();
const snapshotWithForeignKey = {
...source,
tables: source.tables.map((table) =>
table.name === 'orders'
? {
...table,
foreignKeys: [
{
fromColumn: 'account_id',
toCatalog: null,
toDb: null,
toTable: 'accounts',
toColumn: 'id',
constraintName: 'orders_account_id_fkey',
},
],
}
: table.name === 'accounts'
? {
...table,
columns: table.columns.map((column) =>
column.name === 'id' ? { ...column, primaryKey: true } : column,
),
}
: table,
),
};
2026-05-10 23:51:24 +02:00
const schema = snapshotToKtxEnrichedSchema(snapshotWithForeignKey);
2026-05-10 23:12:26 +02:00
expect(schema.relationships).toEqual([
{
id: 'orders:(orders.account_id)->accounts:(accounts.id)',
source: 'formal',
from: {
tableId: 'orders',
columnIds: ['orders.account_id'],
table: { catalog: null, db: null, name: 'orders' },
columns: ['account_id'],
},
to: {
tableId: 'accounts',
columnIds: ['accounts.id'],
table: { catalog: null, db: null, name: 'accounts' },
columns: ['id'],
},
relationshipType: 'many_to_one',
confidence: 1,
isPrimaryKeyReference: true,
},
]);
});
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
it('uses the supplied snapshot without calling connector.introspect', async () => {
const scanConnector = connector();
const introspect = vi.mocked(scanConnector.introspect);
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'structural',
connector: scanConnector,
snapshot,
context: { runId: 'scan-run-snapshot' },
providers: null,
});
expect(result.snapshot).toEqual(snapshot);
expect(introspect).not.toHaveBeenCalled();
});
it('falls back to connector.introspect when no snapshot is supplied', async () => {
const scanConnector = connector();
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'structural',
connector: scanConnector,
context: { runId: 'scan-run-introspect' },
providers: null,
});
expect(result.snapshot).toEqual(snapshot);
expect(scanConnector.introspect).toHaveBeenCalledTimes(1);
});
test: split cli tests from source tree (#216) * feat(cli): define full warehouse dialect contract * test(cli): keep dialect edge tests focused * fix(cli): stabilize dialect contract foundation * refactor(connectors): own read-only query preparation * refactor(connectors): resolve dialects through registry * refactor(connectors): keep concrete dialect classes internal * chore(workspace): enforce dialect import boundary * refactor(cli): resolve relationship dialect at scan boundary * refactor(cli): use dialect display parsing for entity details * refactor(cli): use dialect display parsing for warehouse catalog * refactor(cli): use dialect SQL in relationship workflows * test(cli): verify solid dialect scan workflow closure * test: split cli tests from source tree * refactor(cli): standardize BigQuery scope listing * feat(sqlite): implement connector scope listing * test(connectors): cover required table listing * feat(cli): add warehouse driver registry * refactor(setup): route scope discovery through driver registry * refactor(cli): route local query execution through driver registry * refactor(historic-sql): route dialect support through driver registry * refactor(cli): test warehouse connections through driver registry * fix(cli): close driver registry type export gaps * Improve setup daemon diagnostics * refactor(setup): centralize rail-prefixed diagnostics + query-history fallback Extract errorMessage, writePrefixedLines, and flushPrefixedBufferedCommandOutput into clack.ts so the setup wizard, managed daemons, and embedding/agent steps share one rail-formatted writer. setup-databases.ts also adds a "disable query history and retry" option when the schema-context build fails and query history is the likely culprit, surfaced via a new failed-query-history-unavailable status. * fix(cli): carry catalog through the picker so BigQuery/Snowflake/SQL Server scope filters match The setup picker's KtxTableListEntry was a 2-level { schema, name }, so qualifiedTableId always wrote db.name into enabled_tables. When BigQuery, Snowflake, or SQL Server later ran fast ingest, their introspect step filtered the scope set with scopedTableNames(scope, { catalog: projectId|database, db }) — catalog was non-null on the introspect side but null in the scope refs, so every entry was rejected, the live-database adapter staged zero table files, and detect() failed with 'Adapter "live-database" did not recognize fetched source output'. Align the picker boundary with the canonical 3-level KtxTableRef: - Add catalog: string | null to KtxTableListEntry. - BigQuery/Snowflake/SQL Server listTables populate catalog from the resolved projectId / database; Postgres/MySQL/ClickHouse/SQLite set null. - qualifiedTableId emits catalog.schema.name when catalog is non-null (resolveEnabledTables already accepts the 3-part shape) and schemasFromEnabledTables now goes through parseDottedTableEntry so it recovers the schema correctly from both 2-part and 3-part entries. - Export parseDottedTableEntry from enabled-tables.ts (@internal) for picker reuse. Update listTables expectations in all seven connector tests and the setup / picker test fixtures. Add a picker regression test that covers the catalog-bearing round-trip (save + refine). * fix(cli): allow debug telemetry under opt-out env
2026-05-26 08:49:05 +02:00
it('fails when connector driver and snapshot driver differ', async () => {
const mismatchedConnector: KtxScanConnector = {
...connector(),
driver: 'mysql',
};
await expect(
runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'relationships',
detectRelationships: true,
connector: mismatchedConnector,
snapshot,
context: { runId: 'scan-run-driver-mismatch' },
providers: null,
}),
).rejects.toThrow(
'ktx scan connector driver "mysql" does not match snapshot driver "postgres" for connection "warehouse"',
);
});
2026-05-10 23:12:26 +02:00
it('runs deterministic relationship detection for relationship scans', async () => {
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'relationships',
detectRelationships: true,
connector: connector(),
context: { runId: 'scan-run-1' },
providers: null,
});
expect(result.summary).toMatchObject({
deterministicRelationships: 'completed',
llmRelationshipValidation: 'skipped',
embeddings: 'skipped',
});
expect(result.relationships).toEqual({ accepted: 0, review: 1, rejected: 0, skipped: 0 });
expect(result.summary.statisticalValidation).toBe('skipped');
expect(result.warnings).toContainEqual({
code: 'relationship_validation_failed',
2026-05-10 23:51:24 +02:00
message: 'KTX scan connector advertises readOnlySql but does not expose executeReadOnly',
2026-05-10 23:12:26 +02:00
recoverable: true,
metadata: { capability: 'readOnlySql' },
});
});
it('runs relationship discovery with connector SQL evidence', async () => {
const executor = new InMemorySqliteExecutor();
try {
executor.db.exec(`
CREATE TABLE accounts (id INTEGER NOT NULL);
CREATE TABLE orders (id INTEGER NOT NULL, account_id INTEGER NOT NULL);
INSERT INTO accounts (id) VALUES (1), (2);
INSERT INTO orders (id, account_id) VALUES (10, 1), (11, 1), (12, 2);
`);
const scanConnector = {
...connector(),
driver: 'sqlite' as const,
2026-05-10 23:51:24 +02:00
capabilities: createKtxConnectorCapabilities({ readOnlySql: true, columnStats: true }),
2026-05-10 23:12:26 +02:00
introspect: vi.fn(async () => noDeclaredRelationshipSnapshot()),
executeReadOnly: executor.executeReadOnly.bind(executor),
};
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'relationships',
detectRelationships: true,
connector: scanConnector,
context: { runId: 'scan-run-relationship-discovery' },
providers: null,
});
expect(result.relationships).toEqual({ accepted: 1, review: 0, rejected: 0, skipped: 0 });
expect(result.summary.statisticalValidation).toBe('completed');
expect(result.relationshipProfile).toMatchObject({ sqlAvailable: true });
expect(result.resolvedRelationships).toEqual([
expect.objectContaining({
status: 'accepted',
from: expect.objectContaining({ table: expect.objectContaining({ name: 'orders' }), columns: ['account_id'] }),
to: expect.objectContaining({ table: expect.objectContaining({ name: 'accounts' }), columns: ['id'] }),
}),
]);
expect(result.relationshipUpdate?.accepted).toHaveLength(1);
} finally {
executor.close();
}
});
it('honors scan relationship config when LLM proposals are disabled', async () => {
const providers = createDeterministicLocalScanEnrichmentProviders();
feat: add claude-code llm backend with runtime port (#115) * docs: revise claude-code ingest backend spec * docs: keep claude-code spec focused on ingest * docs: expand claude-code spec to full llm parity * Refine claude-code backend spec after adversarial review iteration 1 * Refine claude-code backend spec after adversarial review iteration 2 * Refine claude-code backend spec after adversarial review iteration 3 * feat: recognize claude-code llm backend * feat: add ktx llm runtime port * feat: add claude-code llm runtime * feat: route non-agent llm calls through runtime * feat: run ingest agents through llm runtime * feat: support claude-code setup and status * test: verify claude-code backend runtime * docs: add claude-code backend v1 runtime plan * fix: close claude-code runtime isolation checks * fix: warn on claude-code prompt caching during setup * chore: verify claude-code v1 closure * docs: add claude-code backend v1 isolation closure plan * fix: update claude-code ingest setup guidance * docs: add claude-code backend v1 ingest guidance closure plan * docs: align claude-code isolation spec with sdk metadata * test: cover claude-code host discovery metadata * fix: tolerate claude-code host discovery metadata * docs: clarify claude-code host discovery metadata * docs: add claude-code auth-probe isolation fix plan * chore: prepare kaelio ktx rc1 release * chore: add semantic release workflow * fix: unblock ci checks * chore(release): 0.1.0-rc.1 * feat: add Claude Code model selection to setup * fix: keep git maintenance attached in local repos
2026-05-16 12:06:34 +02:00
const generateObject = vi.fn();
2026-05-10 23:12:26 +02:00
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'relationships',
detectRelationships: true,
connector: connector(),
context: { runId: 'scan-run-llm-disabled' },
providers: {
...providers,
feat: add claude-code llm backend with runtime port (#115) * docs: revise claude-code ingest backend spec * docs: keep claude-code spec focused on ingest * docs: expand claude-code spec to full llm parity * Refine claude-code backend spec after adversarial review iteration 1 * Refine claude-code backend spec after adversarial review iteration 2 * Refine claude-code backend spec after adversarial review iteration 3 * feat: recognize claude-code llm backend * feat: add ktx llm runtime port * feat: add claude-code llm runtime * feat: route non-agent llm calls through runtime * feat: run ingest agents through llm runtime * feat: support claude-code setup and status * test: verify claude-code backend runtime * docs: add claude-code backend v1 runtime plan * fix: close claude-code runtime isolation checks * fix: warn on claude-code prompt caching during setup * chore: verify claude-code v1 closure * docs: add claude-code backend v1 isolation closure plan * fix: update claude-code ingest setup guidance * docs: add claude-code backend v1 ingest guidance closure plan * docs: align claude-code isolation spec with sdk metadata * test: cover claude-code host discovery metadata * fix: tolerate claude-code host discovery metadata * docs: clarify claude-code host discovery metadata * docs: add claude-code auth-probe isolation fix plan * chore: prepare kaelio ktx rc1 release * chore: add semantic release workflow * fix: unblock ci checks * chore(release): 0.1.0-rc.1 * feat: add Claude Code model selection to setup * fix: keep git maintenance attached in local repos
2026-05-16 12:06:34 +02:00
llmRuntime: {
...providers.llmRuntime,
generateObject: generateObject as never,
2026-05-10 23:12:26 +02:00
},
},
relationshipSettings: {
...buildDefaultKtxProjectConfig().scan.relationships,
2026-05-10 23:12:26 +02:00
llmProposals: false,
maxLlmTablesPerBatch: 40,
},
});
expect(result.summary.llmRelationshipValidation).toBe('skipped');
feat: add claude-code llm backend with runtime port (#115) * docs: revise claude-code ingest backend spec * docs: keep claude-code spec focused on ingest * docs: expand claude-code spec to full llm parity * Refine claude-code backend spec after adversarial review iteration 1 * Refine claude-code backend spec after adversarial review iteration 2 * Refine claude-code backend spec after adversarial review iteration 3 * feat: recognize claude-code llm backend * feat: add ktx llm runtime port * feat: add claude-code llm runtime * feat: route non-agent llm calls through runtime * feat: run ingest agents through llm runtime * feat: support claude-code setup and status * test: verify claude-code backend runtime * docs: add claude-code backend v1 runtime plan * fix: close claude-code runtime isolation checks * fix: warn on claude-code prompt caching during setup * chore: verify claude-code v1 closure * docs: add claude-code backend v1 isolation closure plan * fix: update claude-code ingest setup guidance * docs: add claude-code backend v1 ingest guidance closure plan * docs: align claude-code isolation spec with sdk metadata * test: cover claude-code host discovery metadata * fix: tolerate claude-code host discovery metadata * docs: clarify claude-code host discovery metadata * docs: add claude-code auth-probe isolation fix plan * chore: prepare kaelio ktx rc1 release * chore: add semantic release workflow * fix: unblock ci checks * chore(release): 0.1.0-rc.1 * feat: add Claude Code model selection to setup * fix: keep git maintenance attached in local repos
2026-05-16 12:06:34 +02:00
expect(generateObject).not.toHaveBeenCalled();
2026-05-10 23:12:26 +02:00
});
it('skips relationship detection when scan relationships are disabled', async () => {
const settings = {
...buildDefaultKtxProjectConfig().scan.relationships,
2026-05-10 23:12:26 +02:00
enabled: false,
};
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
connector: connector(),
context: { runId: 'disabled-relationships' },
providers: createDeterministicLocalScanEnrichmentProviders(),
relationshipSettings: settings,
});
expect(result.summary.deterministicRelationships).toBe('skipped');
expect(result.summary.statisticalValidation).toBe('skipped');
expect(result.summary.llmRelationshipValidation).toBe('skipped');
expect(result.relationships).toEqual({ accepted: 0, review: 0, rejected: 0, skipped: 0 });
expect(result.relationshipUpdate).toBeNull();
expect(result.relationshipProfile).toBeNull();
expect(result.resolvedRelationships).toBeNull();
});
fix(context): merge overlay columns onto manifest columns by name (#94) * fix(context): merge overlay columns onto manifest columns by name composeOverlay was appending overlay columns to the manifest column list, producing duplicate entries when dbt/metabase overlays declared a column just to attach descriptions. The duplicates carried no `type`, so the pydantic SourceDefinition rejected them at semantic-query time and broke `ktx sl query` for every overlay-backed measure. Now overlay columns match base columns by name (case-insensitive): same-name entries merge onto the manifest (overlay fields win, type/role fall back to the base, descriptions merge per source key) and only new names append. * refactor(sl): split overlay columns from column_overrides and enforce TS/Python wire contract Overlay sources now have two distinct collections: `columns:` for computed columns (requiring `expr` + `type`) and `column_overrides:` for metadata patches to inherited manifest columns. Composing or loading an overlay that mixes the two — or references an unknown column — fails with a typed error. Introduce `ResolvedSemanticLayerSource` / `resolvedSourceSchema` / `toResolvedWire` as the strict shape sent to the Python engine, and add a schema contract test that diffs Zod against the Pydantic JSON schema dumped by `python -m semantic_layer dump-schema`. `SourceDefinition` is now `extra="forbid"` on the Python side. `loadAllSources` surfaces per-file load errors instead of swallowing them, so validation/query paths can report manifest shard parse failures. * fix(context): make scan description generation resilient and quiet A transient sampleTable failure during ingest used to take out every table in a connection: generateTableDescription returned a hardcoded 'Table not found' string into descriptions.ai, and KtxDescriptionGenerator was constructed without a logger, so the failure left no trail anywhere. - sampleTable / sampleColumn calls retry 3x with 200/400/800ms backoff, honouring KtxScanContext.signal via a new KtxAbortedError. - On retry exhaustion or missing capability, table generation falls back to a metadata-only prompt built from column name / native type / comment / rawDescriptions. The column path follows the same rule -- call the LLM when any of samples or rawDescriptions are available; skip only when both are absent. - Logger is now threaded from KtxScanContext into the generator. Failures emit structured KtxScanWarning entries (new description_fallback_used code, plus existing sampling_failed / enrichment_failed / connector_capability_missing). ktx scan groups warnings by code so a batch of identical failures collapses to one summary line plus sample. - Returns null on failure instead of the 'Table not found' sentinel; the manifest writer's existing guard already skips empty descriptions, so schema YAML no longer carries misleading text. SCAN_MANAGED_DESCRIPTION_KEYS already strips stale 'ai' on merge, so existing YAML clears on next run. Also suppress AI SDK v6 'system in messages' warning: pull system messages out of KtxMessageBuilder.wrapSimple's output via a new splitKtxSystemMessages helper and pass them top-level to generateText (preserves cacheControl providerOptions on the SystemModelMessage). Agent-runner's local splitSystemPromptMessages dedupes onto the shared helper. * test(docs): align examples-docs assertions with revamped docs PR #103 (setup/guide doc revamp) reworded several CLI examples and connection labels; the assertions in scripts/examples-docs.test.mjs still referenced the pre-revamp wording and were failing in CI on main. Update the regexes to match the post-revamp content: - drop the `--json` flag from the sl-query example expectation - move the `Driver:` / `Status: ok` probe to the connection reference, which is where that output now lives (driver id is lowercase `postgres`, not the display name `PostgreSQL`) - drop the obsolete `Install \`uv\`...` troubleshooting line - accept `<connectionId>` everywhere; the docs no longer use the hyphenated `<connection-id>` form - match the `warehouse` connection id used in the quickstart instead of the `postgres-warehouse` id only used in the README and setup ref * fix(sl): skip TS/Python schema contract test when uv is unavailable The TypeScript checks CI job does not install uv or Python, so the module-level `execFileSync('uv', ...)` in schemas.contract.test.ts threw ENOENT and failed the suite. Wrap the schema dump in a try/catch and guard the describe block with `describe.skipIf` so the test skips in environments without uv. Local dev and any CI job that has uv on PATH still runs the cross-language contract assertion.
2026-05-15 02:11:04 +02:00
it('forwards context.logger and emits warnings when sampleTable fails repeatedly', async () => {
const failingConnector: KtxScanConnector = {
...connector(),
sampleTable: vi.fn(async () => {
throw new Error('pool: ECONNRESET');
}),
};
const logger = {
debug: vi.fn(),
info: vi.fn(),
warn: vi.fn(),
error: vi.fn(),
};
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
detectRelationships: false,
connector: failingConnector,
context: { runId: 'scan-run-warnings', logger },
providers: createDeterministicLocalScanEnrichmentProviders(),
fix(context): merge overlay columns onto manifest columns by name (#94) * fix(context): merge overlay columns onto manifest columns by name composeOverlay was appending overlay columns to the manifest column list, producing duplicate entries when dbt/metabase overlays declared a column just to attach descriptions. The duplicates carried no `type`, so the pydantic SourceDefinition rejected them at semantic-query time and broke `ktx sl query` for every overlay-backed measure. Now overlay columns match base columns by name (case-insensitive): same-name entries merge onto the manifest (overlay fields win, type/role fall back to the base, descriptions merge per source key) and only new names append. * refactor(sl): split overlay columns from column_overrides and enforce TS/Python wire contract Overlay sources now have two distinct collections: `columns:` for computed columns (requiring `expr` + `type`) and `column_overrides:` for metadata patches to inherited manifest columns. Composing or loading an overlay that mixes the two — or references an unknown column — fails with a typed error. Introduce `ResolvedSemanticLayerSource` / `resolvedSourceSchema` / `toResolvedWire` as the strict shape sent to the Python engine, and add a schema contract test that diffs Zod against the Pydantic JSON schema dumped by `python -m semantic_layer dump-schema`. `SourceDefinition` is now `extra="forbid"` on the Python side. `loadAllSources` surfaces per-file load errors instead of swallowing them, so validation/query paths can report manifest shard parse failures. * fix(context): make scan description generation resilient and quiet A transient sampleTable failure during ingest used to take out every table in a connection: generateTableDescription returned a hardcoded 'Table not found' string into descriptions.ai, and KtxDescriptionGenerator was constructed without a logger, so the failure left no trail anywhere. - sampleTable / sampleColumn calls retry 3x with 200/400/800ms backoff, honouring KtxScanContext.signal via a new KtxAbortedError. - On retry exhaustion or missing capability, table generation falls back to a metadata-only prompt built from column name / native type / comment / rawDescriptions. The column path follows the same rule -- call the LLM when any of samples or rawDescriptions are available; skip only when both are absent. - Logger is now threaded from KtxScanContext into the generator. Failures emit structured KtxScanWarning entries (new description_fallback_used code, plus existing sampling_failed / enrichment_failed / connector_capability_missing). ktx scan groups warnings by code so a batch of identical failures collapses to one summary line plus sample. - Returns null on failure instead of the 'Table not found' sentinel; the manifest writer's existing guard already skips empty descriptions, so schema YAML no longer carries misleading text. SCAN_MANAGED_DESCRIPTION_KEYS already strips stale 'ai' on merge, so existing YAML clears on next run. Also suppress AI SDK v6 'system in messages' warning: pull system messages out of KtxMessageBuilder.wrapSimple's output via a new splitKtxSystemMessages helper and pass them top-level to generateText (preserves cacheControl providerOptions on the SystemModelMessage). Agent-runner's local splitSystemPromptMessages dedupes onto the shared helper. * test(docs): align examples-docs assertions with revamped docs PR #103 (setup/guide doc revamp) reworded several CLI examples and connection labels; the assertions in scripts/examples-docs.test.mjs still referenced the pre-revamp wording and were failing in CI on main. Update the regexes to match the post-revamp content: - drop the `--json` flag from the sl-query example expectation - move the `Driver:` / `Status: ok` probe to the connection reference, which is where that output now lives (driver id is lowercase `postgres`, not the display name `PostgreSQL`) - drop the obsolete `Install \`uv\`...` troubleshooting line - accept `<connectionId>` everywhere; the docs no longer use the hyphenated `<connection-id>` form - match the `warehouse` connection id used in the quickstart instead of the `postgres-warehouse` id only used in the README and setup ref * fix(sl): skip TS/Python schema contract test when uv is unavailable The TypeScript checks CI job does not install uv or Python, so the module-level `execFileSync('uv', ...)` in schemas.contract.test.ts threw ENOENT and failed the suite. Wrap the schema dump in a try/catch and guard the describe block with `describe.skipIf` so the test skips in environments without uv. Local dev and any CI job that has uv on PATH still runs the cross-language contract assertion.
2026-05-15 02:11:04 +02:00
});
const codes = result.warnings.map((warning) => warning.code);
expect(codes).toContain('sampling_failed');
expect(codes).toContain('description_fallback_used');
expect(result.warnings.some((warning) => warning.table === 'customers')).toBe(true);
expect(logger.warn).toHaveBeenCalled();
expect(logger.error).toHaveBeenCalled();
// Each of the two tables produced sampling_failed + description_fallback_used, so 2 + 2 = 4 warnings minimum.
expect(result.warnings.length).toBeGreaterThanOrEqual(4);
// Sampling was retried 3× for each of the 2 tables = 6 calls
expect(failingConnector.sampleTable).toHaveBeenCalledTimes(6);
});
it('runs configured deterministic enrichment with descriptions and no embeddings', async () => {
2026-05-10 23:12:26 +02:00
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
detectRelationships: true,
connector: connector(),
context: { runId: 'scan-run-2' },
providers: createDeterministicLocalScanEnrichmentProviders(),
2026-05-10 23:12:26 +02:00
});
expect(result.summary).toMatchObject({
dataDictionary: 'completed',
tableDescriptions: 'completed',
columnDescriptions: 'completed',
embeddings: 'skipped',
2026-05-10 23:12:26 +02:00
deterministicRelationships: 'completed',
});
expect(result.embeddingUpdates).toEqual([]);
2026-05-10 23:12:26 +02:00
expect(result.snapshot).toEqual(snapshot);
expect(result.relationships).toEqual({ accepted: 0, review: 1, rejected: 0, skipped: 0 });
});
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
it('generates batched table descriptions with bounded table-level concurrency', async () => {
const concurrentSnapshot: KtxSchemaSnapshot = {
...snapshot,
tables: Array.from({ length: 8 }, (_, index) => ({
catalog: null,
db: 'public',
name: `table_${index + 1}`,
kind: 'table' as const,
comment: null,
estimatedRows: 2,
foreignKeys: [],
columns: [
{
name: 'id',
nativeType: 'integer',
normalizedType: 'integer',
dimensionType: 'number' as const,
nullable: false,
primaryKey: true,
comment: null,
},
],
})),
};
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
let activeTableSamples = 0;
let maxActiveTableSamples = 0;
const scanConnector = {
...connector(),
introspect: vi.fn(async () => concurrentSnapshot),
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
sampleColumn: vi.fn(async () => ({
values: ['1'],
nullCount: 0,
distinctCount: 1,
})),
sampleTable: vi.fn(async () => {
activeTableSamples += 1;
maxActiveTableSamples = Math.max(maxActiveTableSamples, activeTableSamples);
await new Promise((resolve) => setTimeout(resolve, 10));
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
activeTableSamples -= 1;
return {
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
headers: ['id'],
rows: [[1]],
totalRows: 1,
};
}),
};
const settings = {
...buildDefaultKtxProjectConfig().scan.relationships,
enabled: false,
};
await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
connector: scanConnector,
context: { runId: 'scan-run-concurrent-descriptions' },
providers: createDeterministicLocalScanEnrichmentProviders(),
relationshipSettings: settings,
});
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
expect(maxActiveTableSamples).toBe(4);
expect(scanConnector.sampleColumn).not.toHaveBeenCalled();
});
2026-05-10 23:12:26 +02:00
it('reports enrichment progress for countable stages', async () => {
const events: Array<{ progress: number; message?: string; transient?: boolean }> = [];
const progress = {
async update(progressValue: number, message?: string, options?: { transient?: boolean }) {
events.push({ progress: progressValue, message, transient: options?.transient });
},
startPhase() {
return progress;
},
};
await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
detectRelationships: true,
connector: connector(),
context: { runId: 'scan-run-progress', progress },
providers: {
...createDeterministicLocalScanEnrichmentProviders(),
embedding: fakeScanEmbedding({ dimensions: 6 }),
},
2026-05-10 23:12:26 +02:00
});
expect(events).toEqual(
expect.arrayContaining([
expect.objectContaining({ message: 'Generating descriptions 1/2 tables', transient: true }),
expect.objectContaining({ message: 'Generating descriptions 2/2 tables', transient: true }),
expect.objectContaining({ message: 'Building embeddings 1/1 batches', transient: true }),
expect.objectContaining({ message: 'Detecting relationships' }),
]),
);
});
it('reports progress before enrichment connector introspection starts', async () => {
const events: Array<{ progress: number; message?: string; transient?: boolean }> = [];
const progress = {
async update(progressValue: number, message?: string, options?: { transient?: boolean }) {
events.push({ progress: progressValue, message, transient: options?.transient });
},
startPhase() {
return progress;
},
};
const scanConnector = {
...connector(),
introspect: vi.fn(async () => {
expect(events).toContainEqual(expect.objectContaining({ message: 'Loading enrichment schema snapshot' }));
return snapshot;
}),
};
await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'relationships',
detectRelationships: true,
connector: scanConnector,
context: { runId: 'scan-run-progress-before-introspection', progress },
providers: null,
});
expect(scanConnector.introspect).toHaveBeenCalled();
});
it('splits enrichment embedding requests by provider batch size', async () => {
2026-05-10 23:51:24 +02:00
const manyColumnSnapshot: KtxSchemaSnapshot = {
2026-05-10 23:12:26 +02:00
...snapshot,
tables: [
{
catalog: null,
db: 'public',
name: 'wide_orders',
kind: 'table',
comment: 'Wide order facts',
estimatedRows: 3,
foreignKeys: [],
columns: Array.from({ length: 5 }, (_, index) => ({
name: `metric_${index + 1}`,
nativeType: 'integer',
normalizedType: 'integer',
dimensionType: 'number' as const,
nullable: false,
primaryKey: false,
comment: `Metric ${index + 1}`,
})),
},
],
};
const scanConnector = {
...connector(),
introspect: vi.fn(async () => manyColumnSnapshot),
};
const deterministicProviders = createDeterministicLocalScanEnrichmentProviders();
2026-05-10 23:12:26 +02:00
const embedBatch = vi.fn(async (texts: string[]) => {
if (texts.length > 2) {
throw new Error(`Embedding batch size ${texts.length} exceeds maximum 2`);
}
return texts.map((_, index) => [index, index + 1, index + 2]);
});
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
detectRelationships: false,
connector: scanConnector,
context: { runId: 'scan-run-batched-embeddings' },
providers: {
feat: add claude-code llm backend with runtime port (#115) * docs: revise claude-code ingest backend spec * docs: keep claude-code spec focused on ingest * docs: expand claude-code spec to full llm parity * Refine claude-code backend spec after adversarial review iteration 1 * Refine claude-code backend spec after adversarial review iteration 2 * Refine claude-code backend spec after adversarial review iteration 3 * feat: recognize claude-code llm backend * feat: add ktx llm runtime port * feat: add claude-code llm runtime * feat: route non-agent llm calls through runtime * feat: run ingest agents through llm runtime * feat: support claude-code setup and status * test: verify claude-code backend runtime * docs: add claude-code backend v1 runtime plan * fix: close claude-code runtime isolation checks * fix: warn on claude-code prompt caching during setup * chore: verify claude-code v1 closure * docs: add claude-code backend v1 isolation closure plan * fix: update claude-code ingest setup guidance * docs: add claude-code backend v1 ingest guidance closure plan * docs: align claude-code isolation spec with sdk metadata * test: cover claude-code host discovery metadata * fix: tolerate claude-code host discovery metadata * docs: clarify claude-code host discovery metadata * docs: add claude-code auth-probe isolation fix plan * chore: prepare kaelio ktx rc1 release * chore: add semantic release workflow * fix: unblock ci checks * chore(release): 0.1.0-rc.1 * feat: add Claude Code model selection to setup * fix: keep git maintenance attached in local repos
2026-05-16 12:06:34 +02:00
llmRuntime: deterministicProviders.llmRuntime,
2026-05-10 23:12:26 +02:00
embedding: {
dimensions: 3,
maxBatchSize: 2,
embedBatch,
},
},
});
expect(result.embeddingUpdates).toHaveLength(5);
expect(embedBatch.mock.calls.map(([texts]) => texts).map((texts) => texts.length)).toEqual([2, 2, 1]);
});
it('reuses completed description and embedding stages for the same run id and snapshot hash', async () => {
const stateStore = memoryEnrichmentStateStore();
const scanConnector = connector();
const providers = {
...createDeterministicLocalScanEnrichmentProviders(),
embedding: fakeScanEmbedding({ dimensions: 6 }),
};
2026-05-10 23:12:26 +02:00
const first = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
detectRelationships: true,
connector: scanConnector,
context: { runId: 'scan-run-resume-1' },
providers,
stateStore,
syncId: 'sync-resume-1',
providerIdentity: { provider: 'fake', embeddingDimensions: 6 },
2026-05-10 23:12:26 +02:00
});
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
const generateObject = vi.spyOn(providers.llmRuntime, 'generateObject');
2026-05-10 23:12:26 +02:00
const embedBatch = vi.spyOn(providers.embedding, 'embedBatch');
const second = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
detectRelationships: true,
connector: scanConnector,
context: { runId: 'scan-run-resume-1' },
providers,
stateStore,
syncId: 'sync-resume-1',
providerIdentity: { provider: 'fake', embeddingDimensions: 6 },
2026-05-10 23:12:26 +02:00
});
expect(first.state.completedStages).toEqual(['descriptions', 'embeddings', 'relationships']);
expect(first.state.resumedStages).toEqual([]);
expect(second.state.resumedStages).toEqual(['descriptions', 'embeddings', 'relationships']);
expect(second.state.completedStages).toEqual(['descriptions', 'embeddings', 'relationships']);
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
expect(generateObject).not.toHaveBeenCalled();
2026-05-10 23:12:26 +02:00
expect(embedBatch).not.toHaveBeenCalled();
expect(second.descriptionUpdates).toEqual(first.descriptionUpdates);
expect(second.embeddingUpdates).toEqual(first.embeddingUpdates);
expect(second.relationships).toEqual(first.relationships);
});
it('does not reuse completed stages when the snapshot changes', async () => {
const stateStore = memoryEnrichmentStateStore();
const providers = {
...createDeterministicLocalScanEnrichmentProviders(),
embedding: fakeScanEmbedding({ dimensions: 6 }),
};
2026-05-10 23:12:26 +02:00
const scanConnector = connector();
await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
detectRelationships: false,
connector: scanConnector,
context: { runId: 'scan-run-resume-hash' },
providers,
stateStore,
syncId: 'sync-resume-hash',
providerIdentity: { provider: 'fake', embeddingDimensions: 6 },
2026-05-10 23:12:26 +02:00
});
const firstTable = snapshot.tables[0];
if (!firstTable) {
throw new Error('Expected test snapshot table');
}
const changedConnector = {
...connector(),
introspect: vi.fn(async () => ({
...snapshot,
tables: [{ ...firstTable, name: 'customers' }],
})),
};
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
const generateObject = vi.spyOn(providers.llmRuntime, 'generateObject');
2026-05-10 23:12:26 +02:00
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
detectRelationships: false,
connector: changedConnector,
context: { runId: 'scan-run-resume-hash' },
providers,
stateStore,
syncId: 'sync-resume-hash',
providerIdentity: { provider: 'fake', embeddingDimensions: 6 },
2026-05-10 23:12:26 +02:00
});
expect(result.state.resumedStages).toEqual([]);
expect(result.state.completedStages).toEqual(['descriptions', 'embeddings', 'relationships']);
fix(snowflake): unblock multi-schema ingest and relationship discovery (#204) * feat(setup): drop redundant Snowflake schema prompt; fall back to free-text on listSchemas failure Snowflake setup previously asked for a single schema as free text, then ran a multiselect against the discovered schemas — two schema questions back-to-back, with the first being only a session bootstrap. The SDK's `schema` is optional, so the bootstrap step is unnecessary. - Remove the free-text Snowflake schema prompt; only pass `schema` to snowflake-sdk when one is configured. - When `listSchemas()` fails (e.g. role lacks SHOW SCHEMAS), prompt the user for a comma-separated list, persist it as `schema_names`, and use it as both the table-list filter and the multiselect default. Applies to every driver with a scope-discovery spec, not just Snowflake. - Update docs to lead with `schema_names`; keep `schema_name` as a documented single-schema shorthand. * fix(snowflake): keep introspecting when primary-key discovery is denied The PK query joins INFORMATION_SCHEMA.TABLE_CONSTRAINTS and INFORMATION_SCHEMA.KEY_COLUMN_USAGE, which require grants the connection role may not have. Previously a 'SQL compilation error: Object ANALYTICS.INFORMATION_SCHEMA.KEY_COLUMN_USAGE does not exist or not authorized' aborted the entire introspect — schemas, columns, and row counts were all discarded over a missing nice-to-have. Wrap the constraint query in try/catch, log a one-line warning per schema, and return an empty PK map. Columns end up with primaryKey=false; relationship inference still has FK and profiling to fall back on. * fix(scan): unblock relationship discovery on Snowflake Two adjacent bugs prevented the scan's relationship pipeline from producing any joins on a Snowflake warehouse: - relationship-profiling.ts fell through to a default `GROUP_CONCAT` branch for unknown drivers. Snowflake has no GROUP_CONCAT, so every per-table profile query failed with "Unknown function GROUP_CONCAT". Add an explicit Snowflake branch that uses LISTAGG with a literal '\x1f' delimiter (Snowflake requires the delimiter to be a constant, so CHR(31) is rejected). - description-generation.ts destructured `connector.sampleTable` and `connector.sampleColumn` into bare locals, losing the `this` binding when the class-method connectors (Snowflake, Postgres, MySQL) were invoked. Every sample call threw "Cannot read properties of undefined (reading 'assertConnection')" and degraded LLM descriptions to metadata-only prompts. Call the methods through the connector instead. Without these, even after the primary-key probe is allowed to fail softly, the scan ends up with 0 validated relationships and an empty `joins:` block in every shard YAML. * test(scan): cover table-ref helpers * feat(scan): plumb tableScope through live-database introspection port * feat(scan): apply tableScope during metadata fetch * feat(scan): enforce table scope at fetch boundary * feat(scan): pool Snowflake sessions and batch enrichment for faster ingest (#206) * feat(cli): add RSA key-pair auth option to Snowflake setup wizard Extends the interactive Snowflake setup flow with an authentication-method prompt (password vs RSA/JWT key-pair). The RSA branch collects a private-key path (env/file/absolute) and an optional passphrase; the resulting connection config records `authMethod: 'rsa'` with `privateKey` and `passphrase` instead of `password`. * feat(scan): pool Snowflake sessions * fix(scan): reuse structural snapshots and cleanup connectors * feat(scan): parallelize relationship profiling * feat(scan): batch table description generation * docs: document Snowflake ingest concurrency knobs * fix(scan): close Snowflake ingest perf verification gaps * fix(scan): keep batched description failure bounded * feat(scan): dispatch query-history probes by connection driver Extract historic-sql dialect resolution into a shared helper so the status-project readiness check and the local ingest factory agree on which connections enable query history and which probe to run. The status command now picks the postgres/snowflake/bigquery probe based on the connection's driver instead of always reporting against postgres, which previously caused snowflake connections with queryHistory.enabled to surface a misleading "driver is snowflake" failure. Also drops a noisy console.warn from Snowflake primary-key discovery — INFORMATION_SCHEMA.KEY_COLUMN_USAGE is commonly ungranted for read-only roles and the FK + profiling paths handle the empty PK map already. * fix(llm): allow StructuredOutput tool and raise maxTurns for generateObject The Claude Code agent SDK announces an internal pseudo-tool named StructuredOutput in the system/init message whenever outputFormat is set to { type: 'json_schema' }. The runtime's isolation check built its allowedToolIds set only from MCP tool ids and treated StructuredOutput as an unexpected host-injected tool, so every generateObject call threw "Claude Code runtime isolation failed: tools=StructuredOutput ..." and the table-descriptions and relationship-LLM-proposal enrichment stages recorded null output across the board. Whitelist StructuredOutput specifically in generateObject's allowedToolIds — the check also enforces missing_tools symmetry, so generateText and runAgentLoop, which do not see StructuredOutput, must not require it. generateObject also ran with maxTurns: 1, which the model intermittently breached when it emitted thinking text before the structured response. Raised to 5 to give the schema-bound call enough headroom without allowing unbounded loops. The existing tests now exercise the path with an init message that announces StructuredOutput so the regression cannot slip back in. * chore(scripts): add ktx-reset.sh project-cleanup helper Convenience script for repeatable ingest testing: takes a project directory and prunes everything except ktx.yaml and .ktx/secrets/, so the next ktx setup or ktx ingest run starts from a known-clean state.
2026-05-23 10:41:30 +02:00
expect(generateObject).toHaveBeenCalled();
2026-05-10 23:12:26 +02:00
});
it('runs providerless enriched scans as relationship-only discovery enrichment', async () => {
const executor = new InMemorySqliteExecutor();
try {
executor.db.exec(`
CREATE TABLE accounts (id INTEGER NOT NULL);
CREATE TABLE orders (id INTEGER NOT NULL, account_id INTEGER NOT NULL);
INSERT INTO accounts (id) VALUES (1), (2);
INSERT INTO orders (id, account_id) VALUES (10, 1), (11, 1), (12, 2);
`);
const scanConnector = {
...connector(),
driver: 'sqlite' as const,
2026-05-10 23:51:24 +02:00
capabilities: createKtxConnectorCapabilities({ readOnlySql: true, columnStats: true }),
2026-05-10 23:12:26 +02:00
introspect: vi.fn(async () => noDeclaredRelationshipSnapshot()),
executeReadOnly: executor.executeReadOnly.bind(executor),
};
const result = await runLocalScanEnrichment({
connectionId: 'warehouse',
mode: 'enriched',
detectRelationships: false,
connector: scanConnector,
context: { runId: 'scan-run-providerless-enriched' },
providers: null,
});
expect(result.summary).toEqual({
dataDictionary: 'skipped',
tableDescriptions: 'skipped',
columnDescriptions: 'skipped',
embeddings: 'skipped',
deterministicRelationships: 'completed',
llmRelationshipValidation: 'skipped',
statisticalValidation: 'completed',
});
expect(result.descriptionUpdates).toEqual([]);
expect(result.embeddingUpdates).toEqual([]);
expect(result.relationships).toEqual({ accepted: 1, review: 0, rejected: 0, skipped: 0 });
expect(result.relationshipUpdate?.accepted).toHaveLength(1);
expect(result.relationshipProfile).toMatchObject({ sqlAvailable: true });
expect(result.resolvedRelationships).toEqual([
expect.objectContaining({
status: 'accepted',
from: expect.objectContaining({ table: expect.objectContaining({ name: 'orders' }), columns: ['account_id'] }),
to: expect.objectContaining({ table: expect.objectContaining({ name: 'accounts' }), columns: ['id'] }),
}),
]);
expect(result.warnings).toContainEqual({
code: 'scan_enrichment_backend_not_configured',
message:
'Skipping description and embedding enrichment because scan.enrichment.mode is not configured; relationship discovery still ran.',
recoverable: true,
metadata: {
skippedStages: ['descriptions', 'embeddings'],
relationshipDetection: true,
},
});
} finally {
executor.close();
}
});
});