ktx/packages/cli/src/connectors/postgres/connector.ts

853 lines
28 KiB
TypeScript
Raw Normal View History

2026-05-10 23:12:26 +02:00
import { readFileSync } from 'node:fs';
import { homedir } from 'node:os';
import { resolve } from 'node:path';
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 { getDialectForDriver } from '../../context/connections/dialects.js';
chore(workspace): gate dead-code with knip production mode (#196) * refactor(workspace): relocate @ktx/llm source into packages/cli/src/llm * refactor(workspace): rewrite @ktx/llm imports to relative paths * refactor(workspace): fold internal packages into cli * chore(workspace): gate dead-code with knip production mode Turn on production-mode knip plus an autofix run in pre-commit and the `pnpm dead-code` script, document the `/** @internal */` convention for test-only exports in AGENTS.md, annotate test-only exports across the CLI with that JSDoc, and drop dead exports/wrappers the new gate surfaced (e.g. `cli-project.ts`, `lookerRuntimeSourceToFileAdapterSource`, `createLocalScanEnrichmentProvidersFromConfig`, `PGLITE_OWNER_PROCESS_BACKEND_CAPABILITIES`, stale type re-exports). Replace the loose `ignoreIssues` allowlist in `knip.json` with explicit production entries so cross-package barrel leaks are caught. * refactor(cli): delete internal barrel index.ts files The 34 `index.ts` re-export barrels inside `packages/cli/src/` were holdovers from the pre-fold multi-workspace structure. Post-fold-in they served no production purpose: external consumers go through the single package main entry, and in-repo callers mostly imported through them only because the path was short. Internally, knip flagged most barrel re-exports as production-dead (only reached via tests). This change: - Deletes every internal barrel except `packages/cli/src/index.ts` (the published package entry). - Rewrites ~270 source/test files to import each name directly from the file that defines it. - Moves `tools/warehouse-verification/index.ts` to `create-warehouse-verification-tools.ts` (the function it defined locally) and updates its single consumer. - Renames `search/backend-conformance.ts` → `.test-utils.ts` to match the existing test-helper file convention. - Deletes 13 dead test-only chains (dbt-descriptions/*, live-database/extracted-schema, live-database/structural-sync, relationship-* feedback/review chain) plus their tests and a cascading orphan integration test. - Updates test mocks that pointed at deleted barrel paths (notion-client, connector barrels in scan/local-scan-connectors tests) to mock the source files instead. - Points the maintainer benchmark script (`scripts/relationship-benchmark-report.mjs`) at source files instead of `dist/context/scan/index.js`. - Drops the barrel `!` entries from `knip.json`; adds explicit production entries only for the benchmark code reached via dist by the maintainer script. Net: 413 files changed, ~1.2k insertions, ~9.4k deletions. `pnpm run dead-code` (Biome + knip default + knip production) and `pnpm run type-check` are clean; 2277 tests pass. * refactor(workspace): rename @ktx/cli to @kaelio/ktx and pack it directly Promote the CLI workspace package to the public name `@kaelio/ktx` and drop the separate `scripts/build-public-npm-package.mjs` wrapper. The CLI package is now publishable in place (`publishConfig.access: public`, `provenance: true`), so artifact packing uses `pnpm pack` against `packages/cli/` instead of assembling a parallel package tree. Updates all workspace filter invocations, docs, tests, and release readiness checks to reference the new package name, and folds the tarball-name helper into `scripts/public-npm-release-metadata.mjs`. * docs: align "agent clients" and "data agents" terminology Replace "client agents" with "agent clients" and "database agents" with "data agents" across AGENTS.md, README.md, the docs-site copy, and the matching setup-agents test description, matching the canonical vocabulary in docs/terminology.md. Also moves packages/cli/tsconfig.json's tsBuildInfoFile from node_modules/.cache/ to dist/.tsbuildinfo so incremental builds survive node_modules reinstalls. * refactor(release): single source of truth for package version Make packages/cli/package.json the single source of truth for the @kaelio/ktx version. publicNpmPackageVersion() now reads it directly, so artifact filenames, release-readiness checks, and the Python wheel version all derive from one field. The duplicate release-policy.json.publicNpmPackageVersion is removed. Previously the two fields could drift: tarballs were named kaelio-ktx-0.4.1.tgz while internally containing @kaelio/ktx@0.0.0-private. - update-public-release-version.mjs rewrites both Python pyproject.toml files (ktx-daemon, ktx-sl) alongside the npm package.jsons, normalizing the version for PEP 440 (e.g. 0.1.0-rc.2 -> 0.1.0rc2). - semantic-release-config.cjs adds the two pyproject.toml files to @semantic-release/git assets so the release commit back to main carries every version source in lockstep. - The six "?? '0.0.0-private'" fallback literals across the CLI are replaced with "?? getKtxCliPackageInfo().version", and createDefaultKtxMcpServer makes its version arg required. - docs/release.md describes the actual commit-back model: the dev tree always reflects the most recent release; no sentinel pin to maintain. Verified: pnpm run artifacts:build now produces kaelio-ktx-0.4.1.tgz and kaelio_ktx-0.4.1-py3-none-any.whl with @kaelio/ktx@0.4.1 inside. Full type-check, dead-code, and 2287 vitests + 173 script tests pass. * refactor(cli): inject embedding provider resolution and detect sentence-transformers runtime Make resolveProjectEmbeddingProvider and runtimeIo injectable in ingest and scan command entrypoints so tests can stub them, and teach resolvePublicIngestRuntimeRequirements to flag the local-embeddings runtime feature when ktx.yaml selects sentence-transformers. * chore(cli): mark buildLocalStatsStatus and LocalStatsStatus as @internal Both symbols are consumed only by status-project.test.ts. Annotating with /** @internal */ keeps knip's production-mode check clean without changing runtime behavior. * fix(cli): use real package metadata in print-command-tree The stubbed package name embedded a forbidden product identifier that tripped the boundary check in CI. Read the metadata from package.json instead — keeps the rendered tree unchanged and removes a duplicate source of truth. * feat(cli): show embedding coverage in `ktx status`, drop duplicate disk counts Inline `(N embedded)` next to the Wiki scope counts and Semantic-layer source counts, computed with `SUM(embedding_json IS NOT NULL)` over `knowledge_pages` and `local_sl_sources`. Rename the "Knowledge" label to "Wiki" (canonical per `docs/terminology.md`) and rename the matching `localStats.knowledgePages` field to `localStats.wikiPages`. Drop `wiki=N md` and `semantic-layer=N yaml` from the Disk row — those duplicated the per-surface rows above. Disk now reports only actual byte usage (db, cache, raw-sources). The unused `wikiGlobalMarkdownCount` / `semanticLayerYamlCount` fields, the `isMarkdownEntry` / `isYamlEntry` helpers, and the `filter` arg on `summarizeDir` are removed.
2026-05-21 15:28:58 +02:00
import { assertReadOnlySql, limitSqlForExecution } from '../../context/connections/read-only-sql.js';
import { tryConstraintQuery } from '../../context/scan/constraint-discovery.js';
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
import { scopedTableNames } from '../../context/scan/table-ref.js';
import {
createKtxConnectorCapabilities,
type KtxColumnSampleInput,
type KtxColumnSampleResult,
type KtxColumnStatsInput,
type KtxColumnStatsResult,
type KtxQueryResult,
type KtxReadOnlyQueryInput,
type KtxScanConnector,
type KtxScanContext,
type KtxScanInput,
type KtxScanWarning,
type KtxSchemaColumn,
type KtxSchemaForeignKey,
type KtxSchemaSnapshot,
type KtxSchemaTable,
type KtxTableListEntry,
type KtxTableRef,
type KtxTableSampleInput,
type KtxTableSampleResult,
} from '../../context/scan/types.js';
2026-05-10 23:12:26 +02:00
import { Pool } from 'pg';
const PG_OID_TYPE_MAP: Record<number, string> = {
16: 'boolean',
20: 'bigint',
21: 'smallint',
23: 'integer',
25: 'text',
700: 'real',
701: 'double precision',
1043: 'varchar',
1082: 'date',
1114: 'timestamp',
1184: 'timestamptz',
1700: 'numeric',
2950: 'uuid',
3802: 'jsonb',
114: 'json',
1009: 'text[]',
1007: 'integer[]',
1016: 'bigint[]',
};
2026-05-10 23:51:24 +02:00
export interface KtxPostgresConnectionConfig {
2026-05-10 23:12:26 +02:00
driver?: string;
host?: string;
port?: number;
database?: string;
username?: string;
user?: string;
password?: string;
url?: string;
schema?: string;
schemas?: string[];
ssl?: boolean;
sslmode?: string;
sslMode?: string;
2026-05-10 23:12:26 +02:00
rejectUnauthorized?: boolean;
maxConnections?: number;
2026-05-10 23:12:26 +02:00
[key: string]: unknown;
}
2026-05-10 23:51:24 +02:00
export interface KtxPostgresPoolConfig {
2026-05-10 23:12:26 +02:00
host?: string;
port?: number;
database?: string;
user?: string;
password?: string;
connectionString?: string;
max: number;
idleTimeoutMillis: number;
connectionTimeoutMillis: number;
options?: string;
ssl?: { rejectUnauthorized: boolean };
}
2026-05-10 23:51:24 +02:00
interface KtxPostgresQueryResult {
2026-05-10 23:12:26 +02:00
fields?: Array<{ name: string; dataTypeID: number }>;
rows: Record<string, unknown>[];
}
2026-05-10 23:51:24 +02:00
interface KtxPostgresClient {
query(sql: string, params?: unknown[]): Promise<KtxPostgresQueryResult>;
2026-05-10 23:12:26 +02:00
release(): void;
}
2026-05-10 23:51:24 +02:00
interface KtxPostgresPool {
connect(): Promise<KtxPostgresClient>;
2026-05-10 23:12:26 +02:00
end(): Promise<void>;
on?(event: 'error', listener: (error: Error) => void): void;
2026-05-10 23:12:26 +02:00
}
2026-05-10 23:51:24 +02:00
export interface KtxPostgresPoolFactory {
createPool(config: KtxPostgresPoolConfig): KtxPostgresPool;
2026-05-10 23:12:26 +02:00
}
2026-05-10 23:51:24 +02:00
interface KtxPostgresResolvedEndpoint {
2026-05-10 23:12:26 +02:00
host: string;
port: number;
close?: () => Promise<void>;
}
2026-05-10 23:51:24 +02:00
export interface KtxPostgresEndpointResolver {
2026-05-10 23:12:26 +02:00
resolve(input: {
host: string;
port: number;
2026-05-10 23:51:24 +02:00
connection: KtxPostgresConnectionConfig;
}): Promise<KtxPostgresResolvedEndpoint>;
2026-05-10 23:12:26 +02:00
}
2026-05-10 23:51:24 +02:00
export interface KtxPostgresScanConnectorOptions {
2026-05-10 23:12:26 +02:00
connectionId: string;
2026-05-10 23:51:24 +02:00
connection: KtxPostgresConnectionConfig | undefined;
poolFactory?: KtxPostgresPoolFactory;
endpointResolver?: KtxPostgresEndpointResolver;
2026-05-10 23:12:26 +02:00
env?: NodeJS.ProcessEnv;
now?: () => Date;
}
2026-05-10 23:51:24 +02:00
export interface KtxPostgresReadOnlyQueryInput extends KtxReadOnlyQueryInput {
2026-05-10 23:12:26 +02:00
params?: Record<string, unknown> | unknown[];
}
2026-05-10 23:51:24 +02:00
export interface KtxPostgresColumnDistinctValuesOptions {
2026-05-10 23:12:26 +02:00
maxCardinality: number;
limit: number;
sampleSize?: number;
}
2026-05-10 23:51:24 +02:00
export interface KtxPostgresColumnDistinctValuesResult {
2026-05-10 23:12:26 +02:00
values: string[] | null;
cardinality: number;
}
2026-05-10 23:51:24 +02:00
export interface KtxPostgresColumnStatisticsResult {
2026-05-10 23:12:26 +02:00
cardinalityByColumn: Map<string, number>;
}
2026-05-10 23:51:24 +02:00
export interface KtxPostgresTableSampleResult extends KtxTableSampleResult {
2026-05-10 23:12:26 +02:00
headerTypes?: string[];
}
2026-05-10 23:51:24 +02:00
type PostgresTableRef = Pick<KtxTableRef, 'name'> & Partial<Pick<KtxTableRef, 'catalog' | 'db'>>;
2026-05-10 23:12:26 +02:00
interface PostgresTableRow {
table_name: string;
table_kind: string;
row_count: unknown;
table_comment: string | null;
}
interface PostgresColumnRow {
table_name: string;
column_name: string;
data_type: string;
is_nullable: boolean;
column_comment: string | null;
}
interface PostgresPrimaryKeyRow {
table_name: string;
column_name: string;
}
interface PostgresForeignKeyRow {
table_name: string;
column_name: string;
foreign_table_schema: string | null;
foreign_table_name: string;
foreign_column_name: string;
constraint_name: string | null;
}
interface PostgresSchemaRow {
schema_name: string;
}
2026-05-12 18:22:05 -07:00
interface PostgresTableListRow {
schema_name: string;
table_name: string;
table_kind: string;
}
2026-05-10 23:12:26 +02:00
interface PostgresCountRow {
count?: unknown;
cardinality?: unknown;
}
interface PostgresDistinctValueRow {
val: unknown;
}
interface PostgresStatsRow {
column_name: string;
estimated_cardinality: unknown;
}
2026-05-10 23:51:24 +02:00
class DefaultPostgresPoolFactory implements KtxPostgresPoolFactory {
createPool(config: KtxPostgresPoolConfig): KtxPostgresPool {
2026-05-10 23:12:26 +02:00
return new Pool(config);
}
}
function groupByTable<T extends { table_name: string }>(rows: T[]): Map<string, T[]> {
const grouped = new Map<string, T[]>();
for (const row of rows) {
const tableRows = grouped.get(row.table_name) ?? [];
tableRows.push(row);
grouped.set(row.table_name, tableRows);
}
return grouped;
}
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
/** @internal */
export function preparePostgresReadOnlyQuery(
sql: string,
params?: Record<string, unknown>,
): { sql: string; params?: unknown[] } {
if (!params) {
return { sql, params: undefined };
}
const paramNames = Object.keys(params);
const values: unknown[] = new Array(paramNames.length);
const paramIndexMap = new Map<string, number>();
paramNames.forEach((name, index) => {
paramIndexMap.set(name, index + 1);
values[index] = params[name];
});
const sortedKeys = [...paramNames].sort((a, b) => b.length - a.length);
let parameterizedQuery = sql;
for (const name of sortedKeys) {
parameterizedQuery = parameterizedQuery.replace(new RegExp(`:${name}\\b`, 'g'), `$${paramIndexMap.get(name)}`);
}
return { sql: parameterizedQuery, params: values };
}
2026-05-10 23:12:26 +02:00
function primaryKeyMap(rows: PostgresPrimaryKeyRow[]): Map<string, Set<string>> {
const grouped = new Map<string, Set<string>>();
for (const row of rows) {
const columns = grouped.get(row.table_name) ?? new Set<string>();
columns.add(row.column_name);
grouped.set(row.table_name, columns);
}
return grouped;
}
function isDeniedError(error: unknown): boolean {
if (!error || typeof error !== 'object') {
return false;
}
const code = (error as { code?: unknown }).code;
return code === '42501' || code === '42P01';
}
2026-05-10 23:51:24 +02:00
function queryRows(result: KtxPostgresQueryResult): unknown[][] {
2026-05-10 23:12:26 +02:00
const headers = (result.fields ?? []).map((field) => field.name);
return result.rows.map((row) => headers.map((header) => row[header]));
}
function finiteNumber(value: unknown): number | null {
const parsed = Number(value);
return Number.isFinite(parsed) ? parsed : null;
}
function stringConfigValue(
2026-05-10 23:51:24 +02:00
connection: KtxPostgresConnectionConfig | undefined,
key: keyof KtxPostgresConnectionConfig,
2026-05-10 23:12:26 +02:00
env: NodeJS.ProcessEnv,
): string | undefined {
const value = connection?.[key];
return typeof value === 'string' && value.trim().length > 0 ? resolveStringReference(value.trim(), env) : undefined;
}
function resolveStringReference(value: string, env: NodeJS.ProcessEnv): string {
if (value.startsWith('env:')) {
return env[value.slice('env:'.length)] ?? '';
}
if (value.startsWith('file:')) {
const rawPath = value.slice('file:'.length);
const path = rawPath.startsWith('~') ? resolve(homedir(), rawPath.slice(1)) : rawPath;
return readFileSync(path, 'utf-8').trim();
}
return value;
}
function numberValue(value: unknown): number | undefined {
return typeof value === 'number' && Number.isFinite(value) ? value : undefined;
}
function positiveIntegerConfigValue(input: {
connection: KtxPostgresConnectionConfig;
key: keyof KtxPostgresConnectionConfig;
connectionId: string;
defaultValue: number;
}): number {
const value = input.connection[input.key];
if (value === undefined) {
return input.defaultValue;
}
const numberValue = Number(value);
if (!Number.isInteger(numberValue) || numberValue < 1) {
throw new Error(`connections.${input.connectionId}.${String(input.key)} must be a positive integer`);
}
return numberValue;
}
2026-05-10 23:51:24 +02:00
function parsePostgresUrl(url: string): Partial<KtxPostgresConnectionConfig> {
2026-05-10 23:12:26 +02:00
const parsed = new URL(url);
const sslmode = parsed.searchParams.get('sslmode') ?? undefined;
2026-05-10 23:12:26 +02:00
return {
host: parsed.hostname,
port: parsed.port ? Number(parsed.port) : undefined,
database: parsed.pathname.replace(/^\/+/, '') || undefined,
username: parsed.username ? decodeURIComponent(parsed.username) : undefined,
password: parsed.password ? decodeURIComponent(parsed.password) : undefined,
...(sslmode ? { sslmode } : {}),
2026-05-10 23:12:26 +02:00
};
}
function normalizedSslMode(connection: KtxPostgresConnectionConfig): string | undefined {
const value = connection.sslmode ?? connection.sslMode;
return typeof value === 'string' && value.trim().length > 0 ? value.trim().toLowerCase() : undefined;
}
2026-05-10 23:51:24 +02:00
function schemasFromConnection(connection: KtxPostgresConnectionConfig): string[] {
2026-05-10 23:12:26 +02:00
if (Array.isArray(connection.schemas) && connection.schemas.length > 0) {
return connection.schemas.filter((schema): schema is string => typeof schema === 'string' && schema.length > 0);
}
return typeof connection.schema === 'string' && connection.schema.length > 0 ? [connection.schema] : ['public'];
}
2026-05-10 23:51:24 +02:00
function searchPathSchemasFromConnection(connection: KtxPostgresConnectionConfig): string[] {
2026-05-10 23:12:26 +02:00
const schemas = schemasFromConnection(connection);
return schemas.includes('public') ? schemas : [...schemas, 'public'];
}
export function isKtxPostgresConnectionConfig(
connection: KtxPostgresConnectionConfig | undefined,
): connection is KtxPostgresConnectionConfig {
2026-05-10 23:12:26 +02:00
const driver = String(connection?.driver ?? '').toLowerCase();
return driver === 'postgres';
2026-05-10 23:12:26 +02:00
}
chore(workspace): gate dead-code with knip production mode (#196) * refactor(workspace): relocate @ktx/llm source into packages/cli/src/llm * refactor(workspace): rewrite @ktx/llm imports to relative paths * refactor(workspace): fold internal packages into cli * chore(workspace): gate dead-code with knip production mode Turn on production-mode knip plus an autofix run in pre-commit and the `pnpm dead-code` script, document the `/** @internal */` convention for test-only exports in AGENTS.md, annotate test-only exports across the CLI with that JSDoc, and drop dead exports/wrappers the new gate surfaced (e.g. `cli-project.ts`, `lookerRuntimeSourceToFileAdapterSource`, `createLocalScanEnrichmentProvidersFromConfig`, `PGLITE_OWNER_PROCESS_BACKEND_CAPABILITIES`, stale type re-exports). Replace the loose `ignoreIssues` allowlist in `knip.json` with explicit production entries so cross-package barrel leaks are caught. * refactor(cli): delete internal barrel index.ts files The 34 `index.ts` re-export barrels inside `packages/cli/src/` were holdovers from the pre-fold multi-workspace structure. Post-fold-in they served no production purpose: external consumers go through the single package main entry, and in-repo callers mostly imported through them only because the path was short. Internally, knip flagged most barrel re-exports as production-dead (only reached via tests). This change: - Deletes every internal barrel except `packages/cli/src/index.ts` (the published package entry). - Rewrites ~270 source/test files to import each name directly from the file that defines it. - Moves `tools/warehouse-verification/index.ts` to `create-warehouse-verification-tools.ts` (the function it defined locally) and updates its single consumer. - Renames `search/backend-conformance.ts` → `.test-utils.ts` to match the existing test-helper file convention. - Deletes 13 dead test-only chains (dbt-descriptions/*, live-database/extracted-schema, live-database/structural-sync, relationship-* feedback/review chain) plus their tests and a cascading orphan integration test. - Updates test mocks that pointed at deleted barrel paths (notion-client, connector barrels in scan/local-scan-connectors tests) to mock the source files instead. - Points the maintainer benchmark script (`scripts/relationship-benchmark-report.mjs`) at source files instead of `dist/context/scan/index.js`. - Drops the barrel `!` entries from `knip.json`; adds explicit production entries only for the benchmark code reached via dist by the maintainer script. Net: 413 files changed, ~1.2k insertions, ~9.4k deletions. `pnpm run dead-code` (Biome + knip default + knip production) and `pnpm run type-check` are clean; 2277 tests pass. * refactor(workspace): rename @ktx/cli to @kaelio/ktx and pack it directly Promote the CLI workspace package to the public name `@kaelio/ktx` and drop the separate `scripts/build-public-npm-package.mjs` wrapper. The CLI package is now publishable in place (`publishConfig.access: public`, `provenance: true`), so artifact packing uses `pnpm pack` against `packages/cli/` instead of assembling a parallel package tree. Updates all workspace filter invocations, docs, tests, and release readiness checks to reference the new package name, and folds the tarball-name helper into `scripts/public-npm-release-metadata.mjs`. * docs: align "agent clients" and "data agents" terminology Replace "client agents" with "agent clients" and "database agents" with "data agents" across AGENTS.md, README.md, the docs-site copy, and the matching setup-agents test description, matching the canonical vocabulary in docs/terminology.md. Also moves packages/cli/tsconfig.json's tsBuildInfoFile from node_modules/.cache/ to dist/.tsbuildinfo so incremental builds survive node_modules reinstalls. * refactor(release): single source of truth for package version Make packages/cli/package.json the single source of truth for the @kaelio/ktx version. publicNpmPackageVersion() now reads it directly, so artifact filenames, release-readiness checks, and the Python wheel version all derive from one field. The duplicate release-policy.json.publicNpmPackageVersion is removed. Previously the two fields could drift: tarballs were named kaelio-ktx-0.4.1.tgz while internally containing @kaelio/ktx@0.0.0-private. - update-public-release-version.mjs rewrites both Python pyproject.toml files (ktx-daemon, ktx-sl) alongside the npm package.jsons, normalizing the version for PEP 440 (e.g. 0.1.0-rc.2 -> 0.1.0rc2). - semantic-release-config.cjs adds the two pyproject.toml files to @semantic-release/git assets so the release commit back to main carries every version source in lockstep. - The six "?? '0.0.0-private'" fallback literals across the CLI are replaced with "?? getKtxCliPackageInfo().version", and createDefaultKtxMcpServer makes its version arg required. - docs/release.md describes the actual commit-back model: the dev tree always reflects the most recent release; no sentinel pin to maintain. Verified: pnpm run artifacts:build now produces kaelio-ktx-0.4.1.tgz and kaelio_ktx-0.4.1-py3-none-any.whl with @kaelio/ktx@0.4.1 inside. Full type-check, dead-code, and 2287 vitests + 173 script tests pass. * refactor(cli): inject embedding provider resolution and detect sentence-transformers runtime Make resolveProjectEmbeddingProvider and runtimeIo injectable in ingest and scan command entrypoints so tests can stub them, and teach resolvePublicIngestRuntimeRequirements to flag the local-embeddings runtime feature when ktx.yaml selects sentence-transformers. * chore(cli): mark buildLocalStatsStatus and LocalStatsStatus as @internal Both symbols are consumed only by status-project.test.ts. Annotating with /** @internal */ keeps knip's production-mode check clean without changing runtime behavior. * fix(cli): use real package metadata in print-command-tree The stubbed package name embedded a forbidden product identifier that tripped the boundary check in CI. Read the metadata from package.json instead — keeps the rendered tree unchanged and removes a duplicate source of truth. * feat(cli): show embedding coverage in `ktx status`, drop duplicate disk counts Inline `(N embedded)` next to the Wiki scope counts and Semantic-layer source counts, computed with `SUM(embedding_json IS NOT NULL)` over `knowledge_pages` and `local_sl_sources`. Rename the "Knowledge" label to "Wiki" (canonical per `docs/terminology.md`) and rename the matching `localStats.knowledgePages` field to `localStats.wikiPages`. Drop `wiki=N md` and `semantic-layer=N yaml` from the Disk row — those duplicated the per-surface rows above. Disk now reports only actual byte usage (db, cache, raw-sources). The unused `wikiGlobalMarkdownCount` / `semanticLayerYamlCount` fields, the `isMarkdownEntry` / `isYamlEntry` helpers, and the `filter` arg on `summarizeDir` are removed.
2026-05-21 15:28:58 +02:00
/** @internal */
2026-05-10 23:12:26 +02:00
export function postgresPoolConfigFromConfig(input: {
connectionId: string;
2026-05-10 23:51:24 +02:00
connection: KtxPostgresConnectionConfig | undefined;
2026-05-10 23:12:26 +02:00
env?: NodeJS.ProcessEnv;
2026-05-10 23:51:24 +02:00
}): KtxPostgresPoolConfig {
const inputDriver = input.connection?.driver ?? 'unknown';
2026-05-10 23:51:24 +02:00
if (!isKtxPostgresConnectionConfig(input.connection)) {
throw new Error(`Native PostgreSQL connector cannot run driver "${inputDriver}"`);
2026-05-10 23:12:26 +02:00
}
const env = input.env ?? process.env;
const referencedUrl = stringConfigValue(input.connection, 'url', env);
const urlConfig = referencedUrl ? parsePostgresUrl(referencedUrl) : {};
2026-05-10 23:51:24 +02:00
const merged: KtxPostgresConnectionConfig = { ...urlConfig, ...input.connection };
2026-05-10 23:12:26 +02:00
const host = stringConfigValue(merged, 'host', env);
const database = stringConfigValue(merged, 'database', env);
const user = stringConfigValue(merged, 'username', env) ?? stringConfigValue(merged, 'user', env);
const password = stringConfigValue(merged, 'password', env);
const sslmode = normalizedSslMode(merged);
const maxConnections = positiveIntegerConfigValue({
connection: merged,
key: 'maxConnections',
connectionId: input.connectionId,
defaultValue: 10,
});
2026-05-10 23:12:26 +02:00
if (!referencedUrl && !host) {
throw new Error(`Native PostgreSQL connector requires connections.${input.connectionId}.host or url`);
}
if (!database && !referencedUrl) {
throw new Error(`Native PostgreSQL connector requires connections.${input.connectionId}.database or url`);
}
if (!user && !referencedUrl) {
throw new Error(`Native PostgreSQL connector requires connections.${input.connectionId}.username, user, or url`);
}
2026-05-10 23:51:24 +02:00
const config: KtxPostgresPoolConfig = {
max: maxConnections,
2026-05-10 23:12:26 +02:00
idleTimeoutMillis: 30_000,
connectionTimeoutMillis: 10_000,
...(referencedUrl && sslmode !== 'prefer' && sslmode !== 'disable'
2026-05-10 23:12:26 +02:00
? { connectionString: referencedUrl }
: { host, port: numberValue(merged.port) ?? 5432, database, user, password }),
};
const searchPathSchemas = searchPathSchemasFromConnection(merged);
if (searchPathSchemas.length > 0) {
config.options = `-c search_path=${searchPathSchemas.join(',')}`;
}
if (merged.ssl && sslmode !== 'prefer' && sslmode !== 'disable') {
2026-05-10 23:12:26 +02:00
config.ssl = { rejectUnauthorized: merged.rejectUnauthorized ?? true };
}
return config;
}
2026-05-10 23:51:24 +02:00
export class KtxPostgresScanConnector implements KtxScanConnector {
2026-05-10 23:12:26 +02:00
readonly id: string;
readonly driver = 'postgres' as const;
2026-05-10 23:51:24 +02:00
readonly capabilities = createKtxConnectorCapabilities({
2026-05-10 23:12:26 +02:00
tableSampling: true,
columnSampling: true,
columnStats: true,
readOnlySql: true,
nestedAnalysis: true,
formalForeignKeys: true,
estimatedRowCounts: true,
});
private readonly connectionId: string;
2026-05-10 23:51:24 +02:00
private readonly connection: KtxPostgresConnectionConfig;
private readonly poolConfig: KtxPostgresPoolConfig;
private readonly poolFactory: KtxPostgresPoolFactory;
private readonly endpointResolver?: KtxPostgresEndpointResolver;
2026-05-10 23:12:26 +02:00
private readonly now: () => Date;
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
private readonly dialect = getDialectForDriver('postgres');
2026-05-10 23:51:24 +02:00
private pool: KtxPostgresPool | null = null;
private lastIdlePoolError: Error | null = null;
2026-05-10 23:51:24 +02:00
private resolvedEndpoint: KtxPostgresResolvedEndpoint | null = null;
2026-05-10 23:12:26 +02:00
2026-05-10 23:51:24 +02:00
constructor(options: KtxPostgresScanConnectorOptions) {
2026-05-10 23:12:26 +02:00
this.connectionId = options.connectionId;
this.connection = options.connection ?? {};
this.poolConfig = postgresPoolConfigFromConfig({
connectionId: options.connectionId,
connection: options.connection,
env: options.env,
});
this.poolFactory = options.poolFactory ?? new DefaultPostgresPoolFactory();
this.endpointResolver = options.endpointResolver;
this.now = options.now ?? (() => new Date());
this.id = `postgres:${options.connectionId}`;
}
async testConnection(): Promise<{ success: boolean; error?: string }> {
try {
await this.query('SELECT 1');
return { success: true };
} catch (error) {
return { success: false, error: error instanceof Error ? error.message : String(error) };
}
}
2026-05-10 23:51:24 +02:00
async introspect(input: KtxScanInput, _ctx: KtxScanContext): Promise<KtxSchemaSnapshot> {
2026-05-10 23:12:26 +02:00
this.assertConnection(input.connectionId);
const schemas = schemasFromConnection(this.connection);
2026-05-10 23:51:24 +02:00
const allTables: KtxSchemaTable[] = [];
const snapshotWarnings: KtxScanWarning[] = [];
2026-05-10 23:12:26 +02:00
for (const schema of schemas) {
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 scopedNames = input.tableScope ? scopedTableNames(input.tableScope, { catalog: null, db: schema }) : null;
if (scopedNames && scopedNames.length === 0) continue;
const tables = await this.loadSchemaTables(schema, scopedNames, snapshotWarnings);
2026-05-10 23:12:26 +02:00
allTables.push(...tables);
}
return {
connectionId: this.connectionId,
driver: 'postgres',
extractedAt: this.now().toISOString(),
scope: { schemas },
metadata: {
database: this.poolConfig.database ?? this.connection.database ?? null,
schemas,
host: this.poolConfig.host ?? this.connection.host ?? null,
table_count: allTables.length,
total_columns: allTables.reduce((sum, table) => sum + table.columns.length, 0),
},
tables: allTables,
warnings: snapshotWarnings,
2026-05-10 23:12:26 +02:00
};
}
2026-05-10 23:51:24 +02:00
async sampleTable(input: KtxTableSampleInput, _ctx: KtxScanContext): Promise<KtxPostgresTableSampleResult> {
2026-05-10 23:12:26 +02:00
this.assertConnection(input.connectionId);
const result = await this.query(this.dialect.generateSampleQuery(this.qTableName(input.table), input.limit, input.columns));
return {
headers: result.headers,
headerTypes: result.headerTypes,
rows: result.rows,
totalRows: result.totalRows,
};
}
2026-05-10 23:51:24 +02:00
async sampleColumn(input: KtxColumnSampleInput, _ctx: KtxScanContext): Promise<KtxColumnSampleResult> {
2026-05-10 23:12:26 +02:00
this.assertConnection(input.connectionId);
const result = await this.query(
this.dialect.generateColumnSampleQuery(this.qTableName(input.table), input.column, input.limit),
);
const values = result.rows.filter((row) => row.length > 0 && row[0] !== null).map((row) => row[0]);
return { values, nullCount: null, distinctCount: null };
}
2026-05-10 23:51:24 +02:00
async columnStats(input: KtxColumnStatsInput, _ctx: KtxScanContext): Promise<KtxColumnStatsResult | null> {
2026-05-10 23:12:26 +02:00
const stats = await this.getColumnStatistics(input.table);
const value = stats?.cardinalityByColumn.get(input.column);
return value === undefined
? null
: { min: null, max: null, average: null, nullCount: null, distinctCount: value };
}
2026-05-10 23:51:24 +02:00
async executeReadOnly(input: KtxPostgresReadOnlyQueryInput, _ctx: KtxScanContext): Promise<KtxQueryResult> {
2026-05-10 23:12:26 +02:00
this.assertConnection(input.connectionId);
const limitedSql = limitSqlForExecution(assertReadOnlySql(input.sql), input.maxRows);
const prepared = Array.isArray(input.params)
? { sql: limitedSql, params: input.params }
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
: preparePostgresReadOnlyQuery(limitedSql, input.params);
2026-05-10 23:12:26 +02:00
const result = await this.query(prepared.sql, prepared.params);
return { ...result, rowCount: result.rows.length };
}
async getColumnDistinctValues(
2026-05-10 23:51:24 +02:00
table: KtxTableRef,
2026-05-10 23:12:26 +02:00
columnName: string,
2026-05-10 23:51:24 +02:00
options: KtxPostgresColumnDistinctValuesOptions,
): Promise<KtxPostgresColumnDistinctValuesResult | null> {
2026-05-10 23:12:26 +02:00
const sampleSize = options.sampleSize ?? 10000;
const tableName = this.qTableName(table);
const quotedColumn = this.dialect.quoteIdentifier(columnName);
const cardinalityRows = await this.queryRaw<PostgresCountRow>(
this.dialect.generateCardinalitySampleQuery(tableName, quotedColumn, sampleSize),
);
const cardinality = finiteNumber(cardinalityRows[0]?.cardinality);
if (cardinality === null) {
return null;
}
if (cardinality === 0) {
return { values: [], cardinality: 0 };
}
if (cardinality > options.maxCardinality) {
return { values: null, cardinality };
}
const valuesRows = await this.queryRaw<PostgresDistinctValueRow>(
this.dialect.generateDistinctValuesQuery(tableName, quotedColumn, options.limit),
);
return {
values: valuesRows.filter((row) => row.val !== null).map((row) => String(row.val)),
cardinality,
};
}
2026-05-10 23:51:24 +02:00
async getColumnStatistics(table: KtxTableRef): Promise<KtxPostgresColumnStatisticsResult | null> {
2026-05-10 23:12:26 +02:00
const schema = table.db ?? schemasFromConnection(this.connection)[0] ?? 'public';
const sql = this.dialect.generateColumnStatisticsQuery(schema, table.name);
if (!sql) {
return null;
}
const rows = await this.queryRaw<PostgresStatsRow>(sql);
const cardinalityByColumn = new Map<string, number>();
for (const row of rows) {
const cardinality = finiteNumber(row.estimated_cardinality);
if (cardinality !== null) {
cardinalityByColumn.set(row.column_name, cardinality);
}
}
return cardinalityByColumn.size > 0 ? { cardinalityByColumn } : null;
}
async getTableRowCount(table: string | PostgresTableRef): Promise<number> {
const tableRef =
typeof table === 'string'
? { catalog: null, db: schemasFromConnection(this.connection)[0] ?? 'public', name: table }
: table;
const rows = await this.queryRaw<PostgresCountRow>(`SELECT COUNT(*) AS count FROM ${this.qTableName(tableRef)}`);
return finiteNumber(rows[0]?.count) ?? 0;
}
qTableName(table: PostgresTableRef): string {
return this.dialect.formatTableName(table);
}
quoteIdentifier(identifier: string): string {
return this.dialect.quoteIdentifier(identifier);
}
async listSchemas(): Promise<string[]> {
const rows = await this.queryRaw<PostgresSchemaRow>(`
SELECT schema_name
FROM information_schema.schemata
WHERE schema_name <> 'information_schema'
AND schema_name NOT LIKE 'pg_%'
ORDER BY schema_name
`);
return rows.map((row) => row.schema_name);
}
2026-05-12 18:22:05 -07:00
async listTables(schemas?: string[]): Promise<KtxTableListEntry[]> {
const filterSchemas = schemas ?? (await this.listSchemas());
if (filterSchemas.length === 0) return [];
const rows = await this.queryRaw<PostgresTableListRow>(
`
SELECT n.nspname AS schema_name, c.relname AS table_name, c.relkind AS table_kind
FROM pg_catalog.pg_class c
JOIN pg_catalog.pg_namespace n ON c.relnamespace = n.oid
WHERE n.nspname = ANY($1)
AND c.relkind IN ('r', 'v')
ORDER BY n.nspname, c.relname
`,
[filterSchemas],
);
return rows.map((row) => ({
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
catalog: null,
2026-05-12 18:22:05 -07:00
schema: row.schema_name,
name: row.table_name,
kind: row.table_kind === 'v' ? ('view' as const) : ('table' as const),
}));
}
2026-05-10 23:12:26 +02:00
async cleanup(): Promise<void> {
if (this.pool) {
await this.pool.end();
this.pool = null;
}
if (this.resolvedEndpoint?.close) {
await this.resolvedEndpoint.close();
this.resolvedEndpoint = null;
}
}
private async loadSchemaTables(
schema: string,
scopedNames: readonly string[] | null,
snapshotWarnings: KtxScanWarning[],
): Promise<KtxSchemaTable[]> {
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
if (scopedNames && scopedNames.length === 0) return [];
const pgCatalogScopeClause = scopedNames ? 'AND c.relname = ANY($2)' : '';
const tableConstraintScopeClause = scopedNames ? 'AND tc.table_name = ANY($2)' : '';
const scopeValues = scopedNames ? [scopedNames] : [];
2026-05-10 23:12:26 +02:00
const tables = await this.queryRaw<PostgresTableRow>(
`
SELECT
c.relname AS table_name,
c.relkind AS table_kind,
c.reltuples::bigint AS row_count,
d.description AS table_comment
FROM pg_catalog.pg_class c
JOIN pg_catalog.pg_namespace n ON c.relnamespace = n.oid
LEFT JOIN pg_catalog.pg_description d
ON d.objoid = c.oid AND d.objsubid = 0
WHERE n.nspname = $1
AND c.relkind IN ('r', 'v')
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
${pgCatalogScopeClause}
2026-05-10 23:12:26 +02:00
ORDER BY c.relname
`,
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
[schema, ...scopeValues],
2026-05-10 23:12:26 +02:00
);
const columns = await this.queryRaw<PostgresColumnRow>(
`
SELECT
c.relname AS table_name,
a.attname AS column_name,
format_type(a.atttypid, a.atttypmod) AS data_type,
NOT a.attnotnull AS is_nullable,
d.description AS column_comment
FROM pg_catalog.pg_attribute a
JOIN pg_catalog.pg_class c ON a.attrelid = c.oid
JOIN pg_catalog.pg_namespace n ON c.relnamespace = n.oid
LEFT JOIN pg_catalog.pg_description d
ON d.objoid = c.oid AND d.objsubid = a.attnum
WHERE n.nspname = $1
AND c.relkind IN ('r', 'v')
AND a.attnum > 0
AND NOT a.attisdropped
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
${pgCatalogScopeClause}
2026-05-10 23:12:26 +02:00
ORDER BY c.relname, a.attnum
`,
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
[schema, ...scopeValues],
2026-05-10 23:12:26 +02:00
);
const primaryKeysResult = await tryConstraintQuery(
{ schema, kind: 'primary_key', isDeniedError },
() =>
this.queryRaw<PostgresPrimaryKeyRow>(
`
2026-05-10 23:12:26 +02:00
SELECT tc.table_name, kcu.column_name
FROM information_schema.table_constraints tc
JOIN information_schema.key_column_usage kcu
ON tc.constraint_name = kcu.constraint_name
AND tc.table_schema = kcu.table_schema
WHERE tc.constraint_type = 'PRIMARY KEY'
AND tc.table_schema = $1
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
${tableConstraintScopeClause}
2026-05-10 23:12:26 +02:00
ORDER BY tc.table_name, kcu.ordinal_position
`,
[schema, ...scopeValues],
),
2026-05-10 23:12:26 +02:00
);
const primaryKeys = primaryKeysResult.ok ? primaryKeysResult.value : [];
if (!primaryKeysResult.ok) {
snapshotWarnings.push(primaryKeysResult.warning);
}
const foreignKeysResult = await tryConstraintQuery(
{ schema, kind: 'foreign_key', isDeniedError },
() =>
this.queryRaw<PostgresForeignKeyRow>(
`
2026-05-10 23:12:26 +02:00
SELECT
tc.table_name,
kcu.column_name,
ccu.table_schema AS foreign_table_schema,
ccu.table_name AS foreign_table_name,
ccu.column_name AS foreign_column_name,
tc.constraint_name
FROM information_schema.table_constraints AS tc
JOIN information_schema.key_column_usage AS kcu
ON tc.constraint_name = kcu.constraint_name
AND tc.table_schema = kcu.table_schema
JOIN information_schema.constraint_column_usage AS ccu
ON ccu.constraint_name = tc.constraint_name
AND ccu.table_schema = tc.table_schema
WHERE tc.constraint_type = 'FOREIGN KEY'
AND tc.table_schema = $1
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
${tableConstraintScopeClause}
2026-05-10 23:12:26 +02:00
ORDER BY tc.table_name, kcu.column_name
`,
[schema, ...scopeValues],
),
2026-05-10 23:12:26 +02:00
);
const foreignKeys = foreignKeysResult.ok ? foreignKeysResult.value : [];
if (!foreignKeysResult.ok) {
snapshotWarnings.push(foreignKeysResult.warning);
}
2026-05-10 23:12:26 +02:00
const columnsByTable = groupByTable(columns);
const primaryKeysByTable = primaryKeyMap(primaryKeys);
const foreignKeysByTable = groupByTable(foreignKeys);
return tables.map((table) =>
this.toSchemaTable(
schema,
table,
columnsByTable.get(table.table_name) ?? [],
primaryKeysByTable.get(table.table_name) ?? new Set<string>(),
foreignKeysByTable.get(table.table_name) ?? [],
),
);
}
private toSchemaTable(
schema: string,
table: PostgresTableRow,
columns: PostgresColumnRow[],
primaryKeys: Set<string>,
foreignKeys: PostgresForeignKeyRow[],
2026-05-10 23:51:24 +02:00
): KtxSchemaTable {
2026-05-10 23:12:26 +02:00
const kind = table.table_kind === 'v' ? 'view' : 'table';
return {
catalog: null,
db: schema,
name: table.table_name,
kind,
comment: table.table_comment || null,
estimatedRows: kind === 'view' ? null : finiteNumber(table.row_count),
columns: columns.map((column) => this.toSchemaColumn(column, primaryKeys)),
foreignKeys: foreignKeys.map((foreignKey) => this.toSchemaForeignKey(foreignKey)),
};
}
2026-05-10 23:51:24 +02:00
private toSchemaColumn(column: PostgresColumnRow, primaryKeys: Set<string>): KtxSchemaColumn {
2026-05-10 23:12:26 +02:00
return {
name: column.column_name,
nativeType: column.data_type,
normalizedType: this.dialect.mapDataType(column.data_type),
dimensionType: this.dialect.mapToDimensionType(column.data_type),
nullable: column.is_nullable,
primaryKey: primaryKeys.has(column.column_name),
comment: column.column_comment || null,
};
}
2026-05-10 23:51:24 +02:00
private toSchemaForeignKey(row: PostgresForeignKeyRow): KtxSchemaForeignKey {
2026-05-10 23:12:26 +02:00
return {
fromColumn: row.column_name,
toCatalog: null,
toDb: row.foreign_table_schema,
toTable: row.foreign_table_name,
toColumn: row.foreign_column_name,
constraintName: row.constraint_name || null,
};
}
2026-05-10 23:51:24 +02:00
private async getPool(): Promise<KtxPostgresPool> {
2026-05-10 23:12:26 +02:00
if (!this.pool) {
let config = { ...this.poolConfig };
if (this.endpointResolver) {
const endpoint = await this.endpointResolver.resolve({
host: config.host ?? this.connection.host ?? 'localhost',
port: config.port ?? numberValue(this.connection.port) ?? 5432,
connection: this.connection,
});
this.resolvedEndpoint = endpoint;
config = { ...config, host: endpoint.host, port: endpoint.port };
}
this.pool = this.poolFactory.createPool(config);
this.pool.on?.('error', (error) => {
this.lastIdlePoolError = error;
});
2026-05-10 23:12:26 +02:00
}
return this.pool;
}
private async queryRaw<T>(sql: string, params?: unknown[]): Promise<T[]> {
this.throwIdlePoolErrorIfPresent();
2026-05-10 23:12:26 +02:00
const pool = await this.getPool();
const client = await pool.connect();
try {
const result = await client.query(sql, params);
return result.rows as T[];
} finally {
client.release();
}
}
2026-05-10 23:51:24 +02:00
private async query(sql: string, params?: Record<string, unknown> | unknown[]): Promise<KtxQueryResult> {
this.throwIdlePoolErrorIfPresent();
2026-05-10 23:12:26 +02:00
const pool = await this.getPool();
const client = await pool.connect();
try {
const result = await client.query(assertReadOnlySql(sql), Array.isArray(params) ? params : undefined);
return {
headers: (result.fields ?? []).map((field) => field.name),
headerTypes: (result.fields ?? []).map((field) => PG_OID_TYPE_MAP[field.dataTypeID] ?? `oid:${field.dataTypeID}`),
rows: queryRows(result),
totalRows: result.rows.length,
rowCount: result.rows.length,
};
} finally {
client.release();
}
}
private assertConnection(connectionId: string): void {
if (connectionId !== this.connectionId) {
throw new Error(`PostgreSQL connector ${this.connectionId} cannot run scan for ${connectionId}`);
}
}
private throwIdlePoolErrorIfPresent(): void {
if (!this.lastIdlePoolError) {
return;
}
const error = this.lastIdlePoolError;
this.lastIdlePoolError = null;
throw error;
}
2026-05-10 23:12:26 +02:00
}