@databricks/appkit 0.41.5 → 0.41.7

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (35) hide show
  1. package/README.md +3 -3
  2. package/dist/appkit/package.js +1 -1
  3. package/dist/shared/src/schemas/metric-fqn.js +85 -0
  4. package/dist/shared/src/schemas/metric-fqn.js.map +1 -0
  5. package/dist/telemetry/noop.js +2 -1
  6. package/dist/telemetry/noop.js.map +1 -1
  7. package/dist/type-generator/cache.js +33 -1
  8. package/dist/type-generator/cache.js.map +1 -1
  9. package/dist/type-generator/errors.js +100 -0
  10. package/dist/type-generator/errors.js.map +1 -0
  11. package/dist/type-generator/index.js +190 -2
  12. package/dist/type-generator/index.js.map +1 -1
  13. package/dist/type-generator/mv-registry/config.js +131 -0
  14. package/dist/type-generator/mv-registry/config.js.map +1 -0
  15. package/dist/type-generator/mv-registry/describe.js +230 -0
  16. package/dist/type-generator/mv-registry/describe.js.map +1 -0
  17. package/dist/type-generator/mv-registry/metadata.js +40 -0
  18. package/dist/type-generator/mv-registry/metadata.js.map +1 -0
  19. package/dist/type-generator/mv-registry/render-types.js +118 -0
  20. package/dist/type-generator/mv-registry/render-types.js.map +1 -0
  21. package/dist/type-generator/mv-registry/sync.js +96 -0
  22. package/dist/type-generator/mv-registry/sync.js.map +1 -0
  23. package/dist/type-generator/query-registry.js +126 -88
  24. package/dist/type-generator/query-registry.js.map +1 -1
  25. package/dist/type-generator/statement-result.js +116 -0
  26. package/dist/type-generator/statement-result.js.map +1 -0
  27. package/dist/type-generator/types.js.map +1 -1
  28. package/dist/type-generator/vite-plugin.d.ts +12 -0
  29. package/dist/type-generator/vite-plugin.d.ts.map +1 -1
  30. package/dist/type-generator/vite-plugin.js +11 -5
  31. package/dist/type-generator/vite-plugin.js.map +1 -1
  32. package/docs/development/type-generation.md +4 -0
  33. package/docs/plugins/analytics.md +15 -0
  34. package/package.json +19 -18
  35. package/sbom.cdx.json +1 -1
@@ -1 +1 @@
1
- {"version":3,"file":"query-registry.js","names":[],"sources":["../../src/type-generator/query-registry.ts"],"sourcesContent":["import fs from \"node:fs/promises\";\nimport path from \"node:path\";\nimport { WorkspaceClient } from \"@databricks/sdk-experimental\";\nimport pc from \"picocolors\";\nimport { createLogger } from \"../logging/logger\";\nimport { CACHE_VERSION, hashSQL, loadCache, saveCache } from \"./cache\";\nimport { decidePreflight, type PreflightMode } from \"./preflight\";\nimport { Spinner } from \"./spinner\";\nimport {\n type DatabricksStatementExecutionResponse,\n type QueryFatalError,\n type QueryGenerationResult,\n type QuerySchema,\n type QuerySyntaxError,\n sqlTypeToHelper,\n sqlTypeToMarker,\n} from \"./types\";\nimport {\n getWarehouseState,\n startWarehouse,\n waitUntilRunning,\n} from \"./warehouse-status\";\n\nconst logger = createLogger(\"type-generator:query-registry\");\n\n/**\n * Upper bound on how long a `blocking`-mode preflight will wait for a starting\n * warehouse to reach RUNNING before giving up (~5 min). Generous enough to ride\n * out a cold start without hanging an interactive CLI invocation indefinitely.\n */\nconst PREFLIGHT_WAIT_MAX_MS = 300_000;\n\n/**\n * Regex breakdown:\n * '(?:[^']|'')*' — matches a SQL string literal, including escaped '' pairs\n * | — alternation: whichever branch matches first at a position wins\n * --[^\\n]* — matches a single-line SQL comment\n *\n * Because the regex engine scans left-to-right, a `'` is consumed as a string\n * literal before any `--` inside it could match as a comment — giving us\n * correct single-pass ordering without a manual state machine.\n *\n * V1: no block-comment support (deferred to next PR).\n */\nconst PROTECTED_RANGE_RE = /'(?:[^']|'')*'|--[^\\n]*/g;\n\n/**\n * Numeric-context patterns for positional type inference.\n * Hoisted to module scope — safe because matchAll() clones the regex internally.\n */\nconst NUMERIC_PATTERNS: RegExp[] = [\n /\\bLIMIT\\s+:([a-zA-Z_]\\w*)/gi,\n /\\bOFFSET\\s+:([a-zA-Z_]\\w*)/gi,\n /\\bTOP\\s+:([a-zA-Z_]\\w*)/gi,\n /\\bFETCH\\s+FIRST\\s+:([a-zA-Z_]\\w*)\\s+ROWS/gi,\n // V1 limitation: arithmetic operators may false-positive for date\n // expressions like `:start_date - INTERVAL '1 day'`. A smarter\n // heuristic (e.g. look-ahead for INTERVAL) is deferred to a future PR.\n /[+\\-*/]\\s*:([a-zA-Z_]\\w*)/g,\n /:([a-zA-Z_]\\w*)\\s*[+\\-*/]/g,\n];\n\nexport function getProtectedRanges(sql: string): Array<[number, number]> {\n const ranges: Array<[number, number]> = [];\n for (const m of sql.matchAll(PROTECTED_RANGE_RE)) {\n ranges.push([m.index, m.index + m[0].length]);\n }\n return ranges;\n}\n\nfunction isInsideProtectedRange(\n offset: number,\n ranges: Array<[number, number]>,\n): boolean {\n return ranges.some(([start, end]) => offset >= start && offset < end);\n}\n\n/**\n * Parse a raw API/SDK error into a structured code + message.\n * Handles Databricks-style JSON bodies embedded in the message string,\n * e.g. `Response from server (Bad Request) {\"error_code\":\"...\",\"message\":\"...\"}`.\n */\nfunction parseError(raw: string): { code?: string; message: string } {\n const jsonMatch = raw.match(/\\{[\\s\\S]*\\}/);\n if (jsonMatch) {\n try {\n const parsed = JSON.parse(jsonMatch[0]);\n if (parsed.error_code || parsed.message) {\n return {\n code: parsed.error_code,\n message: parsed.message || raw,\n };\n }\n } catch {\n // not valid JSON, fall through\n }\n }\n return { message: raw };\n}\n\nfunction isObject(value: unknown): value is Record<string, unknown> {\n return typeof value === \"object\" && value !== null;\n}\n\nfunction getErrorMessage(error: unknown): string {\n if (error instanceof Error) return error.message;\n if (isObject(error) && typeof error.message === \"string\") {\n return error.message;\n }\n return String(error);\n}\n\nfunction getErrorDiagnostic(error: unknown): string {\n const seen = new Set<unknown>();\n const messages: string[] = [];\n const stack = [error];\n\n while (stack.length > 0) {\n const current = stack.pop();\n if (current === undefined || seen.has(current)) continue;\n seen.add(current);\n\n const message = getErrorMessage(current);\n if (\n message &&\n message !== \"[object Object]\" &&\n !messages.includes(message)\n ) {\n messages.push(message);\n }\n\n const code = getErrorCode(current);\n if (code && !messages.includes(code)) messages.push(code);\n\n stack.push(...getErrorChildren(current));\n }\n\n return messages.length > 0 ? messages.join(\": \") : getErrorMessage(error);\n}\n\nfunction getErrorCode(error: unknown): string | undefined {\n if (!isObject(error)) return undefined;\n const code = error.code ?? error.errno;\n return typeof code === \"string\" ? code : undefined;\n}\n\nfunction getErrorStatus(error: unknown): number | undefined {\n if (!isObject(error)) return undefined;\n const direct = error.status ?? error.statusCode;\n if (typeof direct === \"number\") return direct;\n if (isObject(error.response) && typeof error.response.status === \"number\") {\n return error.response.status;\n }\n return undefined;\n}\n\nfunction getErrorChildren(error: unknown): unknown[] {\n if (!isObject(error)) return [];\n const children: unknown[] = [];\n if (\"cause\" in error) children.push(error.cause);\n if (error instanceof AggregateError) children.push(...error.errors);\n return children;\n}\n\nconst CONNECTIVITY_ERROR_CODES = new Set([\n \"ECONNREFUSED\",\n \"ECONNRESET\",\n \"ENOTFOUND\",\n \"ETIMEDOUT\",\n \"EAI_AGAIN\",\n \"EAI_NODATA\",\n \"EAI_NONAME\",\n \"EHOSTUNREACH\",\n \"ENETUNREACH\",\n \"CERT_HAS_EXPIRED\",\n \"DEPTH_ZERO_SELF_SIGNED_CERT\",\n \"ERR_TLS_CERT_ALTNAME_INVALID\",\n \"SELF_SIGNED_CERT_IN_CHAIN\",\n \"UNABLE_TO_VERIFY_LEAF_SIGNATURE\",\n]);\n\nfunction isConnectivityMessage(message: string): boolean {\n return (\n /\\bconnection (?:refused|reset|timed out)\\b/i.test(message) ||\n /\\bsocket hang up\\b/i.test(message) ||\n /\\bnetwork error\\b/i.test(message) ||\n /\\bcan'?t connect to\\b/i.test(message) ||\n /\\bcertificate has expired\\b/i.test(message) ||\n /\\bunable to verify the first certificate\\b/i.test(message) ||\n /\\bupstream connect error or disconnect\\/reset before headers\\b/i.test(\n message,\n )\n );\n}\n\nfunction isConnectivityError(error: unknown): boolean {\n const seen = new Set<unknown>();\n const stack = [error];\n\n while (stack.length > 0) {\n const current = stack.pop();\n if (current === undefined || seen.has(current)) continue;\n seen.add(current);\n\n const code = getErrorCode(current);\n if (\n code &&\n (CONNECTIVITY_ERROR_CODES.has(code) || code.startsWith(\"UND_ERR_\"))\n ) {\n return true;\n }\n\n const status = getErrorStatus(current);\n if (status === 502 || status === 503 || status === 504) return true;\n\n if (isConnectivityMessage(getErrorMessage(current))) return true;\n\n stack.push(...getErrorChildren(current));\n }\n\n return false;\n}\n\n/**\n * Extract parameters from a SQL query\n * @param sql - the SQL query to extract parameters from\n * @returns an array of parameter names\n */\nexport function extractParameters(\n sql: string,\n ranges?: Array<[number, number]>,\n): string[] {\n const protectedRanges = ranges ?? getProtectedRanges(sql);\n const matches = sql.matchAll(/(?<!:):([a-zA-Z_]\\w*)/g);\n const params = new Set<string>();\n for (const match of matches) {\n if (!isInsideProtectedRange(match.index, protectedRanges)) {\n params.add(match[1]);\n }\n }\n return Array.from(params);\n}\n\n// parameters that are injected by the server\nexport const SERVER_INJECTED_PARAMS = [\"workspaceId\"];\n\n/**\n * Generates the TypeScript type literal for query parameters from SQL.\n * Shared by both the success and failure paths.\n */\nfunction formatParametersType(sql: string): string {\n const params = extractParameters(sql).filter(\n (p) => !SERVER_INJECTED_PARAMS.includes(p),\n );\n const paramTypes = extractParameterTypes(sql);\n\n return params.length > 0\n ? `{\\n ${params\n .map((p) => {\n const sqlType = paramTypes[p];\n const markerType = sqlType\n ? sqlTypeToMarker[sqlType]\n : \"SQLTypeMarker\";\n const helper = sqlType ? sqlTypeToHelper[sqlType] : \"sql.*()\";\n return `/** ${sqlType || \"any\"} - use ${helper} */\\n ${p}: ${markerType}`;\n })\n .join(\";\\n \")};\\n }`\n : \"Record<string, never>\";\n}\n\nexport function convertToQueryType(\n result: DatabricksStatementExecutionResponse,\n sql: string,\n queryName: string,\n): { type: string; hasResults: boolean } {\n const dataRows = result.result?.data_array || [];\n const columns = dataRows.map((row) => ({\n name: row[0] || \"\",\n type_name: row[1]?.toUpperCase() || \"STRING\",\n comment: row[2] || undefined,\n }));\n\n const paramsType = formatParametersType(sql);\n\n // generate result fields with JSDoc\n const resultFields = columns.map((column) => {\n const normalizedType = normalizeTypeName(column.type_name);\n const mappedType = typeMap[normalizedType] || \"unknown\";\n // validate column name is a valid identifier\n const name = /^[a-zA-Z_$][a-zA-Z0-9_$]*$/.test(column.name)\n ? column.name\n : `\"${column.name}\"`;\n\n // generate comment for column\n const comment = column.comment\n ? `/** ${column.comment} */\\n `\n : `/** @sqlType ${column.type_name} */\\n `;\n\n return `${comment}${name}: ${mappedType}`;\n });\n\n const hasResults = resultFields.length > 0;\n\n const type = `{\n name: \"${queryName}\";\n parameters: ${paramsType};\n result: ${\n hasResults\n ? `Array<{\n ${resultFields.join(\";\\n \")};\n }>`\n : \"unknown\"\n };\n }`;\n\n return { type, hasResults };\n}\n\n/**\n * Used when DESCRIBE QUERY fails so the query still appears in QueryRegistry.\n * Generates a type with unknown result from SQL alone (no warehouse call).\n */\nfunction generateUnknownResultQuery(sql: string, queryName: string): string {\n const paramsType = formatParametersType(sql);\n\n return `{\n name: \"${queryName}\";\n parameters: ${paramsType};\n result: unknown;\n }`;\n}\n\n/**\n * Degrade gracefully when DESCRIBE can't produce a fresh schema (transient\n * connectivity outage, or a warehouse that's reachable but not ready). Reuse\n * the last-good cached type when the SQL hash is unchanged, otherwise emit\n * `unknown` from SQL alone. Never persists `result: unknown`.\n */\nfunction degradedType(\n cache: Awaited<ReturnType<typeof loadCache>>,\n queryName: string,\n sql: string,\n sqlHash: string,\n): string {\n const prior = cache.queries[queryName];\n const canReusePrior = prior?.hash === sqlHash && !prior.retry;\n return canReusePrior\n ? prior.type\n : generateUnknownResultQuery(sql, queryName);\n}\n\nexport function extractParameterTypes(sql: string): Record<string, string> {\n const paramTypes: Record<string, string> = {};\n // Alternation order matters: TIMESTAMP_NTZ must precede TIMESTAMP so the\n // regex engine doesn't greedy-match TIMESTAMP and leave `_NTZ` unconsumed.\n const regex =\n /--\\s*@param\\s+(\\w+)\\s+(STRING|NUMERIC|DECIMAL|BIGINT|TINYINT|SMALLINT|INT|FLOAT|DOUBLE|BOOLEAN|DATE|TIMESTAMP_NTZ|TIMESTAMP|BINARY)\\b/gi;\n const matches = sql.matchAll(regex);\n for (const match of matches) {\n const [, paramName, paramType] = match;\n paramTypes[paramName] = paramType.toUpperCase();\n }\n\n return paramTypes;\n}\n\nexport function defaultForType(sqlType: string | undefined): string {\n switch (sqlType?.toUpperCase()) {\n case \"NUMERIC\":\n case \"DECIMAL\":\n case \"BIGINT\":\n case \"TINYINT\":\n case \"SMALLINT\":\n case \"INT\":\n return \"0\";\n case \"FLOAT\":\n case \"DOUBLE\":\n return \"0.0\";\n case \"STRING\":\n return \"''\";\n case \"BOOLEAN\":\n return \"true\";\n case \"DATE\":\n return \"'2000-01-01'\";\n case \"TIMESTAMP\":\n return \"'2000-01-01T00:00:00Z'\";\n case \"TIMESTAMP_NTZ\":\n return \"'2000-01-01T00:00:00'\";\n case \"BINARY\":\n return \"X'00'\";\n default:\n return \"''\";\n }\n}\n\n/**\n * Infer parameter types from positional context in SQL.\n * V1 only infers NUMERIC from patterns like LIMIT, OFFSET, TOP,\n * FETCH FIRST ... ROWS, and arithmetic operators.\n * Parameters inside string literals or SQL comments are ignored.\n */\nexport function inferParameterTypes(\n sql: string,\n ranges?: Array<[number, number]>,\n): Record<string, string> {\n const inferred: Record<string, string> = {};\n const protectedRanges = ranges ?? getProtectedRanges(sql);\n\n for (const pattern of NUMERIC_PATTERNS) {\n for (const match of sql.matchAll(pattern)) {\n if (!isInsideProtectedRange(match.index, protectedRanges)) {\n inferred[match[1]] = \"NUMERIC\";\n }\n }\n }\n\n return inferred;\n}\n\n/**\n * Generate query schemas from a folder of SQL files\n * It uses DESCRIBE QUERY to get the schema without executing the query\n * @param queryFolder - the folder containing the SQL files\n * @param warehouseId - the warehouse id to use for schema analysis\n * @param options - options for the query generation\n * @param options.noCache - if true, skip the cache and regenerate all types\n * @param options.mode - preflight policy: \"non-blocking\" never probes the\n * warehouse and never describes (emits cached/`unknown` types and returns\n * immediately), \"blocking\" waits for a starting warehouse and starts (then\n * waits for) a stopped one, treating only a deleted/deleting warehouse as\n * fatal. Defaults to \"non-blocking\".\n * @returns an array of query schemas\n */\nexport async function generateQueriesFromDescribe(\n queryFolder: string,\n warehouseId: string,\n options: {\n noCache?: boolean;\n concurrency?: number;\n mode?: PreflightMode;\n } = {},\n): Promise<QueryGenerationResult> {\n const {\n noCache = false,\n concurrency: rawConcurrency = 10,\n mode = \"non-blocking\",\n } = options;\n const concurrency =\n typeof rawConcurrency === \"number\" && Number.isFinite(rawConcurrency)\n ? Math.max(1, Math.floor(rawConcurrency))\n : 10;\n\n // read all query files and cache in parallel\n const [allFiles, cache] = await Promise.all([\n fs.readdir(queryFolder),\n noCache\n ? ({ version: CACHE_VERSION, queries: {} } as Awaited<\n ReturnType<typeof loadCache>\n >)\n : loadCache(),\n ]);\n\n const queryFiles = allFiles.filter((file) => file.endsWith(\".sql\"));\n logger.debug(\"Found %d SQL queries\", queryFiles.length);\n\n const client = new WorkspaceClient({});\n const spinner = new Spinner();\n\n // Read all SQL files in parallel\n const sqlContents = await Promise.all(\n queryFiles.map((file) => fs.readFile(path.join(queryFolder, file), \"utf8\")),\n );\n\n const startTime = performance.now();\n\n // Phase 1: Check cache, separate cached vs uncached\n const cachedResults: Array<{ index: number; schema: QuerySchema }> = [];\n const uncachedQueries: Array<{\n index: number;\n queryName: string;\n sql: string;\n sqlHash: string;\n cleanedSql: string;\n }> = [];\n const logEntries: Array<{\n queryName: string;\n status: \"HIT\" | \"MISS\";\n // Absent for clean hits/misses. \"syntax\" = bad SQL on a reachable warehouse;\n // \"connectivity\" = warehouse unreachable; \"empty\" = described but no columns;\n // \"fatal\" = non-SQL setup/request failure surfaced after .d.ts emission.\n kind?: \"syntax\" | \"connectivity\" | \"empty\" | \"fatal\";\n error?: { code?: string; message: string };\n }> = [];\n\n for (let i = 0; i < queryFiles.length; i++) {\n const file = queryFiles[i];\n const rawName = path.basename(file, \".sql\");\n const queryName = normalizeQueryName(rawName);\n\n const sql = sqlContents[i];\n const sqlHash = hashSQL(sql);\n\n const cached = cache.queries[queryName];\n if (cached && cached.hash === sqlHash && !cached.retry) {\n cachedResults.push({\n index: i,\n schema: { name: queryName, type: cached.type },\n });\n logEntries.push({ queryName, status: \"HIT\" });\n } else {\n const protectedRanges = getProtectedRanges(sql);\n const annotatedTypes = extractParameterTypes(sql);\n const inferredTypes = inferParameterTypes(sql, protectedRanges);\n const parameterTypes = { ...inferredTypes, ...annotatedTypes };\n const sqlWithDefaults = sql.replace(\n /(?<!:):([a-zA-Z_]\\w*)/g,\n (original, paramName, offset) => {\n if (isInsideProtectedRange(offset, protectedRanges)) {\n return original;\n }\n return defaultForType(parameterTypes[paramName]);\n },\n );\n\n // Warn about unresolved parameters\n const allParams = extractParameters(sql, protectedRanges);\n for (const param of allParams) {\n if (SERVER_INJECTED_PARAMS.includes(param)) continue;\n if (parameterTypes[param]) continue;\n logger.warn(\n '%s: parameter \":%s\" has no type annotation or inference. Add %s to the query file.',\n queryFiles[i],\n param,\n `-- @param ${param} <TYPE>`,\n );\n }\n\n const cleanedSql = sqlWithDefaults.trim().replace(/;\\s*$/, \"\");\n uncachedQueries.push({ index: i, queryName, sql, sqlHash, cleanedSql });\n }\n }\n\n // Phase 2: Execute all uncached DESCRIBE calls in parallel\n type DescribeResult =\n | {\n // Described successfully with a result schema — the only case we cache.\n status: \"ok\";\n index: number;\n schema: QuerySchema;\n cacheEntry: { hash: string; type: string; retry: boolean };\n }\n | {\n // Reachable warehouse ran DESCRIBE and rejected the statement — a\n // genuine SQL error. Eligible to fail the build; never cached.\n status: \"syntax\";\n index: number;\n schema: QuerySchema;\n error: { code?: string; message: string };\n }\n | {\n // DESCRIBE succeeded but returned no columns — soft `unknown`. Not a\n // failure, not cached, retried next run.\n status: \"empty\";\n index: number;\n schema: QuerySchema;\n }\n | {\n // Warehouse reachable but returned a non-terminal state (PENDING/\n // RUNNING) with no rows — stopped/cold-starting/busy. Degrade like a\n // transient outage (reuse cache or `unknown`); not cached, not empty.\n status: \"unavailable\";\n index: number;\n schema: QuerySchema;\n };\n\n const freshResults: Array<{ index: number; schema: QuerySchema }> = [];\n // Genuine SQL errors (reachable warehouse). Connectivity failures are NOT\n // recorded here — they degrade silently so a transient outage isn't fatal.\n const syntaxErrors: QuerySyntaxError[] = [];\n const fatalErrors: QueryFatalError[] = [];\n\n if (uncachedQueries.length > 0) {\n // One-time warehouse preflight (before issuing any DESCRIBE). A single\n // warehouses.get classifies the warehouse so we can skip the whole describe\n // batch when it can't serve this run, instead of letting every query fail\n // (and re-fail next run). Reuses this file's degrade/classify helpers so a\n // not-ready warehouse degrades exactly like a per-query outage.\n let decision: ReturnType<typeof decidePreflight> = \"proceed\";\n let fatalMessage = \"\";\n if (mode === \"non-blocking\") {\n // `non-blocking` never describes and must make ZERO warehouse round-trips:\n // skip the probe entirely (no getWarehouseState) and go straight to\n // degradeAll. A foreground/one-shot run can't describe in the background,\n // so it emits best-available types (reused cache or `unknown`) and returns\n // now.\n decision = \"degradeAll\";\n } else {\n try {\n const state = await getWarehouseState(client, warehouseId);\n decision = decidePreflight(state, mode);\n if (decision === \"fatal\") {\n fatalMessage = `warehouse ${warehouseId} is ${state}`;\n }\n if (decision === \"startWaitProceed\") {\n // Stopped/stopping warehouse: nudge it out of the stopped state, then\n // poll to RUNNING. treatStoppedAsTransient rides out the stale\n // pre-start STOPPED/STOPPING reading the start hasn't propagated past\n // yet — only DELETED/DELETING (or the deadline) ends the wait early.\n await startWarehouse(client, warehouseId);\n const final = await waitUntilRunning(client, warehouseId, {\n maxMs: PREFLIGHT_WAIT_MAX_MS,\n treatStoppedAsTransient: true,\n });\n if (final === \"RUNNING\") {\n decision = \"proceed\";\n } else {\n decision = \"fatal\";\n fatalMessage = `warehouse ${warehouseId} did not reach RUNNING (now ${final})`;\n }\n }\n if (decision === \"waitThenProceed\") {\n const final = await waitUntilRunning(client, warehouseId, {\n maxMs: PREFLIGHT_WAIT_MAX_MS,\n });\n if (final === \"RUNNING\") {\n decision = \"proceed\";\n } else {\n decision = \"fatal\";\n fatalMessage = `warehouse ${warehouseId} did not reach RUNNING (now ${final})`;\n }\n }\n } catch (err) {\n if (isConnectivityError(err)) {\n // Warehouse unreachable (transient outage): degrade silently like a\n // per-query connectivity failure — never fail a build on a blip.\n decision = \"degradeAll\";\n } else {\n // Auth, bad warehouse id, malformed config, or a timed-out wait: fatal.\n decision = \"fatal\";\n fatalMessage = `warehouse ${warehouseId}: ${getErrorDiagnostic(err)}`;\n }\n }\n }\n\n if (decision !== \"proceed\") {\n // degradeAll or fatal: skip DESCRIBE entirely. Every uncached query gets a\n // degraded schema (reused cache or `unknown`); fatal additionally records\n // a fatalError per query so the caller fails the build after writing.\n const kind = decision === \"fatal\" ? \"fatal\" : \"connectivity\";\n for (const { index, queryName, sql, sqlHash } of uncachedQueries) {\n freshResults.push({\n index,\n schema: {\n name: queryName,\n type: degradedType(cache, queryName, sql, sqlHash),\n },\n });\n if (decision === \"fatal\") {\n fatalErrors.push({ name: queryName, message: fatalMessage });\n logEntries.push({\n queryName,\n status: \"MISS\",\n kind,\n error: { message: fatalMessage },\n });\n } else {\n logEntries.push({ queryName, status: \"MISS\", kind });\n }\n }\n } else {\n let completed = 0;\n const total = uncachedQueries.length;\n spinner.start(\n `Describing ${total} ${total === 1 ? \"query\" : \"queries\"} (0/${total})`,\n );\n\n const describeOne = async ({\n index,\n queryName,\n sql,\n sqlHash,\n cleanedSql,\n }: (typeof uncachedQueries)[number]): Promise<DescribeResult> => {\n const result = (await client.statementExecution.executeStatement({\n statement: `DESCRIBE QUERY ${cleanedSql}`,\n warehouse_id: warehouseId,\n })) as DatabricksStatementExecutionResponse;\n\n completed++;\n spinner.update(\n `Describing ${total} ${total === 1 ? \"query\" : \"queries\"} (${completed}/${total})`,\n );\n\n logger.debug(\n \"DESCRIBE result for %s: state=%s, rows=%d\",\n queryName,\n result.status.state,\n result.result?.data_array?.length ?? 0,\n );\n\n if (result.status.state === \"FAILED\") {\n // The warehouse was reachable and ran DESCRIBE, but the statement\n // failed — a genuine SQL error (bad table, syntax, incompatible type).\n const sqlError =\n result.status.error?.message || \"Query execution failed\";\n // The failure is surfaced once, formatted, by the aggregated\n // TypegenSyntaxError (and the summary table) — don't also log the raw\n // message here or every SQL error prints twice in dev.\n const type = generateUnknownResultQuery(sql, queryName);\n return {\n status: \"syntax\",\n index,\n schema: { name: queryName, type },\n error: parseError(sqlError),\n };\n }\n\n if (result.status.state !== \"SUCCEEDED\") {\n // Non-terminal state (PENDING/RUNNING) with no result rows: the\n // warehouse is reachable but not ready (stopped, cold-starting, or\n // busy). Degrade like a transient outage — reuse the last-good cached\n // type when the SQL is unchanged, else emit `unknown`. Never \"empty\":\n // treating this as empty would emit `result: unknown` AND discard the\n // good cached type, silently throwing away working types.\n return {\n status: \"unavailable\",\n index,\n schema: {\n name: queryName,\n type: degradedType(cache, queryName, sql, sqlHash),\n },\n };\n }\n\n const { type, hasResults } = convertToQueryType(result, sql, queryName);\n if (!hasResults) {\n // Described, but no result columns. Emit `unknown` and retry next run;\n // do not cache (we never persist `result: unknown`).\n return { status: \"empty\", index, schema: { name: queryName, type } };\n }\n return {\n status: \"ok\",\n index,\n schema: { name: queryName, type },\n cacheEntry: { hash: sqlHash, type, retry: false },\n };\n };\n\n // Process in chunks, saving cache after each chunk\n const processBatchResults = (\n settled: PromiseSettledResult<DescribeResult>[],\n batchOffset: number,\n ) => {\n for (let i = 0; i < settled.length; i++) {\n const entry = settled[i];\n const { queryName } = uncachedQueries[batchOffset + i];\n\n if (entry.status === \"fulfilled\") {\n const res = entry.value;\n freshResults.push({ index: res.index, schema: res.schema });\n\n if (res.status === \"ok\") {\n // Only a successful describe with a result schema is cached.\n cache.queries[queryName] = res.cacheEntry;\n logEntries.push({ queryName, status: \"MISS\" });\n } else if (res.status === \"syntax\") {\n // Genuine SQL error — record it for the caller's prod/dev gate.\n // Not cached: re-described next run so a fixed query recovers.\n syntaxErrors.push({\n name: queryName,\n message: res.error.message,\n });\n logEntries.push({\n queryName,\n status: \"MISS\",\n kind: \"syntax\",\n error: res.error,\n });\n } else if (res.status === \"empty\") {\n // status === \"empty\": described, no columns. Soft unknown, not cached.\n logEntries.push({ queryName, status: \"MISS\", kind: \"empty\" });\n } else {\n // status === \"unavailable\": non-terminal DESCRIBE (warehouse\n // stopped/cold-starting/busy). Degrade like a transient outage:\n // tag OFFLINE, count as degraded, never cache.\n logEntries.push({\n queryName,\n status: \"MISS\",\n kind: \"connectivity\",\n });\n }\n } else {\n // executeStatement rejected without a normal StatementExecution result.\n // Only structured transport/connectivity failures are treated as\n // offline; auth, bad warehouse IDs, malformed requests, and SDK/config\n // failures stay fatal so users fix the underlying setup issue.\n completed++;\n spinner.update(\n `Describing ${total} ${total === 1 ? \"query\" : \"queries\"} (${completed}/${total})`,\n );\n\n const { sql, sqlHash, index } = uncachedQueries[batchOffset + i];\n const reason = getErrorDiagnostic(entry.reason);\n const error = parseError(reason);\n const priorEntry = cache.queries[queryName];\n const canReusePrior =\n priorEntry?.hash === sqlHash && !priorEntry.retry;\n const type = degradedType(cache, queryName, sql, sqlHash);\n freshResults.push({ index, schema: { name: queryName, type } });\n\n if (!isConnectivityError(entry.reason)) {\n fatalErrors.push({ name: queryName, message: error.message });\n logEntries.push({\n queryName,\n status: \"MISS\",\n kind: \"fatal\",\n error,\n });\n continue;\n }\n\n logger.warn(\n \"DESCRIBE unreachable for %s: %s — %s\",\n queryName,\n reason,\n canReusePrior\n ? \"reusing last cached type\"\n : \"emitting unknown (no matching cache)\",\n );\n logEntries.push({\n queryName,\n status: \"MISS\",\n kind: \"connectivity\",\n error,\n });\n }\n }\n };\n\n if (uncachedQueries.length > concurrency) {\n for (let b = 0; b < uncachedQueries.length; b += concurrency) {\n const batch = uncachedQueries.slice(b, b + concurrency);\n const batchResults = await Promise.allSettled(batch.map(describeOne));\n processBatchResults(batchResults, b);\n await saveCache(cache);\n }\n } else {\n const settled = await Promise.allSettled(\n uncachedQueries.map(describeOne),\n );\n processBatchResults(settled, 0);\n await saveCache(cache);\n }\n\n spinner.stop(\"\");\n }\n }\n\n const elapsed = ((performance.now() - startTime) / 1000).toFixed(2);\n\n // Print formatted table\n if (logEntries.length > 0) {\n const maxNameLen = Math.max(...logEntries.map((e) => e.queryName.length));\n const separator = pc.dim(\"─\".repeat(50));\n console.log(\"\");\n console.log(\n ` ${pc.bold(\"Typegen Queries\")} ${pc.dim(`(${logEntries.length})`)}`,\n );\n console.log(` ${separator}`);\n for (const entry of logEntries) {\n let tag: string;\n switch (entry.kind) {\n case \"syntax\":\n tag = pc.bold(pc.red(\"SQL ERR\"));\n break;\n case \"connectivity\":\n tag = pc.bold(pc.yellow(\"OFFLINE\"));\n break;\n case \"empty\":\n tag = pc.dim(\"EMPTY \");\n break;\n case \"fatal\":\n tag = pc.bold(pc.red(\"FATAL \"));\n break;\n default:\n tag =\n entry.status === \"HIT\"\n ? `cache ${pc.bold(pc.green(\"HIT \"))}`\n : `cache ${pc.bold(pc.yellow(\"MISS \"))}`;\n }\n const rawName = entry.queryName.padEnd(maxNameLen);\n // Only genuine SQL errors are struck through. Connectivity/empty kept a\n // usable type (reused or unknown), so they read as degraded, not broken.\n const name =\n entry.kind === \"syntax\" || entry.kind === \"fatal\"\n ? pc.dim(pc.strikethrough(rawName))\n : rawName;\n const errorCode = entry.error?.message.match(/\\[([^\\]]+)\\]/)?.[1];\n const reason = errorCode ? ` ${pc.dim(errorCode)}` : \"\";\n console.log(` ${tag} ${name}${reason}`);\n }\n const newCount = logEntries.filter(\n (e) => e.status === \"MISS\" && !e.kind,\n ).length;\n const cacheCount = logEntries.filter((e) => e.status === \"HIT\").length;\n const syntaxCount = logEntries.filter((e) => e.kind === \"syntax\").length;\n const offlineCount = logEntries.filter(\n (e) => e.kind === \"connectivity\",\n ).length;\n const emptyCount = logEntries.filter((e) => e.kind === \"empty\").length;\n const fatalCount = logEntries.filter((e) => e.kind === \"fatal\").length;\n console.log(` ${separator}`);\n const parts = [`${newCount} new`, `${cacheCount} from cache`];\n if (syntaxCount > 0)\n parts.push(\n `${syntaxCount} SQL ${syntaxCount === 1 ? \"error\" : \"errors\"}`,\n );\n if (offlineCount > 0) parts.push(`${offlineCount} degraded`);\n if (emptyCount > 0) parts.push(`${emptyCount} empty`);\n if (fatalCount > 0)\n parts.push(\n `${fatalCount} fatal ${fatalCount === 1 ? \"error\" : \"errors\"}`,\n );\n console.log(` ${parts.join(\", \")}. ${pc.dim(`${elapsed}s`)}`);\n console.log(\"\");\n }\n\n // Merge and sort by original file index for deterministic output\n const schemas = [...cachedResults, ...freshResults]\n .sort((a, b) => a.index - b.index)\n .map((r) => r.schema);\n\n return { schemas, syntaxErrors, fatalErrors };\n}\n\n/**\n * Normalize query name by removing the .obo extension\n * @param queryName - the query name to normalize\n * @returns the normalized query name\n */\nfunction normalizeQueryName(fileName: string): string {\n return fileName.replace(/\\.obo$/, \"\");\n}\n\n/**\n * Normalize SQL type name by removing parameters/generics\n * Examples:\n * DECIMAL(38,6) -> DECIMAL\n * ARRAY<STRING> -> ARRAY\n * MAP<STRING,INT> -> MAP\n * STRUCT<name:STRING> -> STRUCT\n * INTERVAL DAY TO SECOND -> INTERVAL\n * GEOGRAPHY(4326) -> GEOGRAPHY\n */\nexport function normalizeTypeName(typeName: string): string {\n return typeName\n .replace(/\\(.*\\)$/, \"\") // remove (p, s) eg: DECIMAL(38,6) -> DECIMAL\n .replace(/<.*>$/, \"\") // remove <T> eg: ARRAY<STRING> -> ARRAY\n .split(\" \")[0]; // take first word eg: INTERVAL DAY TO SECOND -> INTERVAL\n}\n\n/** Type Map for Databricks data types to JavaScript types */\nconst typeMap: Record<string, string> = {\n // string types\n STRING: \"string\",\n BINARY: \"string\",\n // boolean\n BOOLEAN: \"boolean\",\n // numeric types\n TINYINT: \"number\",\n SMALLINT: \"number\",\n INT: \"number\",\n BIGINT: \"number\",\n FLOAT: \"number\",\n DOUBLE: \"number\",\n DECIMAL: \"number\",\n // date/time types\n DATE: \"string\",\n TIMESTAMP: \"string\",\n TIMESTAMP_NTZ: \"string\",\n INTERVAL: \"string\",\n // complex types\n ARRAY: \"unknown[]\",\n MAP: \"Record<string, unknown>\",\n STRUCT: \"Record<string, unknown>\",\n OBJECT: \"Record<string, unknown>\",\n VARIANT: \"unknown\",\n // spatial types\n GEOGRAPHY: \"unknown\",\n GEOMETRY: \"unknown\",\n // null type\n VOID: \"null\",\n};\n"],"mappings":";;;;;;;;;;;;AAuBA,MAAM,SAAS,aAAa,gCAAgC;;;;;;AAO5D,MAAM,wBAAwB;;;;;;;;;;;;;AAc9B,MAAM,qBAAqB;;;;;AAM3B,MAAM,mBAA6B;CACjC;CACA;CACA;CACA;CAIA;CACA;CACD;AAED,SAAgB,mBAAmB,KAAsC;CACvE,MAAM,SAAkC,EAAE;AAC1C,MAAK,MAAM,KAAK,IAAI,SAAS,mBAAmB,CAC9C,QAAO,KAAK,CAAC,EAAE,OAAO,EAAE,QAAQ,EAAE,GAAG,OAAO,CAAC;AAE/C,QAAO;;AAGT,SAAS,uBACP,QACA,QACS;AACT,QAAO,OAAO,MAAM,CAAC,OAAO,SAAS,UAAU,SAAS,SAAS,IAAI;;;;;;;AAQvE,SAAS,WAAW,KAAiD;CACnE,MAAM,YAAY,IAAI,MAAM,cAAc;AAC1C,KAAI,UACF,KAAI;EACF,MAAM,SAAS,KAAK,MAAM,UAAU,GAAG;AACvC,MAAI,OAAO,cAAc,OAAO,QAC9B,QAAO;GACL,MAAM,OAAO;GACb,SAAS,OAAO,WAAW;GAC5B;SAEG;AAIV,QAAO,EAAE,SAAS,KAAK;;AAGzB,SAAS,SAAS,OAAkD;AAClE,QAAO,OAAO,UAAU,YAAY,UAAU;;AAGhD,SAAS,gBAAgB,OAAwB;AAC/C,KAAI,iBAAiB,MAAO,QAAO,MAAM;AACzC,KAAI,SAAS,MAAM,IAAI,OAAO,MAAM,YAAY,SAC9C,QAAO,MAAM;AAEf,QAAO,OAAO,MAAM;;AAGtB,SAAS,mBAAmB,OAAwB;CAClD,MAAM,uBAAO,IAAI,KAAc;CAC/B,MAAM,WAAqB,EAAE;CAC7B,MAAM,QAAQ,CAAC,MAAM;AAErB,QAAO,MAAM,SAAS,GAAG;EACvB,MAAM,UAAU,MAAM,KAAK;AAC3B,MAAI,YAAY,UAAa,KAAK,IAAI,QAAQ,CAAE;AAChD,OAAK,IAAI,QAAQ;EAEjB,MAAM,UAAU,gBAAgB,QAAQ;AACxC,MACE,WACA,YAAY,qBACZ,CAAC,SAAS,SAAS,QAAQ,CAE3B,UAAS,KAAK,QAAQ;EAGxB,MAAM,OAAO,aAAa,QAAQ;AAClC,MAAI,QAAQ,CAAC,SAAS,SAAS,KAAK,CAAE,UAAS,KAAK,KAAK;AAEzD,QAAM,KAAK,GAAG,iBAAiB,QAAQ,CAAC;;AAG1C,QAAO,SAAS,SAAS,IAAI,SAAS,KAAK,KAAK,GAAG,gBAAgB,MAAM;;AAG3E,SAAS,aAAa,OAAoC;AACxD,KAAI,CAAC,SAAS,MAAM,CAAE,QAAO;CAC7B,MAAM,OAAO,MAAM,QAAQ,MAAM;AACjC,QAAO,OAAO,SAAS,WAAW,OAAO;;AAG3C,SAAS,eAAe,OAAoC;AAC1D,KAAI,CAAC,SAAS,MAAM,CAAE,QAAO;CAC7B,MAAM,SAAS,MAAM,UAAU,MAAM;AACrC,KAAI,OAAO,WAAW,SAAU,QAAO;AACvC,KAAI,SAAS,MAAM,SAAS,IAAI,OAAO,MAAM,SAAS,WAAW,SAC/D,QAAO,MAAM,SAAS;;AAK1B,SAAS,iBAAiB,OAA2B;AACnD,KAAI,CAAC,SAAS,MAAM,CAAE,QAAO,EAAE;CAC/B,MAAM,WAAsB,EAAE;AAC9B,KAAI,WAAW,MAAO,UAAS,KAAK,MAAM,MAAM;AAChD,KAAI,iBAAiB,eAAgB,UAAS,KAAK,GAAG,MAAM,OAAO;AACnE,QAAO;;AAGT,MAAM,2BAA2B,IAAI,IAAI;CACvC;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACD,CAAC;AAEF,SAAS,sBAAsB,SAA0B;AACvD,QACE,8CAA8C,KAAK,QAAQ,IAC3D,sBAAsB,KAAK,QAAQ,IACnC,qBAAqB,KAAK,QAAQ,IAClC,yBAAyB,KAAK,QAAQ,IACtC,+BAA+B,KAAK,QAAQ,IAC5C,8CAA8C,KAAK,QAAQ,IAC3D,kEAAkE,KAChE,QACD;;AAIL,SAAS,oBAAoB,OAAyB;CACpD,MAAM,uBAAO,IAAI,KAAc;CAC/B,MAAM,QAAQ,CAAC,MAAM;AAErB,QAAO,MAAM,SAAS,GAAG;EACvB,MAAM,UAAU,MAAM,KAAK;AAC3B,MAAI,YAAY,UAAa,KAAK,IAAI,QAAQ,CAAE;AAChD,OAAK,IAAI,QAAQ;EAEjB,MAAM,OAAO,aAAa,QAAQ;AAClC,MACE,SACC,yBAAyB,IAAI,KAAK,IAAI,KAAK,WAAW,WAAW,EAElE,QAAO;EAGT,MAAM,SAAS,eAAe,QAAQ;AACtC,MAAI,WAAW,OAAO,WAAW,OAAO,WAAW,IAAK,QAAO;AAE/D,MAAI,sBAAsB,gBAAgB,QAAQ,CAAC,CAAE,QAAO;AAE5D,QAAM,KAAK,GAAG,iBAAiB,QAAQ,CAAC;;AAG1C,QAAO;;;;;;;AAQT,SAAgB,kBACd,KACA,QACU;CACV,MAAM,kBAAkB,UAAU,mBAAmB,IAAI;CACzD,MAAM,UAAU,IAAI,SAAS,yBAAyB;CACtD,MAAM,yBAAS,IAAI,KAAa;AAChC,MAAK,MAAM,SAAS,QAClB,KAAI,CAAC,uBAAuB,MAAM,OAAO,gBAAgB,CACvD,QAAO,IAAI,MAAM,GAAG;AAGxB,QAAO,MAAM,KAAK,OAAO;;AAI3B,MAAa,yBAAyB,CAAC,cAAc;;;;;AAMrD,SAAS,qBAAqB,KAAqB;CACjD,MAAM,SAAS,kBAAkB,IAAI,CAAC,QACnC,MAAM,CAAC,uBAAuB,SAAS,EAAE,CAC3C;CACD,MAAM,aAAa,sBAAsB,IAAI;AAE7C,QAAO,OAAO,SAAS,IACnB,YAAY,OACT,KAAK,MAAM;EACV,MAAM,UAAU,WAAW;EAC3B,MAAM,aAAa,UACf,gBAAgB,WAChB;EACJ,MAAM,SAAS,UAAU,gBAAgB,WAAW;AACpD,SAAO,OAAO,WAAW,MAAM,SAAS,OAAO,aAAa,EAAE,IAAI;GAClE,CACD,KAAK,YAAY,CAAC,YACrB;;AAGN,SAAgB,mBACd,QACA,KACA,WACuC;CAEvC,MAAM,WADW,OAAO,QAAQ,cAAc,EAAE,EACvB,KAAK,SAAS;EACrC,MAAM,IAAI,MAAM;EAChB,WAAW,IAAI,IAAI,aAAa,IAAI;EACpC,SAAS,IAAI,MAAM;EACpB,EAAE;CAEH,MAAM,aAAa,qBAAqB,IAAI;CAG5C,MAAM,eAAe,QAAQ,KAAK,WAAW;EAE3C,MAAM,aAAa,QADI,kBAAkB,OAAO,UAAU,KACZ;EAE9C,MAAM,OAAO,6BAA6B,KAAK,OAAO,KAAK,GACvD,OAAO,OACP,IAAI,OAAO,KAAK;AAOpB,SAAO,GAJS,OAAO,UACnB,OAAO,OAAO,QAAQ,eACtB,gBAAgB,OAAO,UAAU,eAEjB,KAAK,IAAI;GAC7B;CAEF,MAAM,aAAa,aAAa,SAAS;AAczC,QAAO;EAAE,MAZI;aACF,UAAU;kBACL,WAAW;cAEvB,aACI;QACF,aAAa,KAAK,YAAY,CAAC;UAE7B,UACL;;EAGY;EAAY;;;;;;AAO7B,SAAS,2BAA2B,KAAa,WAA2B;AAG1E,QAAO;aACI,UAAU;kBAHF,qBAAqB,IAAI,CAIjB;;;;;;;;;;AAW7B,SAAS,aACP,OACA,WACA,KACA,SACQ;CACR,MAAM,QAAQ,MAAM,QAAQ;AAE5B,QADsB,OAAO,SAAS,WAAW,CAAC,MAAM,QAEpD,MAAM,OACN,2BAA2B,KAAK,UAAU;;AAGhD,SAAgB,sBAAsB,KAAqC;CACzE,MAAM,aAAqC,EAAE;CAK7C,MAAM,UAAU,IAAI,SADlB,0IACiC;AACnC,MAAK,MAAM,SAAS,SAAS;EAC3B,MAAM,GAAG,WAAW,aAAa;AACjC,aAAW,aAAa,UAAU,aAAa;;AAGjD,QAAO;;AAGT,SAAgB,eAAe,SAAqC;AAClE,SAAQ,SAAS,aAAa,EAA9B;EACE,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK,MACH,QAAO;EACT,KAAK;EACL,KAAK,SACH,QAAO;EACT,KAAK,SACH,QAAO;EACT,KAAK,UACH,QAAO;EACT,KAAK,OACH,QAAO;EACT,KAAK,YACH,QAAO;EACT,KAAK,gBACH,QAAO;EACT,KAAK,SACH,QAAO;EACT,QACE,QAAO;;;;;;;;;AAUb,SAAgB,oBACd,KACA,QACwB;CACxB,MAAM,WAAmC,EAAE;CAC3C,MAAM,kBAAkB,UAAU,mBAAmB,IAAI;AAEzD,MAAK,MAAM,WAAW,iBACpB,MAAK,MAAM,SAAS,IAAI,SAAS,QAAQ,CACvC,KAAI,CAAC,uBAAuB,MAAM,OAAO,gBAAgB,CACvD,UAAS,MAAM,MAAM;AAK3B,QAAO;;;;;;;;;;;;;;;;AAiBT,eAAsB,4BACpB,aACA,aACA,UAII,EAAE,EAC0B;CAChC,MAAM,EACJ,UAAU,OACV,aAAa,iBAAiB,IAC9B,OAAO,mBACL;CACJ,MAAM,cACJ,OAAO,mBAAmB,YAAY,OAAO,SAAS,eAAe,GACjE,KAAK,IAAI,GAAG,KAAK,MAAM,eAAe,CAAC,GACvC;CAGN,MAAM,CAAC,UAAU,SAAS,MAAM,QAAQ,IAAI,CAC1C,GAAG,QAAQ,YAAY,EACvB,UACK;EAAE,SAAS;EAAe,SAAS,EAAE;EAAE,GAGxC,WAAW,CAChB,CAAC;CAEF,MAAM,aAAa,SAAS,QAAQ,SAAS,KAAK,SAAS,OAAO,CAAC;AACnE,QAAO,MAAM,wBAAwB,WAAW,OAAO;CAEvD,MAAM,SAAS,IAAI,gBAAgB,EAAE,CAAC;CACtC,MAAM,UAAU,IAAI,SAAS;CAG7B,MAAM,cAAc,MAAM,QAAQ,IAChC,WAAW,KAAK,SAAS,GAAG,SAAS,KAAK,KAAK,aAAa,KAAK,EAAE,OAAO,CAAC,CAC5E;CAED,MAAM,YAAY,YAAY,KAAK;CAGnC,MAAM,gBAA+D,EAAE;CACvE,MAAM,kBAMD,EAAE;CACP,MAAM,aAQD,EAAE;AAEP,MAAK,IAAI,IAAI,GAAG,IAAI,WAAW,QAAQ,KAAK;EAC1C,MAAM,OAAO,WAAW;EAExB,MAAM,YAAY,mBADF,KAAK,SAAS,MAAM,OAAO,CACE;EAE7C,MAAM,MAAM,YAAY;EACxB,MAAM,UAAU,QAAQ,IAAI;EAE5B,MAAM,SAAS,MAAM,QAAQ;AAC7B,MAAI,UAAU,OAAO,SAAS,WAAW,CAAC,OAAO,OAAO;AACtD,iBAAc,KAAK;IACjB,OAAO;IACP,QAAQ;KAAE,MAAM;KAAW,MAAM,OAAO;KAAM;IAC/C,CAAC;AACF,cAAW,KAAK;IAAE;IAAW,QAAQ;IAAO,CAAC;SACxC;GACL,MAAM,kBAAkB,mBAAmB,IAAI;GAC/C,MAAM,iBAAiB,sBAAsB,IAAI;GAEjD,MAAM,iBAAiB;IAAE,GADH,oBAAoB,KAAK,gBAAgB;IACpB,GAAG;IAAgB;GAC9D,MAAM,kBAAkB,IAAI,QAC1B,2BACC,UAAU,WAAW,WAAW;AAC/B,QAAI,uBAAuB,QAAQ,gBAAgB,CACjD,QAAO;AAET,WAAO,eAAe,eAAe,WAAW;KAEnD;GAGD,MAAM,YAAY,kBAAkB,KAAK,gBAAgB;AACzD,QAAK,MAAM,SAAS,WAAW;AAC7B,QAAI,uBAAuB,SAAS,MAAM,CAAE;AAC5C,QAAI,eAAe,OAAQ;AAC3B,WAAO,KACL,wFACA,WAAW,IACX,OACA,aAAa,MAAM,SACpB;;GAGH,MAAM,aAAa,gBAAgB,MAAM,CAAC,QAAQ,SAAS,GAAG;AAC9D,mBAAgB,KAAK;IAAE,OAAO;IAAG;IAAW;IAAK;IAAS;IAAY,CAAC;;;CAqC3E,MAAM,eAA8D,EAAE;CAGtE,MAAM,eAAmC,EAAE;CAC3C,MAAM,cAAiC,EAAE;AAEzC,KAAI,gBAAgB,SAAS,GAAG;EAM9B,IAAI,WAA+C;EACnD,IAAI,eAAe;AACnB,MAAI,SAAS,eAMX,YAAW;MAEX,KAAI;GACF,MAAM,QAAQ,MAAM,kBAAkB,QAAQ,YAAY;AAC1D,cAAW,gBAAgB,OAAO,KAAK;AACvC,OAAI,aAAa,QACf,gBAAe,aAAa,YAAY,MAAM;AAEhD,OAAI,aAAa,oBAAoB;AAKnC,UAAM,eAAe,QAAQ,YAAY;IACzC,MAAM,QAAQ,MAAM,iBAAiB,QAAQ,aAAa;KACxD,OAAO;KACP,yBAAyB;KAC1B,CAAC;AACF,QAAI,UAAU,UACZ,YAAW;SACN;AACL,gBAAW;AACX,oBAAe,aAAa,YAAY,8BAA8B,MAAM;;;AAGhF,OAAI,aAAa,mBAAmB;IAClC,MAAM,QAAQ,MAAM,iBAAiB,QAAQ,aAAa,EACxD,OAAO,uBACR,CAAC;AACF,QAAI,UAAU,UACZ,YAAW;SACN;AACL,gBAAW;AACX,oBAAe,aAAa,YAAY,8BAA8B,MAAM;;;WAGzE,KAAK;AACZ,OAAI,oBAAoB,IAAI,CAG1B,YAAW;QACN;AAEL,eAAW;AACX,mBAAe,aAAa,YAAY,IAAI,mBAAmB,IAAI;;;AAKzE,MAAI,aAAa,WAAW;GAI1B,MAAM,OAAO,aAAa,UAAU,UAAU;AAC9C,QAAK,MAAM,EAAE,OAAO,WAAW,KAAK,aAAa,iBAAiB;AAChE,iBAAa,KAAK;KAChB;KACA,QAAQ;MACN,MAAM;MACN,MAAM,aAAa,OAAO,WAAW,KAAK,QAAQ;MACnD;KACF,CAAC;AACF,QAAI,aAAa,SAAS;AACxB,iBAAY,KAAK;MAAE,MAAM;MAAW,SAAS;MAAc,CAAC;AAC5D,gBAAW,KAAK;MACd;MACA,QAAQ;MACR;MACA,OAAO,EAAE,SAAS,cAAc;MACjC,CAAC;UAEF,YAAW,KAAK;KAAE;KAAW,QAAQ;KAAQ;KAAM,CAAC;;SAGnD;GACL,IAAI,YAAY;GAChB,MAAM,QAAQ,gBAAgB;AAC9B,WAAQ,MACN,cAAc,MAAM,GAAG,UAAU,IAAI,UAAU,UAAU,MAAM,MAAM,GACtE;GAED,MAAM,cAAc,OAAO,EACzB,OACA,WACA,KACA,SACA,iBAC+D;IAC/D,MAAM,SAAU,MAAM,OAAO,mBAAmB,iBAAiB;KAC/D,WAAW,kBAAkB;KAC7B,cAAc;KACf,CAAC;AAEF;AACA,YAAQ,OACN,cAAc,MAAM,GAAG,UAAU,IAAI,UAAU,UAAU,IAAI,UAAU,GAAG,MAAM,GACjF;AAED,WAAO,MACL,6CACA,WACA,OAAO,OAAO,OACd,OAAO,QAAQ,YAAY,UAAU,EACtC;AAED,QAAI,OAAO,OAAO,UAAU,UAAU;KAGpC,MAAM,WACJ,OAAO,OAAO,OAAO,WAAW;AAKlC,YAAO;MACL,QAAQ;MACR;MACA,QAAQ;OAAE,MAAM;OAAW,MAJhB,2BAA2B,KAAK,UAAU;OAIpB;MACjC,OAAO,WAAW,SAAS;MAC5B;;AAGH,QAAI,OAAO,OAAO,UAAU,YAO1B,QAAO;KACL,QAAQ;KACR;KACA,QAAQ;MACN,MAAM;MACN,MAAM,aAAa,OAAO,WAAW,KAAK,QAAQ;MACnD;KACF;IAGH,MAAM,EAAE,MAAM,eAAe,mBAAmB,QAAQ,KAAK,UAAU;AACvE,QAAI,CAAC,WAGH,QAAO;KAAE,QAAQ;KAAS;KAAO,QAAQ;MAAE,MAAM;MAAW;MAAM;KAAE;AAEtE,WAAO;KACL,QAAQ;KACR;KACA,QAAQ;MAAE,MAAM;MAAW;MAAM;KACjC,YAAY;MAAE,MAAM;MAAS;MAAM,OAAO;MAAO;KAClD;;GAIH,MAAM,uBACJ,SACA,gBACG;AACH,SAAK,IAAI,IAAI,GAAG,IAAI,QAAQ,QAAQ,KAAK;KACvC,MAAM,QAAQ,QAAQ;KACtB,MAAM,EAAE,cAAc,gBAAgB,cAAc;AAEpD,SAAI,MAAM,WAAW,aAAa;MAChC,MAAM,MAAM,MAAM;AAClB,mBAAa,KAAK;OAAE,OAAO,IAAI;OAAO,QAAQ,IAAI;OAAQ,CAAC;AAE3D,UAAI,IAAI,WAAW,MAAM;AAEvB,aAAM,QAAQ,aAAa,IAAI;AAC/B,kBAAW,KAAK;QAAE;QAAW,QAAQ;QAAQ,CAAC;iBACrC,IAAI,WAAW,UAAU;AAGlC,oBAAa,KAAK;QAChB,MAAM;QACN,SAAS,IAAI,MAAM;QACpB,CAAC;AACF,kBAAW,KAAK;QACd;QACA,QAAQ;QACR,MAAM;QACN,OAAO,IAAI;QACZ,CAAC;iBACO,IAAI,WAAW,QAExB,YAAW,KAAK;OAAE;OAAW,QAAQ;OAAQ,MAAM;OAAS,CAAC;UAK7D,YAAW,KAAK;OACd;OACA,QAAQ;OACR,MAAM;OACP,CAAC;YAEC;AAKL;AACA,cAAQ,OACN,cAAc,MAAM,GAAG,UAAU,IAAI,UAAU,UAAU,IAAI,UAAU,GAAG,MAAM,GACjF;MAED,MAAM,EAAE,KAAK,SAAS,UAAU,gBAAgB,cAAc;MAC9D,MAAM,SAAS,mBAAmB,MAAM,OAAO;MAC/C,MAAM,QAAQ,WAAW,OAAO;MAChC,MAAM,aAAa,MAAM,QAAQ;MACjC,MAAM,gBACJ,YAAY,SAAS,WAAW,CAAC,WAAW;MAC9C,MAAM,OAAO,aAAa,OAAO,WAAW,KAAK,QAAQ;AACzD,mBAAa,KAAK;OAAE;OAAO,QAAQ;QAAE,MAAM;QAAW;QAAM;OAAE,CAAC;AAE/D,UAAI,CAAC,oBAAoB,MAAM,OAAO,EAAE;AACtC,mBAAY,KAAK;QAAE,MAAM;QAAW,SAAS,MAAM;QAAS,CAAC;AAC7D,kBAAW,KAAK;QACd;QACA,QAAQ;QACR,MAAM;QACN;QACD,CAAC;AACF;;AAGF,aAAO,KACL,wCACA,WACA,QACA,gBACI,6BACA,uCACL;AACD,iBAAW,KAAK;OACd;OACA,QAAQ;OACR,MAAM;OACN;OACD,CAAC;;;;AAKR,OAAI,gBAAgB,SAAS,YAC3B,MAAK,IAAI,IAAI,GAAG,IAAI,gBAAgB,QAAQ,KAAK,aAAa;IAC5D,MAAM,QAAQ,gBAAgB,MAAM,GAAG,IAAI,YAAY;AAEvD,wBADqB,MAAM,QAAQ,WAAW,MAAM,IAAI,YAAY,CAAC,EACnC,EAAE;AACpC,UAAM,UAAU,MAAM;;QAEnB;AAIL,wBAHgB,MAAM,QAAQ,WAC5B,gBAAgB,IAAI,YAAY,CACjC,EAC4B,EAAE;AAC/B,UAAM,UAAU,MAAM;;AAGxB,WAAQ,KAAK,GAAG;;;CAIpB,MAAM,YAAY,YAAY,KAAK,GAAG,aAAa,KAAM,QAAQ,EAAE;AAGnE,KAAI,WAAW,SAAS,GAAG;EACzB,MAAM,aAAa,KAAK,IAAI,GAAG,WAAW,KAAK,MAAM,EAAE,UAAU,OAAO,CAAC;EACzE,MAAM,YAAY,GAAG,IAAI,IAAI,OAAO,GAAG,CAAC;AACxC,UAAQ,IAAI,GAAG;AACf,UAAQ,IACN,KAAK,GAAG,KAAK,kBAAkB,CAAC,GAAG,GAAG,IAAI,IAAI,WAAW,OAAO,GAAG,GACpE;AACD,UAAQ,IAAI,KAAK,YAAY;AAC7B,OAAK,MAAM,SAAS,YAAY;GAC9B,IAAI;AACJ,WAAQ,MAAM,MAAd;IACE,KAAK;AACH,WAAM,GAAG,KAAK,GAAG,IAAI,UAAU,CAAC;AAChC;IACF,KAAK;AACH,WAAM,GAAG,KAAK,GAAG,OAAO,UAAU,CAAC;AACnC;IACF,KAAK;AACH,WAAM,GAAG,IAAI,UAAU;AACvB;IACF,KAAK;AACH,WAAM,GAAG,KAAK,GAAG,IAAI,SAAS,CAAC;AAC/B;IACF,QACE,OACE,MAAM,WAAW,QACb,SAAS,GAAG,KAAK,GAAG,MAAM,QAAQ,CAAC,KACnC,SAAS,GAAG,KAAK,GAAG,OAAO,QAAQ,CAAC;;GAE9C,MAAM,UAAU,MAAM,UAAU,OAAO,WAAW;GAGlD,MAAM,OACJ,MAAM,SAAS,YAAY,MAAM,SAAS,UACtC,GAAG,IAAI,GAAG,cAAc,QAAQ,CAAC,GACjC;GACN,MAAM,YAAY,MAAM,OAAO,QAAQ,MAAM,eAAe,GAAG;GAC/D,MAAM,SAAS,YAAY,KAAK,GAAG,IAAI,UAAU,KAAK;AACtD,WAAQ,IAAI,KAAK,IAAI,IAAI,OAAO,SAAS;;EAE3C,MAAM,WAAW,WAAW,QACzB,MAAM,EAAE,WAAW,UAAU,CAAC,EAAE,KAClC,CAAC;EACF,MAAM,aAAa,WAAW,QAAQ,MAAM,EAAE,WAAW,MAAM,CAAC;EAChE,MAAM,cAAc,WAAW,QAAQ,MAAM,EAAE,SAAS,SAAS,CAAC;EAClE,MAAM,eAAe,WAAW,QAC7B,MAAM,EAAE,SAAS,eACnB,CAAC;EACF,MAAM,aAAa,WAAW,QAAQ,MAAM,EAAE,SAAS,QAAQ,CAAC;EAChE,MAAM,aAAa,WAAW,QAAQ,MAAM,EAAE,SAAS,QAAQ,CAAC;AAChE,UAAQ,IAAI,KAAK,YAAY;EAC7B,MAAM,QAAQ,CAAC,GAAG,SAAS,OAAO,GAAG,WAAW,aAAa;AAC7D,MAAI,cAAc,EAChB,OAAM,KACJ,GAAG,YAAY,OAAO,gBAAgB,IAAI,UAAU,WACrD;AACH,MAAI,eAAe,EAAG,OAAM,KAAK,GAAG,aAAa,WAAW;AAC5D,MAAI,aAAa,EAAG,OAAM,KAAK,GAAG,WAAW,QAAQ;AACrD,MAAI,aAAa,EACf,OAAM,KACJ,GAAG,WAAW,SAAS,eAAe,IAAI,UAAU,WACrD;AACH,UAAQ,IAAI,KAAK,MAAM,KAAK,KAAK,CAAC,IAAI,GAAG,IAAI,GAAG,QAAQ,GAAG,GAAG;AAC9D,UAAQ,IAAI,GAAG;;AAQjB,QAAO;EAAE,SAJO,CAAC,GAAG,eAAe,GAAG,aAAa,CAChD,MAAM,GAAG,MAAM,EAAE,QAAQ,EAAE,MAAM,CACjC,KAAK,MAAM,EAAE,OAAO;EAEL;EAAc;EAAa;;;;;;;AAQ/C,SAAS,mBAAmB,UAA0B;AACpD,QAAO,SAAS,QAAQ,UAAU,GAAG;;;;;;;;;;;;AAavC,SAAgB,kBAAkB,UAA0B;AAC1D,QAAO,SACJ,QAAQ,WAAW,GAAG,CACtB,QAAQ,SAAS,GAAG,CACpB,MAAM,IAAI,CAAC;;;AAIhB,MAAM,UAAkC;CAEtC,QAAQ;CACR,QAAQ;CAER,SAAS;CAET,SAAS;CACT,UAAU;CACV,KAAK;CACL,QAAQ;CACR,OAAO;CACP,QAAQ;CACR,SAAS;CAET,MAAM;CACN,WAAW;CACX,eAAe;CACf,UAAU;CAEV,OAAO;CACP,KAAK;CACL,QAAQ;CACR,QAAQ;CACR,SAAS;CAET,WAAW;CACX,UAAU;CAEV,MAAM;CACP"}
1
+ {"version":3,"file":"query-registry.js","names":[],"sources":["../../src/type-generator/query-registry.ts"],"sourcesContent":["import fs from \"node:fs/promises\";\nimport path from \"node:path\";\nimport { WorkspaceClient } from \"@databricks/sdk-experimental\";\nimport pc from \"picocolors\";\nimport { createLogger } from \"../logging/logger\";\nimport { CACHE_VERSION, hashSQL, loadCache, saveCache } from \"./cache\";\nimport { getErrorDiagnostic, isConnectivityError } from \"./errors\";\nimport { decidePreflight, type PreflightMode } from \"./preflight\";\nimport { Spinner } from \"./spinner\";\nimport { type DescribeFormatMemo, describeAdaptive } from \"./statement-result\";\nimport {\n type DatabricksStatementExecutionResponse,\n type QueryFatalError,\n type QueryGenerationResult,\n type QuerySchema,\n type QuerySyntaxError,\n sqlTypeToHelper,\n sqlTypeToMarker,\n} from \"./types\";\nimport {\n getWarehouseState,\n startWarehouse,\n waitUntilRunning,\n} from \"./warehouse-status\";\n\nconst logger = createLogger(\"type-generator:query-registry\");\n\n/**\n * Upper bound on how long a `blocking`-mode preflight will wait for a starting\n * warehouse to reach RUNNING before giving up (~5 min). Generous enough to ride\n * out a cold start without hanging an interactive CLI invocation indefinitely.\n */\nconst PREFLIGHT_WAIT_MAX_MS = 300_000;\n\n/**\n * Regex breakdown:\n * '(?:[^']|'')*' — matches a SQL string literal, including escaped '' pairs\n * | — alternation: whichever branch matches first at a position wins\n * --[^\\n]* — matches a single-line SQL comment\n *\n * Because the regex engine scans left-to-right, a `'` is consumed as a string\n * literal before any `--` inside it could match as a comment — giving us\n * correct single-pass ordering without a manual state machine.\n *\n * V1: no block-comment support (deferred to next PR).\n */\nconst PROTECTED_RANGE_RE = /'(?:[^']|'')*'|--[^\\n]*/g;\n\n/**\n * Numeric-context patterns for positional type inference.\n * Hoisted to module scope — safe because matchAll() clones the regex internally.\n */\nconst NUMERIC_PATTERNS: RegExp[] = [\n /\\bLIMIT\\s+:([a-zA-Z_]\\w*)/gi,\n /\\bOFFSET\\s+:([a-zA-Z_]\\w*)/gi,\n /\\bTOP\\s+:([a-zA-Z_]\\w*)/gi,\n /\\bFETCH\\s+FIRST\\s+:([a-zA-Z_]\\w*)\\s+ROWS/gi,\n // V1 limitation: arithmetic operators may false-positive for date\n // expressions like `:start_date - INTERVAL '1 day'`. A smarter\n // heuristic (e.g. look-ahead for INTERVAL) is deferred to a future PR.\n /[+\\-*/]\\s*:([a-zA-Z_]\\w*)/g,\n /:([a-zA-Z_]\\w*)\\s*[+\\-*/]/g,\n];\n\nexport function getProtectedRanges(sql: string): Array<[number, number]> {\n const ranges: Array<[number, number]> = [];\n for (const m of sql.matchAll(PROTECTED_RANGE_RE)) {\n ranges.push([m.index, m.index + m[0].length]);\n }\n return ranges;\n}\n\nfunction isInsideProtectedRange(\n offset: number,\n ranges: Array<[number, number]>,\n): boolean {\n return ranges.some(([start, end]) => offset >= start && offset < end);\n}\n\n/**\n * Parse a raw API/SDK error into a structured code + message.\n * Handles Databricks-style JSON bodies embedded in the message string,\n * e.g. `Response from server (Bad Request) {\"error_code\":\"...\",\"message\":\"...\"}`.\n */\nfunction parseError(raw: string): { code?: string; message: string } {\n const jsonMatch = raw.match(/\\{[\\s\\S]*\\}/);\n if (jsonMatch) {\n try {\n const parsed = JSON.parse(jsonMatch[0]);\n if (parsed.error_code || parsed.message) {\n return {\n code: parsed.error_code,\n message: parsed.message || raw,\n };\n }\n } catch {\n // not valid JSON, fall through\n }\n }\n return { message: raw };\n}\n\n/**\n * Extract parameters from a SQL query\n * @param sql - the SQL query to extract parameters from\n * @returns an array of parameter names\n */\nexport function extractParameters(\n sql: string,\n ranges?: Array<[number, number]>,\n): string[] {\n const protectedRanges = ranges ?? getProtectedRanges(sql);\n const matches = sql.matchAll(/(?<!:):([a-zA-Z_]\\w*)/g);\n const params = new Set<string>();\n for (const match of matches) {\n if (!isInsideProtectedRange(match.index, protectedRanges)) {\n params.add(match[1]);\n }\n }\n return Array.from(params);\n}\n\n// parameters that are injected by the server\nexport const SERVER_INJECTED_PARAMS = [\"workspaceId\"];\n\n/**\n * Generates the TypeScript type literal for query parameters from SQL.\n * Shared by both the success and failure paths.\n */\nfunction formatParametersType(sql: string): string {\n const params = extractParameters(sql).filter(\n (p) => !SERVER_INJECTED_PARAMS.includes(p),\n );\n const paramTypes = extractParameterTypes(sql);\n\n return params.length > 0\n ? `{\\n ${params\n .map((p) => {\n const sqlType = paramTypes[p];\n const markerType = sqlType\n ? sqlTypeToMarker[sqlType]\n : \"SQLTypeMarker\";\n const helper = sqlType ? sqlTypeToHelper[sqlType] : \"sql.*()\";\n return `/** ${sqlType || \"any\"} - use ${helper} */\\n ${p}: ${markerType}`;\n })\n .join(\";\\n \")};\\n }`\n : \"Record<string, never>\";\n}\n\nexport function convertToQueryType(\n result: DatabricksStatementExecutionResponse,\n sql: string,\n queryName: string,\n): { type: string; hasResults: boolean } {\n const dataRows = result.result?.data_array || [];\n const columns = dataRows.map((row) => ({\n name: row[0] || \"\",\n type_name: row[1]?.toUpperCase() || \"STRING\",\n comment: row[2] || undefined,\n }));\n\n const paramsType = formatParametersType(sql);\n\n // generate result fields with JSDoc\n const resultFields = columns.map((column) => {\n const normalizedType = normalizeTypeName(column.type_name);\n const mappedType = typeMap[normalizedType] || \"unknown\";\n // validate column name is a valid identifier\n const name = /^[a-zA-Z_$][a-zA-Z0-9_$]*$/.test(column.name)\n ? column.name\n : `\"${column.name}\"`;\n\n // generate comment for column\n const comment = column.comment\n ? `/** ${column.comment.replace(/\\*\\//g, \"* /\")} */\\n `\n : `/** @sqlType ${column.type_name.replace(/\\*\\//g, \"* /\")} */\\n `;\n\n return `${comment}${name}: ${mappedType}`;\n });\n\n const hasResults = resultFields.length > 0;\n\n const type = `{\n name: \"${queryName}\";\n parameters: ${paramsType};\n result: ${\n hasResults\n ? `Array<{\n ${resultFields.join(\";\\n \")};\n }>`\n : \"unknown\"\n };\n }`;\n\n return { type, hasResults };\n}\n\n/**\n * Used when DESCRIBE QUERY fails so the query still appears in QueryRegistry.\n * Generates a type with unknown result from SQL alone (no warehouse call).\n */\nfunction generateUnknownResultQuery(sql: string, queryName: string): string {\n const paramsType = formatParametersType(sql);\n\n return `{\n name: \"${queryName}\";\n parameters: ${paramsType};\n result: unknown;\n }`;\n}\n\n/**\n * Degrade gracefully when DESCRIBE can't produce a fresh schema (transient\n * connectivity outage, or a warehouse that's reachable but not ready). Reuse\n * the last-good cached type when the SQL hash is unchanged, otherwise emit\n * `unknown` from SQL alone. Never persists `result: unknown`.\n */\nfunction degradedType(\n cache: Awaited<ReturnType<typeof loadCache>>,\n queryName: string,\n sql: string,\n sqlHash: string,\n): string {\n const prior = cache.queries[queryName];\n const canReusePrior = prior?.hash === sqlHash && !prior.retry;\n return canReusePrior\n ? prior.type\n : generateUnknownResultQuery(sql, queryName);\n}\n\n// Single source of truth for the `@param` type alternation, shared by\n// extractParameterTypes and extractParameterDefaults so the two can't drift.\n// Alternation order matters: TIMESTAMP_NTZ must precede TIMESTAMP so the regex\n// engine doesn't greedy-match TIMESTAMP and leave `_NTZ` unconsumed.\nconst PARAM_TYPE_ALTERNATION =\n \"STRING|NUMERIC|DECIMAL|BIGINT|TINYINT|SMALLINT|INT|FLOAT|DOUBLE|BOOLEAN|DATE|TIMESTAMP_NTZ|TIMESTAMP|BINARY\";\n\nexport function extractParameterTypes(sql: string): Record<string, string> {\n const paramTypes: Record<string, string> = {};\n const regex = new RegExp(\n `--\\\\s*@param\\\\s+(\\\\w+)\\\\s+(${PARAM_TYPE_ALTERNATION})\\\\b`,\n \"gi\",\n );\n const matches = sql.matchAll(regex);\n for (const match of matches) {\n const [, paramName, paramType] = match;\n paramTypes[paramName] = paramType.toUpperCase();\n }\n\n return paramTypes;\n}\n\nexport function defaultForType(sqlType: string | undefined): string {\n switch (sqlType?.toUpperCase()) {\n case \"NUMERIC\":\n case \"DECIMAL\":\n case \"BIGINT\":\n case \"TINYINT\":\n case \"SMALLINT\":\n case \"INT\":\n return \"0\";\n case \"FLOAT\":\n case \"DOUBLE\":\n return \"0.0\";\n case \"STRING\":\n return \"''\";\n case \"BOOLEAN\":\n return \"true\";\n case \"DATE\":\n return \"'2000-01-01'\";\n case \"TIMESTAMP\":\n return \"'2000-01-01T00:00:00Z'\";\n case \"TIMESTAMP_NTZ\":\n return \"'2000-01-01T00:00:00'\";\n case \"BINARY\":\n return \"X'00'\";\n default:\n return \"''\";\n }\n}\n\n/**\n * True when `raw` is already a single, well-formed SQL single-quoted string\n * literal — i.e. it opens and closes with `'` and every interior quote is part\n * of an escaped `''` pair. `'2024-01-01'` and `'O''Brien'` qualify;\n * `'a' OR 1=1 OR 'b'` does not (it has lone interior quotes), so it is treated\n * as raw content and re-escaped rather than trusted. A backslash also\n * disqualifies it: Databricks/Spark treats `\\` as an escape inside literals (so\n * `'a\\'` is unterminated), and we never trust such input — it is re-escaped.\n */\nfunction isWellFormedStringLiteral(raw: string): boolean {\n if (raw.length < 2 || !raw.startsWith(\"'\") || !raw.endsWith(\"'\")) {\n return false;\n }\n const inner = raw.slice(1, -1);\n if (inner.includes(\"\\\\\")) return false;\n return !inner.replace(/''/g, \"\").includes(\"'\");\n}\n\n/**\n * Format a user-supplied sample value as a SQL literal for substitution into\n * the build-time DESCRIBE statement. Returns `null` when the value isn't valid\n * for its type, so the caller falls back to the safe type-based placeholder\n * instead of substituting attacker-controllable text.\n *\n * The value comes from a `.sql` file that may be shared via a template or\n * dependency, so it must not be able to inject SQL into `DESCRIBE QUERY`:\n * - string-like types are always emitted as one well-formed, fully-escaped\n * single-quoted literal (a pre-quoted literal is kept as-is, anything else is\n * quoted with both `\\` and `'` doubled so neither can terminate the literal),\n * so the value can never break out of the string;\n * - numeric / boolean / binary values must match a strict literal shape and are\n * rejected (`null`) otherwise, rather than being passed through verbatim.\n */\nfunction formatSampleValue(\n sqlType: string | undefined,\n raw: string,\n): string | null {\n switch (sqlType?.toUpperCase()) {\n case \"STRING\":\n case \"DATE\":\n case \"TIMESTAMP\":\n case \"TIMESTAMP_NTZ\":\n return isWellFormedStringLiteral(raw)\n ? raw\n : `'${raw.replace(/\\\\/g, \"\\\\\\\\\").replace(/'/g, \"''\")}'`;\n case \"NUMERIC\":\n case \"DECIMAL\":\n case \"BIGINT\":\n case \"TINYINT\":\n case \"SMALLINT\":\n case \"INT\":\n case \"FLOAT\":\n case \"DOUBLE\":\n return /^[+-]?\\d+(\\.\\d+)?$/.test(raw) ? raw : null;\n case \"BOOLEAN\":\n return /^(?:true|false)$/i.test(raw) ? raw.toLowerCase() : null;\n case \"BINARY\":\n return /^X'[0-9a-fA-F]*'$/i.test(raw) ? raw : null;\n default:\n return null;\n }\n}\n\n/**\n * Parse optional describe-time sample values from `@param` annotations, e.g.\n * `-- @param target_catalog STRING = main`. The value is substituted into the\n * SQL **only during DESCRIBE QUERY** so type generation can resolve queries\n * whose shape depends on a parameter value — most notably dynamic table names\n * via `IDENTIFIER(:target_catalog || '.schema.table')`, where the empty-string\n * default would otherwise produce malformed SQL. Runtime binding is unaffected:\n * the analytics plugin still binds the real parameter at execution time, so the\n * query stays portable across environments.\n *\n * Returns a map of parameter name to the formatted SQL literal to substitute.\n */\nexport function extractParameterDefaults(sql: string): Record<string, string> {\n const defaults: Record<string, string> = {};\n // Reuses PARAM_TYPE_ALTERNATION, then requires `= <value>` on the same line.\n // All inter-token whitespace is horizontal-only (`[^\\S\\r\\n]`, not `\\s`): `\\s`\n // matches newlines, so a value-less line like `-- @param x STRING =` would let\n // `\\s*=\\s*(.+?)` swallow the *next* line as the sample value. Restricting to\n // same-line whitespace makes such a line simply not match, so it correctly\n // falls back to the type placeholder.\n const regex = new RegExp(\n `--[^\\\\S\\\\r\\\\n]*@param[^\\\\S\\\\r\\\\n]+(\\\\w+)[^\\\\S\\\\r\\\\n]+(${PARAM_TYPE_ALTERNATION})[^\\\\S\\\\r\\\\n]*=[^\\\\S\\\\r\\\\n]*(.+?)[^\\\\S\\\\r\\\\n]*$`,\n \"gim\",\n );\n for (const match of sql.matchAll(regex)) {\n const [, paramName, paramType, rawValue] = match;\n const formatted = formatSampleValue(paramType, rawValue);\n // A value that fails type validation is dropped, not substituted: the param\n // then falls back to the safe type-based placeholder during DESCRIBE.\n if (formatted !== null) {\n defaults[paramName] = formatted;\n }\n }\n return defaults;\n}\n\n/**\n * Replace `:param` placeholders with describe-time literals so `DESCRIBE QUERY`\n * can run without bound parameters. Resolution order per parameter:\n * 1. An explicit `-- @param name TYPE = value` sample value (wins), which lets\n * dynamic table names via `IDENTIFIER(...)` resolve to a real table.\n * 2. Otherwise a placeholder default derived from the annotated/inferred type.\n * Placeholders inside string literals or comments are left untouched.\n */\nexport function substituteParametersForDescribe(sql: string): string {\n const protectedRanges = getProtectedRanges(sql);\n const annotatedTypes = extractParameterTypes(sql);\n const inferredTypes = inferParameterTypes(sql, protectedRanges);\n const parameterTypes = { ...inferredTypes, ...annotatedTypes };\n const parameterDefaults = extractParameterDefaults(sql);\n return sql.replace(\n /(?<!:):([a-zA-Z_]\\w*)/g,\n (original, paramName, offset) => {\n if (isInsideProtectedRange(offset, protectedRanges)) {\n return original;\n }\n const sampleValue = parameterDefaults[paramName];\n if (sampleValue !== undefined) {\n return sampleValue;\n }\n return defaultForType(parameterTypes[paramName]);\n },\n );\n}\n\n/**\n * Append a remediation hint when a DESCRIBE failure looks like a dynamic\n * identifier that couldn't be resolved: the query calls `IDENTIFIER(...)` and\n * has at least one parameter without a describe-time sample value. These fail\n * because typegen substitutes a placeholder default (e.g. `''`) that yields a\n * malformed or non-existent table name. Steering the user to the `= value`\n * annotation turns the fatal error into a one-line fix.\n */\nfunction withIdentifierHint(\n error: { code?: string; message: string },\n sql: string,\n): { code?: string; message: string } {\n if (!/\\bIDENTIFIER\\s*\\(/i.test(sql)) {\n return error;\n }\n const protectedRanges = getProtectedRanges(sql);\n const params = extractParameters(sql, protectedRanges);\n const defaults = extractParameterDefaults(sql);\n const unresolved = params.filter(\n (p) => !SERVER_INJECTED_PARAMS.includes(p) && defaults[p] === undefined,\n );\n if (unresolved.length === 0) {\n return error;\n }\n const example = unresolved[0];\n return {\n ...error,\n message: `${error.message}\\n Hint: this query uses IDENTIFIER() with parameter(s) ${unresolved\n .map((p) => `:${p}`)\n .join(\n \", \",\n )}. Give type generation a sample value so it can resolve the table, e.g. \\`-- @param ${example} STRING = my_catalog\\`. The runtime query still binds the real parameter.`,\n };\n}\n\n/**\n * Infer parameter types from positional context in SQL.\n * V1 only infers NUMERIC from patterns like LIMIT, OFFSET, TOP,\n * FETCH FIRST ... ROWS, and arithmetic operators.\n * Parameters inside string literals or SQL comments are ignored.\n */\nexport function inferParameterTypes(\n sql: string,\n ranges?: Array<[number, number]>,\n): Record<string, string> {\n const inferred: Record<string, string> = {};\n const protectedRanges = ranges ?? getProtectedRanges(sql);\n\n for (const pattern of NUMERIC_PATTERNS) {\n for (const match of sql.matchAll(pattern)) {\n if (!isInsideProtectedRange(match.index, protectedRanges)) {\n inferred[match[1]] = \"NUMERIC\";\n }\n }\n }\n\n return inferred;\n}\n\n/**\n * Generate query schemas from a folder of SQL files\n * It uses DESCRIBE QUERY to get the schema without executing the query\n * @param queryFolder - the folder containing the SQL files\n * @param warehouseId - the warehouse id to use for schema analysis\n * @param options - options for the query generation\n * @param options.noCache - if true, skip the cache and regenerate all types\n * @param options.mode - preflight policy: \"non-blocking\" never probes the\n * warehouse and never describes (emits cached/`unknown` types and returns\n * immediately), \"blocking\" waits for a starting warehouse and starts (then\n * waits for) a stopped one, treating only a deleted/deleting warehouse as\n * fatal. Defaults to \"non-blocking\".\n * @returns an array of query schemas\n */\nexport async function generateQueriesFromDescribe(\n queryFolder: string,\n warehouseId: string,\n options: {\n noCache?: boolean;\n concurrency?: number;\n mode?: PreflightMode;\n } = {},\n): Promise<QueryGenerationResult> {\n const {\n noCache = false,\n concurrency: rawConcurrency = 10,\n mode = \"non-blocking\",\n } = options;\n const concurrency =\n typeof rawConcurrency === \"number\" && Number.isFinite(rawConcurrency)\n ? Math.max(1, Math.floor(rawConcurrency))\n : 10;\n\n // read all query files and cache in parallel\n const [allFiles, cache] = await Promise.all([\n fs.readdir(queryFolder),\n noCache\n ? ({ version: CACHE_VERSION, queries: {} } as Awaited<\n ReturnType<typeof loadCache>\n >)\n : loadCache(),\n ]);\n\n const queryFiles = allFiles.filter((file) => file.endsWith(\".sql\"));\n logger.debug(\"Found %d SQL queries\", queryFiles.length);\n\n const client = new WorkspaceClient({});\n const spinner = new Spinner();\n\n // Read all SQL files in parallel\n const sqlContents = await Promise.all(\n queryFiles.map((file) => fs.readFile(path.join(queryFolder, file), \"utf8\")),\n );\n\n const startTime = performance.now();\n\n // Phase 1: Check cache, separate cached vs uncached\n const cachedResults: Array<{ index: number; schema: QuerySchema }> = [];\n const uncachedQueries: Array<{\n index: number;\n queryName: string;\n sql: string;\n sqlHash: string;\n cleanedSql: string;\n }> = [];\n const logEntries: Array<{\n queryName: string;\n status: \"HIT\" | \"MISS\";\n // Absent for clean hits/misses. \"syntax\" = bad SQL on a reachable warehouse;\n // \"connectivity\" = warehouse unreachable; \"empty\" = described but no columns;\n // \"fatal\" = non-SQL setup/request failure surfaced after .d.ts emission.\n kind?: \"syntax\" | \"connectivity\" | \"empty\" | \"fatal\";\n error?: { code?: string; message: string };\n }> = [];\n\n for (let i = 0; i < queryFiles.length; i++) {\n const file = queryFiles[i];\n const rawName = path.basename(file, \".sql\");\n const queryName = normalizeQueryName(rawName);\n\n const sql = sqlContents[i];\n const sqlHash = hashSQL(sql);\n\n const cached = cache.queries[queryName];\n if (cached && cached.hash === sqlHash && !cached.retry) {\n cachedResults.push({\n index: i,\n schema: { name: queryName, type: cached.type },\n });\n logEntries.push({ queryName, status: \"HIT\" });\n } else {\n const protectedRanges = getProtectedRanges(sql);\n const annotatedTypes = extractParameterTypes(sql);\n const inferredTypes = inferParameterTypes(sql, protectedRanges);\n const parameterTypes = { ...inferredTypes, ...annotatedTypes };\n // substituteParametersForDescribe applies `-- @param name TYPE = value`\n // sample values (e.g. for IDENTIFIER table names) ahead of type defaults;\n // parameterDefaults is recomputed here only to skip them in the warn loop.\n const parameterDefaults = extractParameterDefaults(sql);\n const sqlWithDefaults = substituteParametersForDescribe(sql);\n\n // Warn about unresolved parameters\n const allParams = extractParameters(sql, protectedRanges);\n for (const param of allParams) {\n if (SERVER_INJECTED_PARAMS.includes(param)) continue;\n if (parameterDefaults[param]) continue;\n if (parameterTypes[param]) continue;\n logger.warn(\n '%s: parameter \":%s\" has no type annotation or inference. Add %s to the query file.',\n queryFiles[i],\n param,\n `-- @param ${param} <TYPE>`,\n );\n }\n\n const cleanedSql = sqlWithDefaults.trim().replace(/;\\s*$/, \"\");\n uncachedQueries.push({ index: i, queryName, sql, sqlHash, cleanedSql });\n }\n }\n\n // Phase 2: Execute all uncached DESCRIBE calls in parallel\n type DescribeResult =\n | {\n // Described successfully with a result schema — the only case we cache.\n status: \"ok\";\n index: number;\n schema: QuerySchema;\n cacheEntry: { hash: string; type: string; retry: boolean };\n }\n | {\n // Reachable warehouse ran DESCRIBE and rejected the statement — a\n // genuine SQL error. Eligible to fail the build; never cached.\n status: \"syntax\";\n index: number;\n schema: QuerySchema;\n error: { code?: string; message: string };\n }\n | {\n // DESCRIBE succeeded but returned no columns — soft `unknown`. Not a\n // failure, not cached, retried next run.\n status: \"empty\";\n index: number;\n schema: QuerySchema;\n }\n | {\n // Warehouse reachable but returned a non-terminal state (PENDING/\n // RUNNING) with no rows — stopped/cold-starting/busy. Degrade like a\n // transient outage (reuse cache or `unknown`); not cached, not empty.\n status: \"unavailable\";\n index: number;\n schema: QuerySchema;\n };\n\n const freshResults: Array<{ index: number; schema: QuerySchema }> = [];\n // Genuine SQL errors (reachable warehouse). Connectivity failures are NOT\n // recorded here — they degrade silently so a transient outage isn't fatal.\n const syntaxErrors: QuerySyntaxError[] = [];\n const fatalErrors: QueryFatalError[] = [];\n\n if (uncachedQueries.length > 0) {\n // One-time warehouse preflight (before issuing any DESCRIBE). A single\n // warehouses.get classifies the warehouse so we can skip the whole describe\n // batch when it can't serve this run, instead of letting every query fail\n // (and re-fail next run). Reuses this file's degrade/classify helpers so a\n // not-ready warehouse degrades exactly like a per-query outage.\n let decision: ReturnType<typeof decidePreflight> = \"proceed\";\n let fatalMessage = \"\";\n if (mode === \"non-blocking\") {\n // `non-blocking` never describes and must make ZERO warehouse round-trips:\n // skip the probe entirely (no getWarehouseState) and go straight to\n // degradeAll. A foreground/one-shot run can't describe in the background,\n // so it emits best-available types (reused cache or `unknown`) and returns\n // now.\n decision = \"degradeAll\";\n } else {\n try {\n const state = await getWarehouseState(client, warehouseId);\n decision = decidePreflight(state, mode);\n if (decision === \"fatal\") {\n fatalMessage = `warehouse ${warehouseId} is ${state}`;\n }\n if (decision === \"startWaitProceed\") {\n // Stopped/stopping warehouse: nudge it out of the stopped state, then\n // poll to RUNNING. treatStoppedAsTransient rides out the stale\n // pre-start STOPPED/STOPPING reading the start hasn't propagated past\n // yet — only DELETED/DELETING (or the deadline) ends the wait early.\n await startWarehouse(client, warehouseId);\n const final = await waitUntilRunning(client, warehouseId, {\n maxMs: PREFLIGHT_WAIT_MAX_MS,\n treatStoppedAsTransient: true,\n });\n if (final === \"RUNNING\") {\n decision = \"proceed\";\n } else {\n decision = \"fatal\";\n fatalMessage = `warehouse ${warehouseId} did not reach RUNNING (now ${final})`;\n }\n }\n if (decision === \"waitThenProceed\") {\n const final = await waitUntilRunning(client, warehouseId, {\n maxMs: PREFLIGHT_WAIT_MAX_MS,\n });\n if (final === \"RUNNING\") {\n decision = \"proceed\";\n } else {\n decision = \"fatal\";\n fatalMessage = `warehouse ${warehouseId} did not reach RUNNING (now ${final})`;\n }\n }\n } catch (err) {\n if (isConnectivityError(err)) {\n // Warehouse unreachable (transient outage): degrade silently like a\n // per-query connectivity failure — never fail a build on a blip.\n decision = \"degradeAll\";\n } else {\n // Auth, bad warehouse id, malformed config, or a timed-out wait: fatal.\n decision = \"fatal\";\n fatalMessage = `warehouse ${warehouseId}: ${getErrorDiagnostic(err)}`;\n }\n }\n }\n\n if (decision !== \"proceed\") {\n // degradeAll or fatal: skip DESCRIBE entirely. Every uncached query gets a\n // degraded schema (reused cache or `unknown`); fatal additionally records\n // a fatalError per query so the caller fails the build after writing.\n const kind = decision === \"fatal\" ? \"fatal\" : \"connectivity\";\n for (const { index, queryName, sql, sqlHash } of uncachedQueries) {\n freshResults.push({\n index,\n schema: {\n name: queryName,\n type: degradedType(cache, queryName, sql, sqlHash),\n },\n });\n if (decision === \"fatal\") {\n fatalErrors.push({ name: queryName, message: fatalMessage });\n logEntries.push({\n queryName,\n status: \"MISS\",\n kind,\n error: { message: fatalMessage },\n });\n } else {\n logEntries.push({ queryName, status: \"MISS\", kind });\n }\n }\n } else {\n let completed = 0;\n const total = uncachedQueries.length;\n spinner.start(\n `Describing ${total} ${total === 1 ? \"query\" : \"queries\"} (0/${total})`,\n );\n\n // Shared across this run's DESCRIBE QUERY calls: discover the warehouse's\n // result format once, then reuse it.\n const describeFormat: DescribeFormatMemo = {};\n\n const describeOne = async ({\n index,\n queryName,\n sql,\n sqlHash,\n cleanedSql,\n }: (typeof uncachedQueries)[number]): Promise<DescribeResult> => {\n // describeAdaptive negotiates the result format with the warehouse\n // (JSON_ARRAY for standard DBSQL, ARROW_STREAM for Reyden) and returns\n // an already-normalized response, so the state classification below\n // reads it directly.\n const result = await describeAdaptive(\n client,\n `DESCRIBE QUERY ${cleanedSql}`,\n warehouseId,\n describeFormat,\n );\n\n completed++;\n spinner.update(\n `Describing ${total} ${total === 1 ? \"query\" : \"queries\"} (${completed}/${total})`,\n );\n\n logger.debug(\n \"DESCRIBE result for %s: state=%s, rows=%d\",\n queryName,\n result.status.state,\n result.result?.data_array?.length ?? 0,\n );\n\n if (result.status.state === \"FAILED\") {\n // The warehouse was reachable and ran DESCRIBE, but the statement\n // failed — a genuine SQL error (bad table, syntax, incompatible type).\n const sqlError =\n result.status.error?.message || \"Query execution failed\";\n // The failure is surfaced once, formatted, by the aggregated\n // TypegenSyntaxError (and the summary table) — don't also log the raw\n // message here or every SQL error prints twice in dev.\n const type = generateUnknownResultQuery(sql, queryName);\n return {\n status: \"syntax\",\n index,\n schema: { name: queryName, type },\n error: withIdentifierHint(parseError(sqlError), sql),\n };\n }\n\n if (result.status.state !== \"SUCCEEDED\") {\n // Non-terminal state (PENDING/RUNNING) with no result rows: the\n // warehouse is reachable but not ready (stopped, cold-starting, or\n // busy). Degrade like a transient outage — reuse the last-good cached\n // type when the SQL is unchanged, else emit `unknown`. Never \"empty\":\n // treating this as empty would emit `result: unknown` AND discard the\n // good cached type, silently throwing away working types.\n return {\n status: \"unavailable\",\n index,\n schema: {\n name: queryName,\n type: degradedType(cache, queryName, sql, sqlHash),\n },\n };\n }\n\n const { type, hasResults } = convertToQueryType(result, sql, queryName);\n if (!hasResults) {\n // Described, but no result columns. Emit `unknown` and retry next run;\n // do not cache (we never persist `result: unknown`).\n return { status: \"empty\", index, schema: { name: queryName, type } };\n }\n return {\n status: \"ok\",\n index,\n schema: { name: queryName, type },\n cacheEntry: { hash: sqlHash, type, retry: false },\n };\n };\n\n // Process in chunks, saving cache after each chunk\n const processBatchResults = (\n settled: PromiseSettledResult<DescribeResult>[],\n batchOffset: number,\n ) => {\n for (let i = 0; i < settled.length; i++) {\n const entry = settled[i];\n const { queryName } = uncachedQueries[batchOffset + i];\n\n if (entry.status === \"fulfilled\") {\n const res = entry.value;\n freshResults.push({ index: res.index, schema: res.schema });\n\n if (res.status === \"ok\") {\n // Only a successful describe with a result schema is cached.\n cache.queries[queryName] = res.cacheEntry;\n logEntries.push({ queryName, status: \"MISS\" });\n } else if (res.status === \"syntax\") {\n // Genuine SQL error — record it for the caller's prod/dev gate.\n // Not cached: re-described next run so a fixed query recovers.\n syntaxErrors.push({\n name: queryName,\n message: res.error.message,\n });\n logEntries.push({\n queryName,\n status: \"MISS\",\n kind: \"syntax\",\n error: res.error,\n });\n } else if (res.status === \"empty\") {\n // status === \"empty\": described, no columns. Soft unknown, not cached.\n logEntries.push({ queryName, status: \"MISS\", kind: \"empty\" });\n } else {\n // status === \"unavailable\": non-terminal DESCRIBE (warehouse\n // stopped/cold-starting/busy). Degrade like a transient outage:\n // tag OFFLINE, count as degraded, never cache.\n logEntries.push({\n queryName,\n status: \"MISS\",\n kind: \"connectivity\",\n });\n }\n } else {\n // executeStatement rejected without a normal StatementExecution result.\n // Only structured transport/connectivity failures are treated as\n // offline; auth, bad warehouse IDs, malformed requests, and SDK/config\n // failures stay fatal so users fix the underlying setup issue.\n completed++;\n spinner.update(\n `Describing ${total} ${total === 1 ? \"query\" : \"queries\"} (${completed}/${total})`,\n );\n\n const { sql, sqlHash, index } = uncachedQueries[batchOffset + i];\n const reason = getErrorDiagnostic(entry.reason);\n const error = parseError(reason);\n const priorEntry = cache.queries[queryName];\n const canReusePrior =\n priorEntry?.hash === sqlHash && !priorEntry.retry;\n const type = degradedType(cache, queryName, sql, sqlHash);\n freshResults.push({ index, schema: { name: queryName, type } });\n\n if (!isConnectivityError(entry.reason)) {\n fatalErrors.push({ name: queryName, message: error.message });\n logEntries.push({\n queryName,\n status: \"MISS\",\n kind: \"fatal\",\n error,\n });\n continue;\n }\n\n logger.warn(\n \"DESCRIBE unreachable for %s: %s — %s\",\n queryName,\n reason,\n canReusePrior\n ? \"reusing last cached type\"\n : \"emitting unknown (no matching cache)\",\n );\n logEntries.push({\n queryName,\n status: \"MISS\",\n kind: \"connectivity\",\n error,\n });\n }\n }\n };\n\n if (uncachedQueries.length > concurrency) {\n for (let b = 0; b < uncachedQueries.length; b += concurrency) {\n const batch = uncachedQueries.slice(b, b + concurrency);\n const batchResults = await Promise.allSettled(batch.map(describeOne));\n processBatchResults(batchResults, b);\n await saveCache(cache);\n }\n } else {\n const settled = await Promise.allSettled(\n uncachedQueries.map(describeOne),\n );\n processBatchResults(settled, 0);\n await saveCache(cache);\n }\n\n spinner.stop(\"\");\n }\n }\n\n const elapsed = ((performance.now() - startTime) / 1000).toFixed(2);\n\n // Print formatted table\n if (logEntries.length > 0) {\n const maxNameLen = Math.max(...logEntries.map((e) => e.queryName.length));\n const separator = pc.dim(\"─\".repeat(50));\n console.log(\"\");\n console.log(\n ` ${pc.bold(\"Typegen Queries\")} ${pc.dim(`(${logEntries.length})`)}`,\n );\n console.log(` ${separator}`);\n for (const entry of logEntries) {\n let tag: string;\n switch (entry.kind) {\n case \"syntax\":\n tag = pc.bold(pc.red(\"SQL ERR\"));\n break;\n case \"connectivity\":\n tag = pc.bold(pc.yellow(\"OFFLINE\"));\n break;\n case \"empty\":\n tag = pc.dim(\"EMPTY \");\n break;\n case \"fatal\":\n tag = pc.bold(pc.red(\"FATAL \"));\n break;\n default:\n tag =\n entry.status === \"HIT\"\n ? `cache ${pc.bold(pc.green(\"HIT \"))}`\n : `cache ${pc.bold(pc.yellow(\"MISS \"))}`;\n }\n const rawName = entry.queryName.padEnd(maxNameLen);\n // Only genuine SQL errors are struck through. Connectivity/empty kept a\n // usable type (reused or unknown), so they read as degraded, not broken.\n const name =\n entry.kind === \"syntax\" || entry.kind === \"fatal\"\n ? pc.dim(pc.strikethrough(rawName))\n : rawName;\n const errorCode = entry.error?.message.match(/\\[([^\\]]+)\\]/)?.[1];\n const reason = errorCode ? ` ${pc.dim(errorCode)}` : \"\";\n console.log(` ${tag} ${name}${reason}`);\n }\n const newCount = logEntries.filter(\n (e) => e.status === \"MISS\" && !e.kind,\n ).length;\n const cacheCount = logEntries.filter((e) => e.status === \"HIT\").length;\n const syntaxCount = logEntries.filter((e) => e.kind === \"syntax\").length;\n const offlineCount = logEntries.filter(\n (e) => e.kind === \"connectivity\",\n ).length;\n const emptyCount = logEntries.filter((e) => e.kind === \"empty\").length;\n const fatalCount = logEntries.filter((e) => e.kind === \"fatal\").length;\n console.log(` ${separator}`);\n const parts = [`${newCount} new`, `${cacheCount} from cache`];\n if (syntaxCount > 0)\n parts.push(\n `${syntaxCount} SQL ${syntaxCount === 1 ? \"error\" : \"errors\"}`,\n );\n if (offlineCount > 0) parts.push(`${offlineCount} degraded`);\n if (emptyCount > 0) parts.push(`${emptyCount} empty`);\n if (fatalCount > 0)\n parts.push(\n `${fatalCount} fatal ${fatalCount === 1 ? \"error\" : \"errors\"}`,\n );\n console.log(` ${parts.join(\", \")}. ${pc.dim(`${elapsed}s`)}`);\n console.log(\"\");\n }\n\n // Merge and sort by original file index for deterministic output\n const schemas = [...cachedResults, ...freshResults]\n .sort((a, b) => a.index - b.index)\n .map((r) => r.schema);\n\n return { schemas, syntaxErrors, fatalErrors };\n}\n\n/**\n * Normalize query name by removing the .obo extension\n * @param queryName - the query name to normalize\n * @returns the normalized query name\n */\nfunction normalizeQueryName(fileName: string): string {\n return fileName.replace(/\\.obo$/, \"\");\n}\n\n/**\n * Normalize SQL type name by removing parameters/generics\n * Examples:\n * DECIMAL(38,6) -> DECIMAL\n * ARRAY<STRING> -> ARRAY\n * MAP<STRING,INT> -> MAP\n * STRUCT<name:STRING> -> STRUCT\n * INTERVAL DAY TO SECOND -> INTERVAL\n * GEOGRAPHY(4326) -> GEOGRAPHY\n */\nexport function normalizeTypeName(typeName: string): string {\n return typeName\n .replace(/\\(.*\\)$/, \"\") // remove (p, s) eg: DECIMAL(38,6) -> DECIMAL\n .replace(/<.*>$/, \"\") // remove <T> eg: ARRAY<STRING> -> ARRAY\n .split(\" \")[0]; // take first word eg: INTERVAL DAY TO SECOND -> INTERVAL\n}\n\n/** Type Map for Databricks data types to JavaScript types */\nconst typeMap: Record<string, string> = {\n // string types\n STRING: \"string\",\n BINARY: \"string\",\n // boolean\n BOOLEAN: \"boolean\",\n // numeric types\n TINYINT: \"number\",\n SMALLINT: \"number\",\n INT: \"number\",\n BIGINT: \"number\",\n FLOAT: \"number\",\n DOUBLE: \"number\",\n DECIMAL: \"number\",\n // date/time types\n DATE: \"string\",\n TIMESTAMP: \"string\",\n TIMESTAMP_NTZ: \"string\",\n INTERVAL: \"string\",\n // complex types\n ARRAY: \"unknown[]\",\n MAP: \"Record<string, unknown>\",\n STRUCT: \"Record<string, unknown>\",\n OBJECT: \"Record<string, unknown>\",\n VARIANT: \"unknown\",\n // spatial types\n GEOGRAPHY: \"unknown\",\n GEOMETRY: \"unknown\",\n // null type\n VOID: \"null\",\n};\n"],"mappings":";;;;;;;;;;;;;;AAyBA,MAAM,SAAS,aAAa,gCAAgC;;;;;;AAO5D,MAAM,wBAAwB;;;;;;;;;;;;;AAc9B,MAAM,qBAAqB;;;;;AAM3B,MAAM,mBAA6B;CACjC;CACA;CACA;CACA;CAIA;CACA;CACD;AAED,SAAgB,mBAAmB,KAAsC;CACvE,MAAM,SAAkC,EAAE;AAC1C,MAAK,MAAM,KAAK,IAAI,SAAS,mBAAmB,CAC9C,QAAO,KAAK,CAAC,EAAE,OAAO,EAAE,QAAQ,EAAE,GAAG,OAAO,CAAC;AAE/C,QAAO;;AAGT,SAAS,uBACP,QACA,QACS;AACT,QAAO,OAAO,MAAM,CAAC,OAAO,SAAS,UAAU,SAAS,SAAS,IAAI;;;;;;;AAQvE,SAAS,WAAW,KAAiD;CACnE,MAAM,YAAY,IAAI,MAAM,cAAc;AAC1C,KAAI,UACF,KAAI;EACF,MAAM,SAAS,KAAK,MAAM,UAAU,GAAG;AACvC,MAAI,OAAO,cAAc,OAAO,QAC9B,QAAO;GACL,MAAM,OAAO;GACb,SAAS,OAAO,WAAW;GAC5B;SAEG;AAIV,QAAO,EAAE,SAAS,KAAK;;;;;;;AAQzB,SAAgB,kBACd,KACA,QACU;CACV,MAAM,kBAAkB,UAAU,mBAAmB,IAAI;CACzD,MAAM,UAAU,IAAI,SAAS,yBAAyB;CACtD,MAAM,yBAAS,IAAI,KAAa;AAChC,MAAK,MAAM,SAAS,QAClB,KAAI,CAAC,uBAAuB,MAAM,OAAO,gBAAgB,CACvD,QAAO,IAAI,MAAM,GAAG;AAGxB,QAAO,MAAM,KAAK,OAAO;;AAI3B,MAAa,yBAAyB,CAAC,cAAc;;;;;AAMrD,SAAS,qBAAqB,KAAqB;CACjD,MAAM,SAAS,kBAAkB,IAAI,CAAC,QACnC,MAAM,CAAC,uBAAuB,SAAS,EAAE,CAC3C;CACD,MAAM,aAAa,sBAAsB,IAAI;AAE7C,QAAO,OAAO,SAAS,IACnB,YAAY,OACT,KAAK,MAAM;EACV,MAAM,UAAU,WAAW;EAC3B,MAAM,aAAa,UACf,gBAAgB,WAChB;EACJ,MAAM,SAAS,UAAU,gBAAgB,WAAW;AACpD,SAAO,OAAO,WAAW,MAAM,SAAS,OAAO,aAAa,EAAE,IAAI;GAClE,CACD,KAAK,YAAY,CAAC,YACrB;;AAGN,SAAgB,mBACd,QACA,KACA,WACuC;CAEvC,MAAM,WADW,OAAO,QAAQ,cAAc,EAAE,EACvB,KAAK,SAAS;EACrC,MAAM,IAAI,MAAM;EAChB,WAAW,IAAI,IAAI,aAAa,IAAI;EACpC,SAAS,IAAI,MAAM;EACpB,EAAE;CAEH,MAAM,aAAa,qBAAqB,IAAI;CAG5C,MAAM,eAAe,QAAQ,KAAK,WAAW;EAE3C,MAAM,aAAa,QADI,kBAAkB,OAAO,UAAU,KACZ;EAE9C,MAAM,OAAO,6BAA6B,KAAK,OAAO,KAAK,GACvD,OAAO,OACP,IAAI,OAAO,KAAK;AAOpB,SAAO,GAJS,OAAO,UACnB,OAAO,OAAO,QAAQ,QAAQ,SAAS,MAAM,CAAC,eAC9C,gBAAgB,OAAO,UAAU,QAAQ,SAAS,MAAM,CAAC,eAEzC,KAAK,IAAI;GAC7B;CAEF,MAAM,aAAa,aAAa,SAAS;AAczC,QAAO;EAAE,MAZI;aACF,UAAU;kBACL,WAAW;cAEvB,aACI;QACF,aAAa,KAAK,YAAY,CAAC;UAE7B,UACL;;EAGY;EAAY;;;;;;AAO7B,SAAS,2BAA2B,KAAa,WAA2B;AAG1E,QAAO;aACI,UAAU;kBAHF,qBAAqB,IAAI,CAIjB;;;;;;;;;;AAW7B,SAAS,aACP,OACA,WACA,KACA,SACQ;CACR,MAAM,QAAQ,MAAM,QAAQ;AAE5B,QADsB,OAAO,SAAS,WAAW,CAAC,MAAM,QAEpD,MAAM,OACN,2BAA2B,KAAK,UAAU;;AAOhD,MAAM,yBACJ;AAEF,SAAgB,sBAAsB,KAAqC;CACzE,MAAM,aAAqC,EAAE;CAC7C,MAAM,QAAQ,IAAI,OAChB,8BAA8B,uBAAuB,OACrD,KACD;CACD,MAAM,UAAU,IAAI,SAAS,MAAM;AACnC,MAAK,MAAM,SAAS,SAAS;EAC3B,MAAM,GAAG,WAAW,aAAa;AACjC,aAAW,aAAa,UAAU,aAAa;;AAGjD,QAAO;;AAGT,SAAgB,eAAe,SAAqC;AAClE,SAAQ,SAAS,aAAa,EAA9B;EACE,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK,MACH,QAAO;EACT,KAAK;EACL,KAAK,SACH,QAAO;EACT,KAAK,SACH,QAAO;EACT,KAAK,UACH,QAAO;EACT,KAAK,OACH,QAAO;EACT,KAAK,YACH,QAAO;EACT,KAAK,gBACH,QAAO;EACT,KAAK,SACH,QAAO;EACT,QACE,QAAO;;;;;;;;;;;;AAab,SAAS,0BAA0B,KAAsB;AACvD,KAAI,IAAI,SAAS,KAAK,CAAC,IAAI,WAAW,IAAI,IAAI,CAAC,IAAI,SAAS,IAAI,CAC9D,QAAO;CAET,MAAM,QAAQ,IAAI,MAAM,GAAG,GAAG;AAC9B,KAAI,MAAM,SAAS,KAAK,CAAE,QAAO;AACjC,QAAO,CAAC,MAAM,QAAQ,OAAO,GAAG,CAAC,SAAS,IAAI;;;;;;;;;;;;;;;;;AAkBhD,SAAS,kBACP,SACA,KACe;AACf,SAAQ,SAAS,aAAa,EAA9B;EACE,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK,gBACH,QAAO,0BAA0B,IAAI,GACjC,MACA,IAAI,IAAI,QAAQ,OAAO,OAAO,CAAC,QAAQ,MAAM,KAAK,CAAC;EACzD,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK;EACL,KAAK,SACH,QAAO,qBAAqB,KAAK,IAAI,GAAG,MAAM;EAChD,KAAK,UACH,QAAO,oBAAoB,KAAK,IAAI,GAAG,IAAI,aAAa,GAAG;EAC7D,KAAK,SACH,QAAO,qBAAqB,KAAK,IAAI,GAAG,MAAM;EAChD,QACE,QAAO;;;;;;;;;;;;;;;AAgBb,SAAgB,yBAAyB,KAAqC;CAC5E,MAAM,WAAmC,EAAE;CAO3C,MAAM,QAAQ,IAAI,OAChB,yDAAyD,uBAAuB,kDAChF,MACD;AACD,MAAK,MAAM,SAAS,IAAI,SAAS,MAAM,EAAE;EACvC,MAAM,GAAG,WAAW,WAAW,YAAY;EAC3C,MAAM,YAAY,kBAAkB,WAAW,SAAS;AAGxD,MAAI,cAAc,KAChB,UAAS,aAAa;;AAG1B,QAAO;;;;;;;;;;AAWT,SAAgB,gCAAgC,KAAqB;CACnE,MAAM,kBAAkB,mBAAmB,IAAI;CAC/C,MAAM,iBAAiB,sBAAsB,IAAI;CAEjD,MAAM,iBAAiB;EAAE,GADH,oBAAoB,KAAK,gBAAgB;EACpB,GAAG;EAAgB;CAC9D,MAAM,oBAAoB,yBAAyB,IAAI;AACvD,QAAO,IAAI,QACT,2BACC,UAAU,WAAW,WAAW;AAC/B,MAAI,uBAAuB,QAAQ,gBAAgB,CACjD,QAAO;EAET,MAAM,cAAc,kBAAkB;AACtC,MAAI,gBAAgB,OAClB,QAAO;AAET,SAAO,eAAe,eAAe,WAAW;GAEnD;;;;;;;;;;AAWH,SAAS,mBACP,OACA,KACoC;AACpC,KAAI,CAAC,qBAAqB,KAAK,IAAI,CACjC,QAAO;CAGT,MAAM,SAAS,kBAAkB,KADT,mBAAmB,IAAI,CACO;CACtD,MAAM,WAAW,yBAAyB,IAAI;CAC9C,MAAM,aAAa,OAAO,QACvB,MAAM,CAAC,uBAAuB,SAAS,EAAE,IAAI,SAAS,OAAO,OAC/D;AACD,KAAI,WAAW,WAAW,EACxB,QAAO;CAET,MAAM,UAAU,WAAW;AAC3B,QAAO;EACL,GAAG;EACH,SAAS,GAAG,MAAM,QAAQ,2DAA2D,WAClF,KAAK,MAAM,IAAI,IAAI,CACnB,KACC,KACD,CAAC,sFAAsF,QAAQ;EACnG;;;;;;;;AASH,SAAgB,oBACd,KACA,QACwB;CACxB,MAAM,WAAmC,EAAE;CAC3C,MAAM,kBAAkB,UAAU,mBAAmB,IAAI;AAEzD,MAAK,MAAM,WAAW,iBACpB,MAAK,MAAM,SAAS,IAAI,SAAS,QAAQ,CACvC,KAAI,CAAC,uBAAuB,MAAM,OAAO,gBAAgB,CACvD,UAAS,MAAM,MAAM;AAK3B,QAAO;;;;;;;;;;;;;;;;AAiBT,eAAsB,4BACpB,aACA,aACA,UAII,EAAE,EAC0B;CAChC,MAAM,EACJ,UAAU,OACV,aAAa,iBAAiB,IAC9B,OAAO,mBACL;CACJ,MAAM,cACJ,OAAO,mBAAmB,YAAY,OAAO,SAAS,eAAe,GACjE,KAAK,IAAI,GAAG,KAAK,MAAM,eAAe,CAAC,GACvC;CAGN,MAAM,CAAC,UAAU,SAAS,MAAM,QAAQ,IAAI,CAC1C,GAAG,QAAQ,YAAY,EACvB,UACK;EAAE,SAAS;EAAe,SAAS,EAAE;EAAE,GAGxC,WAAW,CAChB,CAAC;CAEF,MAAM,aAAa,SAAS,QAAQ,SAAS,KAAK,SAAS,OAAO,CAAC;AACnE,QAAO,MAAM,wBAAwB,WAAW,OAAO;CAEvD,MAAM,SAAS,IAAI,gBAAgB,EAAE,CAAC;CACtC,MAAM,UAAU,IAAI,SAAS;CAG7B,MAAM,cAAc,MAAM,QAAQ,IAChC,WAAW,KAAK,SAAS,GAAG,SAAS,KAAK,KAAK,aAAa,KAAK,EAAE,OAAO,CAAC,CAC5E;CAED,MAAM,YAAY,YAAY,KAAK;CAGnC,MAAM,gBAA+D,EAAE;CACvE,MAAM,kBAMD,EAAE;CACP,MAAM,aAQD,EAAE;AAEP,MAAK,IAAI,IAAI,GAAG,IAAI,WAAW,QAAQ,KAAK;EAC1C,MAAM,OAAO,WAAW;EAExB,MAAM,YAAY,mBADF,KAAK,SAAS,MAAM,OAAO,CACE;EAE7C,MAAM,MAAM,YAAY;EACxB,MAAM,UAAU,QAAQ,IAAI;EAE5B,MAAM,SAAS,MAAM,QAAQ;AAC7B,MAAI,UAAU,OAAO,SAAS,WAAW,CAAC,OAAO,OAAO;AACtD,iBAAc,KAAK;IACjB,OAAO;IACP,QAAQ;KAAE,MAAM;KAAW,MAAM,OAAO;KAAM;IAC/C,CAAC;AACF,cAAW,KAAK;IAAE;IAAW,QAAQ;IAAO,CAAC;SACxC;GACL,MAAM,kBAAkB,mBAAmB,IAAI;GAC/C,MAAM,iBAAiB,sBAAsB,IAAI;GAEjD,MAAM,iBAAiB;IAAE,GADH,oBAAoB,KAAK,gBAAgB;IACpB,GAAG;IAAgB;GAI9D,MAAM,oBAAoB,yBAAyB,IAAI;GACvD,MAAM,kBAAkB,gCAAgC,IAAI;GAG5D,MAAM,YAAY,kBAAkB,KAAK,gBAAgB;AACzD,QAAK,MAAM,SAAS,WAAW;AAC7B,QAAI,uBAAuB,SAAS,MAAM,CAAE;AAC5C,QAAI,kBAAkB,OAAQ;AAC9B,QAAI,eAAe,OAAQ;AAC3B,WAAO,KACL,wFACA,WAAW,IACX,OACA,aAAa,MAAM,SACpB;;GAGH,MAAM,aAAa,gBAAgB,MAAM,CAAC,QAAQ,SAAS,GAAG;AAC9D,mBAAgB,KAAK;IAAE,OAAO;IAAG;IAAW;IAAK;IAAS;IAAY,CAAC;;;CAqC3E,MAAM,eAA8D,EAAE;CAGtE,MAAM,eAAmC,EAAE;CAC3C,MAAM,cAAiC,EAAE;AAEzC,KAAI,gBAAgB,SAAS,GAAG;EAM9B,IAAI,WAA+C;EACnD,IAAI,eAAe;AACnB,MAAI,SAAS,eAMX,YAAW;MAEX,KAAI;GACF,MAAM,QAAQ,MAAM,kBAAkB,QAAQ,YAAY;AAC1D,cAAW,gBAAgB,OAAO,KAAK;AACvC,OAAI,aAAa,QACf,gBAAe,aAAa,YAAY,MAAM;AAEhD,OAAI,aAAa,oBAAoB;AAKnC,UAAM,eAAe,QAAQ,YAAY;IACzC,MAAM,QAAQ,MAAM,iBAAiB,QAAQ,aAAa;KACxD,OAAO;KACP,yBAAyB;KAC1B,CAAC;AACF,QAAI,UAAU,UACZ,YAAW;SACN;AACL,gBAAW;AACX,oBAAe,aAAa,YAAY,8BAA8B,MAAM;;;AAGhF,OAAI,aAAa,mBAAmB;IAClC,MAAM,QAAQ,MAAM,iBAAiB,QAAQ,aAAa,EACxD,OAAO,uBACR,CAAC;AACF,QAAI,UAAU,UACZ,YAAW;SACN;AACL,gBAAW;AACX,oBAAe,aAAa,YAAY,8BAA8B,MAAM;;;WAGzE,KAAK;AACZ,OAAI,oBAAoB,IAAI,CAG1B,YAAW;QACN;AAEL,eAAW;AACX,mBAAe,aAAa,YAAY,IAAI,mBAAmB,IAAI;;;AAKzE,MAAI,aAAa,WAAW;GAI1B,MAAM,OAAO,aAAa,UAAU,UAAU;AAC9C,QAAK,MAAM,EAAE,OAAO,WAAW,KAAK,aAAa,iBAAiB;AAChE,iBAAa,KAAK;KAChB;KACA,QAAQ;MACN,MAAM;MACN,MAAM,aAAa,OAAO,WAAW,KAAK,QAAQ;MACnD;KACF,CAAC;AACF,QAAI,aAAa,SAAS;AACxB,iBAAY,KAAK;MAAE,MAAM;MAAW,SAAS;MAAc,CAAC;AAC5D,gBAAW,KAAK;MACd;MACA,QAAQ;MACR;MACA,OAAO,EAAE,SAAS,cAAc;MACjC,CAAC;UAEF,YAAW,KAAK;KAAE;KAAW,QAAQ;KAAQ;KAAM,CAAC;;SAGnD;GACL,IAAI,YAAY;GAChB,MAAM,QAAQ,gBAAgB;AAC9B,WAAQ,MACN,cAAc,MAAM,GAAG,UAAU,IAAI,UAAU,UAAU,MAAM,MAAM,GACtE;GAID,MAAM,iBAAqC,EAAE;GAE7C,MAAM,cAAc,OAAO,EACzB,OACA,WACA,KACA,SACA,iBAC+D;IAK/D,MAAM,SAAS,MAAM,iBACnB,QACA,kBAAkB,cAClB,aACA,eACD;AAED;AACA,YAAQ,OACN,cAAc,MAAM,GAAG,UAAU,IAAI,UAAU,UAAU,IAAI,UAAU,GAAG,MAAM,GACjF;AAED,WAAO,MACL,6CACA,WACA,OAAO,OAAO,OACd,OAAO,QAAQ,YAAY,UAAU,EACtC;AAED,QAAI,OAAO,OAAO,UAAU,UAAU;KAGpC,MAAM,WACJ,OAAO,OAAO,OAAO,WAAW;AAKlC,YAAO;MACL,QAAQ;MACR;MACA,QAAQ;OAAE,MAAM;OAAW,MAJhB,2BAA2B,KAAK,UAAU;OAIpB;MACjC,OAAO,mBAAmB,WAAW,SAAS,EAAE,IAAI;MACrD;;AAGH,QAAI,OAAO,OAAO,UAAU,YAO1B,QAAO;KACL,QAAQ;KACR;KACA,QAAQ;MACN,MAAM;MACN,MAAM,aAAa,OAAO,WAAW,KAAK,QAAQ;MACnD;KACF;IAGH,MAAM,EAAE,MAAM,eAAe,mBAAmB,QAAQ,KAAK,UAAU;AACvE,QAAI,CAAC,WAGH,QAAO;KAAE,QAAQ;KAAS;KAAO,QAAQ;MAAE,MAAM;MAAW;MAAM;KAAE;AAEtE,WAAO;KACL,QAAQ;KACR;KACA,QAAQ;MAAE,MAAM;MAAW;MAAM;KACjC,YAAY;MAAE,MAAM;MAAS;MAAM,OAAO;MAAO;KAClD;;GAIH,MAAM,uBACJ,SACA,gBACG;AACH,SAAK,IAAI,IAAI,GAAG,IAAI,QAAQ,QAAQ,KAAK;KACvC,MAAM,QAAQ,QAAQ;KACtB,MAAM,EAAE,cAAc,gBAAgB,cAAc;AAEpD,SAAI,MAAM,WAAW,aAAa;MAChC,MAAM,MAAM,MAAM;AAClB,mBAAa,KAAK;OAAE,OAAO,IAAI;OAAO,QAAQ,IAAI;OAAQ,CAAC;AAE3D,UAAI,IAAI,WAAW,MAAM;AAEvB,aAAM,QAAQ,aAAa,IAAI;AAC/B,kBAAW,KAAK;QAAE;QAAW,QAAQ;QAAQ,CAAC;iBACrC,IAAI,WAAW,UAAU;AAGlC,oBAAa,KAAK;QAChB,MAAM;QACN,SAAS,IAAI,MAAM;QACpB,CAAC;AACF,kBAAW,KAAK;QACd;QACA,QAAQ;QACR,MAAM;QACN,OAAO,IAAI;QACZ,CAAC;iBACO,IAAI,WAAW,QAExB,YAAW,KAAK;OAAE;OAAW,QAAQ;OAAQ,MAAM;OAAS,CAAC;UAK7D,YAAW,KAAK;OACd;OACA,QAAQ;OACR,MAAM;OACP,CAAC;YAEC;AAKL;AACA,cAAQ,OACN,cAAc,MAAM,GAAG,UAAU,IAAI,UAAU,UAAU,IAAI,UAAU,GAAG,MAAM,GACjF;MAED,MAAM,EAAE,KAAK,SAAS,UAAU,gBAAgB,cAAc;MAC9D,MAAM,SAAS,mBAAmB,MAAM,OAAO;MAC/C,MAAM,QAAQ,WAAW,OAAO;MAChC,MAAM,aAAa,MAAM,QAAQ;MACjC,MAAM,gBACJ,YAAY,SAAS,WAAW,CAAC,WAAW;MAC9C,MAAM,OAAO,aAAa,OAAO,WAAW,KAAK,QAAQ;AACzD,mBAAa,KAAK;OAAE;OAAO,QAAQ;QAAE,MAAM;QAAW;QAAM;OAAE,CAAC;AAE/D,UAAI,CAAC,oBAAoB,MAAM,OAAO,EAAE;AACtC,mBAAY,KAAK;QAAE,MAAM;QAAW,SAAS,MAAM;QAAS,CAAC;AAC7D,kBAAW,KAAK;QACd;QACA,QAAQ;QACR,MAAM;QACN;QACD,CAAC;AACF;;AAGF,aAAO,KACL,wCACA,WACA,QACA,gBACI,6BACA,uCACL;AACD,iBAAW,KAAK;OACd;OACA,QAAQ;OACR,MAAM;OACN;OACD,CAAC;;;;AAKR,OAAI,gBAAgB,SAAS,YAC3B,MAAK,IAAI,IAAI,GAAG,IAAI,gBAAgB,QAAQ,KAAK,aAAa;IAC5D,MAAM,QAAQ,gBAAgB,MAAM,GAAG,IAAI,YAAY;AAEvD,wBADqB,MAAM,QAAQ,WAAW,MAAM,IAAI,YAAY,CAAC,EACnC,EAAE;AACpC,UAAM,UAAU,MAAM;;QAEnB;AAIL,wBAHgB,MAAM,QAAQ,WAC5B,gBAAgB,IAAI,YAAY,CACjC,EAC4B,EAAE;AAC/B,UAAM,UAAU,MAAM;;AAGxB,WAAQ,KAAK,GAAG;;;CAIpB,MAAM,YAAY,YAAY,KAAK,GAAG,aAAa,KAAM,QAAQ,EAAE;AAGnE,KAAI,WAAW,SAAS,GAAG;EACzB,MAAM,aAAa,KAAK,IAAI,GAAG,WAAW,KAAK,MAAM,EAAE,UAAU,OAAO,CAAC;EACzE,MAAM,YAAY,GAAG,IAAI,IAAI,OAAO,GAAG,CAAC;AACxC,UAAQ,IAAI,GAAG;AACf,UAAQ,IACN,KAAK,GAAG,KAAK,kBAAkB,CAAC,GAAG,GAAG,IAAI,IAAI,WAAW,OAAO,GAAG,GACpE;AACD,UAAQ,IAAI,KAAK,YAAY;AAC7B,OAAK,MAAM,SAAS,YAAY;GAC9B,IAAI;AACJ,WAAQ,MAAM,MAAd;IACE,KAAK;AACH,WAAM,GAAG,KAAK,GAAG,IAAI,UAAU,CAAC;AAChC;IACF,KAAK;AACH,WAAM,GAAG,KAAK,GAAG,OAAO,UAAU,CAAC;AACnC;IACF,KAAK;AACH,WAAM,GAAG,IAAI,UAAU;AACvB;IACF,KAAK;AACH,WAAM,GAAG,KAAK,GAAG,IAAI,SAAS,CAAC;AAC/B;IACF,QACE,OACE,MAAM,WAAW,QACb,SAAS,GAAG,KAAK,GAAG,MAAM,QAAQ,CAAC,KACnC,SAAS,GAAG,KAAK,GAAG,OAAO,QAAQ,CAAC;;GAE9C,MAAM,UAAU,MAAM,UAAU,OAAO,WAAW;GAGlD,MAAM,OACJ,MAAM,SAAS,YAAY,MAAM,SAAS,UACtC,GAAG,IAAI,GAAG,cAAc,QAAQ,CAAC,GACjC;GACN,MAAM,YAAY,MAAM,OAAO,QAAQ,MAAM,eAAe,GAAG;GAC/D,MAAM,SAAS,YAAY,KAAK,GAAG,IAAI,UAAU,KAAK;AACtD,WAAQ,IAAI,KAAK,IAAI,IAAI,OAAO,SAAS;;EAE3C,MAAM,WAAW,WAAW,QACzB,MAAM,EAAE,WAAW,UAAU,CAAC,EAAE,KAClC,CAAC;EACF,MAAM,aAAa,WAAW,QAAQ,MAAM,EAAE,WAAW,MAAM,CAAC;EAChE,MAAM,cAAc,WAAW,QAAQ,MAAM,EAAE,SAAS,SAAS,CAAC;EAClE,MAAM,eAAe,WAAW,QAC7B,MAAM,EAAE,SAAS,eACnB,CAAC;EACF,MAAM,aAAa,WAAW,QAAQ,MAAM,EAAE,SAAS,QAAQ,CAAC;EAChE,MAAM,aAAa,WAAW,QAAQ,MAAM,EAAE,SAAS,QAAQ,CAAC;AAChE,UAAQ,IAAI,KAAK,YAAY;EAC7B,MAAM,QAAQ,CAAC,GAAG,SAAS,OAAO,GAAG,WAAW,aAAa;AAC7D,MAAI,cAAc,EAChB,OAAM,KACJ,GAAG,YAAY,OAAO,gBAAgB,IAAI,UAAU,WACrD;AACH,MAAI,eAAe,EAAG,OAAM,KAAK,GAAG,aAAa,WAAW;AAC5D,MAAI,aAAa,EAAG,OAAM,KAAK,GAAG,WAAW,QAAQ;AACrD,MAAI,aAAa,EACf,OAAM,KACJ,GAAG,WAAW,SAAS,eAAe,IAAI,UAAU,WACrD;AACH,UAAQ,IAAI,KAAK,MAAM,KAAK,KAAK,CAAC,IAAI,GAAG,IAAI,GAAG,QAAQ,GAAG,GAAG;AAC9D,UAAQ,IAAI,GAAG;;AAQjB,QAAO;EAAE,SAJO,CAAC,GAAG,eAAe,GAAG,aAAa,CAChD,MAAM,GAAG,MAAM,EAAE,QAAQ,EAAE,MAAM,CACjC,KAAK,MAAM,EAAE,OAAO;EAEL;EAAc;EAAa;;;;;;;AAQ/C,SAAS,mBAAmB,UAA0B;AACpD,QAAO,SAAS,QAAQ,UAAU,GAAG;;;;;;;;;;;;AAavC,SAAgB,kBAAkB,UAA0B;AAC1D,QAAO,SACJ,QAAQ,WAAW,GAAG,CACtB,QAAQ,SAAS,GAAG,CACpB,MAAM,IAAI,CAAC;;;AAIhB,MAAM,UAAkC;CAEtC,QAAQ;CACR,QAAQ;CAER,SAAS;CAET,SAAS;CACT,UAAU;CACV,KAAK;CACL,QAAQ;CACR,OAAO;CACP,QAAQ;CACR,SAAS;CAET,MAAM;CACN,WAAW;CACX,eAAe;CACf,UAAU;CAEV,OAAO;CACP,KAAK;CACL,QAAQ;CACR,QAAQ;CACR,SAAS;CAET,WAAW;CACX,UAAU;CAEV,MAAM;CACP"}
@@ -0,0 +1,116 @@
1
+ import { createLogger } from "../logging/logger.js";
2
+ import { getErrorMessage } from "./errors.js";
3
+
4
+ //#region src/type-generator/statement-result.ts
5
+ const logger = createLogger("type-generator:statement-result");
6
+ /**
7
+ * Normalize a Statement Execution response so downstream parsers can always
8
+ * read rows from `result.data_array`, regardless of the wire format the
9
+ * warehouse chose.
10
+ *
11
+ * `@databricks/sdk-experimental`'s `executeStatement` defaults to an
12
+ * `ARROW_STREAM` disposition. With an `INLINE` disposition the single
13
+ * DESCRIBE row is returned as a base64-encoded Arrow IPC stream in
14
+ * `result.attachment` and `result.data_array` is left undefined. The metric
15
+ * and query type generators only ever read `result.data_array`, so without
16
+ * this normalization an Arrow response reads as "returned no rows" — the
17
+ * registry ships empty and the runtime fail-closed gate 503s every affected
18
+ * metric/query. (A warehouse configured to return `JSON_ARRAY` populates
19
+ * `data_array` directly and needs no decoding — that path, and every mocked
20
+ * test, flows through here unchanged.)
21
+ */
22
+ async function normalizeResultRows(response) {
23
+ if (response.result?.next_chunk_index != null || response.result?.next_chunk_internal_link != null) throw new Error("DESCRIBE result is multi-chunk (truncated); refusing to emit partial types — see next_chunk_index");
24
+ if (response.result?.data_array !== void 0) return response;
25
+ const attachment = response.result?.attachment;
26
+ if (attachment === void 0) return response;
27
+ try {
28
+ const { tableFromIPC } = await import("apache-arrow");
29
+ const dataArray = tableFromIPC(Buffer.from(attachment, "base64")).toArray().map((row) => [...row].map(([, value]) => value == null ? null : String(value)));
30
+ return {
31
+ ...response,
32
+ result: {
33
+ ...response.result,
34
+ data_array: dataArray
35
+ }
36
+ };
37
+ } catch (err) {
38
+ logger.warn("failed to decode ARROW_STREAM DESCRIBE attachment (%s); emitting no rows — metric/query types may degrade", getErrorMessage(err));
39
+ return response;
40
+ }
41
+ }
42
+ /**
43
+ * True when a failure means the warehouse REJECTED the requested result format
44
+ * (so another format is worth trying), not that it ran the statement and hit a
45
+ * real error. There is no single structured signal for this, so we combine two:
46
+ *
47
+ * - `errorCode` — a statement that actually RAN and failed carries a SQL error
48
+ * code (`TABLE_OR_VIEW_NOT_FOUND`, `PARSE_SYNTAX_ERROR`, …); a request-shape
49
+ * rejection comes back as `INVALID_PARAMETER_VALUE`. Any other code means the
50
+ * statement ran, so it is never a format rejection — whatever its message
51
+ * text happens to contain. This gate matters because DESCRIBE runs over
52
+ * user-supplied SQL and analysis errors echo the offending SQL back: a source
53
+ * with columns named `disposition`/`format` would otherwise trip the message
54
+ * match below and get its real diagnostic masked by a pointless retry.
55
+ * - the message signatures — Reyden's `merge_json_arrays` (its `JSON_ARRAY`
56
+ * assembly fails on a `… AS JSON` single-cell result) and standard DBSQL's
57
+ * rejection of `ARROW_STREAM` under an `INLINE` disposition.
58
+ *
59
+ * A genuine SQL error, connectivity failure, or not-ready warehouse does NOT
60
+ * match — those are returned/propagated for the caller's normal handling, never
61
+ * re-tried. `errorCode` is optional: a thrown SDK error carries no reliable
62
+ * structured code, so the catch-path caller passes message only.
63
+ */
64
+ function isFormatRejection(message, errorCode) {
65
+ if (!message) return false;
66
+ if (errorCode && errorCode !== "INVALID_PARAMETER_VALUE") return false;
67
+ const m = message.toLowerCase();
68
+ return m.includes("merge_json_arrays") || m.includes("must be json_array") || m.includes("disposition") && m.includes("format");
69
+ }
70
+ /**
71
+ * Run a DESCRIBE and return a response whose rows are readable via
72
+ * `result.data_array`, adapting to the warehouse's result-format capability.
73
+ *
74
+ * No single format is portable: standard DBSQL (PRO/CLASSIC) serves
75
+ * `INLINE`+`JSON_ARRAY` and rejects `INLINE`+`ARROW_STREAM`; the Reyden engine
76
+ * is the inverse — it rejects `JSON_ARRAY` on a `… AS JSON` result
77
+ * (`merge_json_arrays`) and only returns rows as an `INLINE`+`ARROW_STREAM`
78
+ * attachment. We try `JSON_ARRAY` first (the documented default) and ONLY when
79
+ * the warehouse rejects that format ({@link isFormatRejection}) fall back to
80
+ * `ARROW_STREAM` (decoded by {@link normalizeResultRows}); the accepted format
81
+ * is memoized so the rest of the run skips the probe. Any other outcome —
82
+ * success, SQL error, degrade, connectivity failure — is returned or propagated
83
+ * unchanged, exactly as a single executeStatement would.
84
+ */
85
+ async function describeAdaptive(client, statement, warehouseId, memo) {
86
+ const formats = memo.format ? [memo.format] : ["JSON_ARRAY", "ARROW_STREAM"];
87
+ let lastResponse;
88
+ let lastError;
89
+ for (const format of formats) try {
90
+ const normalized = await normalizeResultRows(await client.statementExecution.executeStatement({
91
+ statement,
92
+ warehouse_id: warehouseId,
93
+ wait_timeout: "30s",
94
+ format,
95
+ disposition: "INLINE"
96
+ }));
97
+ if (normalized.status?.state === "FAILED" && isFormatRejection(normalized.status.error?.message, normalized.status.error?.error_code)) {
98
+ lastResponse = normalized;
99
+ continue;
100
+ }
101
+ if (normalized.status?.state === "SUCCEEDED") memo.format = format;
102
+ return normalized;
103
+ } catch (error) {
104
+ if (isFormatRejection(getErrorMessage(error))) {
105
+ lastError = error;
106
+ continue;
107
+ }
108
+ throw error;
109
+ }
110
+ if (lastResponse !== void 0) return lastResponse;
111
+ throw lastError;
112
+ }
113
+
114
+ //#endregion
115
+ export { describeAdaptive };
116
+ //# sourceMappingURL=statement-result.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"statement-result.js","names":[],"sources":["../../src/type-generator/statement-result.ts"],"sourcesContent":["import type { WorkspaceClient } from \"@databricks/sdk-experimental\";\nimport { createLogger } from \"../logging/logger\";\nimport { getErrorMessage } from \"./errors\";\nimport type { DatabricksStatementExecutionResponse } from \"./types\";\n\nconst logger = createLogger(\"type-generator:statement-result\");\n\n/**\n * Normalize a Statement Execution response so downstream parsers can always\n * read rows from `result.data_array`, regardless of the wire format the\n * warehouse chose.\n *\n * `@databricks/sdk-experimental`'s `executeStatement` defaults to an\n * `ARROW_STREAM` disposition. With an `INLINE` disposition the single\n * DESCRIBE row is returned as a base64-encoded Arrow IPC stream in\n * `result.attachment` and `result.data_array` is left undefined. The metric\n * and query type generators only ever read `result.data_array`, so without\n * this normalization an Arrow response reads as \"returned no rows\" — the\n * registry ships empty and the runtime fail-closed gate 503s every affected\n * metric/query. (A warehouse configured to return `JSON_ARRAY` populates\n * `data_array` directly and needs no decoding — that path, and every mocked\n * test, flows through here unchanged.)\n */\nexport async function normalizeResultRows(\n response: DatabricksStatementExecutionResponse,\n): Promise<DatabricksStatementExecutionResponse> {\n // Truncation guard, above the passthrough so it runs on either transport. A\n // result exceeding INLINE's size limit is paginated (`next_chunk_*` set) and\n // we hold only the first chunk; emitting types from it would cache partial\n // types. A deliberate throw — unlike the best-effort decode below — that both\n // callers catch per-entry as a loud per-key/per-query failure.\n if (\n response.result?.next_chunk_index != null ||\n response.result?.next_chunk_internal_link != null\n ) {\n throw new Error(\n \"DESCRIBE result is multi-chunk (truncated); refusing to emit partial types — see next_chunk_index\",\n );\n }\n\n // Passthrough: rows already materialized (JSON_ARRAY warehouses + every\n // mocked test). `data_array` being an empty array still counts as present —\n // that is a genuine \"no rows\" answer we must not overwrite with a decode.\n if (response.result?.data_array !== undefined) {\n return response;\n }\n\n const attachment = response.result?.attachment;\n if (attachment === undefined) {\n // No rows, no attachment: let the downstream \"no rows\" degrade path fire.\n return response;\n }\n\n try {\n // Lazy import: only pull apache-arrow into the process when an attachment\n // genuinely needs decoding.\n const { tableFromIPC } = await import(\"apache-arrow\");\n const bytes = Buffer.from(attachment, \"base64\");\n const table = tableFromIPC(bytes);\n // Extract each cell in SCHEMA/FIELD order. Spreading a StructRow\n // (`[...row]`) drives apache-arrow's StructRowIterator, which walks the\n // struct's children by positional index and yields `[fieldName, value]`\n // pairs in field order. We deliberately do NOT use `Object.values(row)`\n // nor `row.toArray()`: both funnel through `Object.values` (toArray() is\n // literally `Object.values(this.toJSON())` in apache-arrow@21), and\n // `Object.values` re-sorts integer-like keys ascending per the ECMAScript\n // spec — so an integer-named DESCRIBE column would scramble the positional\n // `[col_name, data_type, comment]` order. The iterator is immune to that.\n const dataArray: (string | null)[][] = table\n .toArray()\n .map((row) =>\n [...(row as Iterable<[unknown, unknown]>)].map(([, value]) =>\n value == null ? null : String(value),\n ),\n );\n\n return {\n ...response,\n result: {\n ...response.result,\n data_array: dataArray,\n },\n };\n } catch (err) {\n // Best-effort: a corrupt/partial Arrow payload — or a missing apache-arrow\n // module — must not crash the pass. Warn so a Reyden user whose decode failed\n // gets a breadcrumb instead of mysteriously-empty types, then return the\n // response unchanged: it routes into the deterministic \"no rows\" degrade.\n logger.warn(\n \"failed to decode ARROW_STREAM DESCRIBE attachment (%s); emitting no rows — metric/query types may degrade\",\n getErrorMessage(err),\n );\n return response;\n }\n}\n\n/** Result format the typegen requests for a DESCRIBE. */\ntype DescribeFormat = \"JSON_ARRAY\" | \"ARROW_STREAM\";\n\n/**\n * Per-path memo of the result format a warehouse accepts for a DESCRIBE shape.\n * Create one per describe path (metric / query) and reuse it across that path's\n * statements: a typegen run targets a single warehouse, so the working format\n * is discovered once and every later DESCRIBE skips the probe. NOT shared\n * across paths — `DESCRIBE QUERY` and `DESCRIBE … AS JSON` can differ (Reyden\n * fails `JSON_ARRAY` only for the single-cell `AS JSON` result).\n */\nexport interface DescribeFormatMemo {\n format?: \"JSON_ARRAY\" | \"ARROW_STREAM\";\n}\n\n/**\n * True when a failure means the warehouse REJECTED the requested result format\n * (so another format is worth trying), not that it ran the statement and hit a\n * real error. There is no single structured signal for this, so we combine two:\n *\n * - `errorCode` — a statement that actually RAN and failed carries a SQL error\n * code (`TABLE_OR_VIEW_NOT_FOUND`, `PARSE_SYNTAX_ERROR`, …); a request-shape\n * rejection comes back as `INVALID_PARAMETER_VALUE`. Any other code means the\n * statement ran, so it is never a format rejection — whatever its message\n * text happens to contain. This gate matters because DESCRIBE runs over\n * user-supplied SQL and analysis errors echo the offending SQL back: a source\n * with columns named `disposition`/`format` would otherwise trip the message\n * match below and get its real diagnostic masked by a pointless retry.\n * - the message signatures — Reyden's `merge_json_arrays` (its `JSON_ARRAY`\n * assembly fails on a `… AS JSON` single-cell result) and standard DBSQL's\n * rejection of `ARROW_STREAM` under an `INLINE` disposition.\n *\n * A genuine SQL error, connectivity failure, or not-ready warehouse does NOT\n * match — those are returned/propagated for the caller's normal handling, never\n * re-tried. `errorCode` is optional: a thrown SDK error carries no reliable\n * structured code, so the catch-path caller passes message only.\n */\nfunction isFormatRejection(\n message: string | undefined,\n errorCode?: string,\n): boolean {\n if (!message) return false;\n // It ran and produced a SQL error — never a format rejection, whatever the\n // message text happens to contain.\n if (errorCode && errorCode !== \"INVALID_PARAMETER_VALUE\") return false;\n const m = message.toLowerCase();\n return (\n m.includes(\"merge_json_arrays\") ||\n m.includes(\"must be json_array\") ||\n (m.includes(\"disposition\") && m.includes(\"format\"))\n );\n}\n\n/**\n * Run a DESCRIBE and return a response whose rows are readable via\n * `result.data_array`, adapting to the warehouse's result-format capability.\n *\n * No single format is portable: standard DBSQL (PRO/CLASSIC) serves\n * `INLINE`+`JSON_ARRAY` and rejects `INLINE`+`ARROW_STREAM`; the Reyden engine\n * is the inverse — it rejects `JSON_ARRAY` on a `… AS JSON` result\n * (`merge_json_arrays`) and only returns rows as an `INLINE`+`ARROW_STREAM`\n * attachment. We try `JSON_ARRAY` first (the documented default) and ONLY when\n * the warehouse rejects that format ({@link isFormatRejection}) fall back to\n * `ARROW_STREAM` (decoded by {@link normalizeResultRows}); the accepted format\n * is memoized so the rest of the run skips the probe. Any other outcome —\n * success, SQL error, degrade, connectivity failure — is returned or propagated\n * unchanged, exactly as a single executeStatement would.\n */\nexport async function describeAdaptive(\n client: WorkspaceClient,\n statement: string,\n warehouseId: string,\n memo: DescribeFormatMemo,\n): Promise<DatabricksStatementExecutionResponse> {\n const formats: DescribeFormat[] = memo.format\n ? [memo.format]\n : [\"JSON_ARRAY\", \"ARROW_STREAM\"];\n let lastResponse: DatabricksStatementExecutionResponse | undefined;\n let lastError: unknown;\n for (const format of formats) {\n try {\n const response = (await client.statementExecution.executeStatement({\n statement,\n warehouse_id: warehouseId,\n // Synchronous wait: without it the call can return PENDING/RUNNING with\n // no rows, which downstream misreads as a no-result degrade.\n wait_timeout: \"30s\",\n format,\n disposition: \"INLINE\",\n })) as DatabricksStatementExecutionResponse;\n const normalized = await normalizeResultRows(response);\n if (\n normalized.status?.state === \"FAILED\" &&\n isFormatRejection(\n normalized.status.error?.message,\n normalized.status.error?.error_code,\n )\n ) {\n lastResponse = normalized;\n continue; // warehouse rejected this format — try the next\n }\n if (normalized.status?.state === \"SUCCEEDED\") {\n memo.format = format;\n }\n return normalized;\n } catch (error) {\n if (isFormatRejection(getErrorMessage(error))) {\n lastError = error;\n continue; // format rejected via a thrown error — try the next\n }\n throw error;\n }\n }\n // Every attempted format was rejected. Surface the last outcome so the caller\n // degrades / reports as usual.\n if (lastResponse !== undefined) return lastResponse;\n throw lastError;\n}\n"],"mappings":";;;;AAKA,MAAM,SAAS,aAAa,kCAAkC;;;;;;;;;;;;;;;;;AAkB9D,eAAsB,oBACpB,UAC+C;AAM/C,KACE,SAAS,QAAQ,oBAAoB,QACrC,SAAS,QAAQ,4BAA4B,KAE7C,OAAM,IAAI,MACR,oGACD;AAMH,KAAI,SAAS,QAAQ,eAAe,OAClC,QAAO;CAGT,MAAM,aAAa,SAAS,QAAQ;AACpC,KAAI,eAAe,OAEjB,QAAO;AAGT,KAAI;EAGF,MAAM,EAAE,iBAAiB,MAAM,OAAO;EAYtC,MAAM,YAVQ,aADA,OAAO,KAAK,YAAY,SAAS,CACd,CAW9B,SAAS,CACT,KAAK,QACJ,CAAC,GAAI,IAAqC,CAAC,KAAK,GAAG,WACjD,SAAS,OAAO,OAAO,OAAO,MAAM,CACrC,CACF;AAEH,SAAO;GACL,GAAG;GACH,QAAQ;IACN,GAAG,SAAS;IACZ,YAAY;IACb;GACF;UACM,KAAK;AAKZ,SAAO,KACL,6GACA,gBAAgB,IAAI,CACrB;AACD,SAAO;;;;;;;;;;;;;;;;;;;;;;;;;AAyCX,SAAS,kBACP,SACA,WACS;AACT,KAAI,CAAC,QAAS,QAAO;AAGrB,KAAI,aAAa,cAAc,0BAA2B,QAAO;CACjE,MAAM,IAAI,QAAQ,aAAa;AAC/B,QACE,EAAE,SAAS,oBAAoB,IAC/B,EAAE,SAAS,qBAAqB,IAC/B,EAAE,SAAS,cAAc,IAAI,EAAE,SAAS,SAAS;;;;;;;;;;;;;;;;;AAmBtD,eAAsB,iBACpB,QACA,WACA,aACA,MAC+C;CAC/C,MAAM,UAA4B,KAAK,SACnC,CAAC,KAAK,OAAO,GACb,CAAC,cAAc,eAAe;CAClC,IAAI;CACJ,IAAI;AACJ,MAAK,MAAM,UAAU,QACnB,KAAI;EAUF,MAAM,aAAa,MAAM,oBATP,MAAM,OAAO,mBAAmB,iBAAiB;GACjE;GACA,cAAc;GAGd,cAAc;GACd;GACA,aAAa;GACd,CAAC,CACoD;AACtD,MACE,WAAW,QAAQ,UAAU,YAC7B,kBACE,WAAW,OAAO,OAAO,SACzB,WAAW,OAAO,OAAO,WAC1B,EACD;AACA,kBAAe;AACf;;AAEF,MAAI,WAAW,QAAQ,UAAU,YAC/B,MAAK,SAAS;AAEhB,SAAO;UACA,OAAO;AACd,MAAI,kBAAkB,gBAAgB,MAAM,CAAC,EAAE;AAC7C,eAAY;AACZ;;AAEF,QAAM;;AAKV,KAAI,iBAAiB,OAAW,QAAO;AACvC,OAAM"}
@@ -1 +1 @@
1
- {"version":3,"file":"types.js","names":[],"sources":["../../src/type-generator/types.ts"],"sourcesContent":["/**\n * Databricks statement execution response interface for DESCRIBE QUERY\n * @property statement_id - the id of the statement\n * @property status - the status of the statement\n * @property result - the result containing column schema as rows [col_name, data_type, comment]\n */\nexport interface DatabricksStatementExecutionResponse {\n statement_id: string;\n status: {\n state: string;\n error?: { error_code?: string; message?: string };\n };\n result?: {\n data_array?: (string | null)[][];\n };\n}\n\n/**\n * Map of SQL types to their corresponding marker types\n * Used to convert SQL types to their corresponding marker types\n */\nexport const sqlTypeToMarker: Record<string, string> = {\n // string\n STRING: \"SQLStringMarker\",\n BINARY: \"SQLBinaryMarker\",\n // boolean\n BOOLEAN: \"SQLBooleanMarker\",\n // numeric\n NUMERIC: \"SQLNumberMarker\",\n INT: \"SQLNumberMarker\",\n BIGINT: \"SQLNumberMarker\",\n TINYINT: \"SQLNumberMarker\",\n SMALLINT: \"SQLNumberMarker\",\n FLOAT: \"SQLNumberMarker\",\n DOUBLE: \"SQLNumberMarker\",\n DECIMAL: \"SQLNumberMarker\",\n // date/time\n DATE: \"SQLDateMarker\",\n TIMESTAMP: \"SQLTimestampMarker\",\n TIMESTAMP_NTZ: \"SQLTimestampMarker\",\n};\n\n/**\n * Map of SQL types to their corresponding helper function names\n * Used to generate JSDoc hints for parameters\n */\nexport const sqlTypeToHelper: Record<string, string> = {\n // string\n STRING: \"sql.string()\",\n BINARY: \"sql.binary()\",\n // boolean\n BOOLEAN: \"sql.boolean()\",\n // numeric — route each SQL type to its closest typed helper. INT/BIGINT\n // are critical for LIMIT/OFFSET; FLOAT/DOUBLE preserve precision intent;\n // NUMERIC/DECIMAL route to sql.numeric() for exact-decimal columns.\n NUMERIC: \"sql.numeric()\",\n DECIMAL: \"sql.numeric()\",\n BIGINT: \"sql.bigint()\",\n INT: \"sql.int()\",\n TINYINT: \"sql.int()\",\n SMALLINT: \"sql.int()\",\n FLOAT: \"sql.float()\",\n DOUBLE: \"sql.double()\",\n // date/time\n DATE: \"sql.date()\",\n TIMESTAMP: \"sql.timestamp()\",\n TIMESTAMP_NTZ: \"sql.timestamp()\",\n};\n\n/**\n * Query schema interface\n * @property name - the name of the query\n * @property type - the type of the query (string, number, boolean, object, array, etc.)\n */\nexport interface QuerySchema {\n name: string;\n type: string;\n}\n\n/**\n * A genuine SQL error: `DESCRIBE QUERY` ran against a *reachable* warehouse and\n * the warehouse reported the statement as FAILED (bad table, syntax error,\n * incompatible type, …). Distinct from a connectivity failure (warehouse\n * unreachable), which is non-fatal and never recorded here.\n * @property name - the query name\n * @property message - the SQL error message reported by the warehouse\n */\nexport interface QuerySyntaxError {\n name: string;\n message: string;\n}\n\n/**\n * A non-SQL fatal error while attempting to describe a query: authentication,\n * authorization, invalid warehouse/configuration, malformed SDK request, or\n * any other setup problem that should not be treated as an offline warehouse.\n * @property name - the query name\n * @property message - the fatal error message\n */\nexport interface QueryFatalError {\n name: string;\n message: string;\n}\n\n/**\n * Result of describing a folder of queries.\n * @property schemas - one schema per query, in original file order. Queries that\n * could not be described carry `result: unknown` so output stays valid.\n * @property syntaxErrors - queries whose DESCRIBE failed against a reachable\n * warehouse (genuine SQL errors). Connectivity failures are deliberately NOT\n * included: they degrade silently (reuse last-known-good type or emit\n * `unknown`) so a transient outage never fails a build.\n * @property fatalErrors - non-SQL fatal describe request failures. These still\n * produce `result: unknown` schemas so callers can write declarations before\n * surfacing the error.\n */\nexport interface QueryGenerationResult {\n schemas: QuerySchema[];\n syntaxErrors: QuerySyntaxError[];\n fatalErrors: QueryFatalError[];\n}\n"],"mappings":";;;;;AAqBA,MAAa,kBAA0C;CAErD,QAAQ;CACR,QAAQ;CAER,SAAS;CAET,SAAS;CACT,KAAK;CACL,QAAQ;CACR,SAAS;CACT,UAAU;CACV,OAAO;CACP,QAAQ;CACR,SAAS;CAET,MAAM;CACN,WAAW;CACX,eAAe;CAChB;;;;;AAMD,MAAa,kBAA0C;CAErD,QAAQ;CACR,QAAQ;CAER,SAAS;CAIT,SAAS;CACT,SAAS;CACT,QAAQ;CACR,KAAK;CACL,SAAS;CACT,UAAU;CACV,OAAO;CACP,QAAQ;CAER,MAAM;CACN,WAAW;CACX,eAAe;CAChB"}
1
+ {"version":3,"file":"types.js","names":[],"sources":["../../src/type-generator/types.ts"],"sourcesContent":["/**\n * Databricks statement execution response interface for DESCRIBE QUERY /\n * DESCRIBE TABLE EXTENDED.\n *\n * Two result shapes matter here:\n * - `result.data_array` — rows already materialized as JSON arrays. Present\n * when the warehouse returns `JSON_ARRAY` (and what every mocked test\n * builds).\n * - `result.attachment` — a base64-encoded Arrow IPC stream. Present when the\n * statement runs with `format: \"ARROW_STREAM\"` + `disposition: \"INLINE\"`,\n * which is the SDK's default disposition. The single row lands here and\n * `data_array` is left undefined. {@link normalizeResultRows} decodes this\n * back into `data_array` so downstream parsers stay shape-agnostic.\n *\n * @property statement_id - the id of the statement\n * @property status - the status of the statement\n * @property manifest - result metadata; `manifest.format` echoes the wire\n * format (`ARROW_STREAM`, `JSON_ARRAY`, ...) the warehouse chose.\n * @property result - the result; either `data_array` (rows as\n * `[col_name, data_type, comment]` arrays) or `attachment` (base64 Arrow IPC)\n */\nexport interface DatabricksStatementExecutionResponse {\n statement_id: string;\n status: {\n state: string;\n error?: { error_code?: string; message?: string };\n };\n manifest?: {\n format?: string;\n };\n result?: {\n data_array?: (string | null)[][];\n /** Base64-encoded Arrow IPC stream (ARROW_STREAM + INLINE disposition). */\n attachment?: string;\n /**\n * Set when the result spans multiple chunks (rows exceeded INLINE's size\n * limit). Its presence means this response holds only the FIRST chunk;\n * {@link normalizeResultRows} throws rather than emit truncated types.\n */\n next_chunk_index?: number;\n /** Companion to {@link next_chunk_index}: link to fetch the next chunk. */\n next_chunk_internal_link?: string;\n };\n}\n\n/**\n * Map of SQL types to their corresponding marker types\n * Used to convert SQL types to their corresponding marker types\n */\nexport const sqlTypeToMarker: Record<string, string> = {\n // string\n STRING: \"SQLStringMarker\",\n BINARY: \"SQLBinaryMarker\",\n // boolean\n BOOLEAN: \"SQLBooleanMarker\",\n // numeric\n NUMERIC: \"SQLNumberMarker\",\n INT: \"SQLNumberMarker\",\n BIGINT: \"SQLNumberMarker\",\n TINYINT: \"SQLNumberMarker\",\n SMALLINT: \"SQLNumberMarker\",\n FLOAT: \"SQLNumberMarker\",\n DOUBLE: \"SQLNumberMarker\",\n DECIMAL: \"SQLNumberMarker\",\n // date/time\n DATE: \"SQLDateMarker\",\n TIMESTAMP: \"SQLTimestampMarker\",\n TIMESTAMP_NTZ: \"SQLTimestampMarker\",\n};\n\n/**\n * Map of SQL types to their corresponding helper function names\n * Used to generate JSDoc hints for parameters\n */\nexport const sqlTypeToHelper: Record<string, string> = {\n // string\n STRING: \"sql.string()\",\n BINARY: \"sql.binary()\",\n // boolean\n BOOLEAN: \"sql.boolean()\",\n // numeric — route each SQL type to its closest typed helper. INT/BIGINT\n // are critical for LIMIT/OFFSET; FLOAT/DOUBLE preserve precision intent;\n // NUMERIC/DECIMAL route to sql.numeric() for exact-decimal columns.\n NUMERIC: \"sql.numeric()\",\n DECIMAL: \"sql.numeric()\",\n BIGINT: \"sql.bigint()\",\n INT: \"sql.int()\",\n TINYINT: \"sql.int()\",\n SMALLINT: \"sql.int()\",\n FLOAT: \"sql.float()\",\n DOUBLE: \"sql.double()\",\n // date/time\n DATE: \"sql.date()\",\n TIMESTAMP: \"sql.timestamp()\",\n TIMESTAMP_NTZ: \"sql.timestamp()\",\n};\n\n/**\n * Query schema interface\n * @property name - the name of the query\n * @property type - the type of the query (string, number, boolean, object, array, etc.)\n */\nexport interface QuerySchema {\n name: string;\n type: string;\n}\n\n/**\n * A genuine SQL error: `DESCRIBE QUERY` ran against a *reachable* warehouse and\n * the warehouse reported the statement as FAILED (bad table, syntax error,\n * incompatible type, …). Distinct from a connectivity failure (warehouse\n * unreachable), which is non-fatal and never recorded here.\n * @property name - the query name\n * @property message - the SQL error message reported by the warehouse\n */\nexport interface QuerySyntaxError {\n name: string;\n message: string;\n}\n\n/**\n * A non-SQL fatal error while attempting to describe a query: authentication,\n * authorization, invalid warehouse/configuration, malformed SDK request, or\n * any other setup problem that should not be treated as an offline warehouse.\n * @property name - the query name\n * @property message - the fatal error message\n */\nexport interface QueryFatalError {\n name: string;\n message: string;\n}\n\n/**\n * Result of describing a folder of queries.\n * @property schemas - one schema per query, in original file order. Queries that\n * could not be described carry `result: unknown` so output stays valid.\n * @property syntaxErrors - queries whose DESCRIBE failed against a reachable\n * warehouse (genuine SQL errors). Connectivity failures are deliberately NOT\n * included: they degrade silently (reuse last-known-good type or emit\n * `unknown`) so a transient outage never fails a build.\n * @property fatalErrors - non-SQL fatal describe request failures. These still\n * produce `result: unknown` schemas so callers can write declarations before\n * surfacing the error.\n */\nexport interface QueryGenerationResult {\n schemas: QuerySchema[];\n syntaxErrors: QuerySyntaxError[];\n fatalErrors: QueryFatalError[];\n}\n"],"mappings":";;;;;AAiDA,MAAa,kBAA0C;CAErD,QAAQ;CACR,QAAQ;CAER,SAAS;CAET,SAAS;CACT,KAAK;CACL,QAAQ;CACR,SAAS;CACT,UAAU;CACV,OAAO;CACP,QAAQ;CACR,SAAS;CAET,MAAM;CACN,WAAW;CACX,eAAe;CAChB;;;;;AAMD,MAAa,kBAA0C;CAErD,QAAQ;CACR,QAAQ;CAER,SAAS;CAIT,SAAS;CACT,SAAS;CACT,QAAQ;CACR,KAAK;CACL,SAAS;CACT,UAAU;CACV,OAAO;CACP,QAAQ;CAER,MAAM;CACN,WAAW;CACX,eAAe;CAChB"}
@@ -6,6 +6,18 @@ import { Plugin } from "vite";
6
6
  */
7
7
  interface AppKitTypesPluginOptions {
8
8
  outFile?: string;
9
+ /**
10
+ * Path to the metric registry d.ts file (relative to client folder).
11
+ * Defaults to a sibling of `outFile`, computed by the generator.
12
+ */
13
+ mvOutFile?: string;
14
+ /**
15
+ * Path to the metric semantic-metadata JSON file (relative to client folder).
16
+ * Build-time artifact — defaults to a sibling of {@link mvOutFile}
17
+ * (itself a sibling of `outFile`), computed by the generator. Skipped
18
+ * automatically when `metric-views.json` is absent.
19
+ */
20
+ mvMetadataOutFile?: string;
9
21
  /** Folders to watch for changes. */
10
22
  watchFolders?: string[];
11
23
  }
@@ -1 +1 @@
1
- {"version":3,"file":"vite-plugin.d.ts","names":[],"sources":["../../src/type-generator/vite-plugin.ts"],"mappings":";;;;;AAGmC;UA6BzB,wBAAA;EAER,OAAA;EAAA;EAEA,YAAA;AAAA;;;;;;;iBASc,iBAAA,CAAkB,OAAA,GAAU,wBAAA,GAA2B,MAAA"}
1
+ {"version":3,"file":"vite-plugin.d.ts","names":[],"sources":["../../src/type-generator/vite-plugin.ts"],"mappings":";;;;;AAGmC;UA6BzB,wBAAA;EAER,OAAA;EAFgC;;;;EAOhC,SAAA;EASY;;AASd;;;;EAXE,iBAAA;EAWgC;EAThC,YAAA;AAAA;;;;;;;iBASc,iBAAA,CAAkB,OAAA,GAAU,wBAAA,GAA2B,MAAA"}
@@ -22,6 +22,8 @@ const DEV_WAREHOUSE_WATCH_MAX_MS = 6e4;
22
22
  */
23
23
  function appKitTypesPlugin(options) {
24
24
  let outFile;
25
+ let mvOutFile;
26
+ let mvMetadataOutFile;
25
27
  let watchFolders;
26
28
  let inFlight = null;
27
29
  let queued = false;
@@ -50,7 +52,9 @@ function appKitTypesPlugin(options) {
50
52
  queryFolder: watchFolders[0],
51
53
  warehouseId,
52
54
  noCache: false,
53
- mode
55
+ mode,
56
+ mvOutFile,
57
+ mvMetadataOutFile
54
58
  });
55
59
  } catch (error) {
56
60
  const isTypegenError = error instanceof TypegenSyntaxError || error instanceof TypegenFatalError;
@@ -106,9 +110,9 @@ function appKitTypesPlugin(options) {
106
110
  *
107
111
  * Post-probe behaviour by state:
108
112
  * - RUNNING → describe right away (the dev foreground degraded, so a running
109
- * warehouse would otherwise never get real types — this is the case Phase 3
110
- * restores). `waitUntilRunning` returns immediately for an already-running
111
- * warehouse, then the blocking regenerate fires.
113
+ * warehouse would otherwise never get real types). `waitUntilRunning`
114
+ * returns immediately for an already-running warehouse, then the blocking
115
+ * regenerate fires.
112
116
  * - STARTING → it's already coming up; just wait for RUNNING, then describe.
113
117
  * - STOPPED / STOPPING → kick off a start, wait for RUNNING, then describe.
114
118
  * - DELETED / DELETING → return (a deleted warehouse can't be started, and
@@ -171,6 +175,8 @@ function appKitTypesPlugin(options) {
171
175
  configResolved(config) {
172
176
  const projectRoot = path.resolve(config.root, "..");
173
177
  outFile = path.resolve(projectRoot, options?.outFile ?? `shared/${TYPES_DIR}/${ANALYTICS_TYPES_FILE}`);
178
+ mvOutFile = options?.mvOutFile !== void 0 ? path.resolve(projectRoot, options.mvOutFile) : void 0;
179
+ mvMetadataOutFile = options?.mvMetadataOutFile !== void 0 ? path.resolve(projectRoot, options.mvMetadataOutFile) : void 0;
174
180
  watchFolders = options?.watchFolders ?? [path.join(process.cwd(), "config", "queries")];
175
181
  },
176
182
  buildStart() {
@@ -181,7 +187,7 @@ function appKitTypesPlugin(options) {
181
187
  configureServer(server) {
182
188
  server.watcher.add(watchFolders);
183
189
  server.watcher.on("change", (changedFile) => {
184
- if (watchFolders.some((folder) => changedFile.startsWith(folder)) && changedFile.endsWith(".sql")) {
190
+ if (watchFolders.some((folder) => changedFile.startsWith(folder)) && (changedFile.endsWith(".sql") || path.basename(changedFile) === "metric-views.json")) {
185
191
  runGenerate("non-blocking");
186
192
  armWarehouseWatch();
187
193
  }
@@ -1 +1 @@
1
- {"version":3,"file":"vite-plugin.js","names":[],"sources":["../../src/type-generator/vite-plugin.ts"],"sourcesContent":["import { existsSync } from \"node:fs\";\nimport path from \"node:path\";\nimport { WorkspaceClient } from \"@databricks/sdk-experimental\";\nimport type { Plugin } from \"vite\";\nimport { createLogger } from \"../logging/logger\";\nimport {\n ANALYTICS_TYPES_FILE,\n generateFromEntryPoint,\n TYPES_DIR,\n TypegenFatalError,\n TypegenSyntaxError,\n} from \"./index\";\nimport type { PreflightMode } from \"./preflight\";\nimport {\n getWarehouseState,\n startWarehouse,\n waitUntilRunning,\n} from \"./warehouse-status\";\n\nconst logger = createLogger(\"type-generator:vite-plugin\");\n\n/**\n * How long the DEV background watcher waits for a STARTING warehouse to reach\n * RUNNING before giving up. Short relative to the CLI's preflight budget: this\n * is a best-effort \"regenerate once the warehouse warms up\" convenience, not a\n * gate, so we'd rather stop polling than hold a detached task open for minutes.\n */\nconst DEV_WAREHOUSE_WATCH_MAX_MS = 60_000;\n\n/**\n * Options for the AppKit types plugin.\n */\ninterface AppKitTypesPluginOptions {\n /* Path to the output d.ts file (relative to client folder). */\n outFile?: string;\n /** Folders to watch for changes. */\n watchFolders?: string[];\n}\n\n/**\n * Vite plugin to generate types for AppKit queries.\n * Calls generateFromEntryPoint under the hood.\n * @param options - Options to override default values.\n * @returns Vite plugin to generate types for AppKit queries.\n */\nexport function appKitTypesPlugin(options?: AppKitTypesPluginOptions): Plugin {\n let outFile: string;\n let watchFolders: string[];\n\n // Single-flight state for runGenerate(). `inFlight` is the promise of the\n // currently-running drain (null when idle); `queued` records that a trigger\n // arrived while a run was active so exactly ONE trailing run fires afterwards\n // (latest-wins — coalesces any number of overlapping triggers into a single\n // rerun). `queued` is read/cleared synchronously inside the drain loop so a\n // trigger landing in any window is caught before the drain exits.\n //\n // `pendingMode` is the mode the next generate should run in (latest-wins, like\n // `queued`): the foreground build runs non-blocking in dev (instant degrade)\n // while the background warehouse watch runs blocking (real DESCRIBEs). A\n // blocking watch trigger that lands while a non-blocking foreground run is in\n // flight therefore still describes when its trailing run fires.\n let inFlight: Promise<void> | null = null;\n let queued = false;\n let pendingMode: PreflightMode = \"non-blocking\";\n\n // The currently-armed DEV background warehouse watch, if any. Aborting it\n // stops a pending waitUntilRunning (server shutdown, or a newer arm replacing\n // an older one).\n let watchController: AbortController | null = null;\n\n /**\n * Generate types once in the given preflight {@link PreflightMode}. Never\n * throws in dev (logs instead); in production it rethrows so the build fails.\n * This is the un-guarded core — callers should go through {@link runGenerate}\n * so concurrent triggers can't race-write the .d.ts.\n *\n * @param mode - preflight policy for this run. The foreground build passes a\n * NODE_ENV-derived mode (blocking in production, non-blocking in dev so it\n * degrades instantly); the background warehouse watch passes \"blocking\" so\n * its regenerate actually DESCRIBEs and lands real (non-degraded) types.\n */\n async function generateOnce(mode: PreflightMode) {\n try {\n const warehouseId = process.env.DATABRICKS_WAREHOUSE_ID || \"\";\n\n if (!warehouseId) {\n logger.debug(\"Warehouse ID not found. Skipping type generation.\");\n return;\n }\n\n await generateFromEntryPoint({\n outFile,\n queryFolder: watchFolders[0],\n warehouseId,\n noCache: false,\n mode,\n });\n } catch (error) {\n // TypegenSyntaxError / TypegenFatalError carry a complete, actionable\n // report in their message. Their stack frames and attached query arrays\n // point into appkit internals and only add noise, so surface just the\n // message — both when failing the prod build and when logging in dev.\n const isTypegenError =\n error instanceof TypegenSyntaxError ||\n error instanceof TypegenFatalError;\n\n // throw in production to fail the build\n if (process.env.NODE_ENV === \"production\") {\n if (isTypegenError) error.stack = error.message;\n throw error;\n }\n\n if (isTypegenError) {\n logger.error(\"%s\", error.message);\n } else {\n logger.error(\"Error generating types: %O\", error);\n }\n }\n }\n\n /**\n * Single-flight wrapper around {@link generateOnce}. The initial build, the\n * .sql watcher, and the DEV warehouse watch all route through here so they can\n * never run typegen concurrently (which would race-write the .d.ts).\n *\n * If a run is already in flight, this does NOT start a second one — it records\n * the requested mode and sets a trailing flag so exactly one more run fires\n * after the current finishes, coalescing any number of overlapping triggers\n * (latest-wins, including the mode: a blocking watch trigger that arrives mid\n * non-blocking foreground run still describes when its trailing run fires).\n *\n * @param mode - preflight policy for this run. Recorded into `pendingMode`,\n * which the drain reads for each generate (latest trigger wins).\n * @returns A promise that resolves when this trigger's work (including any\n * trailing run it scheduled) has completed.\n */\n function runGenerate(mode: PreflightMode): Promise<void> {\n pendingMode = mode;\n\n if (inFlight) {\n // A run is active: remember that another trigger arrived and ride out the\n // current run. One trailing run then covers all coalesced triggers and\n // runs in the latest requested mode (recorded above).\n queued = true;\n return inFlight;\n }\n\n // Drain in a loop rather than recursing after a single queued-check: a\n // trigger can land in the window between generateOnce() resolving and the\n // check, so we re-test `queued` until it's clear. Critically, `inFlight` is\n // cleared synchronously in the SAME tick as the final `queued === false`\n // observation — never deferred to a .finally microtask — so there's no\n // window where a trigger sees `inFlight` set but the drain has already\n // decided to exit. The guard stays held for the whole drain, so concurrent\n // triggers only ever set the flag; they never start a parallel generate.\n const drain = async (): Promise<void> => {\n while (true) {\n queued = false;\n // Snapshot the mode synchronously alongside clearing `queued` so a\n // trigger landing during this generate is observed (via `queued`) on the\n // next loop with its own mode, not silently dropped.\n const runMode = pendingMode;\n await generateOnce(runMode);\n // Synchronous check + clear, atomic w.r.t. other (synchronous) callers.\n if (!queued) {\n inFlight = null;\n return;\n }\n }\n };\n\n inFlight = drain();\n return inFlight;\n }\n\n /**\n * DEV-only: get the warehouse to RUNNING in the background and regenerate with\n * real (non-degraded) types once it is — without blocking dev startup. The\n * foreground build only ever degrades in dev (instant `unknown`/cached types),\n * so this is what lands actual DESCRIBE results in the editor for EVERY\n * reachable warehouse state, not just one that happens to already be warm.\n *\n * Post-probe behaviour by state:\n * - RUNNING → describe right away (the dev foreground degraded, so a running\n * warehouse would otherwise never get real types — this is the case Phase 3\n * restores). `waitUntilRunning` returns immediately for an already-running\n * warehouse, then the blocking regenerate fires.\n * - STARTING → it's already coming up; just wait for RUNNING, then describe.\n * - STOPPED / STOPPING → kick off a start, wait for RUNNING, then describe.\n * - DELETED / DELETING → return (a deleted warehouse can't be started, and\n * blocking typegen would treat it as fatal); leave the degraded types.\n *\n * No-op in production or without a warehouse id. Replaces any previously-armed\n * watch (aborting it first). Fully self-contained: it never throws into the\n * caller and never re-arms itself. The whole lifecycle is abortable via the\n * shared {@link watchController} — its signal is threaded into\n * `waitUntilRunning`, so a dev-server shutdown cancels a pending wait — and the\n * regenerate routes through {@link runGenerate} so it can't race-write the\n * .d.ts with the foreground degrade or a `.sql` re-trigger.\n *\n * The regenerate runs in \"blocking\" mode (not the foreground's non-blocking)\n * so it actually DESCRIBEs the now-RUNNING warehouse and lands real types —\n * the whole point of warming the warehouse in the background.\n */\n function armWarehouseWatch(): void {\n if (process.env.NODE_ENV === \"production\") return;\n\n const warehouseId = process.env.DATABRICKS_WAREHOUSE_ID || \"\";\n if (!warehouseId) return;\n\n // Supersede any in-flight watch so we never run two concurrently.\n watchController?.abort();\n const controller = new AbortController();\n watchController = controller;\n const { signal } = controller;\n\n void (async () => {\n try {\n const client = new WorkspaceClient({});\n const state = await getWarehouseState(client, warehouseId);\n\n // A deleted/deleting warehouse can't be started and blocking typegen\n // would treat it as fatal — leave the degraded types and stop. Every\n // other state (including RUNNING) proceeds to wait-then-describe so the\n // dev editor gets real types, not just the foreground's degraded ones.\n if (state === \"DELETED\" || state === \"DELETING\") {\n return;\n }\n\n // Stopped/stopping won't reach RUNNING on its own — nudge it. RUNNING and\n // STARTING need no start (RUNNING is already up; STARTING is coming up),\n // so don't issue a redundant one. A failed start is non-fatal: give up\n // silently rather than throw out of the detached task (the developer\n // still has degraded/cached types).\n let startedByUs = false;\n if (state === \"STOPPED\" || state === \"STOPPING\") {\n try {\n logger.debug(\"Warehouse is %s; starting it.\", state);\n await startWarehouse(client, warehouseId);\n startedByUs = true;\n } catch {\n return;\n }\n }\n\n // Wait for RUNNING. For an already-RUNNING warehouse this returns on the\n // first poll; for STARTING/STOPPED it polls (abortably) until the\n // warehouse warms up, a terminal state, or the deadline.\n const final = await waitUntilRunning(client, warehouseId, {\n maxMs: DEV_WAREHOUSE_WATCH_MAX_MS,\n signal,\n // We just issued the start, so the first poll(s) often still report\n // STOPPED/STOPPING before the start propagates. Poll through those\n // instead of bailing, or the regenerate would never fire. When we\n // didn't start it (RUNNING/STARTING branch), keep the default terminal\n // states.\n treatStoppedAsTransient: startedByUs,\n });\n\n if (final === \"RUNNING\" && !signal.aborted) {\n logger.debug(\"Warehouse is RUNNING; regenerating types.\");\n // Blocking: the warehouse is RUNNING now, so describe it and emit real\n // (non-degraded) types — unlike the foreground dev run, which degraded.\n // Routed through the single-flight guard so it coalesces with the\n // foreground degrade / any `.sql` re-trigger instead of racing them.\n await runGenerate(\"blocking\");\n }\n } catch {\n // Detached background task: any failure (timeout, abort, connectivity,\n // auth) is non-fatal — the developer still has degraded/cached types.\n }\n })();\n }\n\n return {\n name: \"appkit-types\",\n\n apply() {\n const warehouseId = process.env.DATABRICKS_WAREHOUSE_ID || \"\";\n\n if (!warehouseId) {\n logger.debug(\"Warehouse ID not found. Skipping type generation.\");\n return false;\n }\n\n if (!existsSync(path.join(process.cwd(), \"config\", \"queries\"))) {\n return false;\n }\n\n return true;\n },\n\n configResolved(config) {\n const projectRoot = path.resolve(config.root, \"..\");\n outFile = path.resolve(\n projectRoot,\n options?.outFile ?? `shared/${TYPES_DIR}/${ANALYTICS_TYPES_FILE}`,\n );\n watchFolders = options?.watchFolders ?? [\n path.join(process.cwd(), \"config\", \"queries\"),\n ];\n },\n\n buildStart() {\n // Production: block the build on this generate (and surface failures).\n // The watch is a dev-only no-op, so just run typegen.\n if (process.env.NODE_ENV === \"production\") {\n return runGenerate(\"blocking\");\n }\n\n // Dev: don't block startup waiting on typegen. The foreground generate runs\n // non-blocking — it skips the warehouse entirely and writes degraded\n // (cached/`unknown`) types instantly. Then arm the warehouse watch so the\n // warehouse gets a one-shot BLOCKING regenerate (real types) in the\n // background for EVERY reachable state: RUNNING describes right away, while\n // STARTING/STOPPED are waited (and started) until they reach RUNNING.\n void runGenerate(\"non-blocking\");\n armWarehouseWatch();\n },\n\n configureServer(server) {\n server.watcher.add(watchFolders);\n\n server.watcher.on(\"change\", (changedFile) => {\n const isWatchedFile = watchFolders.some((folder) =>\n changedFile.startsWith(folder),\n );\n\n if (isWatchedFile && changedFile.endsWith(\".sql\")) {\n // Route through the single-flight runner (was fire-and-forget\n // generate(), which could race the initial build / watch). This is a\n // dev-only hook, so degrade instantly (non-blocking), then re-arm the\n // warehouse watch so the edited query is re-described in the background\n // against the running warehouse (or once a still-starting one warms\n // up), landing fresh blocking-described types.\n void runGenerate(\"non-blocking\");\n armWarehouseWatch();\n }\n });\n\n // Tear down any pending warehouse watch when the dev server closes so a\n // long backoff can't keep the process alive after shutdown.\n server.httpServer?.once(\"close\", () => {\n watchController?.abort();\n });\n },\n };\n}\n"],"mappings":";;;;;;;;AAmBA,MAAM,SAAS,aAAa,6BAA6B;;;;;;;AAQzD,MAAM,6BAA6B;;;;;;;AAkBnC,SAAgB,kBAAkB,SAA4C;CAC5E,IAAI;CACJ,IAAI;CAcJ,IAAI,WAAiC;CACrC,IAAI,SAAS;CACb,IAAI,cAA6B;CAKjC,IAAI,kBAA0C;;;;;;;;;;;;CAa9C,eAAe,aAAa,MAAqB;AAC/C,MAAI;GACF,MAAM,cAAc,QAAQ,IAAI,2BAA2B;AAE3D,OAAI,CAAC,aAAa;AAChB,WAAO,MAAM,oDAAoD;AACjE;;AAGF,SAAM,uBAAuB;IAC3B;IACA,aAAa,aAAa;IAC1B;IACA,SAAS;IACT;IACD,CAAC;WACK,OAAO;GAKd,MAAM,iBACJ,iBAAiB,sBACjB,iBAAiB;AAGnB,OAAI,QAAQ,IAAI,aAAa,cAAc;AACzC,QAAI,eAAgB,OAAM,QAAQ,MAAM;AACxC,UAAM;;AAGR,OAAI,eACF,QAAO,MAAM,MAAM,MAAM,QAAQ;OAEjC,QAAO,MAAM,8BAA8B,MAAM;;;;;;;;;;;;;;;;;;;CAqBvD,SAAS,YAAY,MAAoC;AACvD,gBAAc;AAEd,MAAI,UAAU;AAIZ,YAAS;AACT,UAAO;;EAWT,MAAM,QAAQ,YAA2B;AACvC,UAAO,MAAM;AACX,aAAS;AAKT,UAAM,aADU,YACW;AAE3B,QAAI,CAAC,QAAQ;AACX,gBAAW;AACX;;;;AAKN,aAAW,OAAO;AAClB,SAAO;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CAgCT,SAAS,oBAA0B;AACjC,MAAI,QAAQ,IAAI,aAAa,aAAc;EAE3C,MAAM,cAAc,QAAQ,IAAI,2BAA2B;AAC3D,MAAI,CAAC,YAAa;AAGlB,mBAAiB,OAAO;EACxB,MAAM,aAAa,IAAI,iBAAiB;AACxC,oBAAkB;EAClB,MAAM,EAAE,WAAW;AAEnB,GAAM,YAAY;AAChB,OAAI;IACF,MAAM,SAAS,IAAI,gBAAgB,EAAE,CAAC;IACtC,MAAM,QAAQ,MAAM,kBAAkB,QAAQ,YAAY;AAM1D,QAAI,UAAU,aAAa,UAAU,WACnC;IAQF,IAAI,cAAc;AAClB,QAAI,UAAU,aAAa,UAAU,WACnC,KAAI;AACF,YAAO,MAAM,iCAAiC,MAAM;AACpD,WAAM,eAAe,QAAQ,YAAY;AACzC,mBAAc;YACR;AACN;;AAkBJ,QAXc,MAAM,iBAAiB,QAAQ,aAAa;KACxD,OAAO;KACP;KAMA,yBAAyB;KAC1B,CAAC,KAEY,aAAa,CAAC,OAAO,SAAS;AAC1C,YAAO,MAAM,4CAA4C;AAKzD,WAAM,YAAY,WAAW;;WAEzB;MAIN;;AAGN,QAAO;EACL,MAAM;EAEN,QAAQ;AAGN,OAAI,EAFgB,QAAQ,IAAI,2BAA2B,KAEzC;AAChB,WAAO,MAAM,oDAAoD;AACjE,WAAO;;AAGT,OAAI,CAAC,WAAW,KAAK,KAAK,QAAQ,KAAK,EAAE,UAAU,UAAU,CAAC,CAC5D,QAAO;AAGT,UAAO;;EAGT,eAAe,QAAQ;GACrB,MAAM,cAAc,KAAK,QAAQ,OAAO,MAAM,KAAK;AACnD,aAAU,KAAK,QACb,aACA,SAAS,WAAW,UAAU,UAAU,GAAG,uBAC5C;AACD,kBAAe,SAAS,gBAAgB,CACtC,KAAK,KAAK,QAAQ,KAAK,EAAE,UAAU,UAAU,CAC9C;;EAGH,aAAa;AAGX,OAAI,QAAQ,IAAI,aAAa,aAC3B,QAAO,YAAY,WAAW;AAShC,GAAK,YAAY,eAAe;AAChC,sBAAmB;;EAGrB,gBAAgB,QAAQ;AACtB,UAAO,QAAQ,IAAI,aAAa;AAEhC,UAAO,QAAQ,GAAG,WAAW,gBAAgB;AAK3C,QAJsB,aAAa,MAAM,WACvC,YAAY,WAAW,OAAO,CAC/B,IAEoB,YAAY,SAAS,OAAO,EAAE;AAOjD,KAAK,YAAY,eAAe;AAChC,wBAAmB;;KAErB;AAIF,UAAO,YAAY,KAAK,eAAe;AACrC,qBAAiB,OAAO;KACxB;;EAEL"}
1
+ {"version":3,"file":"vite-plugin.js","names":[],"sources":["../../src/type-generator/vite-plugin.ts"],"sourcesContent":["import { existsSync } from \"node:fs\";\nimport path from \"node:path\";\nimport { WorkspaceClient } from \"@databricks/sdk-experimental\";\nimport type { Plugin } from \"vite\";\nimport { createLogger } from \"../logging/logger\";\nimport {\n ANALYTICS_TYPES_FILE,\n generateFromEntryPoint,\n TYPES_DIR,\n TypegenFatalError,\n TypegenSyntaxError,\n} from \"./index\";\nimport type { PreflightMode } from \"./preflight\";\nimport {\n getWarehouseState,\n startWarehouse,\n waitUntilRunning,\n} from \"./warehouse-status\";\n\nconst logger = createLogger(\"type-generator:vite-plugin\");\n\n/**\n * How long the DEV background watcher waits for a STARTING warehouse to reach\n * RUNNING before giving up. Short relative to the CLI's preflight budget: this\n * is a best-effort \"regenerate once the warehouse warms up\" convenience, not a\n * gate, so we'd rather stop polling than hold a detached task open for minutes.\n */\nconst DEV_WAREHOUSE_WATCH_MAX_MS = 60_000;\n\n/**\n * Options for the AppKit types plugin.\n */\ninterface AppKitTypesPluginOptions {\n /* Path to the output d.ts file (relative to client folder). */\n outFile?: string;\n /**\n * Path to the metric registry d.ts file (relative to client folder).\n * Defaults to a sibling of `outFile`, computed by the generator.\n */\n mvOutFile?: string;\n /**\n * Path to the metric semantic-metadata JSON file (relative to client folder).\n * Build-time artifact — defaults to a sibling of {@link mvOutFile}\n * (itself a sibling of `outFile`), computed by the generator. Skipped\n * automatically when `metric-views.json` is absent.\n */\n mvMetadataOutFile?: string;\n /** Folders to watch for changes. */\n watchFolders?: string[];\n}\n\n/**\n * Vite plugin to generate types for AppKit queries.\n * Calls generateFromEntryPoint under the hood.\n * @param options - Options to override default values.\n * @returns Vite plugin to generate types for AppKit queries.\n */\nexport function appKitTypesPlugin(options?: AppKitTypesPluginOptions): Plugin {\n let outFile: string;\n let mvOutFile: string | undefined;\n let mvMetadataOutFile: string | undefined;\n let watchFolders: string[];\n\n // Single-flight state for runGenerate(). `inFlight` is the promise of the\n // currently-running drain (null when idle); `queued` records that a trigger\n // arrived while a run was active so exactly ONE trailing run fires afterwards\n // (latest-wins — coalesces any number of overlapping triggers into a single\n // rerun). `queued` is read/cleared synchronously inside the drain loop so a\n // trigger landing in any window is caught before the drain exits.\n //\n // `pendingMode` is the mode the next generate should run in (latest-wins, like\n // `queued`): the foreground build runs non-blocking in dev (instant degrade)\n // while the background warehouse watch runs blocking (real DESCRIBEs). A\n // blocking watch trigger that lands while a non-blocking foreground run is in\n // flight therefore still describes when its trailing run fires.\n let inFlight: Promise<void> | null = null;\n let queued = false;\n let pendingMode: PreflightMode = \"non-blocking\";\n\n // The currently-armed DEV background warehouse watch, if any. Aborting it\n // stops a pending waitUntilRunning (server shutdown, or a newer arm replacing\n // an older one).\n let watchController: AbortController | null = null;\n\n /**\n * Generate types once in the given preflight {@link PreflightMode}. Never\n * throws in dev (logs instead); in production it rethrows so the build fails.\n * This is the un-guarded core — callers should go through {@link runGenerate}\n * so concurrent triggers can't race-write the .d.ts.\n *\n * @param mode - preflight policy for this run. The foreground build passes a\n * NODE_ENV-derived mode (blocking in production, non-blocking in dev so it\n * degrades instantly); the background warehouse watch passes \"blocking\" so\n * its regenerate actually DESCRIBEs and lands real (non-degraded) types.\n */\n async function generateOnce(mode: PreflightMode) {\n try {\n const warehouseId = process.env.DATABRICKS_WAREHOUSE_ID || \"\";\n\n if (!warehouseId) {\n logger.debug(\"Warehouse ID not found. Skipping type generation.\");\n return;\n }\n\n await generateFromEntryPoint({\n outFile,\n queryFolder: watchFolders[0],\n warehouseId,\n noCache: false,\n mode,\n mvOutFile,\n mvMetadataOutFile,\n });\n } catch (error) {\n // TypegenSyntaxError / TypegenFatalError carry a complete, actionable\n // report in their message. Their stack frames and attached query arrays\n // point into appkit internals and only add noise, so surface just the\n // message — both when failing the prod build and when logging in dev.\n const isTypegenError =\n error instanceof TypegenSyntaxError ||\n error instanceof TypegenFatalError;\n\n // throw in production to fail the build\n if (process.env.NODE_ENV === \"production\") {\n if (isTypegenError) error.stack = error.message;\n throw error;\n }\n\n if (isTypegenError) {\n logger.error(\"%s\", error.message);\n } else {\n logger.error(\"Error generating types: %O\", error);\n }\n }\n }\n\n /**\n * Single-flight wrapper around {@link generateOnce}. The initial build, the\n * .sql watcher, and the DEV warehouse watch all route through here so they can\n * never run typegen concurrently (which would race-write the .d.ts).\n *\n * If a run is already in flight, this does NOT start a second one — it records\n * the requested mode and sets a trailing flag so exactly one more run fires\n * after the current finishes, coalescing any number of overlapping triggers\n * (latest-wins, including the mode: a blocking watch trigger that arrives mid\n * non-blocking foreground run still describes when its trailing run fires).\n *\n * @param mode - preflight policy for this run. Recorded into `pendingMode`,\n * which the drain reads for each generate (latest trigger wins).\n * @returns A promise that resolves when this trigger's work (including any\n * trailing run it scheduled) has completed.\n */\n function runGenerate(mode: PreflightMode): Promise<void> {\n pendingMode = mode;\n\n if (inFlight) {\n // A run is active: remember that another trigger arrived and ride out the\n // current run. One trailing run then covers all coalesced triggers and\n // runs in the latest requested mode (recorded above).\n queued = true;\n return inFlight;\n }\n\n // Drain in a loop rather than recursing after a single queued-check: a\n // trigger can land in the window between generateOnce() resolving and the\n // check, so we re-test `queued` until it's clear. Critically, `inFlight` is\n // cleared synchronously in the SAME tick as the final `queued === false`\n // observation — never deferred to a .finally microtask — so there's no\n // window where a trigger sees `inFlight` set but the drain has already\n // decided to exit. The guard stays held for the whole drain, so concurrent\n // triggers only ever set the flag; they never start a parallel generate.\n const drain = async (): Promise<void> => {\n while (true) {\n queued = false;\n // Snapshot the mode synchronously alongside clearing `queued` so a\n // trigger landing during this generate is observed (via `queued`) on the\n // next loop with its own mode, not silently dropped.\n const runMode = pendingMode;\n await generateOnce(runMode);\n // Synchronous check + clear, atomic w.r.t. other (synchronous) callers.\n if (!queued) {\n inFlight = null;\n return;\n }\n }\n };\n\n inFlight = drain();\n return inFlight;\n }\n\n /**\n * DEV-only: get the warehouse to RUNNING in the background and regenerate with\n * real (non-degraded) types once it is — without blocking dev startup. The\n * foreground build only ever degrades in dev (instant `unknown`/cached types),\n * so this is what lands actual DESCRIBE results in the editor for EVERY\n * reachable warehouse state, not just one that happens to already be warm.\n *\n * Post-probe behaviour by state:\n * - RUNNING → describe right away (the dev foreground degraded, so a running\n * warehouse would otherwise never get real types). `waitUntilRunning`\n * returns immediately for an already-running warehouse, then the blocking\n * regenerate fires.\n * - STARTING → it's already coming up; just wait for RUNNING, then describe.\n * - STOPPED / STOPPING → kick off a start, wait for RUNNING, then describe.\n * - DELETED / DELETING → return (a deleted warehouse can't be started, and\n * blocking typegen would treat it as fatal); leave the degraded types.\n *\n * No-op in production or without a warehouse id. Replaces any previously-armed\n * watch (aborting it first). Fully self-contained: it never throws into the\n * caller and never re-arms itself. The whole lifecycle is abortable via the\n * shared {@link watchController} — its signal is threaded into\n * `waitUntilRunning`, so a dev-server shutdown cancels a pending wait — and the\n * regenerate routes through {@link runGenerate} so it can't race-write the\n * .d.ts with the foreground degrade or a `.sql` re-trigger.\n *\n * The regenerate runs in \"blocking\" mode (not the foreground's non-blocking)\n * so it actually DESCRIBEs the now-RUNNING warehouse and lands real types —\n * the whole point of warming the warehouse in the background.\n */\n function armWarehouseWatch(): void {\n if (process.env.NODE_ENV === \"production\") return;\n\n const warehouseId = process.env.DATABRICKS_WAREHOUSE_ID || \"\";\n if (!warehouseId) return;\n\n // Supersede any in-flight watch so we never run two concurrently.\n watchController?.abort();\n const controller = new AbortController();\n watchController = controller;\n const { signal } = controller;\n\n void (async () => {\n try {\n const client = new WorkspaceClient({});\n const state = await getWarehouseState(client, warehouseId);\n\n // A deleted/deleting warehouse can't be started and blocking typegen\n // would treat it as fatal — leave the degraded types and stop. Every\n // other state (including RUNNING) proceeds to wait-then-describe so the\n // dev editor gets real types, not just the foreground's degraded ones.\n if (state === \"DELETED\" || state === \"DELETING\") {\n return;\n }\n\n // Stopped/stopping won't reach RUNNING on its own — nudge it. RUNNING and\n // STARTING need no start (RUNNING is already up; STARTING is coming up),\n // so don't issue a redundant one. A failed start is non-fatal: give up\n // silently rather than throw out of the detached task (the developer\n // still has degraded/cached types).\n let startedByUs = false;\n if (state === \"STOPPED\" || state === \"STOPPING\") {\n try {\n logger.debug(\"Warehouse is %s; starting it.\", state);\n await startWarehouse(client, warehouseId);\n startedByUs = true;\n } catch {\n return;\n }\n }\n\n // Wait for RUNNING. For an already-RUNNING warehouse this returns on the\n // first poll; for STARTING/STOPPED it polls (abortably) until the\n // warehouse warms up, a terminal state, or the deadline.\n const final = await waitUntilRunning(client, warehouseId, {\n maxMs: DEV_WAREHOUSE_WATCH_MAX_MS,\n signal,\n // We just issued the start, so the first poll(s) often still report\n // STOPPED/STOPPING before the start propagates. Poll through those\n // instead of bailing, or the regenerate would never fire. When we\n // didn't start it (RUNNING/STARTING branch), keep the default terminal\n // states.\n treatStoppedAsTransient: startedByUs,\n });\n\n if (final === \"RUNNING\" && !signal.aborted) {\n logger.debug(\"Warehouse is RUNNING; regenerating types.\");\n // Blocking: the warehouse is RUNNING now, so describe it and emit real\n // (non-degraded) types — unlike the foreground dev run, which degraded.\n // Routed through the single-flight guard so it coalesces with the\n // foreground degrade / any `.sql` re-trigger instead of racing them.\n await runGenerate(\"blocking\");\n }\n } catch {\n // Detached background task: any failure (timeout, abort, connectivity,\n // auth) is non-fatal — the developer still has degraded/cached types.\n }\n })();\n }\n\n return {\n name: \"appkit-types\",\n\n apply() {\n const warehouseId = process.env.DATABRICKS_WAREHOUSE_ID || \"\";\n\n if (!warehouseId) {\n logger.debug(\"Warehouse ID not found. Skipping type generation.\");\n return false;\n }\n\n if (!existsSync(path.join(process.cwd(), \"config\", \"queries\"))) {\n return false;\n }\n\n return true;\n },\n\n configResolved(config) {\n const projectRoot = path.resolve(config.root, \"..\");\n outFile = path.resolve(\n projectRoot,\n options?.outFile ?? `shared/${TYPES_DIR}/${ANALYTICS_TYPES_FILE}`,\n );\n // Metric out-paths resolve against projectRoot only when explicitly\n // provided; unset options pass through as undefined so the generator\n // computes its sibling-of-outFile defaults. In the all-defaults case\n // the final paths are identical (the default outFile above lives in\n // shared/<TYPES_DIR>/), and a customized outFile now keeps its metric\n // siblings next to it instead of pinning them under shared/.\n mvOutFile =\n options?.mvOutFile !== undefined\n ? path.resolve(projectRoot, options.mvOutFile)\n : undefined;\n mvMetadataOutFile =\n options?.mvMetadataOutFile !== undefined\n ? path.resolve(projectRoot, options.mvMetadataOutFile)\n : undefined;\n watchFolders = options?.watchFolders ?? [\n path.join(process.cwd(), \"config\", \"queries\"),\n ];\n },\n\n buildStart() {\n // Production: block the build on this generate (and surface failures).\n // The watch is a dev-only no-op, so just run typegen.\n if (process.env.NODE_ENV === \"production\") {\n return runGenerate(\"blocking\");\n }\n\n // Dev: don't block startup waiting on typegen. The foreground generate runs\n // non-blocking — it skips the warehouse entirely and writes degraded\n // (cached/`unknown`) types instantly. Then arm the warehouse watch so the\n // warehouse gets a one-shot BLOCKING regenerate (real types) in the\n // background for EVERY reachable state: RUNNING describes right away, while\n // STARTING/STOPPED are waited (and started) until they reach RUNNING.\n void runGenerate(\"non-blocking\");\n armWarehouseWatch();\n },\n\n configureServer(server) {\n server.watcher.add(watchFolders);\n\n server.watcher.on(\"change\", (changedFile) => {\n const isWatchedFile = watchFolders.some((folder) =>\n changedFile.startsWith(folder),\n );\n\n if (\n isWatchedFile &&\n (changedFile.endsWith(\".sql\") ||\n // Basename equality, not endsWith: a sibling like\n // \"legacy-metric-views.json\" must not trigger a regenerate —\n // only the real config file does.\n path.basename(changedFile) === \"metric-views.json\")\n ) {\n // Route through the single-flight runner (was fire-and-forget\n // generate(), which could race the initial build / watch). This is a\n // dev-only hook, so degrade instantly (non-blocking), then re-arm the\n // warehouse watch so the edited query or metric-view source is\n // re-described in the background against the running warehouse (or\n // once a still-starting one warms up), landing fresh\n // blocking-described types.\n void runGenerate(\"non-blocking\");\n armWarehouseWatch();\n }\n });\n\n // Tear down any pending warehouse watch when the dev server closes so a\n // long backoff can't keep the process alive after shutdown.\n server.httpServer?.once(\"close\", () => {\n watchController?.abort();\n });\n },\n };\n}\n"],"mappings":";;;;;;;;AAmBA,MAAM,SAAS,aAAa,6BAA6B;;;;;;;AAQzD,MAAM,6BAA6B;;;;;;;AA8BnC,SAAgB,kBAAkB,SAA4C;CAC5E,IAAI;CACJ,IAAI;CACJ,IAAI;CACJ,IAAI;CAcJ,IAAI,WAAiC;CACrC,IAAI,SAAS;CACb,IAAI,cAA6B;CAKjC,IAAI,kBAA0C;;;;;;;;;;;;CAa9C,eAAe,aAAa,MAAqB;AAC/C,MAAI;GACF,MAAM,cAAc,QAAQ,IAAI,2BAA2B;AAE3D,OAAI,CAAC,aAAa;AAChB,WAAO,MAAM,oDAAoD;AACjE;;AAGF,SAAM,uBAAuB;IAC3B;IACA,aAAa,aAAa;IAC1B;IACA,SAAS;IACT;IACA;IACA;IACD,CAAC;WACK,OAAO;GAKd,MAAM,iBACJ,iBAAiB,sBACjB,iBAAiB;AAGnB,OAAI,QAAQ,IAAI,aAAa,cAAc;AACzC,QAAI,eAAgB,OAAM,QAAQ,MAAM;AACxC,UAAM;;AAGR,OAAI,eACF,QAAO,MAAM,MAAM,MAAM,QAAQ;OAEjC,QAAO,MAAM,8BAA8B,MAAM;;;;;;;;;;;;;;;;;;;CAqBvD,SAAS,YAAY,MAAoC;AACvD,gBAAc;AAEd,MAAI,UAAU;AAIZ,YAAS;AACT,UAAO;;EAWT,MAAM,QAAQ,YAA2B;AACvC,UAAO,MAAM;AACX,aAAS;AAKT,UAAM,aADU,YACW;AAE3B,QAAI,CAAC,QAAQ;AACX,gBAAW;AACX;;;;AAKN,aAAW,OAAO;AAClB,SAAO;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CAgCT,SAAS,oBAA0B;AACjC,MAAI,QAAQ,IAAI,aAAa,aAAc;EAE3C,MAAM,cAAc,QAAQ,IAAI,2BAA2B;AAC3D,MAAI,CAAC,YAAa;AAGlB,mBAAiB,OAAO;EACxB,MAAM,aAAa,IAAI,iBAAiB;AACxC,oBAAkB;EAClB,MAAM,EAAE,WAAW;AAEnB,GAAM,YAAY;AAChB,OAAI;IACF,MAAM,SAAS,IAAI,gBAAgB,EAAE,CAAC;IACtC,MAAM,QAAQ,MAAM,kBAAkB,QAAQ,YAAY;AAM1D,QAAI,UAAU,aAAa,UAAU,WACnC;IAQF,IAAI,cAAc;AAClB,QAAI,UAAU,aAAa,UAAU,WACnC,KAAI;AACF,YAAO,MAAM,iCAAiC,MAAM;AACpD,WAAM,eAAe,QAAQ,YAAY;AACzC,mBAAc;YACR;AACN;;AAkBJ,QAXc,MAAM,iBAAiB,QAAQ,aAAa;KACxD,OAAO;KACP;KAMA,yBAAyB;KAC1B,CAAC,KAEY,aAAa,CAAC,OAAO,SAAS;AAC1C,YAAO,MAAM,4CAA4C;AAKzD,WAAM,YAAY,WAAW;;WAEzB;MAIN;;AAGN,QAAO;EACL,MAAM;EAEN,QAAQ;AAGN,OAAI,EAFgB,QAAQ,IAAI,2BAA2B,KAEzC;AAChB,WAAO,MAAM,oDAAoD;AACjE,WAAO;;AAGT,OAAI,CAAC,WAAW,KAAK,KAAK,QAAQ,KAAK,EAAE,UAAU,UAAU,CAAC,CAC5D,QAAO;AAGT,UAAO;;EAGT,eAAe,QAAQ;GACrB,MAAM,cAAc,KAAK,QAAQ,OAAO,MAAM,KAAK;AACnD,aAAU,KAAK,QACb,aACA,SAAS,WAAW,UAAU,UAAU,GAAG,uBAC5C;AAOD,eACE,SAAS,cAAc,SACnB,KAAK,QAAQ,aAAa,QAAQ,UAAU,GAC5C;AACN,uBACE,SAAS,sBAAsB,SAC3B,KAAK,QAAQ,aAAa,QAAQ,kBAAkB,GACpD;AACN,kBAAe,SAAS,gBAAgB,CACtC,KAAK,KAAK,QAAQ,KAAK,EAAE,UAAU,UAAU,CAC9C;;EAGH,aAAa;AAGX,OAAI,QAAQ,IAAI,aAAa,aAC3B,QAAO,YAAY,WAAW;AAShC,GAAK,YAAY,eAAe;AAChC,sBAAmB;;EAGrB,gBAAgB,QAAQ;AACtB,UAAO,QAAQ,IAAI,aAAa;AAEhC,UAAO,QAAQ,GAAG,WAAW,gBAAgB;AAK3C,QAJsB,aAAa,MAAM,WACvC,YAAY,WAAW,OAAO,CAC/B,KAIE,YAAY,SAAS,OAAO,IAI3B,KAAK,SAAS,YAAY,KAAK,sBACjC;AAQA,KAAK,YAAY,eAAe;AAChC,wBAAmB;;KAErB;AAIF,UAAO,YAAY,KAAK,eAAe;AACrC,qBAAiB,OAAO;KACxB;;EAEL"}
@@ -96,6 +96,10 @@ The type generator:
96
96
  4. Generates TypeScript interfaces for query parameters and results
97
97
  5. Creates a `QueryRegistry` type for type-safe query execution
98
98
 
99
+ ### Parameters during `DESCRIBE QUERY`[​](#parameters-during-describe-query "Direct link to parameters-during-describe-query")
100
+
101
+ Type generation describes each query without binding real parameters, so it substitutes a placeholder default for every `:param` (e.g. `''` for a string). That breaks queries whose shape depends on a value — most notably dynamic table names via `IDENTIFIER(:catalog || '.schema.table')`. Annotate such parameters with a sample value (`-- @param catalog STRING = main`) so the describe call can resolve a real table. The sample value is used only at type-generation time; the runtime query still binds the actual parameter. See [SQL parameters → Sample values](./docs/plugins/analytics.md#sample-values-for-type-generation).
102
+
99
103
  ## Using generated types[​](#using-generated-types "Direct link to Using generated types")
100
104
 
101
105
  Once types are generated, your IDE will provide autocomplete and type checking: