@murumets-ee/imports 0.33.0 → 0.34.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/{cleanup-WNOeVksI.mjs → cleanup-BzqyX9Jo.mjs} +2 -2
- package/dist/{cleanup-WNOeVksI.mjs.map → cleanup-BzqyX9Jo.mjs.map} +1 -1
- package/dist/index.d.mts +3 -3
- package/dist/index.mjs +1 -1
- package/dist/mapping/index.d.mts +2 -2
- package/dist/mapping/index.mjs +1 -1
- package/dist/{parts-schema-B7Cbkl9T.d.mts → parts-schema-Bqau0MrJ.d.mts} +12 -2
- package/dist/{parts-schema-B7Cbkl9T.d.mts.map → parts-schema-Bqau0MrJ.d.mts.map} +1 -1
- package/dist/{parts-schema-Cuaq2_At.mjs → parts-schema-Vv0EJCjj.mjs} +2 -2
- package/dist/{parts-schema-Cuaq2_At.mjs.map → parts-schema-Vv0EJCjj.mjs.map} +1 -1
- package/dist/plugin.d.mts +1 -1
- package/dist/plugin.mjs +1 -1
- package/dist/runner-0r6UICjo.mjs +2 -0
- package/dist/runner-0r6UICjo.mjs.map +1 -0
- package/dist/{schema-registry-DtXbLhOg.mjs → schema-registry-JU4gsC8E.mjs} +2 -2
- package/dist/{schema-registry-DtXbLhOg.mjs.map → schema-registry-JU4gsC8E.mjs.map} +1 -1
- package/dist/storage-resolver.d.mts +1 -1
- package/dist/{transform-dO4z0Ze6.d.mts → transform-BRJJLSXE.d.mts} +22 -2
- package/dist/transform-BRJJLSXE.d.mts.map +1 -0
- package/dist/{transform-DdhsE3ZI.mjs → transform-BzacGXbo.mjs} +2 -2
- package/dist/{transform-DdhsE3ZI.mjs.map → transform-BzacGXbo.mjs.map} +1 -1
- package/dist/transforms.d.mts +5 -2
- package/dist/transforms.d.mts.map +1 -1
- package/dist/transforms.mjs +1 -1
- package/dist/{worker-D2Pf-R9o.d.mts → worker-Ck67AOnh.d.mts} +35 -3
- package/dist/worker-Ck67AOnh.d.mts.map +1 -0
- package/dist/worker.d.mts +1 -1
- package/dist/worker.mjs +1 -1
- package/dist/worker.mjs.map +1 -1
- package/dist/{xlsx-reader-Co4ulrnu.mjs → xlsx-reader-oApgZRJd.mjs} +2 -2
- package/dist/{xlsx-reader-Co4ulrnu.mjs.map → xlsx-reader-oApgZRJd.mjs.map} +1 -1
- package/package.json +8 -8
- package/dist/runner-Cb1QPSt8.mjs +0 -2
- package/dist/runner-Cb1QPSt8.mjs.map +0 -1
- package/dist/transform-dO4z0Ze6.d.mts.map +0 -1
- package/dist/worker-D2Pf-R9o.d.mts.map +0 -1
|
@@ -1,2 +1,2 @@
|
|
|
1
|
-
import{t as e}from"./rolldown-runtime-DK3Fl9T5.mjs";import{getImportsStorage as t}from"./storage.mjs";var n=e({cleanupOrphanedImports:()=>a,cleanupProcessedImports:()=>s});const r=1e3;function i(e){let{ttlOrRetention:t,ttlLabel:n,limit:i=100}=e;if(!Number.isFinite(t)||t<0)throw Error(`imports cleanup: ${n} must be a finite, non-negative number (got ${t})`);if(!Number.isFinite(i)||!Number.isInteger(i)||i<1||i>r)throw Error(`imports cleanup: limit must be an integer in [1, ${r}] (got ${i})`);return{ttlOrRetention:t,limit:i}}async function a(e){let{app:n,bucket:r,logger:a}=e,o=n.db.readWrite,{ttlOrRetention:s,limit:c}=i({ttlOrRetention:e.ttlHours,ttlLabel:`ttlHours`,...e.limit!==void 0&&{limit:e.limit}}),{and:l,asc:u,eq:d,isNull:f,lt:p,getTableColumns:m}=await import(`drizzle-orm`),{toolkitFiles:h}=await import(`@murumets-ee/storage`),{createAdminClient:g}=await import(`@murumets-ee/core/clients`),{ImportRun:_}=await import(`./transform-
|
|
2
|
-
//# sourceMappingURL=cleanup-
|
|
1
|
+
import{t as e}from"./rolldown-runtime-DK3Fl9T5.mjs";import{getImportsStorage as t}from"./storage.mjs";var n=e({cleanupOrphanedImports:()=>a,cleanupProcessedImports:()=>s});const r=1e3;function i(e){let{ttlOrRetention:t,ttlLabel:n,limit:i=100}=e;if(!Number.isFinite(t)||t<0)throw Error(`imports cleanup: ${n} must be a finite, non-negative number (got ${t})`);if(!Number.isFinite(i)||!Number.isInteger(i)||i<1||i>r)throw Error(`imports cleanup: limit must be an integer in [1, ${r}] (got ${i})`);return{ttlOrRetention:t,limit:i}}async function a(e){let{app:n,bucket:r,logger:a}=e,o=n.db.readWrite,{ttlOrRetention:s,limit:c}=i({ttlOrRetention:e.ttlHours,ttlLabel:`ttlHours`,...e.limit!==void 0&&{limit:e.limit}}),{and:l,asc:u,eq:d,isNull:f,lt:p,getTableColumns:m}=await import(`drizzle-orm`),{toolkitFiles:h}=await import(`@murumets-ee/storage`),{createAdminClient:g}=await import(`@murumets-ee/core/clients`),{ImportRun:_}=await import(`./transform-BzacGXbo.mjs`).then(e=>e.o),v=g(_).getTable(),y=m(h),b=y.key,x=y.bucket,S=y.createdAt;if(!b||!x||!S)throw Error(`cleanupOrphanedImports: toolkit_files schema missing expected columns (key/bucket/createdAt)`);let C=m(v),w=C.filePath,T=C.id;if(!w||!T)throw Error(`cleanupOrphanedImports: import_run schema missing expected columns (filePath/id)`);let E=new Date(Date.now()-s*60*60*1e3),D=await o.select({key:b}).from(h).leftJoin(v,d(w,b)).where(l(d(x,r),p(S,E),f(T))).orderBy(u(S)).limit(c);if(D.length===0)return a.debug({bucket:r,ttlHours:s},`No orphaned imports uploads found`),0;let O=await t(),k=0;for(let e of D){let t=e.key;try{await O.delete(t),k++}catch(e){a.warn({key:t,err:e},`Failed to delete orphaned imports upload`)}}return a.info({cleaned:k,total:D.length,bucket:r,ttlHours:s},`Cleaned up orphaned imports uploads`),k}const o=[`succeeded`,`failed`,`cancelled`];async function s(e){let{app:n,bucket:r,logger:a}=e,s=n.db.readWrite,{ttlOrRetention:c,limit:l}=i({ttlOrRetention:e.retentionDays,ttlLabel:`retentionDays`,...e.limit!==void 0&&{limit:e.limit}}),{and:u,asc:d,eq:f,getTableColumns:p,inArray:m,lt:h}=await import(`drizzle-orm`),{toolkitFiles:g}=await import(`@murumets-ee/storage`),{createAdminClient:_}=await import(`@murumets-ee/core/clients`),{ImportRun:v}=await import(`./transform-BzacGXbo.mjs`).then(e=>e.o),y=_(v).getTable(),b=p(g),x=b.key,S=b.bucket;if(!x||!S)throw Error(`cleanupProcessedImports: toolkit_files schema missing expected columns (key/bucket)`);let C=p(y),w=C.filePath,T=C.status,E=C.finishedAt;if(!w||!T||!E)throw Error(`cleanupProcessedImports: import_run schema missing expected columns (filePath/status/finishedAt)`);let D=new Date(Date.now()-c*24*60*60*1e3),O=await s.selectDistinctOn([x],{key:x}).from(g).innerJoin(y,f(w,x)).where(u(f(S,r),m(T,[...o]),h(E,D))).orderBy(d(x),d(E)).limit(l);if(O.length===0)return a.debug({bucket:r,retentionDays:c},`No processed-imports uploads ready for cleanup`),0;let k=await t(),A=0;for(let e of O){let t=e.key;try{await k.delete(t),A++}catch(e){a.warn({key:t,err:e},`Failed to delete processed-imports upload`)}}return a.info({cleaned:A,total:O.length,bucket:r,retentionDays:c},`Cleaned up processed imports uploads`),A}export{s as n,n as r,a as t};
|
|
2
|
+
//# sourceMappingURL=cleanup-BzqyX9Jo.mjs.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"cleanup-WNOeVksI.mjs","names":[],"sources":["../src/admin/cleanup.ts"],"sourcesContent":["/**\n * Cleanup orphaned imports uploads.\n *\n * Finds rows in `toolkit_files` within the configured imports bucket\n * that:\n * - were created more than `ttlHours` ago, AND\n * - have no matching `import_run.filePath` reference\n *\n * For each match, calls `StorageClient.delete(key)` — which removes\n * BOTH the adapter bytes (local file / R2 object) AND the\n * `toolkit_files` metadata row in one transactional unit (see\n * `@murumets-ee/storage` `StorageClient.delete`).\n *\n * Why this exists: the Tus upload flow stores bytes + metadata BEFORE\n * the client's downstream `register` step inserts an `import_run` row.\n * If the register call fails (validation, MIME mismatch, network\n * blip, agent abandons the workflow), the storage row is left\n * dangling. The commerce-side register handler does opportunistic\n * cleanup on its OWN failure paths, but rejections from the Zod\n * schema (before the handler runs) and rejections where the agent\n * never calls register at all both leak.\n *\n * Scheduling: driven by the `defineJob({ schedule: '@every 1h' })`\n * declaration in `plugin.ts`. The queue worker bootstraps a\n * `toolkit_scheduled_jobs` row for this job at startup and its cron\n * loop (`processDueSchedules`) fires the registered handler on each\n * tick. The handler does NOT re-enqueue itself — the schedule is\n * driven by the worker, not the handler.\n */\n\nimport type { ToolkitApp } from '@murumets-ee/core'\nimport type { Logger } from '@murumets-ee/logging'\nimport { getImportsStorage } from '../storage.js'\n\nexport interface CleanupOrphanedImportsArgs {\n /** Toolkit app (provides db). */\n app: ToolkitApp\n /**\n * Bucket name the imports flow uses. Rows in `toolkit_files` are\n * filtered to this bucket so the cleanup doesn't touch media,\n * ticketing attachments, or any other consumer's uploads even if\n * they share the same Postgres instance.\n */\n bucket: string\n /**\n * Hours after which an unreferenced upload is considered orphaned.\n * Operator-grace window: should be long enough that a normal\n * upload-then-register round-trip never wipes a fresh upload, but\n * short enough that storage waste from failed uploads doesn't\n * accumulate. 2 hours is the default in the imports plugin\n * config — calibrated against the typical \"operator clicks\n * upload, walks away, comes back to check\" workflow.\n */\n ttlHours: number\n /**\n * Maximum rows to process per run. Bounded so a sudden flood of\n * orphans (e.g. after a brand-new supplier with unrecognised\n * filenames floods the queue with rejected uploads) doesn't\n * block the queue worker on a single multi-hour cleanup.\n */\n limit?: number\n /** Logger for diagnostics. */\n logger: Pick<Logger, 'debug' | 'info' | 'warn'>\n}\n\n/** Default per-run cap. Tuned against ticketing's `cleanupOrphanedUploads` (also 100). */\nconst DEFAULT_LIMIT = 100\n\n/** Hard cap on per-run limit. Protects against pathological caller input. */\nconst MAX_LIMIT = 1000\n\n/**\n * Validate the numeric inputs every cleanup helper accepts. Returns\n * the sanitised values OR throws — the queue worker logs thrown\n * errors against the scheduled-job row so a misconfigured TTL\n * doesn't silently delete every bucket file (which would happen if\n * a negative `ttlHours` reached the date arithmetic and pushed the\n * cutoff into the future).\n */\nfunction validateCleanupArgs(\n raw: { ttlOrRetention: number; ttlLabel: string; limit?: number },\n): { ttlOrRetention: number; limit: number } {\n const { ttlOrRetention, ttlLabel, limit = DEFAULT_LIMIT } = raw\n if (!Number.isFinite(ttlOrRetention) || ttlOrRetention < 0) {\n throw new Error(\n `imports cleanup: ${ttlLabel} must be a finite, non-negative number (got ${ttlOrRetention})`,\n )\n }\n if (!Number.isFinite(limit) || !Number.isInteger(limit) || limit < 1 || limit > MAX_LIMIT) {\n throw new Error(\n `imports cleanup: limit must be an integer in [1, ${MAX_LIMIT}] (got ${limit})`,\n )\n }\n return { ttlOrRetention, limit }\n}\n\n/**\n * Run one cleanup pass. Returns the count of rows actually deleted\n * (which may be less than the count of orphans found if some\n * `storage.delete` calls failed — see the `warn` log inside the loop).\n */\nexport async function cleanupOrphanedImports(args: CleanupOrphanedImportsArgs): Promise<number> {\n const { app, bucket, logger } = args\n const db = app.db.readWrite\n const { ttlOrRetention: ttlHours, limit } = validateCleanupArgs({\n ttlOrRetention: args.ttlHours,\n ttlLabel: 'ttlHours',\n // `exactOptionalPropertyTypes` rejects an explicit `undefined`\n // for the optional `limit?: number` slot. Spread the field only\n // when defined so the validator's default-via-destructure kicks in.\n ...(args.limit !== undefined && { limit: args.limit }),\n })\n\n // Cross-table query: `toolkit_files` LEFT JOIN `import_run` — same\n // documented exception for cross-entity JOINs that ticketing's\n // cleanup uses. Two-table joins can't go through a single-entity\n // AdminClient (per CLAUDE.md \"Database Queries\").\n const { and, asc, eq, isNull, lt, getTableColumns } = await import('drizzle-orm')\n const { toolkitFiles } = await import('@murumets-ee/storage')\n const { createAdminClient } = await import('@murumets-ee/core/clients')\n const { ImportRun } = await import('../entities/import-run.js')\n type AnyPgColumn = import('drizzle-orm/pg-core').AnyPgColumn\n\n const importRunsTable = createAdminClient(ImportRun).getTable()\n\n // Both `toolkitFiles` and the `importRunsTable` are exported as\n // generic PgTables so direct dot-access widens to `T | undefined`.\n // Pull the column refs through `getTableColumns` (same shape as\n // ticketing's cleanup + the entity-version helper).\n const fileCols = getTableColumns(toolkitFiles) as Record<string, AnyPgColumn>\n const fileKey = fileCols.key\n const fileBucket = fileCols.bucket\n const fileCreatedAt = fileCols.createdAt\n if (!fileKey || !fileBucket || !fileCreatedAt) {\n throw new Error(\n 'cleanupOrphanedImports: toolkit_files schema missing expected columns (key/bucket/createdAt)',\n )\n }\n const importRunCols = getTableColumns(importRunsTable) as Record<string, AnyPgColumn>\n const irFilePath = importRunCols.filePath\n const irId = importRunCols.id\n if (!irFilePath || !irId) {\n throw new Error(\n 'cleanupOrphanedImports: import_run schema missing expected columns (filePath/id)',\n )\n }\n\n const cutoff = new Date(Date.now() - ttlHours * 60 * 60 * 1000)\n\n // Find toolkit_files rows in the imports bucket, older than cutoff,\n // where the storage key is NOT referenced by any import_run row.\n // ORDER BY createdAt ASC delivers strict FIFO so a sustained\n // orphan stream makes deterministic progress on the oldest rows\n // first — without the orderBy, Postgres returns rows in heap\n // order and a LIMIT 100 sweep could keep recycling the same\n // young rows while ancient orphans never get touched.\n const orphans = await db\n .select({ key: fileKey })\n .from(toolkitFiles)\n .leftJoin(importRunsTable, eq(irFilePath, fileKey))\n .where(and(eq(fileBucket, bucket), lt(fileCreatedAt, cutoff), isNull(irId)))\n .orderBy(asc(fileCreatedAt))\n .limit(limit)\n\n if (orphans.length === 0) {\n logger.debug({ bucket, ttlHours }, 'No orphaned imports uploads found')\n return 0\n }\n\n const storage = await getImportsStorage()\n let cleaned = 0\n for (const row of orphans) {\n const key = row.key as unknown as string\n try {\n await storage.delete(key)\n cleaned++\n } catch (err) {\n logger.warn({ key, err }, 'Failed to delete orphaned imports upload')\n }\n }\n\n logger.info(\n { cleaned, total: orphans.length, bucket, ttlHours },\n 'Cleaned up orphaned imports uploads',\n )\n return cleaned\n}\n\n/** Statuses considered \"done\" for the post-success/failure cleanup. */\nconst TERMINAL_IMPORT_RUN_STATUSES = ['succeeded', 'failed', 'cancelled'] as const\n\nexport interface CleanupProcessedImportsArgs {\n /** Toolkit app (provides db). */\n app: ToolkitApp\n /** Bucket name to scope the cleanup. Same semantics as `cleanupOrphanedImports`. */\n bucket: string\n /**\n * Days after which a terminal-status `import_run`'s source file is\n * deleted. Operator-grace window: long enough that an operator can\n * inspect the source file if a transform bug surfaces shortly after\n * a successful import; short enough that storage doesn't accumulate\n * months of supplier feeds.\n *\n * The plugin default is 7. Set to 0 to delete the source file\n * immediately at the next cleanup tick after the import row\n * reaches terminal status — useful when storage cost matters more\n * than forensic reprocess capability.\n */\n retentionDays: number\n /** Max rows per run. Defaults to 100 (same cap as the orphan sweep). */\n limit?: number\n logger: Pick<Logger, 'debug' | 'info' | 'warn'>\n}\n\n/**\n * Cleanup source files for `import_run`s that finished (succeeded /\n * failed / cancelled) more than `retentionDays` days ago.\n *\n * Counterpart to `cleanupOrphanedImports`: that one targets uploads\n * with NO matching `import_run` (handshake never completed); this one\n * targets uploads whose `import_run` row IS present and terminal.\n * Together they cover every \"useless storage row\" case the imports\n * flow can produce.\n *\n * Deletes ONLY the storage row (via `StorageClient.delete`); the\n * `import_run` row is kept so the History panel + audit trail still\n * surface the past run. The `import_run.filePath` value points at a\n * now-deleted key after this runs — consumers that try to re-download\n * the source file (e.g. a future \"reprocess\" button) will get a\n * `getMetadata` → null and should show \"source file unavailable\" in\n * the UI rather than crashing.\n */\nexport async function cleanupProcessedImports(args: CleanupProcessedImportsArgs): Promise<number> {\n const { app, bucket, logger } = args\n const db = app.db.readWrite\n const { ttlOrRetention: retentionDays, limit } = validateCleanupArgs({\n ttlOrRetention: args.retentionDays,\n ttlLabel: 'retentionDays',\n // `exactOptionalPropertyTypes` rejects an explicit `undefined`\n // for the optional `limit?: number` slot. Spread the field only\n // when defined so the validator's default-via-destructure kicks in.\n ...(args.limit !== undefined && { limit: args.limit }),\n })\n\n const { and, asc, eq, getTableColumns, inArray, lt } = await import('drizzle-orm')\n const { toolkitFiles } = await import('@murumets-ee/storage')\n const { createAdminClient } = await import('@murumets-ee/core/clients')\n const { ImportRun } = await import('../entities/import-run.js')\n type AnyPgColumn = import('drizzle-orm/pg-core').AnyPgColumn\n\n const importRunsTable = createAdminClient(ImportRun).getTable()\n\n const fileCols = getTableColumns(toolkitFiles) as Record<string, AnyPgColumn>\n const fileKey = fileCols.key\n const fileBucket = fileCols.bucket\n if (!fileKey || !fileBucket) {\n throw new Error(\n 'cleanupProcessedImports: toolkit_files schema missing expected columns (key/bucket)',\n )\n }\n const importRunCols = getTableColumns(importRunsTable) as Record<string, AnyPgColumn>\n const irFilePath = importRunCols.filePath\n const irStatus = importRunCols.status\n const irFinishedAt = importRunCols.finishedAt\n if (!irFilePath || !irStatus || !irFinishedAt) {\n throw new Error(\n 'cleanupProcessedImports: import_run schema missing expected columns (filePath/status/finishedAt)',\n )\n }\n\n const cutoff = new Date(Date.now() - retentionDays * 24 * 60 * 60 * 1000)\n\n // INNER JOIN so we only consider keys that DO have a matching\n // import_run row — the inverse of the orphan-cleanup query. Filter\n // on terminal status + finishedAt cutoff so a long-running batch\n // doesn't get its still-being-processed file ripped out from under\n // the worker.\n //\n // `selectDistinctOn([fileKey])` is the right primitive here:\n //\n // (a) **Dedupe before LIMIT.** `toolkit_files.key` is UNIQUE\n // within its own table, but `import_run.filePath` has no\n // unique constraint — a future re-run / retry flow that\n // creates two terminal `import_run` rows pointing at the\n // same `storageKey` would make the INNER JOIN project the\n // key twice. Without dedupe in SQL, a pathological cluster\n // of 100 duplicates sharing 1 storageKey would burn the\n // LIMIT cap to clean 1 file.\n //\n // (b) **Deterministic ordering** (NOT strict global FIFO).\n // `DISTINCT ON` requires the ORDER BY to start with the\n // distinct columns (Postgres rule). So we sort by `fileKey`\n // first, then `irFinishedAt` within each key group. The\n // LIMIT then picks the first N keys in `fileKey` order —\n // i.e. alphabetical by storage path, not globally oldest by\n // finishedAt. In practice this is roughly chronological\n // anyway: storage keys are `uploads/YYYY/MM/<uuid>/...` so\n // earlier months sort first. Within a month, ordering is\n // UUID-random — but every key eventually gets cleaned\n // (no starvation), so over ~10 ticks the entire backlog\n // drains either way. A subquery-wrapped \"true global FIFO\"\n // (DISTINCT ON inner, ORDER BY finishedAt outer) is\n // possible but adds complexity for marginal benefit; the\n // practical-FIFO-via-key-prefix property is acceptable.\n const candidates = await db\n .selectDistinctOn([fileKey], { key: fileKey })\n .from(toolkitFiles)\n .innerJoin(importRunsTable, eq(irFilePath, fileKey))\n .where(\n and(\n eq(fileBucket, bucket),\n inArray(irStatus, [...TERMINAL_IMPORT_RUN_STATUSES]),\n lt(irFinishedAt, cutoff),\n ),\n )\n .orderBy(asc(fileKey), asc(irFinishedAt))\n .limit(limit)\n\n if (candidates.length === 0) {\n logger.debug({ bucket, retentionDays }, 'No processed-imports uploads ready for cleanup')\n return 0\n }\n\n const storage = await getImportsStorage()\n let cleaned = 0\n for (const row of candidates) {\n const key = row.key as unknown as string\n try {\n await storage.delete(key)\n cleaned++\n } catch (err) {\n logger.warn({ key, err }, 'Failed to delete processed-imports upload')\n }\n }\n\n logger.info(\n { cleaned, total: candidates.length, bucket, retentionDays },\n 'Cleaned up processed imports uploads',\n )\n return cleaned\n}\n"],"mappings":"4KAkEA,MAGM,EAAY,IAUlB,SAAS,EACP,EAC2C,CAC3C,GAAM,CAAE,iBAAgB,WAAU,QAAQ,KAAkB,EAC5D,GAAI,CAAC,OAAO,SAAS,CAAc,GAAK,EAAiB,EACvD,MAAU,MACR,oBAAoB,EAAS,8CAA8C,EAAe,EAC5F,EAEF,GAAI,CAAC,OAAO,SAAS,CAAK,GAAK,CAAC,OAAO,UAAU,CAAK,GAAK,EAAQ,GAAK,EAAQ,EAC9E,MAAU,MACR,oDAAoD,EAAU,SAAS,EAAM,EAC/E,EAEF,MAAO,CAAE,iBAAgB,OAAM,CACjC,CAOA,eAAsB,EAAuB,EAAmD,CAC9F,GAAM,CAAE,MAAK,SAAQ,UAAW,EAC1B,EAAK,EAAI,GAAG,UACZ,CAAE,eAAgB,EAAU,SAAU,EAAoB,CAC9D,eAAgB,EAAK,SACrB,SAAU,WAIV,GAAI,EAAK,QAAU,IAAA,IAAa,CAAE,MAAO,EAAK,KAAM,CACtD,CAAC,EAMK,CAAE,MAAK,MAAK,KAAI,SAAQ,KAAI,mBAAoB,MAAM,OAAO,eAC7D,CAAE,gBAAiB,MAAM,OAAO,wBAChC,CAAE,qBAAsB,MAAM,OAAO,6BACrC,CAAE,aAAc,MAAM,OAAO,2BAA4B,CAAA,KAAA,GAAA,EAAA,CAAA,EAGzD,EAAkB,EAAkB,CAAS,CAAC,CAAC,SAAS,EAMxD,EAAW,EAAgB,CAAY,EACvC,EAAU,EAAS,IACnB,EAAa,EAAS,OACtB,EAAgB,EAAS,UAC/B,GAAI,CAAC,GAAW,CAAC,GAAc,CAAC,EAC9B,MAAU,MACR,8FACF,EAEF,IAAM,EAAgB,EAAgB,CAAe,EAC/C,EAAa,EAAc,SAC3B,EAAO,EAAc,GAC3B,GAAI,CAAC,GAAc,CAAC,EAClB,MAAU,MACR,kFACF,EAGF,IAAM,EAAS,IAAI,KAAK,KAAK,IAAI,EAAI,EAAW,GAAK,GAAK,GAAI,EASxD,EAAU,MAAM,EACnB,OAAO,CAAE,IAAK,CAAQ,CAAC,CAAC,CACxB,KAAK,CAAY,CAAC,CAClB,SAAS,EAAiB,EAAG,EAAY,CAAO,CAAC,CAAC,CAClD,MAAM,EAAI,EAAG,EAAY,CAAM,EAAG,EAAG,EAAe,CAAM,EAAG,EAAO,CAAI,CAAC,CAAC,CAAC,CAC3E,QAAQ,EAAI,CAAa,CAAC,CAAC,CAC3B,MAAM,CAAK,EAEd,GAAI,EAAQ,SAAW,EAErB,OADA,EAAO,MAAM,CAAE,SAAQ,UAAS,EAAG,mCAAmC,EAC/D,EAGT,IAAM,EAAU,MAAM,EAAkB,EACpC,EAAU,EACd,IAAK,IAAM,KAAO,EAAS,CACzB,IAAM,EAAM,EAAI,IAChB,GAAI,CACF,MAAM,EAAQ,OAAO,CAAG,EACxB,GACF,OAAS,EAAK,CACZ,EAAO,KAAK,CAAE,MAAK,KAAI,EAAG,0CAA0C,CACtE,CACF,CAMA,OAJA,EAAO,KACL,CAAE,UAAS,MAAO,EAAQ,OAAQ,SAAQ,UAAS,EACnD,qCACF,EACO,CACT,CAGA,MAAM,EAA+B,CAAC,YAAa,SAAU,WAAW,EA2CxE,eAAsB,EAAwB,EAAoD,CAChG,GAAM,CAAE,MAAK,SAAQ,UAAW,EAC1B,EAAK,EAAI,GAAG,UACZ,CAAE,eAAgB,EAAe,SAAU,EAAoB,CACnE,eAAgB,EAAK,cACrB,SAAU,gBAIV,GAAI,EAAK,QAAU,IAAA,IAAa,CAAE,MAAO,EAAK,KAAM,CACtD,CAAC,EAEK,CAAE,MAAK,MAAK,KAAI,kBAAiB,UAAS,MAAO,MAAM,OAAO,eAC9D,CAAE,gBAAiB,MAAM,OAAO,wBAChC,CAAE,qBAAsB,MAAM,OAAO,6BACrC,CAAE,aAAc,MAAM,OAAO,2BAA4B,CAAA,KAAA,GAAA,EAAA,CAAA,EAGzD,EAAkB,EAAkB,CAAS,CAAC,CAAC,SAAS,EAExD,EAAW,EAAgB,CAAY,EACvC,EAAU,EAAS,IACnB,EAAa,EAAS,OAC5B,GAAI,CAAC,GAAW,CAAC,EACf,MAAU,MACR,qFACF,EAEF,IAAM,EAAgB,EAAgB,CAAe,EAC/C,EAAa,EAAc,SAC3B,EAAW,EAAc,OACzB,EAAe,EAAc,WACnC,GAAI,CAAC,GAAc,CAAC,GAAY,CAAC,EAC/B,MAAU,MACR,kGACF,EAGF,IAAM,EAAS,IAAI,KAAK,KAAK,IAAI,EAAI,EAAgB,GAAK,GAAK,GAAK,GAAI,EAkClE,EAAa,MAAM,EACtB,iBAAiB,CAAC,CAAO,EAAG,CAAE,IAAK,CAAQ,CAAC,CAAC,CAC7C,KAAK,CAAY,CAAC,CAClB,UAAU,EAAiB,EAAG,EAAY,CAAO,CAAC,CAAC,CACnD,MACC,EACE,EAAG,EAAY,CAAM,EACrB,EAAQ,EAAU,CAAC,GAAG,CAA4B,CAAC,EACnD,EAAG,EAAc,CAAM,CACzB,CACF,CAAC,CACA,QAAQ,EAAI,CAAO,EAAG,EAAI,CAAY,CAAC,CAAC,CACxC,MAAM,CAAK,EAEd,GAAI,EAAW,SAAW,EAExB,OADA,EAAO,MAAM,CAAE,SAAQ,eAAc,EAAG,gDAAgD,EACjF,EAGT,IAAM,EAAU,MAAM,EAAkB,EACpC,EAAU,EACd,IAAK,IAAM,KAAO,EAAY,CAC5B,IAAM,EAAM,EAAI,IAChB,GAAI,CACF,MAAM,EAAQ,OAAO,CAAG,EACxB,GACF,OAAS,EAAK,CACZ,EAAO,KAAK,CAAE,MAAK,KAAI,EAAG,2CAA2C,CACvE,CACF,CAMA,OAJA,EAAO,KACL,CAAE,UAAS,MAAO,EAAW,OAAQ,SAAQ,eAAc,EAC3D,sCACF,EACO,CACT"}
|
|
1
|
+
{"version":3,"file":"cleanup-BzqyX9Jo.mjs","names":[],"sources":["../src/admin/cleanup.ts"],"sourcesContent":["/**\n * Cleanup orphaned imports uploads.\n *\n * Finds rows in `toolkit_files` within the configured imports bucket\n * that:\n * - were created more than `ttlHours` ago, AND\n * - have no matching `import_run.filePath` reference\n *\n * For each match, calls `StorageClient.delete(key)` — which removes\n * BOTH the adapter bytes (local file / R2 object) AND the\n * `toolkit_files` metadata row in one transactional unit (see\n * `@murumets-ee/storage` `StorageClient.delete`).\n *\n * Why this exists: the Tus upload flow stores bytes + metadata BEFORE\n * the client's downstream `register` step inserts an `import_run` row.\n * If the register call fails (validation, MIME mismatch, network\n * blip, agent abandons the workflow), the storage row is left\n * dangling. The commerce-side register handler does opportunistic\n * cleanup on its OWN failure paths, but rejections from the Zod\n * schema (before the handler runs) and rejections where the agent\n * never calls register at all both leak.\n *\n * Scheduling: driven by the `defineJob({ schedule: '@every 1h' })`\n * declaration in `plugin.ts`. The queue worker bootstraps a\n * `toolkit_scheduled_jobs` row for this job at startup and its cron\n * loop (`processDueSchedules`) fires the registered handler on each\n * tick. The handler does NOT re-enqueue itself — the schedule is\n * driven by the worker, not the handler.\n */\n\nimport type { ToolkitApp } from '@murumets-ee/core'\nimport type { Logger } from '@murumets-ee/logging'\nimport { getImportsStorage } from '../storage.js'\n\nexport interface CleanupOrphanedImportsArgs {\n /** Toolkit app (provides db). */\n app: ToolkitApp\n /**\n * Bucket name the imports flow uses. Rows in `toolkit_files` are\n * filtered to this bucket so the cleanup doesn't touch media,\n * ticketing attachments, or any other consumer's uploads even if\n * they share the same Postgres instance.\n */\n bucket: string\n /**\n * Hours after which an unreferenced upload is considered orphaned.\n * Operator-grace window: should be long enough that a normal\n * upload-then-register round-trip never wipes a fresh upload, but\n * short enough that storage waste from failed uploads doesn't\n * accumulate. 2 hours is the default in the imports plugin\n * config — calibrated against the typical \"operator clicks\n * upload, walks away, comes back to check\" workflow.\n */\n ttlHours: number\n /**\n * Maximum rows to process per run. Bounded so a sudden flood of\n * orphans (e.g. after a brand-new supplier with unrecognised\n * filenames floods the queue with rejected uploads) doesn't\n * block the queue worker on a single multi-hour cleanup.\n */\n limit?: number\n /** Logger for diagnostics. */\n logger: Pick<Logger, 'debug' | 'info' | 'warn'>\n}\n\n/** Default per-run cap. Tuned against ticketing's `cleanupOrphanedUploads` (also 100). */\nconst DEFAULT_LIMIT = 100\n\n/** Hard cap on per-run limit. Protects against pathological caller input. */\nconst MAX_LIMIT = 1000\n\n/**\n * Validate the numeric inputs every cleanup helper accepts. Returns\n * the sanitised values OR throws — the queue worker logs thrown\n * errors against the scheduled-job row so a misconfigured TTL\n * doesn't silently delete every bucket file (which would happen if\n * a negative `ttlHours` reached the date arithmetic and pushed the\n * cutoff into the future).\n */\nfunction validateCleanupArgs(\n raw: { ttlOrRetention: number; ttlLabel: string; limit?: number },\n): { ttlOrRetention: number; limit: number } {\n const { ttlOrRetention, ttlLabel, limit = DEFAULT_LIMIT } = raw\n if (!Number.isFinite(ttlOrRetention) || ttlOrRetention < 0) {\n throw new Error(\n `imports cleanup: ${ttlLabel} must be a finite, non-negative number (got ${ttlOrRetention})`,\n )\n }\n if (!Number.isFinite(limit) || !Number.isInteger(limit) || limit < 1 || limit > MAX_LIMIT) {\n throw new Error(\n `imports cleanup: limit must be an integer in [1, ${MAX_LIMIT}] (got ${limit})`,\n )\n }\n return { ttlOrRetention, limit }\n}\n\n/**\n * Run one cleanup pass. Returns the count of rows actually deleted\n * (which may be less than the count of orphans found if some\n * `storage.delete` calls failed — see the `warn` log inside the loop).\n */\nexport async function cleanupOrphanedImports(args: CleanupOrphanedImportsArgs): Promise<number> {\n const { app, bucket, logger } = args\n const db = app.db.readWrite\n const { ttlOrRetention: ttlHours, limit } = validateCleanupArgs({\n ttlOrRetention: args.ttlHours,\n ttlLabel: 'ttlHours',\n // `exactOptionalPropertyTypes` rejects an explicit `undefined`\n // for the optional `limit?: number` slot. Spread the field only\n // when defined so the validator's default-via-destructure kicks in.\n ...(args.limit !== undefined && { limit: args.limit }),\n })\n\n // Cross-table query: `toolkit_files` LEFT JOIN `import_run` — same\n // documented exception for cross-entity JOINs that ticketing's\n // cleanup uses. Two-table joins can't go through a single-entity\n // AdminClient (per CLAUDE.md \"Database Queries\").\n const { and, asc, eq, isNull, lt, getTableColumns } = await import('drizzle-orm')\n const { toolkitFiles } = await import('@murumets-ee/storage')\n const { createAdminClient } = await import('@murumets-ee/core/clients')\n const { ImportRun } = await import('../entities/import-run.js')\n type AnyPgColumn = import('drizzle-orm/pg-core').AnyPgColumn\n\n const importRunsTable = createAdminClient(ImportRun).getTable()\n\n // Both `toolkitFiles` and the `importRunsTable` are exported as\n // generic PgTables so direct dot-access widens to `T | undefined`.\n // Pull the column refs through `getTableColumns` (same shape as\n // ticketing's cleanup + the entity-version helper).\n const fileCols = getTableColumns(toolkitFiles) as Record<string, AnyPgColumn>\n const fileKey = fileCols.key\n const fileBucket = fileCols.bucket\n const fileCreatedAt = fileCols.createdAt\n if (!fileKey || !fileBucket || !fileCreatedAt) {\n throw new Error(\n 'cleanupOrphanedImports: toolkit_files schema missing expected columns (key/bucket/createdAt)',\n )\n }\n const importRunCols = getTableColumns(importRunsTable) as Record<string, AnyPgColumn>\n const irFilePath = importRunCols.filePath\n const irId = importRunCols.id\n if (!irFilePath || !irId) {\n throw new Error(\n 'cleanupOrphanedImports: import_run schema missing expected columns (filePath/id)',\n )\n }\n\n const cutoff = new Date(Date.now() - ttlHours * 60 * 60 * 1000)\n\n // Find toolkit_files rows in the imports bucket, older than cutoff,\n // where the storage key is NOT referenced by any import_run row.\n // ORDER BY createdAt ASC delivers strict FIFO so a sustained\n // orphan stream makes deterministic progress on the oldest rows\n // first — without the orderBy, Postgres returns rows in heap\n // order and a LIMIT 100 sweep could keep recycling the same\n // young rows while ancient orphans never get touched.\n const orphans = await db\n .select({ key: fileKey })\n .from(toolkitFiles)\n .leftJoin(importRunsTable, eq(irFilePath, fileKey))\n .where(and(eq(fileBucket, bucket), lt(fileCreatedAt, cutoff), isNull(irId)))\n .orderBy(asc(fileCreatedAt))\n .limit(limit)\n\n if (orphans.length === 0) {\n logger.debug({ bucket, ttlHours }, 'No orphaned imports uploads found')\n return 0\n }\n\n const storage = await getImportsStorage()\n let cleaned = 0\n for (const row of orphans) {\n const key = row.key as unknown as string\n try {\n await storage.delete(key)\n cleaned++\n } catch (err) {\n logger.warn({ key, err }, 'Failed to delete orphaned imports upload')\n }\n }\n\n logger.info(\n { cleaned, total: orphans.length, bucket, ttlHours },\n 'Cleaned up orphaned imports uploads',\n )\n return cleaned\n}\n\n/** Statuses considered \"done\" for the post-success/failure cleanup. */\nconst TERMINAL_IMPORT_RUN_STATUSES = ['succeeded', 'failed', 'cancelled'] as const\n\nexport interface CleanupProcessedImportsArgs {\n /** Toolkit app (provides db). */\n app: ToolkitApp\n /** Bucket name to scope the cleanup. Same semantics as `cleanupOrphanedImports`. */\n bucket: string\n /**\n * Days after which a terminal-status `import_run`'s source file is\n * deleted. Operator-grace window: long enough that an operator can\n * inspect the source file if a transform bug surfaces shortly after\n * a successful import; short enough that storage doesn't accumulate\n * months of supplier feeds.\n *\n * The plugin default is 7. Set to 0 to delete the source file\n * immediately at the next cleanup tick after the import row\n * reaches terminal status — useful when storage cost matters more\n * than forensic reprocess capability.\n */\n retentionDays: number\n /** Max rows per run. Defaults to 100 (same cap as the orphan sweep). */\n limit?: number\n logger: Pick<Logger, 'debug' | 'info' | 'warn'>\n}\n\n/**\n * Cleanup source files for `import_run`s that finished (succeeded /\n * failed / cancelled) more than `retentionDays` days ago.\n *\n * Counterpart to `cleanupOrphanedImports`: that one targets uploads\n * with NO matching `import_run` (handshake never completed); this one\n * targets uploads whose `import_run` row IS present and terminal.\n * Together they cover every \"useless storage row\" case the imports\n * flow can produce.\n *\n * Deletes ONLY the storage row (via `StorageClient.delete`); the\n * `import_run` row is kept so the History panel + audit trail still\n * surface the past run. The `import_run.filePath` value points at a\n * now-deleted key after this runs — consumers that try to re-download\n * the source file (e.g. a future \"reprocess\" button) will get a\n * `getMetadata` → null and should show \"source file unavailable\" in\n * the UI rather than crashing.\n */\nexport async function cleanupProcessedImports(args: CleanupProcessedImportsArgs): Promise<number> {\n const { app, bucket, logger } = args\n const db = app.db.readWrite\n const { ttlOrRetention: retentionDays, limit } = validateCleanupArgs({\n ttlOrRetention: args.retentionDays,\n ttlLabel: 'retentionDays',\n // `exactOptionalPropertyTypes` rejects an explicit `undefined`\n // for the optional `limit?: number` slot. Spread the field only\n // when defined so the validator's default-via-destructure kicks in.\n ...(args.limit !== undefined && { limit: args.limit }),\n })\n\n const { and, asc, eq, getTableColumns, inArray, lt } = await import('drizzle-orm')\n const { toolkitFiles } = await import('@murumets-ee/storage')\n const { createAdminClient } = await import('@murumets-ee/core/clients')\n const { ImportRun } = await import('../entities/import-run.js')\n type AnyPgColumn = import('drizzle-orm/pg-core').AnyPgColumn\n\n const importRunsTable = createAdminClient(ImportRun).getTable()\n\n const fileCols = getTableColumns(toolkitFiles) as Record<string, AnyPgColumn>\n const fileKey = fileCols.key\n const fileBucket = fileCols.bucket\n if (!fileKey || !fileBucket) {\n throw new Error(\n 'cleanupProcessedImports: toolkit_files schema missing expected columns (key/bucket)',\n )\n }\n const importRunCols = getTableColumns(importRunsTable) as Record<string, AnyPgColumn>\n const irFilePath = importRunCols.filePath\n const irStatus = importRunCols.status\n const irFinishedAt = importRunCols.finishedAt\n if (!irFilePath || !irStatus || !irFinishedAt) {\n throw new Error(\n 'cleanupProcessedImports: import_run schema missing expected columns (filePath/status/finishedAt)',\n )\n }\n\n const cutoff = new Date(Date.now() - retentionDays * 24 * 60 * 60 * 1000)\n\n // INNER JOIN so we only consider keys that DO have a matching\n // import_run row — the inverse of the orphan-cleanup query. Filter\n // on terminal status + finishedAt cutoff so a long-running batch\n // doesn't get its still-being-processed file ripped out from under\n // the worker.\n //\n // `selectDistinctOn([fileKey])` is the right primitive here:\n //\n // (a) **Dedupe before LIMIT.** `toolkit_files.key` is UNIQUE\n // within its own table, but `import_run.filePath` has no\n // unique constraint — a future re-run / retry flow that\n // creates two terminal `import_run` rows pointing at the\n // same `storageKey` would make the INNER JOIN project the\n // key twice. Without dedupe in SQL, a pathological cluster\n // of 100 duplicates sharing 1 storageKey would burn the\n // LIMIT cap to clean 1 file.\n //\n // (b) **Deterministic ordering** (NOT strict global FIFO).\n // `DISTINCT ON` requires the ORDER BY to start with the\n // distinct columns (Postgres rule). So we sort by `fileKey`\n // first, then `irFinishedAt` within each key group. The\n // LIMIT then picks the first N keys in `fileKey` order —\n // i.e. alphabetical by storage path, not globally oldest by\n // finishedAt. In practice this is roughly chronological\n // anyway: storage keys are `uploads/YYYY/MM/<uuid>/...` so\n // earlier months sort first. Within a month, ordering is\n // UUID-random — but every key eventually gets cleaned\n // (no starvation), so over ~10 ticks the entire backlog\n // drains either way. A subquery-wrapped \"true global FIFO\"\n // (DISTINCT ON inner, ORDER BY finishedAt outer) is\n // possible but adds complexity for marginal benefit; the\n // practical-FIFO-via-key-prefix property is acceptable.\n const candidates = await db\n .selectDistinctOn([fileKey], { key: fileKey })\n .from(toolkitFiles)\n .innerJoin(importRunsTable, eq(irFilePath, fileKey))\n .where(\n and(\n eq(fileBucket, bucket),\n inArray(irStatus, [...TERMINAL_IMPORT_RUN_STATUSES]),\n lt(irFinishedAt, cutoff),\n ),\n )\n .orderBy(asc(fileKey), asc(irFinishedAt))\n .limit(limit)\n\n if (candidates.length === 0) {\n logger.debug({ bucket, retentionDays }, 'No processed-imports uploads ready for cleanup')\n return 0\n }\n\n const storage = await getImportsStorage()\n let cleaned = 0\n for (const row of candidates) {\n const key = row.key as unknown as string\n try {\n await storage.delete(key)\n cleaned++\n } catch (err) {\n logger.warn({ key, err }, 'Failed to delete processed-imports upload')\n }\n }\n\n logger.info(\n { cleaned, total: candidates.length, bucket, retentionDays },\n 'Cleaned up processed imports uploads',\n )\n return cleaned\n}\n"],"mappings":"4KAkEA,MAGM,EAAY,IAUlB,SAAS,EACP,EAC2C,CAC3C,GAAM,CAAE,iBAAgB,WAAU,QAAQ,KAAkB,EAC5D,GAAI,CAAC,OAAO,SAAS,CAAc,GAAK,EAAiB,EACvD,MAAU,MACR,oBAAoB,EAAS,8CAA8C,EAAe,EAC5F,EAEF,GAAI,CAAC,OAAO,SAAS,CAAK,GAAK,CAAC,OAAO,UAAU,CAAK,GAAK,EAAQ,GAAK,EAAQ,EAC9E,MAAU,MACR,oDAAoD,EAAU,SAAS,EAAM,EAC/E,EAEF,MAAO,CAAE,iBAAgB,OAAM,CACjC,CAOA,eAAsB,EAAuB,EAAmD,CAC9F,GAAM,CAAE,MAAK,SAAQ,UAAW,EAC1B,EAAK,EAAI,GAAG,UACZ,CAAE,eAAgB,EAAU,SAAU,EAAoB,CAC9D,eAAgB,EAAK,SACrB,SAAU,WAIV,GAAI,EAAK,QAAU,IAAA,IAAa,CAAE,MAAO,EAAK,KAAM,CACtD,CAAC,EAMK,CAAE,MAAK,MAAK,KAAI,SAAQ,KAAI,mBAAoB,MAAM,OAAO,eAC7D,CAAE,gBAAiB,MAAM,OAAO,wBAChC,CAAE,qBAAsB,MAAM,OAAO,6BACrC,CAAE,aAAc,MAAM,OAAO,2BAA4B,CAAA,KAAA,GAAA,EAAA,CAAA,EAGzD,EAAkB,EAAkB,CAAS,CAAC,CAAC,SAAS,EAMxD,EAAW,EAAgB,CAAY,EACvC,EAAU,EAAS,IACnB,EAAa,EAAS,OACtB,EAAgB,EAAS,UAC/B,GAAI,CAAC,GAAW,CAAC,GAAc,CAAC,EAC9B,MAAU,MACR,8FACF,EAEF,IAAM,EAAgB,EAAgB,CAAe,EAC/C,EAAa,EAAc,SAC3B,EAAO,EAAc,GAC3B,GAAI,CAAC,GAAc,CAAC,EAClB,MAAU,MACR,kFACF,EAGF,IAAM,EAAS,IAAI,KAAK,KAAK,IAAI,EAAI,EAAW,GAAK,GAAK,GAAI,EASxD,EAAU,MAAM,EACnB,OAAO,CAAE,IAAK,CAAQ,CAAC,CAAC,CACxB,KAAK,CAAY,CAAC,CAClB,SAAS,EAAiB,EAAG,EAAY,CAAO,CAAC,CAAC,CAClD,MAAM,EAAI,EAAG,EAAY,CAAM,EAAG,EAAG,EAAe,CAAM,EAAG,EAAO,CAAI,CAAC,CAAC,CAAC,CAC3E,QAAQ,EAAI,CAAa,CAAC,CAAC,CAC3B,MAAM,CAAK,EAEd,GAAI,EAAQ,SAAW,EAErB,OADA,EAAO,MAAM,CAAE,SAAQ,UAAS,EAAG,mCAAmC,EAC/D,EAGT,IAAM,EAAU,MAAM,EAAkB,EACpC,EAAU,EACd,IAAK,IAAM,KAAO,EAAS,CACzB,IAAM,EAAM,EAAI,IAChB,GAAI,CACF,MAAM,EAAQ,OAAO,CAAG,EACxB,GACF,OAAS,EAAK,CACZ,EAAO,KAAK,CAAE,MAAK,KAAI,EAAG,0CAA0C,CACtE,CACF,CAMA,OAJA,EAAO,KACL,CAAE,UAAS,MAAO,EAAQ,OAAQ,SAAQ,UAAS,EACnD,qCACF,EACO,CACT,CAGA,MAAM,EAA+B,CAAC,YAAa,SAAU,WAAW,EA2CxE,eAAsB,EAAwB,EAAoD,CAChG,GAAM,CAAE,MAAK,SAAQ,UAAW,EAC1B,EAAK,EAAI,GAAG,UACZ,CAAE,eAAgB,EAAe,SAAU,EAAoB,CACnE,eAAgB,EAAK,cACrB,SAAU,gBAIV,GAAI,EAAK,QAAU,IAAA,IAAa,CAAE,MAAO,EAAK,KAAM,CACtD,CAAC,EAEK,CAAE,MAAK,MAAK,KAAI,kBAAiB,UAAS,MAAO,MAAM,OAAO,eAC9D,CAAE,gBAAiB,MAAM,OAAO,wBAChC,CAAE,qBAAsB,MAAM,OAAO,6BACrC,CAAE,aAAc,MAAM,OAAO,2BAA4B,CAAA,KAAA,GAAA,EAAA,CAAA,EAGzD,EAAkB,EAAkB,CAAS,CAAC,CAAC,SAAS,EAExD,EAAW,EAAgB,CAAY,EACvC,EAAU,EAAS,IACnB,EAAa,EAAS,OAC5B,GAAI,CAAC,GAAW,CAAC,EACf,MAAU,MACR,qFACF,EAEF,IAAM,EAAgB,EAAgB,CAAe,EAC/C,EAAa,EAAc,SAC3B,EAAW,EAAc,OACzB,EAAe,EAAc,WACnC,GAAI,CAAC,GAAc,CAAC,GAAY,CAAC,EAC/B,MAAU,MACR,kGACF,EAGF,IAAM,EAAS,IAAI,KAAK,KAAK,IAAI,EAAI,EAAgB,GAAK,GAAK,GAAK,GAAI,EAkClE,EAAa,MAAM,EACtB,iBAAiB,CAAC,CAAO,EAAG,CAAE,IAAK,CAAQ,CAAC,CAAC,CAC7C,KAAK,CAAY,CAAC,CAClB,UAAU,EAAiB,EAAG,EAAY,CAAO,CAAC,CAAC,CACnD,MACC,EACE,EAAG,EAAY,CAAM,EACrB,EAAQ,EAAU,CAAC,GAAG,CAA4B,CAAC,EACnD,EAAG,EAAc,CAAM,CACzB,CACF,CAAC,CACA,QAAQ,EAAI,CAAO,EAAG,EAAI,CAAY,CAAC,CAAC,CACxC,MAAM,CAAK,EAEd,GAAI,EAAW,SAAW,EAExB,OADA,EAAO,MAAM,CAAE,SAAQ,eAAc,EAAG,gDAAgD,EACjF,EAGT,IAAM,EAAU,MAAM,EAAkB,EACpC,EAAU,EACd,IAAK,IAAM,KAAO,EAAY,CAC5B,IAAM,EAAM,EAAI,IAChB,GAAI,CACF,MAAM,EAAQ,OAAO,CAAG,EACxB,GACF,OAAS,EAAK,CACZ,EAAO,KAAK,CAAE,MAAK,KAAI,EAAG,2CAA2C,CACvE,CACF,CAMA,OAJA,EAAO,KACL,CAAE,UAAS,MAAO,EAAW,OAAQ,SAAQ,eAAc,EAC3D,sCACF,EACO,CACT"}
|
package/dist/index.d.mts
CHANGED
|
@@ -1,5 +1,5 @@
|
|
|
1
|
-
import { A as ErrorTrackerConfig, C as countDataRows, D as ErrorPattern, E as DEFAULT_MAX_SAMPLES_PER_PATTERN, M as ImportRun, N as ImportRunClient, O as ErrorSample, P as ImportRunStatus, S as StreamFeedRow, T as DEFAULT_MAX_PATTERNS, _ as FeedReaderConfig, b as readFeed, f as DEFAULT_BATCH_SIZE, g as runImport, h as RunImportResult, j as IMPORT_RUN_STATUSES, k as ErrorTracker, m as RunImportOptions, n as EsClientResolver, p as ImportRunProgress, r as FilePathResolver, t as EnsureIndexResolver, v as ReadFeedOptions, w as streamFeed, x as StreamFeedOptions, y as feedReaderConfigSchema } from "./worker-
|
|
2
|
-
import { a as
|
|
1
|
+
import { A as ErrorTrackerConfig, C as countDataRows, D as ErrorPattern, E as DEFAULT_MAX_SAMPLES_PER_PATTERN, M as ImportRun, N as ImportRunClient, O as ErrorSample, P as ImportRunStatus, S as StreamFeedRow, T as DEFAULT_MAX_PATTERNS, _ as FeedReaderConfig, b as readFeed, f as DEFAULT_BATCH_SIZE, g as runImport, h as RunImportResult, j as IMPORT_RUN_STATUSES, k as ErrorTracker, m as RunImportOptions, n as EsClientResolver, p as ImportRunProgress, r as FilePathResolver, t as EnsureIndexResolver, v as ReadFeedOptions, w as streamFeed, x as StreamFeedOptions, y as feedReaderConfigSchema } from "./worker-Ck67AOnh.mjs";
|
|
2
|
+
import { a as TransformContext, c as getTransformRegistry, i as RowWarning, l as registerImportTransform, n as RowResult, o as TransformName, r as RowTransform, s as TransformRegistry, t as RowError } from "./transform-BRJJLSXE.mjs";
|
|
3
3
|
import { Behavior } from "@murumets-ee/entity";
|
|
4
4
|
import { ToolkitApp } from "@murumets-ee/core";
|
|
5
5
|
import { Logger as Logger$1 } from "@murumets-ee/logging";
|
|
@@ -219,5 +219,5 @@ interface FeedPreview {
|
|
|
219
219
|
*/
|
|
220
220
|
declare function previewFeed(options: PreviewFeedOptions): Promise<FeedPreview>;
|
|
221
221
|
//#endregion
|
|
222
|
-
export { type CleanupOrphanedImportsArgs, type CleanupProcessedImportsArgs, DEFAULT_BATCH_SIZE, DEFAULT_MAX_PATTERNS, DEFAULT_MAX_SAMPLES_PER_PATTERN, DEFAULT_PREVIEW_ROWS, type EnsureIndexResolver, type ErrorPattern, type ErrorSample, ErrorTracker, type ErrorTrackerConfig, type EsClientResolver, type FeedPreview, type FeedReaderConfig, type FilePathResolver, IMPORT_MAPPING_FORMATS, IMPORT_MAPPING_TARGET_SCHEMAS, IMPORT_RUN_STATUSES, ImportMapping, type ImportMappingClient, type ImportMappingFormat, type ImportMappingTargetSchema, ImportRun, type ImportRunClient, type ImportRunProgress, type ImportRunStatus, MAX_PREVIEW_ROWS, type PreviewFeedOptions, type ReadFeedOptions, type RowError, type RowResult, type RowTransform, type RunImportOptions, type RunImportResult, type StreamFeedOptions, type StreamFeedRow, type TransformContext, type TransformName, TransformRegistry, assertValidMapping, cleanupOrphanedImports, cleanupProcessedImports, countDataRows, feedReaderConfigSchema, getTransformRegistry, importMappingValidation, previewFeed, readFeed, registerImportTransform, runImport, streamFeed };
|
|
222
|
+
export { type CleanupOrphanedImportsArgs, type CleanupProcessedImportsArgs, DEFAULT_BATCH_SIZE, DEFAULT_MAX_PATTERNS, DEFAULT_MAX_SAMPLES_PER_PATTERN, DEFAULT_PREVIEW_ROWS, type EnsureIndexResolver, type ErrorPattern, type ErrorSample, ErrorTracker, type ErrorTrackerConfig, type EsClientResolver, type FeedPreview, type FeedReaderConfig, type FilePathResolver, IMPORT_MAPPING_FORMATS, IMPORT_MAPPING_TARGET_SCHEMAS, IMPORT_RUN_STATUSES, ImportMapping, type ImportMappingClient, type ImportMappingFormat, type ImportMappingTargetSchema, ImportRun, type ImportRunClient, type ImportRunProgress, type ImportRunStatus, MAX_PREVIEW_ROWS, type PreviewFeedOptions, type ReadFeedOptions, type RowError, type RowResult, type RowTransform, type RowWarning, type RunImportOptions, type RunImportResult, type StreamFeedOptions, type StreamFeedRow, type TransformContext, type TransformName, TransformRegistry, assertValidMapping, cleanupOrphanedImports, cleanupProcessedImports, countDataRows, feedReaderConfigSchema, getTransformRegistry, importMappingValidation, previewFeed, readFeed, registerImportTransform, runImport, streamFeed };
|
|
223
223
|
//# sourceMappingURL=index.d.mts.map
|
package/dist/index.mjs
CHANGED
|
@@ -1,2 +1,2 @@
|
|
|
1
|
-
import{n as e,t}from"./cleanup-
|
|
1
|
+
import{n as e,t}from"./cleanup-BzqyX9Jo.mjs";import{a as n,c as r,d as i,i as a,l as o,n as s,r as c,s as l,t as u,u as d}from"./transform-BzacGXbo.mjs";import{c as f,i as p,l as m,n as h,o as g,r as _,s as v,t as y,u as b}from"./runner-0r6UICjo.mjs";const x=100,S=20;async function C(e){let{maxRows:t,...n}=e,r=t??20,i=Math.min(100,Math.max(1,Number.isFinite(r)?Math.floor(r):20)),a=[],o=new Map;for await(let{row:e}of p({...n,relaxColumnCount:!0})){for(let t of Object.keys(e))o.set(t,!0);if(a.push(e),a.length>=i)break}return{columns:[...o.keys()],rows:a}}export{y as DEFAULT_BATCH_SIZE,f as DEFAULT_MAX_PATTERNS,m as DEFAULT_MAX_SAMPLES_PER_PATTERN,S as DEFAULT_PREVIEW_ROWS,b as ErrorTracker,l as IMPORT_MAPPING_FORMATS,r as IMPORT_MAPPING_TARGET_SCHEMAS,a as IMPORT_RUN_STATUSES,o as ImportMapping,n as ImportRun,x as MAX_PREVIEW_ROWS,u as TransformRegistry,d as assertValidMapping,t as cleanupOrphanedImports,e as cleanupProcessedImports,g as countDataRows,_ as feedReaderConfigSchema,s as getTransformRegistry,i as importMappingValidation,C as previewFeed,p as readFeed,c as registerImportTransform,h as runImport,v as streamFeed};
|
|
2
2
|
//# sourceMappingURL=index.mjs.map
|
package/dist/mapping/index.d.mts
CHANGED
|
@@ -1,5 +1,5 @@
|
|
|
1
|
-
import { r as RowTransform } from "../transform-
|
|
2
|
-
import { S as toMinorUnits, _ as applyColumnTransforms, a as MappedRow, b as parseEuropeanNumber, c as TargetSchema, d as textOrNull, f as COLUMN_TRANSFORM_OPS, g as ColumnTransformOpDescriptor, h as ColumnTransformOp, i as BuildRowContext, l as numberCell, m as ColumnTransform, n as partsParamsSchema, o as NumberOutcome, p as CellValue, r as partsSchema, s as TargetField, t as PartsParams, u as rowError, v as columnTransformSchema, x as parseNumber, y as NumberFormat } from "../parts-schema-
|
|
1
|
+
import { r as RowTransform } from "../transform-BRJJLSXE.mjs";
|
|
2
|
+
import { S as toMinorUnits, _ as applyColumnTransforms, a as MappedRow, b as parseEuropeanNumber, c as TargetSchema, d as textOrNull, f as COLUMN_TRANSFORM_OPS, g as ColumnTransformOpDescriptor, h as ColumnTransformOp, i as BuildRowContext, l as numberCell, m as ColumnTransform, n as partsParamsSchema, o as NumberOutcome, p as CellValue, r as partsSchema, s as TargetField, t as PartsParams, u as rowError, v as columnTransformSchema, x as parseNumber, y as NumberFormat } from "../parts-schema-Bqau0MrJ.mjs";
|
|
3
3
|
import { z } from "zod";
|
|
4
4
|
|
|
5
5
|
//#region src/mapping/discount-table-schema.d.ts
|
package/dist/mapping/index.mjs
CHANGED
|
@@ -1,2 +1,2 @@
|
|
|
1
|
-
import{a as e,c as t,d as n,f as r,g as i,h as a,i as o,l as s,m as c,n as l,o as u,p as d,r as f,s as p,t as m,u as h}from"../parts-schema-
|
|
1
|
+
import{a as e,c as t,d as n,f as r,g as i,h as a,i as o,l as s,m as c,n as l,o as u,p as d,r as f,s as p,t as m,u as h}from"../parts-schema-Vv0EJCjj.mjs";import{i as g,n as _,r as v,t as y}from"../schema-registry-JU4gsC8E.mjs";import{z as b}from"zod";const x=`commerce:mapping`,S=b.object({mapping:t}),C=new WeakMap;function w(e){let t=C.get(e);if(t)return t;let{mapping:n}=S.parse(e),r=_(n.targetSchema);if(!r)throw Error(`Unknown target schema "${n.targetSchema}" (registered: ${v().join(`, `)})`);let i=u(n,r);return C.set(e,i),i}const T=async(e,t)=>w(t.params)(e,t);export{n as COLUMN_TRANSFORM_OPS,x as MAPPING_TRANSFORM_NAME,r as applyColumnTransforms,u as buildMappingTransform,d as columnTransformSchema,y as describeTargetSchemas,g as discountTableSchema,_ as getTargetSchema,v as listTargetSchemas,p as mappingColumnSchema,t as mappingConfigSchema,T as mappingTransform,f as numberCell,s as numberFormatSchema,c as parseEuropeanNumber,a as parseNumber,m as partsParamsSchema,l as partsSchema,h as resolveMapping,o as rowError,e as textOrNull,i as toMinorUnits};
|
|
2
2
|
//# sourceMappingURL=index.mjs.map
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { n as RowResult } from "./transform-
|
|
1
|
+
import { n as RowResult } from "./transform-BRJJLSXE.mjs";
|
|
2
2
|
import { z } from "zod";
|
|
3
3
|
import { PartsDocument } from "@murumets-ee/search-elasticsearch";
|
|
4
4
|
|
|
@@ -248,6 +248,14 @@ declare const partsParamsSchema: z.ZodObject<{
|
|
|
248
248
|
supplierDisplayName: z.ZodString;
|
|
249
249
|
importBatchId: z.ZodString; /** Supplier code prefix to strip before normalization (e.g. `'ME-'`). Optional. */
|
|
250
250
|
supplierCodePrefix: z.ZodOptional<z.ZodString>;
|
|
251
|
+
/**
|
|
252
|
+
* Purchase-discount map for this supplier (commerce-demo PR4, phase 03 §2):
|
|
253
|
+
* `{ groupCode → discountPct }`, resolved from `supplier_discount_group` at
|
|
254
|
+
* enqueue time and snapshotted onto the run so `buildRow` stays a pure function
|
|
255
|
+
* (no per-row DB I/O; `@murumets-ee/imports` stays a leaf). Absent for direct-net
|
|
256
|
+
* suppliers + the carmaker path. Percentages are 0–100 (the entity constrains them).
|
|
257
|
+
*/
|
|
258
|
+
discountGroups: z.ZodOptional<z.ZodRecord<z.ZodString, z.ZodNumber>>;
|
|
251
259
|
}, "strip", z.ZodTypeAny, {
|
|
252
260
|
supplierId: string;
|
|
253
261
|
brandId: string;
|
|
@@ -255,6 +263,7 @@ declare const partsParamsSchema: z.ZodObject<{
|
|
|
255
263
|
supplierDisplayName: string;
|
|
256
264
|
importBatchId: string;
|
|
257
265
|
supplierCodePrefix?: string | undefined;
|
|
266
|
+
discountGroups?: Record<string, number> | undefined;
|
|
258
267
|
}, {
|
|
259
268
|
supplierId: string;
|
|
260
269
|
brandId: string;
|
|
@@ -262,6 +271,7 @@ declare const partsParamsSchema: z.ZodObject<{
|
|
|
262
271
|
supplierDisplayName: string;
|
|
263
272
|
importBatchId: string;
|
|
264
273
|
supplierCodePrefix?: string | undefined;
|
|
274
|
+
discountGroups?: Record<string, number> | undefined;
|
|
265
275
|
}>;
|
|
266
276
|
type PartsParams = z.infer<typeof partsParamsSchema>;
|
|
267
277
|
/**
|
|
@@ -272,4 +282,4 @@ type PartsParams = z.infer<typeof partsParamsSchema>;
|
|
|
272
282
|
declare const partsSchema: TargetSchema<PartsDocument>;
|
|
273
283
|
//#endregion
|
|
274
284
|
export { toMinorUnits as S, applyColumnTransforms as _, MappedRow as a, parseEuropeanNumber as b, TargetSchema as c, textOrNull as d, COLUMN_TRANSFORM_OPS as f, ColumnTransformOpDescriptor as g, ColumnTransformOp as h, BuildRowContext as i, numberCell as l, ColumnTransform as m, partsParamsSchema as n, NumberOutcome as o, CellValue as p, partsSchema as r, TargetField as s, PartsParams as t, rowError as u, columnTransformSchema as v, parseNumber as x, NumberFormat as y };
|
|
275
|
-
//# sourceMappingURL=parts-schema-
|
|
285
|
+
//# sourceMappingURL=parts-schema-Bqau0MrJ.d.mts.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"parts-schema-
|
|
1
|
+
{"version":3,"file":"parts-schema-Bqau0MrJ.d.mts","names":[],"sources":["../src/mapping/number-format.ts","../src/mapping/column-transforms.ts","../src/mapping/target-schema.ts","../src/mapping/parts-schema.ts"],"mappings":";;;;;;;;;;AAqBA;;;;AAIW;AAsBX;;;;;;;;AAA6D;AA+B7D;AAAA,UAzDiB,YAAA;;EAEf,OAAA;EAuDwC;EArDxC,SAAS;AAAA;;;AA+EkC;;;;AC9E7C;;;;;;;;;;iBDqBgB,WAAA,CAAY,GAAA,UAAa,MAAA,EAAQ,YAAY;;;;;;;;iBA+B7C,YAAA,CAAa,KAAa;;;;;;;;;;;;;;;;;;;;;;;iBA0B1B,mBAAA,CAAoB,CAAS;;;;;AAzDgB;AA+B7D;;cCpDa,qBAAA,EAAqB,CAAA,CAAA,qBAAA,QAAA,CAAA,CAAA,SAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;KAYtB,eAAA,GAAkB,CAAA,CAAE,KAAK,QAAQ,qBAAA;;KAGjC,iBAAA,GAAoB,eAAe;;UAG9B,2BAAA;EACf,EAAA,EAAI,iBAAiB;EACrB,KAAA;;EAEA,QAAA;AAAA;;;;;;cAQW,oBAAA,WAA+B,2BAA2B;;KAY3D,SAAA;;;;;;;;iBASI,qBAAA,CACd,GAAA,UACA,UAAA,WAAqB,eAAA,IACrB,MAAA,EAAQ,YAAA,GACP,SAAA;;;;UC/Dc,WAAA;EF6B4C;EE3B3D,GAAA;EF0Dc;EExDd,KAAA;;EAEA,QAAA;EFsDwC;EEpDxC,IAAA;AAAA;;KAIU,SAAA,GAAY,MAAM,SAAS,SAAA;AF0EM;AAAA,UEvE5B,eAAA;EACf,WAAA;EACA,SAAA;EDTW;;;;ECcX,MAAA,EAAQ,MAAM;AAAA;;UAIC,YAAA;;EAEf,IAAA;;EAEA,MAAA,WAAiB,WAAA;;EAEjB,QAAA,CAAS,MAAA,EAAQ,SAAA,EAAW,GAAA,EAAK,eAAA,GAAkB,SAAA,CAAU,IAAA;AAAA;;;;;;iBAY/C,UAAA,CAAW,CAAwB,EAArB,SAAS;;KAO3B,aAAA;EACN,IAAA;EAAe,KAAA;AAAA;EACf,IAAA;AAAA;EACA,IAAA;AAAA;;;;;;iBAOU,UAAA,CAAW,CAAA,EAAG,SAAA,eAAwB,aAAa;;iBASnD,QAAA,CAAS,SAAA,UAAmB,OAAA,UAAiB,KAAA;;;;;;;;;;AFgBhB;;;;AC9E7C;;AD8E6C,cG1DhC,iBAAA,EAAiB,CAAA,CAAA,SAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;KAqBlB,WAAA,GAAc,CAAA,CAAE,KAAK,QAAQ,iBAAA;;;;;;cA6D5B,WAAA,EAAa,YAAY,CAAC,aAAA"}
|
|
@@ -1,2 +1,2 @@
|
|
|
1
|
-
import{z as e}from"zod";const t=/^-?\d+(\.\d+)?$/;function n(e,n){let r=e.trim();if(r===``||r.toUpperCase()===`NA`)return null;let{decimal:i,thousands:a}=n,o=r;if(a!==``&&(o=o.split(a).join(``)),i===`,`){if(o.includes(`.`))return NaN;o=o.split(`,`).join(`.`)}else if(o.includes(`,`))return NaN;return t.test(o)?Number.parseFloat(o):NaN}function r(e){return Math.round(e*100)}function i(e){let n=e.trim();if(n===``||n.toUpperCase()===`NA`)return null;let r=n.includes(`,`),i=n.includes(`.`),a;if(r&&i)a=n.replace(/\./g,``).replace(`,`,`.`);else if(r)a=n.replace(`,`,`.`);else if(i)return NaN;else a=n;return t.test(a)?Number.parseFloat(a):NaN}const a=e.discriminatedUnion(`op`,[e.object({op:e.literal(`trim`)}),e.object({op:e.literal(`uppercase`)}),e.object({op:e.literal(`lowercase`)}),e.object({op:e.literal(`stripPrefix`),value:e.string().min(1).max(32)}),e.object({op:e.literal(`stripSuffix`),value:e.string().min(1).max(32)}),e.object({op:e.literal(`defaultIfEmpty`),value:e.string().max(255)}),e.object({op:e.literal(`number`)}),e.object({op:e.literal(`europeanNumber`)})]),o=[{op:`trim`,label:`Trim whitespace`,hasValue:!1},{op:`uppercase`,label:`Uppercase`,hasValue:!1},{op:`lowercase`,label:`Lowercase`,hasValue:!1},{op:`stripPrefix`,label:`Strip prefix`,hasValue:!0},{op:`stripSuffix`,label:`Strip suffix`,hasValue:!0},{op:`defaultIfEmpty`,label:`Default if empty`,hasValue:!0},{op:`number`,label:`Parse number (uses number format)`,hasValue:!1},{op:`europeanNumber`,label:`Parse European number (1.234,56)`,hasValue:!1}];function s(e,t,r){let a=e;for(let e of t)if(typeof a==`string`)switch(e.op){case`trim`:a=a.trim();break;case`uppercase`:a=a.toUpperCase();break;case`lowercase`:a=a.toLowerCase();break;case`stripPrefix`:a=a.startsWith(e.value)?a.slice(e.value.length):a;break;case`stripSuffix`:a=a.endsWith(e.value)?a.slice(0,a.length-e.value.length):a;break;case`defaultIfEmpty`:a=a.trim()===``?e.value:a;break;case`number`:a=n(a,r);break;case`europeanNumber`:a=i(a);break}return a}const c=e.object({source:e.string().min(1).max(256),target:e.string().min(1).max(64),transforms:e.array(a).default([])}),l=e.object({decimal:e.enum([`.`,`,`]),thousands:e.enum([``,`.`,`,`,` `])}).refine(e=>e.decimal!==e.thousands,{message:`decimal and thousands separators must differ`,path:[`thousands`]}),u=e.object({targetSchema:e.string().min(1).max(64),numberFormat:l,columnMap:e.array(c)});function d(e,t){let n=new Set(t.fields.map(e=>e.key));for(let r of e.columnMap)if(!n.has(r.target))throw Error(`Mapping target "${r.target}" is not a field of schema "${t.name}" (allowed: ${[...n].join(`, `)})`);let r=e.numberFormat,i=e.columnMap;return async(e,n)=>{let a={};for(let t of i){let n=e[t.source]??``;a[t.target]=s(n,t.transforms,r)}return t.buildRow(a,{importRunId:n.importRunId,rowNumber:n.rowNumber,params:n.params})}}function f(e,t){let n=e.filter(e=>e.active);if(t!==null){let e=n.find(e=>e.brandId===t);if(e)return e}return n.find(e=>e.brandId===null)}function p(e){if(typeof e!=`string`)return null;let t=e.trim();return t===``?null:t}function m(e){return e==null?{kind:`absent`}:typeof e==`number`?Number.isNaN(e)?{kind:`invalid`}:{kind:`value`,value:e}:e.trim()===``?{kind:`absent`}:{kind:`invalid`}}function h(e,t,n){return{kind:`error`,error:{errorType:e,message:t,field:n}}}const g=e.object({brandId:e.string().uuid(),brandSlug:e.string().min(1).max(64),supplierId:e.string().uuid(),supplierDisplayName:e.string().min(1).max(255),importBatchId:e.string().uuid(),supplierCodePrefix:e.string().max(8).regex(/^[A-Z0-9-]+$/,`must be uppercase alphanumeric or "-"`).optional()}),_=new WeakMap;function v(e){let t=_.get(e);if(t)return t;let n=g.parse(e);return _.set(e,n),n}const y=[`de`,`en`,`es`,`fr`,`it`,`nl`,`pt`,`ro`],b=e=>`name_${e}`,x=[{key:`code`,label:`Code`,required:!0,type:`string`},{key:`net_price`,label:`Net price`,required:!1,type:`number`},{key:`gross_price`,label:`Gross price`,required:!1,type:`number`},{key:`list_price`,label:`List price`,required:!1,type:`number`},{key:`currency`,label:`Currency`,required:!1,type:`string`},{key:`barcode`,label:`Barcode`,required:!1,type:`string`},{key:`description1`,label:`Description 1`,required:!1,type:`string`},{key:`description2`,label:`Description 2`,required:!1,type:`string`},{key:`discount_group`,label:`Discount group`,required:!1,type:`string`},{key:`superseding_code`,label:`Superseding code`,required:!1,type:`string`},...y.map(e=>({key:b(e),label:`Name (${e})`,required:!1,type:`string`}))],S=/[\s-]/g;function C(e){return e.toUpperCase().replace(S,``)}const
|
|
2
|
-
//# sourceMappingURL=parts-schema-
|
|
1
|
+
import{z as e}from"zod";const t=/^-?\d+(\.\d+)?$/;function n(e,n){let r=e.trim();if(r===``||r.toUpperCase()===`NA`)return null;let{decimal:i,thousands:a}=n,o=r;if(a!==``&&(o=o.split(a).join(``)),i===`,`){if(o.includes(`.`))return NaN;o=o.split(`,`).join(`.`)}else if(o.includes(`,`))return NaN;return t.test(o)?Number.parseFloat(o):NaN}function r(e){return Math.round(e*100)}function i(e){let n=e.trim();if(n===``||n.toUpperCase()===`NA`)return null;let r=n.includes(`,`),i=n.includes(`.`),a;if(r&&i)a=n.replace(/\./g,``).replace(`,`,`.`);else if(r)a=n.replace(`,`,`.`);else if(i)return NaN;else a=n;return t.test(a)?Number.parseFloat(a):NaN}const a=e.discriminatedUnion(`op`,[e.object({op:e.literal(`trim`)}),e.object({op:e.literal(`uppercase`)}),e.object({op:e.literal(`lowercase`)}),e.object({op:e.literal(`stripPrefix`),value:e.string().min(1).max(32)}),e.object({op:e.literal(`stripSuffix`),value:e.string().min(1).max(32)}),e.object({op:e.literal(`defaultIfEmpty`),value:e.string().max(255)}),e.object({op:e.literal(`number`)}),e.object({op:e.literal(`europeanNumber`)})]),o=[{op:`trim`,label:`Trim whitespace`,hasValue:!1},{op:`uppercase`,label:`Uppercase`,hasValue:!1},{op:`lowercase`,label:`Lowercase`,hasValue:!1},{op:`stripPrefix`,label:`Strip prefix`,hasValue:!0},{op:`stripSuffix`,label:`Strip suffix`,hasValue:!0},{op:`defaultIfEmpty`,label:`Default if empty`,hasValue:!0},{op:`number`,label:`Parse number (uses number format)`,hasValue:!1},{op:`europeanNumber`,label:`Parse European number (1.234,56)`,hasValue:!1}];function s(e,t,r){let a=e;for(let e of t)if(typeof a==`string`)switch(e.op){case`trim`:a=a.trim();break;case`uppercase`:a=a.toUpperCase();break;case`lowercase`:a=a.toLowerCase();break;case`stripPrefix`:a=a.startsWith(e.value)?a.slice(e.value.length):a;break;case`stripSuffix`:a=a.endsWith(e.value)?a.slice(0,a.length-e.value.length):a;break;case`defaultIfEmpty`:a=a.trim()===``?e.value:a;break;case`number`:a=n(a,r);break;case`europeanNumber`:a=i(a);break}return a}const c=e.object({source:e.string().min(1).max(256),target:e.string().min(1).max(64),transforms:e.array(a).default([])}),l=e.object({decimal:e.enum([`.`,`,`]),thousands:e.enum([``,`.`,`,`,` `])}).refine(e=>e.decimal!==e.thousands,{message:`decimal and thousands separators must differ`,path:[`thousands`]}),u=e.object({targetSchema:e.string().min(1).max(64),numberFormat:l,columnMap:e.array(c)});function d(e,t){let n=new Set(t.fields.map(e=>e.key));for(let r of e.columnMap)if(!n.has(r.target))throw Error(`Mapping target "${r.target}" is not a field of schema "${t.name}" (allowed: ${[...n].join(`, `)})`);let r=e.numberFormat,i=e.columnMap;return async(e,n)=>{let a={};for(let t of i){let n=e[t.source]??``;a[t.target]=s(n,t.transforms,r)}return t.buildRow(a,{importRunId:n.importRunId,rowNumber:n.rowNumber,params:n.params})}}function f(e,t){let n=e.filter(e=>e.active);if(t!==null){let e=n.find(e=>e.brandId===t);if(e)return e}return n.find(e=>e.brandId===null)}function p(e){if(typeof e!=`string`)return null;let t=e.trim();return t===``?null:t}function m(e){return e==null?{kind:`absent`}:typeof e==`number`?Number.isNaN(e)?{kind:`invalid`}:{kind:`value`,value:e}:e.trim()===``?{kind:`absent`}:{kind:`invalid`}}function h(e,t,n){return{kind:`error`,error:{errorType:e,message:t,field:n}}}const g=e.object({brandId:e.string().uuid(),brandSlug:e.string().min(1).max(64),supplierId:e.string().uuid(),supplierDisplayName:e.string().min(1).max(255),importBatchId:e.string().uuid(),supplierCodePrefix:e.string().max(8).regex(/^[A-Z0-9-]+$/,`must be uppercase alphanumeric or "-"`).optional(),discountGroups:e.record(e.string(),e.number().min(0).max(100)).optional()}),_=new WeakMap;function v(e){let t=_.get(e);if(t)return t;let n=g.parse(e);return _.set(e,n),n}const y=[`de`,`en`,`es`,`fr`,`it`,`nl`,`pt`,`ro`],b=e=>`name_${e}`,x=[{key:`code`,label:`Code`,required:!0,type:`string`},{key:`net_price`,label:`Net price`,required:!1,type:`number`},{key:`gross_price`,label:`Gross price`,required:!1,type:`number`},{key:`list_price`,label:`List price`,required:!1,type:`number`},{key:`currency`,label:`Currency`,required:!1,type:`string`},{key:`barcode`,label:`Barcode`,required:!1,type:`string`},{key:`description1`,label:`Description 1`,required:!1,type:`string`},{key:`description2`,label:`Description 2`,required:!1,type:`string`},{key:`discount_group`,label:`Discount group`,required:!1,type:`string`},{key:`superseding_code`,label:`Superseding code`,required:!1,type:`string`},...y.map(e=>({key:b(e),label:`Name (${e})`,required:!1,type:`string`}))],S=/[\s-]/g;function C(e){return e.toUpperCase().replace(S,``)}function w(e,t){return Math.round(e*(1-t/100))}const T={name:`parts`,fields:x,buildRow(e,t){let n=v(t.params),i=typeof e.code==`string`?e.code.trim():``;if(!i)return h(`missing_code`,`code is empty`,`code`);let a=C(n.supplierCodePrefix&&i.startsWith(n.supplierCodePrefix)?i.slice(n.supplierCodePrefix.length):i);if(!a)return h(`empty_normalized_code`,`Code "${i}" is empty after normalization`,`code`);let o=p(e.currency)??`EUR`;if(o!==`EUR`)return h(`unsupported_currency`,`Currency "${o}" not supported (EUR-only)`,`currency`);let s=m(e.net_price);if(s.kind===`invalid`)return h(`invalid_net_price`,`net_price is not a number`,`net_price`);let c=m(e.gross_price);if(c.kind===`invalid`)return h(`invalid_gross_price`,`gross_price is not a number`,`gross_price`);let l=m(e.list_price);if(l.kind===`invalid`)return h(`invalid_list_price`,`list_price is not a number`,`list_price`);let u=s.kind===`value`?s.value:null,d=l.kind===`value`?l.value:null,f=u??d;if(f===null)return h(`missing_price`,`no net_price or list_price provided`,`price`);let g=p(e.discount_group),_=d===null?null:r(d),y=null,b=[];if(u!==null)y=r(u);else if(_!==null){let e=g===null?void 0:n.discountGroups?.[g];e===void 0?b.push({warningType:`no_purchase_price`,message:g===null?`list price present but no discount group on the row — cost pending`:`no discount group "${g}" configured for this supplier — cost pending`,field:`discount_group`}):y=w(_,e)}let x=`${a}__${n.supplierId}`,S={doc_id:x,code:i,code_normalized:a,brand_id:n.brandId,brand_slug:n.brandSlug,supplier_id:n.supplierId,supplier_display_name:n.supplierDisplayName,net_price_eur:f,gross_price_eur:c.kind===`value`?c.value:null,currency:`EUR`,barcode:p(e.barcode),name_de:p(e.name_de),name_en:p(e.name_en),name_es:p(e.name_es),name_fr:p(e.name_fr),name_it:p(e.name_it),name_nl:p(e.name_nl),name_pt:p(e.name_pt),name_ro:p(e.name_ro),description1:p(e.description1),description2:p(e.description2),list_price_minor:_,discount_group:g,superseding_code:p(e.superseding_code),purchase_price_minor:y,core_charge_minor:null,core_charge_currency:null,core_charge_refundable:null,core_charge_tax_class:null,core_charge_tax_inclusive:null,import_batch_id:n.importBatchId,imported_at:new Date().toISOString()};return b.length>0?{kind:`success`,id:x,doc:S,warnings:b}:{kind:`success`,id:x,doc:S}}};export{p as a,u as c,o as d,s as f,r as g,n as h,h as i,l,i as m,T as n,d as o,a as p,m as r,c as s,g as t,f as u};
|
|
2
|
+
//# sourceMappingURL=parts-schema-Vv0EJCjj.mjs.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"parts-schema-Cuaq2_At.mjs","names":[],"sources":["../src/mapping/number-format.ts","../src/mapping/column-transforms.ts","../src/mapping/engine.ts","../src/mapping/target-schema.ts","../src/mapping/parts-schema.ts"],"sourcesContent":["/**\n * Number parsing for the configurable mapping engine (commerce-demo phase 02).\n *\n * Real supplier feeds (R03) disagree on decimal AND thousands separators:\n * `55.02` (dot decimal), `326,88` (comma decimal), `1.234,56` (EU\n * thousands+decimal), `1,234.56` (US), `1 234,56` (space-grouped). The mapping\n * stores an explicit `{ decimal, thousands }` per profile so there is no guessing\n * — `parseNumber` trusts the configured format.\n *\n * Two parsers live here:\n * - `parseNumber(raw, format)` — the format-driven parser used by the generic\n * `number` per-column transform.\n * - `parseEuropeanNumber(raw)` — the carmaker-feed legacy parser (comma decimal,\n * dot thousands, **bare-dot rejected as ambiguous**). Carries the carmaker feed's\n * documented format with two intentional broadenings vs the old inline parser:\n * it trims surrounding whitespace, and returns `null` (not `NaN`) for an empty /\n * `'NA'` cell so both number ops share absent-vs-malformed semantics. The\n * carmaker doc-shape suite (27 tests) is the regression net.\n */\n\n/** Separator config for {@link parseNumber}. Stored per `import_mapping`. */\nexport interface NumberFormat {\n /** Decimal separator. */\n decimal: '.' | ','\n /** Thousands grouping separator. `''` = none. */\n thousands: '' | '.' | ',' | ' '\n}\n\n/** Post-cleanup numeric form: optional sign, digits, optional single decimal. */\nconst STRICT_NUMBER_RE = /^-?\\d+(\\.\\d+)?$/\n\n/**\n * Parse a feed cell into a number using an EXPLICIT `{ decimal, thousands }`\n * format. No ambiguity: thousands separators are stripped, the decimal separator\n * is normalised to `.`, and the result is strict-validated.\n *\n * Returns:\n * - `null` for an empty / `'NA'` cell (a legitimately-absent value — the caller\n * decides whether that's allowed for a given target field).\n * - `NaN` for a non-empty but unparseable cell (malformed — the caller surfaces\n * a row-level error rather than coercing to 0).\n * - the parsed number otherwise.\n *\n * A character that is neither a digit, the configured decimal, the configured\n * thousands separator, nor a leading `-` makes the cell unparseable (`NaN`) —\n * e.g. a literal `.` when the format's decimal is `,` and thousands is not `.`.\n */\nexport function parseNumber(raw: string, format: NumberFormat): number | null {\n const trimmed = raw.trim()\n if (trimmed === '' || trimmed.toUpperCase() === 'NA') return null\n\n const { decimal, thousands } = format\n let s = trimmed\n // 1. Remove every thousands separator (when one is configured).\n if (thousands !== '') s = s.split(thousands).join('')\n\n // 2. Normalise the decimal separator to `.`. After thousands removal, the\n // \"other\" separator must not appear — if it does, the cell doesn't match\n // the declared format and is unparseable.\n if (decimal === ',') {\n if (s.includes('.')) return Number.NaN\n s = s.split(',').join('.')\n } else {\n // decimal === '.'\n if (s.includes(',')) return Number.NaN\n }\n\n if (!STRICT_NUMBER_RE.test(s)) return Number.NaN\n return Number.parseFloat(s)\n}\n\n/**\n * Convert a major-unit money value (e.g. `55.02` EUR) to integer MINOR units\n * (`5502`). Rounds to the nearest cent to absorb float drift. The commerce-demo\n * PartsDocument money fields (`list_price_minor`, `purchase_price_minor`,\n * `core_charge_minor`) are integer minor units so derived prices (PR4) don't\n * accrue rounding error — see F003 in progress.md.\n */\nexport function toMinorUnits(value: number): number {\n return Math.round(value * 100)\n}\n\n/**\n * Parse a European-formatted decimal: `,` is the decimal separator, `.` is the\n * (optional) thousands separator. Returns `null` for an empty / `'NA'` cell and\n * `NaN` on unparseable (malformed) input.\n *\n * This is the carmaker-feed legacy parser, preserved from the hardcoded transform.\n * Crucially it **rejects bare-dot** input (`1.98`, `1.234`) as ambiguous rather\n * than guessing — the carmaker feeds are documented as comma-decimal, and a `.`\n * without a `,` could mean a US-format decimal OR a thousands marker. The\n * format-driven {@link parseNumber} doesn't need this guard (its separators are\n * explicit); the carmaker path keeps it.\n *\n * The `null` (absent) vs `NaN` (malformed) split mirrors {@link parseNumber} so\n * the mapping engine treats both number ops uniformly: a target field's `buildRow`\n * reads `null` as \"not provided\" and `NaN` as \"invalid\". This is what lets the\n * reimplemented carmaker transform keep its `GrossPrice 'NA' → null` behaviour\n * (optional field) while still erroring on a malformed `NetPrice` (required).\n *\n * Cases: `'1,98'`→`1.98`, `'1.234,56'`→`1234.56`, `'5'`→`5`, `'-5,00'`→`-5`,\n * empty / `'NA'` → `null`. Rejected (`NaN`): bare-dot (`'1.98'`, `'1.234'`),\n * trailing junk (`'1,98EUR'`), multiple commas (`'1,2,3'`), other non-numeric.\n */\nexport function parseEuropeanNumber(s: string): number | null {\n const trimmed = s.trim()\n if (trimmed === '' || trimmed.toUpperCase() === 'NA') return null\n const hasComma = trimmed.includes(',')\n const hasDot = trimmed.includes('.')\n let cleaned: string\n if (hasComma && hasDot) {\n // European thousands+decimal: strip every `.`, then swap the (sole) `,` for\n // `.`. The strict regex below catches multi-comma input.\n cleaned = trimmed.replace(/\\./g, '').replace(',', '.')\n } else if (hasComma) {\n cleaned = trimmed.replace(',', '.')\n } else if (hasDot) {\n return Number.NaN\n } else {\n cleaned = trimmed\n }\n if (!STRICT_NUMBER_RE.test(cleaned)) return Number.NaN\n return Number.parseFloat(cleaned)\n}\n","/**\n * Per-column transforms for the mapping engine (commerce-demo phase 02, D005).\n *\n * A **bounded, enum-keyed** library of cleaning operations applied to a source\n * cell before it lands on a target field. This is the security model from R02:\n * transforms are declarative operations with Zod-validated params, NEVER\n * user-supplied executable code. The carmaker feed's logic decomposes exactly\n * into `stripPrefix` + `europeanNumber` (+ the schema's code-normalize).\n *\n * Number-producing ops (`number`, `europeanNumber`) are TERMINAL by convention —\n * they convert the string cell into `number | null`; string ops after them are\n * no-ops. The engine reads the result as `string | number | null` and hands it to\n * the target schema's `buildRow`.\n */\n\nimport { z } from 'zod'\nimport { type NumberFormat, parseEuropeanNumber, parseNumber } from './number-format.js'\n\n/** Max length for a strip-affix value — bounds the param, no ReDoS surface. */\nconst MAX_AFFIX_LEN = 32\n\n/**\n * Discriminated union of bounded transform ops. The TS type is INFERRED from\n * this schema so there's a single source of truth + the snapshot is validated at\n * the engine boundary.\n */\nexport const columnTransformSchema = z.discriminatedUnion('op', [\n z.object({ op: z.literal('trim') }),\n z.object({ op: z.literal('uppercase') }),\n z.object({ op: z.literal('lowercase') }),\n z.object({ op: z.literal('stripPrefix'), value: z.string().min(1).max(MAX_AFFIX_LEN) }),\n z.object({ op: z.literal('stripSuffix'), value: z.string().min(1).max(MAX_AFFIX_LEN) }),\n z.object({ op: z.literal('defaultIfEmpty'), value: z.string().max(255) }),\n z.object({ op: z.literal('number') }),\n z.object({ op: z.literal('europeanNumber') }),\n])\n\n/** One declarative cleaning operation. */\nexport type ColumnTransform = z.infer<typeof columnTransformSchema>\n\n/** The discriminator literal of a {@link ColumnTransform}. */\nexport type ColumnTransformOp = ColumnTransform['op']\n\n/** UI descriptor for one transform op: its key, a label, and whether it takes a `value` param. */\nexport interface ColumnTransformOpDescriptor {\n op: ColumnTransformOp\n label: string\n /** `true` → the op carries a `value` string param (`stripPrefix`/`stripSuffix`/`defaultIfEmpty`). */\n hasValue: boolean\n}\n\n/**\n * Static catalogue of the available per-column transform ops, for the PR3 wizard's\n * transform picker. The ORDER is the suggested UI order; `number`/`europeanNumber`\n * are terminal (see {@link applyColumnTransforms}) and labelled accordingly.\n */\nexport const COLUMN_TRANSFORM_OPS: readonly ColumnTransformOpDescriptor[] = [\n { op: 'trim', label: 'Trim whitespace', hasValue: false },\n { op: 'uppercase', label: 'Uppercase', hasValue: false },\n { op: 'lowercase', label: 'Lowercase', hasValue: false },\n { op: 'stripPrefix', label: 'Strip prefix', hasValue: true },\n { op: 'stripSuffix', label: 'Strip suffix', hasValue: true },\n { op: 'defaultIfEmpty', label: 'Default if empty', hasValue: true },\n { op: 'number', label: 'Parse number (uses number format)', hasValue: false },\n { op: 'europeanNumber', label: 'Parse European number (1.234,56)', hasValue: false },\n] as const\n\n/** The value a column carries as it moves through the transform pipeline. */\nexport type CellValue = string | number | null\n\n/**\n * Apply a transform pipeline to a raw cell, left-to-right. String ops act on the\n * string value; the `number` / `europeanNumber` ops are terminal (convert to\n * `number | null`, with `NaN` for a malformed numeric — the schema's `buildRow`\n * surfaces that as a row error). Once the value is non-string, later string ops\n * pass through unchanged.\n */\nexport function applyColumnTransforms(\n raw: string,\n transforms: readonly ColumnTransform[],\n format: NumberFormat,\n): CellValue {\n let value: CellValue = raw\n for (const t of transforms) {\n if (typeof value !== 'string') {\n // A prior number op already produced number|null — string ops are no-ops\n // past this point (number ops are terminal by convention).\n continue\n }\n switch (t.op) {\n case 'trim':\n value = value.trim()\n break\n case 'uppercase':\n value = value.toUpperCase()\n break\n case 'lowercase':\n value = value.toLowerCase()\n break\n case 'stripPrefix':\n value = value.startsWith(t.value) ? value.slice(t.value.length) : value\n break\n case 'stripSuffix':\n value = value.endsWith(t.value) ? value.slice(0, value.length - t.value.length) : value\n break\n case 'defaultIfEmpty':\n value = value.trim() === '' ? t.value : value\n break\n case 'number':\n value = parseNumber(value, format)\n break\n case 'europeanNumber':\n value = parseEuropeanNumber(value)\n break\n }\n }\n return value\n}\n","/**\n * Mapping engine (commerce-demo phase 02, D005).\n *\n * Turns a declarative `MappingConfig` (which source column → which whitelisted\n * target field, plus per-column transforms + a number format) into a\n * `RowTransform` the existing runner drives. The pipeline per row is:\n *\n * raw cells → applyColumnTransforms(per column) → MappedRow → schema.buildRow\n *\n * Two whitelists guard the boundary (R02 injection model): the `target_schema`\n * name resolves to a registered `TargetSchema`, and every `column_map[].target`\n * must be one of that schema's fields — enforced here at build time (throws),\n * before any row is processed.\n *\n * `resolveMapping` implements the D002 most-specific ladder used at enqueue time\n * to pick the `(supplier, brand)` mapping over the `(supplier, NULL)` default.\n */\n\nimport { z } from 'zod'\nimport type { RowTransform } from '../transform.js'\nimport { applyColumnTransforms, columnTransformSchema } from './column-transforms.js'\nimport type { MappedRow, TargetSchema } from './target-schema.js'\n\n/** One column-map entry. `target` is checked against the schema whitelist separately. */\nexport const mappingColumnSchema = z.object({\n /** Source header name (or positional index string for headerless feeds). */\n source: z.string().min(1).max(256),\n /** Target field key — must be in the chosen schema's whitelist (enforced in `buildMappingTransform`). */\n target: z.string().min(1).max(64),\n /** Ordered per-column cleaning ops. Defaults to none. */\n transforms: z.array(columnTransformSchema).default([]),\n})\nexport type MappingColumn = z.infer<typeof mappingColumnSchema>\n\n/**\n * Decimal + thousands separators for the `number` transform.\n *\n * The two MUST differ: `parseNumber` strips every thousands separator FIRST, so a\n * config where `decimal === thousands` (e.g. both `,`) would strip the decimal\n * separator too — `326,88` → `32688`, a silent 100× error rather than a row error.\n * Rejected at save (`assertValidMapping`) AND at run (`mappingConfigSchema`).\n */\nexport const numberFormatSchema = z\n .object({\n decimal: z.enum(['.', ',']),\n thousands: z.enum(['', '.', ',', ' ']),\n })\n .refine((f) => f.decimal !== f.thousands, {\n message: 'decimal and thousands separators must differ',\n path: ['thousands'],\n })\n\n/**\n * The portable mapping config — the subset of an `import_mapping` row the engine\n * needs. The enqueuer SNAPSHOTS this into `import_run.params.mapping` so a run\n * records the exact mapping it used (reproducible) and the transform stays a pure\n * function of `(row, ctx)` with no per-row DB I/O.\n */\nexport const mappingConfigSchema = z.object({\n targetSchema: z.string().min(1).max(64),\n numberFormat: numberFormatSchema,\n columnMap: z.array(mappingColumnSchema),\n})\nexport type MappingConfig = z.infer<typeof mappingConfigSchema>\n\n/**\n * Build a `RowTransform` from a validated mapping config + its target schema.\n *\n * Throws at BUILD time if any `column_map` target is outside the schema's\n * whitelist (deny-by-default, R02) — so a misconfigured mapping fails loudly\n * before a single row is written, rather than silently dropping a column.\n *\n * Generic over the schema's output type, so passing the concrete `partsSchema`\n * yields a `RowTransform<PartsDocument>` (the carmaker path keeps its typing); the\n * registry-resolved `TargetSchema<unknown>` yields `RowTransform<unknown>` (the\n * `commerce:mapping` path, where the output type isn't known until runtime).\n */\nexport function buildMappingTransform<TOut>(\n config: MappingConfig,\n schema: TargetSchema<TOut>,\n): RowTransform<TOut> {\n const allowed = new Set(schema.fields.map((f) => f.key))\n for (const col of config.columnMap) {\n if (!allowed.has(col.target)) {\n throw new Error(\n `Mapping target \"${col.target}\" is not a field of schema \"${schema.name}\" ` +\n `(allowed: ${[...allowed].join(', ')})`,\n )\n }\n }\n\n const format = config.numberFormat\n const columnMap = config.columnMap\n\n return async (row, ctx) => {\n const mapped: MappedRow = {}\n for (const col of columnMap) {\n // `source` is read as data from the parsed row map — never used as a key\n // into framework objects (R02). A missing source column yields `''`.\n const raw = row[col.source] ?? ''\n mapped[col.target] = applyColumnTransforms(raw, col.transforms, format)\n }\n return schema.buildRow(mapped, {\n importRunId: ctx.importRunId,\n rowNumber: ctx.rowNumber,\n params: ctx.params,\n })\n }\n}\n\n/** The minimal shape `resolveMapping` needs — satisfied by an `import_mapping` row or a test fixture. */\nexport interface ResolvableMapping {\n /** `null` = supplier default (applies to all brands). */\n brandId: string | null\n active: boolean\n}\n\n/**\n * Most-specific resolution (D002 ladder): given the candidate mappings for one\n * `(supplier, target_schema)` and the requested brand, return the brand-specific\n * mapping if one exists, else the supplier-default (`brandId === null`). Inactive\n * mappings are skipped. Returns `undefined` when nothing matches.\n */\nexport function resolveMapping<M extends ResolvableMapping>(\n mappings: readonly M[],\n brandId: string | null,\n): M | undefined {\n const active = mappings.filter((m) => m.active)\n if (brandId !== null) {\n const specific = active.find((m) => m.brandId === brandId)\n if (specific) return specific\n }\n return active.find((m) => m.brandId === null)\n}\n","/**\n * TargetSchema — the contract a mapping writes INTO (commerce-demo phase 02, D005).\n *\n * A mapping's `column_map` points source columns at a schema's whitelisted target\n * fields; the engine applies the per-column transforms to build a `MappedRow`, then\n * hands it to `schema.buildRow` to produce the typed output document (an ES\n * `PartsDocument` for `parts`, a discount row for `discount_table`).\n *\n * The `fields` list is the SECOND injection-boundary whitelist (the first is the\n * registered schema name): a `column_map` entry whose `target` is not a field key\n * here is rejected at validation time (R02). Two schemas ship in PR2 — `parts` and\n * `discount_table`.\n */\n\nimport type { RowResult } from '../transform.js'\nimport type { CellValue } from './column-transforms.js'\n\n/** One whitelisted target field on a schema. */\nexport interface TargetField {\n /** Stable target key referenced by `column_map[].target`. */\n key: string\n /** Human label for the wizard dropdown (PR3). */\n label: string\n /** Whether a mapping for this schema MUST include this target. */\n required: boolean\n /** Coarse type — drives the wizard's transform suggestions; `buildRow` does the real coercion. */\n type: 'string' | 'number'\n}\n\n/** The per-row product of the engine: target field key → cleaned value. */\nexport type MappedRow = Record<string, CellValue>\n\n/** Per-row context handed to `buildRow`. `params` is the opaque per-run params object. */\nexport interface BuildRowContext {\n importRunId: string\n rowNumber: number\n /**\n * Opaque per-run params (the parts schema reads brand/supplier identity from\n * it; `discount_table` ignores it). The schema validates the shape it needs.\n */\n params: Record<string, unknown>\n}\n\n/** A registered target schema: a whitelist + a row builder. */\nexport interface TargetSchema<TOut> {\n /** Registered name — also the `import_mapping.target_schema` value. */\n name: string\n /** The target-field whitelist (injection boundary). */\n fields: readonly TargetField[]\n /** Build a typed output (success) / skip / error from a mapped row. */\n buildRow(mapped: MappedRow, ctx: BuildRowContext): RowResult<TOut>\n}\n\n// ---------------------------------------------------------------------------\n// Shared `buildRow` helpers (used by every concrete schema).\n// ---------------------------------------------------------------------------\n\n/**\n * A nullable text cell → trimmed string, or `null` for empty / non-string. Accepts\n * `undefined` (an UNMAPPED target — indexed access under `noUncheckedIndexedAccess`)\n * and treats it as absent.\n */\nexport function textOrNull(v: CellValue | undefined): string | null {\n if (typeof v !== 'string') return null\n const t = v.trim()\n return t === '' ? null : t\n}\n\n/** The three outcomes a `number` / `europeanNumber` op produces for a cell. */\nexport type NumberOutcome =\n | { kind: 'value'; value: number }\n | { kind: 'absent' }\n | { kind: 'invalid' }\n\n/**\n * Classify a cell that should hold a number. `undefined` (unmapped) / `null` →\n * absent; a real number → value (`NaN` → invalid); a non-empty raw string (no\n * number op declared) → invalid; a blank raw string → absent.\n */\nexport function numberCell(v: CellValue | undefined): NumberOutcome {\n if (v === undefined || v === null) return { kind: 'absent' }\n if (typeof v === 'number') {\n return Number.isNaN(v) ? { kind: 'invalid' } : { kind: 'value', value: v }\n }\n return v.trim() === '' ? { kind: 'absent' } : { kind: 'invalid' }\n}\n\n/** Build a `RowResult` error variant (assignable to any `TargetSchema<TOut>`). */\nexport function rowError(errorType: string, message: string, field: string) {\n return { kind: 'error' as const, error: { errorType, message, field } }\n}\n","/**\n * `parts` target schema — maps a cleaned feed row into the ES `PartsDocument`\n * (commerce-demo phase 02, D005). This is the SHARED assembly + validation logic\n * the reimplemented carmaker transform and every operator-built `parts` mapping\n * run through, so the row taxonomy (code normalization, currency gate, price\n * validation) is uniform across all supplier feeds.\n *\n * **PR2 field coverage.** This schema populates the direct, no-derivation fields:\n * the existing `PartsDocument` columns plus `list_price_minor`, `discount_group`,\n * and `superseding_code` (string/value passthroughs that PR4 / PR6 consume). The\n * DERIVED `purchase_price_minor` (PR4) and all `core_charge_*` fields (PR7) stay\n * null; `core_charge` is deliberately NOT in the PR2 whitelist (deny-by-default —\n * PR7 adds it with its `*S`-row + column ingestion). Replacement-pair and\n * core-charge SIDE EFFECTS (separate entities) are PR6/PR7.\n *\n * **Money.** `list_price_minor` is integer minor units via `toMinorUnits`\n * (F003). The legacy `net_price_eur` stays a major-unit float for back-compat; it\n * holds the direct net price, or — for a two-file LIST feed with no direct net —\n * the list price as a display placeholder until PR4 derives `purchase_price_minor`.\n */\n\nimport type { PartsDocument } from '@murumets-ee/search-elasticsearch'\nimport { z } from 'zod'\nimport { toMinorUnits } from './number-format.js'\nimport {\n type BuildRowContext,\n type MappedRow,\n numberCell,\n rowError,\n type TargetField,\n type TargetSchema,\n textOrNull,\n} from './target-schema.js'\n\n/**\n * Params the `parts` schema reads from the run — the per-(supplier, brand)\n * identity resolved at enqueue time. Identical to the carmaker transform's params\n * (the carmaker path re-exports this), so brand/supplier injection + the optional\n * supplier code-prefix strip are validated the same way everywhere.\n */\nexport const partsParamsSchema = z.object({\n brandId: z.string().uuid(),\n brandSlug: z.string().min(1).max(64),\n supplierId: z.string().uuid(),\n supplierDisplayName: z.string().min(1).max(255),\n importBatchId: z.string().uuid(),\n /** Supplier code prefix to strip before normalization (e.g. `'ME-'`). Optional. */\n supplierCodePrefix: z\n .string()\n .max(8)\n .regex(/^[A-Z0-9-]+$/, 'must be uppercase alphanumeric or \"-\"')\n .optional(),\n})\nexport type PartsParams = z.infer<typeof partsParamsSchema>\n\n/** Per-run memoization of the parsed params (keyed on the `ctx.params` reference). */\nconst paramsCache = new WeakMap<object, PartsParams>()\n\nfunction resolveParams(raw: Record<string, unknown>): PartsParams {\n const cached = paramsCache.get(raw)\n if (cached) return cached\n const parsed = partsParamsSchema.parse(raw)\n paramsCache.set(raw, parsed)\n return parsed\n}\n\n/** Per-locale display-name target fields (8 locales). */\nconst NAME_LOCALES = ['de', 'en', 'es', 'fr', 'it', 'nl', 'pt', 'ro'] as const\ntype NameLocale = (typeof NAME_LOCALES)[number]\nconst nameKey = (l: NameLocale): `name_${NameLocale}` => `name_${l}`\n\n/** The `parts` target-field whitelist (PR2 subset — `core_charge`/`purchase_price` are later PRs). */\nconst PARTS_FIELDS: readonly TargetField[] = [\n { key: 'code', label: 'Code', required: true, type: 'string' },\n { key: 'net_price', label: 'Net price', required: false, type: 'number' },\n { key: 'gross_price', label: 'Gross price', required: false, type: 'number' },\n { key: 'list_price', label: 'List price', required: false, type: 'number' },\n { key: 'currency', label: 'Currency', required: false, type: 'string' },\n { key: 'barcode', label: 'Barcode', required: false, type: 'string' },\n { key: 'description1', label: 'Description 1', required: false, type: 'string' },\n { key: 'description2', label: 'Description 2', required: false, type: 'string' },\n { key: 'discount_group', label: 'Discount group', required: false, type: 'string' },\n { key: 'superseding_code', label: 'Superseding code', required: false, type: 'string' },\n ...NAME_LOCALES.map(\n (l): TargetField => ({\n key: nameKey(l),\n label: `Name (${l})`,\n required: false,\n type: 'string',\n }),\n ),\n] as const\n\nconst NORMALIZE_CODE_RE = /[\\s-]/g\n/** Uppercase + dash/whitespace-stripped (D2). `*`/other chars are preserved (e.g. `*S` markers). */\nfunction normalizeCode(s: string): string {\n return s.toUpperCase().replace(NORMALIZE_CODE_RE, '')\n}\n\n/**\n * The `parts` schema. `buildRow` THROWS on invalid run params (so a misconfigured\n * run surfaces as a transform error, matching the carmaker contract) and returns a\n * typed `RowResult` per row otherwise.\n */\nexport const partsSchema: TargetSchema<PartsDocument> = {\n name: 'parts',\n fields: PARTS_FIELDS,\n buildRow(mapped: MappedRow, ctx: BuildRowContext) {\n const params = resolveParams(ctx.params)\n\n // 1. Code (required) → raw `code` + normalized `code_normalized` + doc id.\n const codeRaw = typeof mapped.code === 'string' ? mapped.code.trim() : ''\n if (!codeRaw) return rowError('missing_code', 'code is empty', 'code')\n const stripped =\n params.supplierCodePrefix && codeRaw.startsWith(params.supplierCodePrefix)\n ? codeRaw.slice(params.supplierCodePrefix.length)\n : codeRaw\n const codeNormalized = normalizeCode(stripped)\n if (!codeNormalized) {\n return rowError('empty_normalized_code', `Code \"${codeRaw}\" is empty after normalization`, 'code')\n }\n\n // 2. Currency gate (D9 EUR-only). Unmapped/blank defaults to EUR.\n const currencyText = textOrNull(mapped.currency)\n const currency = currencyText ?? 'EUR'\n if (currency !== 'EUR') {\n return rowError('unsupported_currency', `Currency \"${currency}\" not supported (EUR-only)`, 'currency')\n }\n\n // 3. Prices. net/gross/list each: absent → null, invalid → error, value → number.\n const net = numberCell(mapped.net_price)\n if (net.kind === 'invalid') return rowError('invalid_net_price', 'net_price is not a number', 'net_price')\n const gross = numberCell(mapped.gross_price)\n if (gross.kind === 'invalid') {\n return rowError('invalid_gross_price', 'gross_price is not a number', 'gross_price')\n }\n const list = numberCell(mapped.list_price)\n if (list.kind === 'invalid') return rowError('invalid_list_price', 'list_price is not a number', 'list_price')\n\n const netValue = net.kind === 'value' ? net.value : null\n const listValue = list.kind === 'value' ? list.value : null\n // At least one price must resolve. `net_price_eur` is non-nullable on the doc,\n // so it takes the direct net or — for a list-only two-file feed — the list\n // price as a display placeholder (PR4 derives the true purchase price).\n const displayPrice = netValue ?? listValue\n if (displayPrice === null) {\n return rowError('missing_price', 'no net_price or list_price provided', 'price')\n }\n\n const docId = `${codeNormalized}__${params.supplierId}`\n const doc: PartsDocument = {\n doc_id: docId,\n code: codeRaw,\n code_normalized: codeNormalized,\n brand_id: params.brandId,\n brand_slug: params.brandSlug,\n supplier_id: params.supplierId,\n supplier_display_name: params.supplierDisplayName,\n net_price_eur: displayPrice,\n gross_price_eur: gross.kind === 'value' ? gross.value : null,\n currency: 'EUR',\n barcode: textOrNull(mapped.barcode),\n name_de: textOrNull(mapped.name_de),\n name_en: textOrNull(mapped.name_en),\n name_es: textOrNull(mapped.name_es),\n name_fr: textOrNull(mapped.name_fr),\n name_it: textOrNull(mapped.name_it),\n name_nl: textOrNull(mapped.name_nl),\n name_pt: textOrNull(mapped.name_pt),\n name_ro: textOrNull(mapped.name_ro),\n description1: textOrNull(mapped.description1),\n description2: textOrNull(mapped.description2),\n // commerce-demo direct fields (PR2): list price → minor units; group + code\n // passthroughs consumed by PR4 / PR6.\n list_price_minor: listValue === null ? null : toMinorUnits(listValue),\n discount_group: textOrNull(mapped.discount_group),\n superseding_code: textOrNull(mapped.superseding_code),\n // Derived / later-PR fields stay null in PR2.\n purchase_price_minor: null,\n core_charge_minor: null,\n core_charge_currency: null,\n core_charge_refundable: null,\n core_charge_tax_class: null,\n core_charge_tax_inclusive: null,\n import_batch_id: params.importBatchId,\n imported_at: new Date().toISOString(),\n }\n\n return { kind: 'success', id: docId, doc }\n },\n}\n"],"mappings":"wBA6BA,MAAM,EAAmB,kBAkBzB,SAAgB,EAAY,EAAa,EAAqC,CAC5E,IAAM,EAAU,EAAI,KAAK,EACzB,GAAI,IAAY,IAAM,EAAQ,YAAY,IAAM,KAAM,OAAO,KAE7D,GAAM,CAAE,UAAS,aAAc,EAC3B,EAAI,EAOR,GALI,IAAc,KAAI,EAAI,EAAE,MAAM,CAAS,CAAC,CAAC,KAAK,EAAE,GAKhD,IAAY,IAAK,CACnB,GAAI,EAAE,SAAS,GAAG,EAAG,MAAO,KAC5B,EAAI,EAAE,MAAM,GAAG,CAAC,CAAC,KAAK,GAAG,CAC3B,MAEE,GAAI,EAAE,SAAS,GAAG,EAAG,MAAO,KAI9B,OADK,EAAiB,KAAK,CAAC,EACrB,OAAO,WAAW,CAAC,EADY,GAExC,CASA,SAAgB,EAAa,EAAuB,CAClD,OAAO,KAAK,MAAM,EAAQ,GAAG,CAC/B,CAwBA,SAAgB,EAAoB,EAA0B,CAC5D,IAAM,EAAU,EAAE,KAAK,EACvB,GAAI,IAAY,IAAM,EAAQ,YAAY,IAAM,KAAM,OAAO,KAC7D,IAAM,EAAW,EAAQ,SAAS,GAAG,EAC/B,EAAS,EAAQ,SAAS,GAAG,EAC/B,EACJ,GAAI,GAAY,EAGd,EAAU,EAAQ,QAAQ,MAAO,EAAE,CAAC,CAAC,QAAQ,IAAK,GAAG,OAChD,GAAI,EACT,EAAU,EAAQ,QAAQ,IAAK,GAAG,OAC7B,GAAI,EACT,MAAO,SAEP,GAAU,EAGZ,OADK,EAAiB,KAAK,CAAO,EAC3B,OAAO,WAAW,CAAO,EADY,GAE9C,CCxGA,MAOa,EAAwB,EAAE,mBAAmB,KAAM,CAC9D,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,MAAM,CAAE,CAAC,EAClC,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,WAAW,CAAE,CAAC,EACvC,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,WAAW,CAAE,CAAC,EACvC,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,aAAa,EAAG,MAAO,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAa,CAAE,CAAC,EACtF,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,aAAa,EAAG,MAAO,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAa,CAAE,CAAC,EACtF,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,gBAAgB,EAAG,MAAO,EAAE,OAAO,CAAC,CAAC,IAAI,GAAG,CAAE,CAAC,EACxE,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,QAAQ,CAAE,CAAC,EACpC,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,gBAAgB,CAAE,CAAC,CAC9C,CAAC,EAqBY,EAA+D,CAC1E,CAAE,GAAI,OAAQ,MAAO,kBAAmB,SAAU,EAAM,EACxD,CAAE,GAAI,YAAa,MAAO,YAAa,SAAU,EAAM,EACvD,CAAE,GAAI,YAAa,MAAO,YAAa,SAAU,EAAM,EACvD,CAAE,GAAI,cAAe,MAAO,eAAgB,SAAU,EAAK,EAC3D,CAAE,GAAI,cAAe,MAAO,eAAgB,SAAU,EAAK,EAC3D,CAAE,GAAI,iBAAkB,MAAO,mBAAoB,SAAU,EAAK,EAClE,CAAE,GAAI,SAAU,MAAO,oCAAqC,SAAU,EAAM,EAC5E,CAAE,GAAI,iBAAkB,MAAO,mCAAoC,SAAU,EAAM,CACrF,EAYA,SAAgB,EACd,EACA,EACA,EACW,CACX,IAAI,EAAmB,EACvB,IAAK,IAAM,KAAK,EACV,UAAO,GAAU,SAKrB,OAAQ,EAAE,GAAV,CACE,IAAK,OACH,EAAQ,EAAM,KAAK,EACnB,MACF,IAAK,YACH,EAAQ,EAAM,YAAY,EAC1B,MACF,IAAK,YACH,EAAQ,EAAM,YAAY,EAC1B,MACF,IAAK,cACH,EAAQ,EAAM,WAAW,EAAE,KAAK,EAAI,EAAM,MAAM,EAAE,MAAM,MAAM,EAAI,EAClE,MACF,IAAK,cACH,EAAQ,EAAM,SAAS,EAAE,KAAK,EAAI,EAAM,MAAM,EAAG,EAAM,OAAS,EAAE,MAAM,MAAM,EAAI,EAClF,MACF,IAAK,iBACH,EAAQ,EAAM,KAAK,IAAM,GAAK,EAAE,MAAQ,EACxC,MACF,IAAK,SACH,EAAQ,EAAY,EAAO,CAAM,EACjC,MACF,IAAK,iBACH,EAAQ,EAAoB,CAAK,EACjC,KACJ,CAEF,OAAO,CACT,CC7FA,MAAa,EAAsB,EAAE,OAAO,CAE1C,OAAQ,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,GAAG,EAEjC,OAAQ,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,EAEhC,WAAY,EAAE,MAAM,CAAqB,CAAC,CAAC,QAAQ,CAAC,CAAC,CACvD,CAAC,EAWY,EAAqB,EAC/B,OAAO,CACN,QAAS,EAAE,KAAK,CAAC,IAAK,GAAG,CAAC,EAC1B,UAAW,EAAE,KAAK,CAAC,GAAI,IAAK,IAAK,GAAG,CAAC,CACvC,CAAC,CAAC,CACD,OAAQ,GAAM,EAAE,UAAY,EAAE,UAAW,CACxC,QAAS,+CACT,KAAM,CAAC,WAAW,CACpB,CAAC,EAQU,EAAsB,EAAE,OAAO,CAC1C,aAAc,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,EACtC,aAAc,EACd,UAAW,EAAE,MAAM,CAAmB,CACxC,CAAC,EAeD,SAAgB,EACd,EACA,EACoB,CACpB,IAAM,EAAU,IAAI,IAAI,EAAO,OAAO,IAAK,GAAM,EAAE,GAAG,CAAC,EACvD,IAAK,IAAM,KAAO,EAAO,UACvB,GAAI,CAAC,EAAQ,IAAI,EAAI,MAAM,EACzB,MAAU,MACR,mBAAmB,EAAI,OAAO,8BAA8B,EAAO,KAAK,cACzD,CAAC,GAAG,CAAO,CAAC,CAAC,KAAK,IAAI,EAAE,EACzC,EAIJ,IAAM,EAAS,EAAO,aAChB,EAAY,EAAO,UAEzB,OAAO,MAAO,EAAK,IAAQ,CACzB,IAAM,EAAoB,CAAC,EAC3B,IAAK,IAAM,KAAO,EAAW,CAG3B,IAAM,EAAM,EAAI,EAAI,SAAW,GAC/B,EAAO,EAAI,QAAU,EAAsB,EAAK,EAAI,WAAY,CAAM,CACxE,CACA,OAAO,EAAO,SAAS,EAAQ,CAC7B,YAAa,EAAI,YACjB,UAAW,EAAI,UACf,OAAQ,EAAI,MACd,CAAC,CACH,CACF,CAeA,SAAgB,EACd,EACA,EACe,CACf,IAAM,EAAS,EAAS,OAAQ,GAAM,EAAE,MAAM,EAC9C,GAAI,IAAY,KAAM,CACpB,IAAM,EAAW,EAAO,KAAM,GAAM,EAAE,UAAY,CAAO,EACzD,GAAI,EAAU,OAAO,CACvB,CACA,OAAO,EAAO,KAAM,GAAM,EAAE,UAAY,IAAI,CAC9C,CCvEA,SAAgB,EAAW,EAAyC,CAClE,GAAI,OAAO,GAAM,SAAU,OAAO,KAClC,IAAM,EAAI,EAAE,KAAK,EACjB,OAAO,IAAM,GAAK,KAAO,CAC3B,CAaA,SAAgB,EAAW,EAAyC,CAKlE,OAJI,GAAyB,KAAa,CAAE,KAAM,QAAS,EACvD,OAAO,GAAM,SACR,OAAO,MAAM,CAAC,EAAI,CAAE,KAAM,SAAU,EAAI,CAAE,KAAM,QAAS,MAAO,CAAE,EAEpE,EAAE,KAAK,IAAM,GAAK,CAAE,KAAM,QAAS,EAAI,CAAE,KAAM,SAAU,CAClE,CAGA,SAAgB,EAAS,EAAmB,EAAiB,EAAe,CAC1E,MAAO,CAAE,KAAM,QAAkB,MAAO,CAAE,YAAW,UAAS,OAAM,CAAE,CACxE,CClDA,MAAa,EAAoB,EAAE,OAAO,CACxC,QAAS,EAAE,OAAO,CAAC,CAAC,KAAK,EACzB,UAAW,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,EACnC,WAAY,EAAE,OAAO,CAAC,CAAC,KAAK,EAC5B,oBAAqB,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,GAAG,EAC9C,cAAe,EAAE,OAAO,CAAC,CAAC,KAAK,EAE/B,mBAAoB,EACjB,OAAO,CAAC,CACR,IAAI,CAAC,CAAC,CACN,MAAM,eAAgB,uCAAuC,CAAC,CAC9D,SAAS,CACd,CAAC,EAIK,EAAc,IAAI,QAExB,SAAS,EAAc,EAA2C,CAChE,IAAM,EAAS,EAAY,IAAI,CAAG,EAClC,GAAI,EAAQ,OAAO,EACnB,IAAM,EAAS,EAAkB,MAAM,CAAG,EAE1C,OADA,EAAY,IAAI,EAAK,CAAM,EACpB,CACT,CAGA,MAAM,EAAe,CAAC,KAAM,KAAM,KAAM,KAAM,KAAM,KAAM,KAAM,IAAI,EAE9D,EAAW,GAAwC,QAAQ,IAG3D,EAAuC,CAC3C,CAAE,IAAK,OAAQ,MAAO,OAAQ,SAAU,GAAM,KAAM,QAAS,EAC7D,CAAE,IAAK,YAAa,MAAO,YAAa,SAAU,GAAO,KAAM,QAAS,EACxE,CAAE,IAAK,cAAe,MAAO,cAAe,SAAU,GAAO,KAAM,QAAS,EAC5E,CAAE,IAAK,aAAc,MAAO,aAAc,SAAU,GAAO,KAAM,QAAS,EAC1E,CAAE,IAAK,WAAY,MAAO,WAAY,SAAU,GAAO,KAAM,QAAS,EACtE,CAAE,IAAK,UAAW,MAAO,UAAW,SAAU,GAAO,KAAM,QAAS,EACpE,CAAE,IAAK,eAAgB,MAAO,gBAAiB,SAAU,GAAO,KAAM,QAAS,EAC/E,CAAE,IAAK,eAAgB,MAAO,gBAAiB,SAAU,GAAO,KAAM,QAAS,EAC/E,CAAE,IAAK,iBAAkB,MAAO,iBAAkB,SAAU,GAAO,KAAM,QAAS,EAClF,CAAE,IAAK,mBAAoB,MAAO,mBAAoB,SAAU,GAAO,KAAM,QAAS,EACtF,GAAG,EAAa,IACb,IAAoB,CACnB,IAAK,EAAQ,CAAC,EACd,MAAO,SAAS,EAAE,GAClB,SAAU,GACV,KAAM,QACR,EACF,CACF,EAEM,EAAoB,SAE1B,SAAS,EAAc,EAAmB,CACxC,OAAO,EAAE,YAAY,CAAC,CAAC,QAAQ,EAAmB,EAAE,CACtD,CAOA,MAAa,EAA2C,CACtD,KAAM,QACN,OAAQ,EACR,SAAS,EAAmB,EAAsB,CAChD,IAAM,EAAS,EAAc,EAAI,MAAM,EAGjC,EAAU,OAAO,EAAO,MAAS,SAAW,EAAO,KAAK,KAAK,EAAI,GACvE,GAAI,CAAC,EAAS,OAAO,EAAS,eAAgB,gBAAiB,MAAM,EAKrE,IAAM,EAAiB,EAHrB,EAAO,oBAAsB,EAAQ,WAAW,EAAO,kBAAkB,EACrE,EAAQ,MAAM,EAAO,mBAAmB,MAAM,EAC9C,CACuC,EAC7C,GAAI,CAAC,EACH,OAAO,EAAS,wBAAyB,SAAS,EAAQ,gCAAiC,MAAM,EAKnG,IAAM,EADe,EAAW,EAAO,QACX,GAAK,MACjC,GAAI,IAAa,MACf,OAAO,EAAS,uBAAwB,aAAa,EAAS,4BAA6B,UAAU,EAIvG,IAAM,EAAM,EAAW,EAAO,SAAS,EACvC,GAAI,EAAI,OAAS,UAAW,OAAO,EAAS,oBAAqB,4BAA6B,WAAW,EACzG,IAAM,EAAQ,EAAW,EAAO,WAAW,EAC3C,GAAI,EAAM,OAAS,UACjB,OAAO,EAAS,sBAAuB,8BAA+B,aAAa,EAErF,IAAM,EAAO,EAAW,EAAO,UAAU,EACzC,GAAI,EAAK,OAAS,UAAW,OAAO,EAAS,qBAAsB,6BAA8B,YAAY,EAE7G,IAAM,EAAW,EAAI,OAAS,QAAU,EAAI,MAAQ,KAC9C,EAAY,EAAK,OAAS,QAAU,EAAK,MAAQ,KAIjD,EAAe,GAAY,EACjC,GAAI,IAAiB,KACnB,OAAO,EAAS,gBAAiB,sCAAuC,OAAO,EAGjF,IAAM,EAAQ,GAAG,EAAe,IAAI,EAAO,aAuC3C,MAAO,CAAE,KAAM,UAAW,GAAI,EAAO,IAAA,CArCnC,OAAQ,EACR,KAAM,EACN,gBAAiB,EACjB,SAAU,EAAO,QACjB,WAAY,EAAO,UACnB,YAAa,EAAO,WACpB,sBAAuB,EAAO,oBAC9B,cAAe,EACf,gBAAiB,EAAM,OAAS,QAAU,EAAM,MAAQ,KACxD,SAAU,MACV,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,aAAc,EAAW,EAAO,YAAY,EAC5C,aAAc,EAAW,EAAO,YAAY,EAG5C,iBAAkB,IAAc,KAAO,KAAO,EAAa,CAAS,EACpE,eAAgB,EAAW,EAAO,cAAc,EAChD,iBAAkB,EAAW,EAAO,gBAAgB,EAEpD,qBAAsB,KACtB,kBAAmB,KACnB,qBAAsB,KACtB,uBAAwB,KACxB,sBAAuB,KACvB,0BAA2B,KAC3B,gBAAiB,EAAO,cACxB,YAAa,IAAI,KAAK,CAAA,CAAE,YAAY,CAGC,CAAE,CAC3C,CACF"}
|
|
1
|
+
{"version":3,"file":"parts-schema-Vv0EJCjj.mjs","names":[],"sources":["../src/mapping/number-format.ts","../src/mapping/column-transforms.ts","../src/mapping/engine.ts","../src/mapping/target-schema.ts","../src/mapping/parts-schema.ts"],"sourcesContent":["/**\n * Number parsing for the configurable mapping engine (commerce-demo phase 02).\n *\n * Real supplier feeds (R03) disagree on decimal AND thousands separators:\n * `55.02` (dot decimal), `326,88` (comma decimal), `1.234,56` (EU\n * thousands+decimal), `1,234.56` (US), `1 234,56` (space-grouped). The mapping\n * stores an explicit `{ decimal, thousands }` per profile so there is no guessing\n * — `parseNumber` trusts the configured format.\n *\n * Two parsers live here:\n * - `parseNumber(raw, format)` — the format-driven parser used by the generic\n * `number` per-column transform.\n * - `parseEuropeanNumber(raw)` — the carmaker-feed legacy parser (comma decimal,\n * dot thousands, **bare-dot rejected as ambiguous**). Carries the carmaker feed's\n * documented format with two intentional broadenings vs the old inline parser:\n * it trims surrounding whitespace, and returns `null` (not `NaN`) for an empty /\n * `'NA'` cell so both number ops share absent-vs-malformed semantics. The\n * carmaker doc-shape suite (27 tests) is the regression net.\n */\n\n/** Separator config for {@link parseNumber}. Stored per `import_mapping`. */\nexport interface NumberFormat {\n /** Decimal separator. */\n decimal: '.' | ','\n /** Thousands grouping separator. `''` = none. */\n thousands: '' | '.' | ',' | ' '\n}\n\n/** Post-cleanup numeric form: optional sign, digits, optional single decimal. */\nconst STRICT_NUMBER_RE = /^-?\\d+(\\.\\d+)?$/\n\n/**\n * Parse a feed cell into a number using an EXPLICIT `{ decimal, thousands }`\n * format. No ambiguity: thousands separators are stripped, the decimal separator\n * is normalised to `.`, and the result is strict-validated.\n *\n * Returns:\n * - `null` for an empty / `'NA'` cell (a legitimately-absent value — the caller\n * decides whether that's allowed for a given target field).\n * - `NaN` for a non-empty but unparseable cell (malformed — the caller surfaces\n * a row-level error rather than coercing to 0).\n * - the parsed number otherwise.\n *\n * A character that is neither a digit, the configured decimal, the configured\n * thousands separator, nor a leading `-` makes the cell unparseable (`NaN`) —\n * e.g. a literal `.` when the format's decimal is `,` and thousands is not `.`.\n */\nexport function parseNumber(raw: string, format: NumberFormat): number | null {\n const trimmed = raw.trim()\n if (trimmed === '' || trimmed.toUpperCase() === 'NA') return null\n\n const { decimal, thousands } = format\n let s = trimmed\n // 1. Remove every thousands separator (when one is configured).\n if (thousands !== '') s = s.split(thousands).join('')\n\n // 2. Normalise the decimal separator to `.`. After thousands removal, the\n // \"other\" separator must not appear — if it does, the cell doesn't match\n // the declared format and is unparseable.\n if (decimal === ',') {\n if (s.includes('.')) return Number.NaN\n s = s.split(',').join('.')\n } else {\n // decimal === '.'\n if (s.includes(',')) return Number.NaN\n }\n\n if (!STRICT_NUMBER_RE.test(s)) return Number.NaN\n return Number.parseFloat(s)\n}\n\n/**\n * Convert a major-unit money value (e.g. `55.02` EUR) to integer MINOR units\n * (`5502`). Rounds to the nearest cent to absorb float drift. The commerce-demo\n * PartsDocument money fields (`list_price_minor`, `purchase_price_minor`,\n * `core_charge_minor`) are integer minor units so derived prices (PR4) don't\n * accrue rounding error — see F003 in progress.md.\n */\nexport function toMinorUnits(value: number): number {\n return Math.round(value * 100)\n}\n\n/**\n * Parse a European-formatted decimal: `,` is the decimal separator, `.` is the\n * (optional) thousands separator. Returns `null` for an empty / `'NA'` cell and\n * `NaN` on unparseable (malformed) input.\n *\n * This is the carmaker-feed legacy parser, preserved from the hardcoded transform.\n * Crucially it **rejects bare-dot** input (`1.98`, `1.234`) as ambiguous rather\n * than guessing — the carmaker feeds are documented as comma-decimal, and a `.`\n * without a `,` could mean a US-format decimal OR a thousands marker. The\n * format-driven {@link parseNumber} doesn't need this guard (its separators are\n * explicit); the carmaker path keeps it.\n *\n * The `null` (absent) vs `NaN` (malformed) split mirrors {@link parseNumber} so\n * the mapping engine treats both number ops uniformly: a target field's `buildRow`\n * reads `null` as \"not provided\" and `NaN` as \"invalid\". This is what lets the\n * reimplemented carmaker transform keep its `GrossPrice 'NA' → null` behaviour\n * (optional field) while still erroring on a malformed `NetPrice` (required).\n *\n * Cases: `'1,98'`→`1.98`, `'1.234,56'`→`1234.56`, `'5'`→`5`, `'-5,00'`→`-5`,\n * empty / `'NA'` → `null`. Rejected (`NaN`): bare-dot (`'1.98'`, `'1.234'`),\n * trailing junk (`'1,98EUR'`), multiple commas (`'1,2,3'`), other non-numeric.\n */\nexport function parseEuropeanNumber(s: string): number | null {\n const trimmed = s.trim()\n if (trimmed === '' || trimmed.toUpperCase() === 'NA') return null\n const hasComma = trimmed.includes(',')\n const hasDot = trimmed.includes('.')\n let cleaned: string\n if (hasComma && hasDot) {\n // European thousands+decimal: strip every `.`, then swap the (sole) `,` for\n // `.`. The strict regex below catches multi-comma input.\n cleaned = trimmed.replace(/\\./g, '').replace(',', '.')\n } else if (hasComma) {\n cleaned = trimmed.replace(',', '.')\n } else if (hasDot) {\n return Number.NaN\n } else {\n cleaned = trimmed\n }\n if (!STRICT_NUMBER_RE.test(cleaned)) return Number.NaN\n return Number.parseFloat(cleaned)\n}\n","/**\n * Per-column transforms for the mapping engine (commerce-demo phase 02, D005).\n *\n * A **bounded, enum-keyed** library of cleaning operations applied to a source\n * cell before it lands on a target field. This is the security model from R02:\n * transforms are declarative operations with Zod-validated params, NEVER\n * user-supplied executable code. The carmaker feed's logic decomposes exactly\n * into `stripPrefix` + `europeanNumber` (+ the schema's code-normalize).\n *\n * Number-producing ops (`number`, `europeanNumber`) are TERMINAL by convention —\n * they convert the string cell into `number | null`; string ops after them are\n * no-ops. The engine reads the result as `string | number | null` and hands it to\n * the target schema's `buildRow`.\n */\n\nimport { z } from 'zod'\nimport { type NumberFormat, parseEuropeanNumber, parseNumber } from './number-format.js'\n\n/** Max length for a strip-affix value — bounds the param, no ReDoS surface. */\nconst MAX_AFFIX_LEN = 32\n\n/**\n * Discriminated union of bounded transform ops. The TS type is INFERRED from\n * this schema so there's a single source of truth + the snapshot is validated at\n * the engine boundary.\n */\nexport const columnTransformSchema = z.discriminatedUnion('op', [\n z.object({ op: z.literal('trim') }),\n z.object({ op: z.literal('uppercase') }),\n z.object({ op: z.literal('lowercase') }),\n z.object({ op: z.literal('stripPrefix'), value: z.string().min(1).max(MAX_AFFIX_LEN) }),\n z.object({ op: z.literal('stripSuffix'), value: z.string().min(1).max(MAX_AFFIX_LEN) }),\n z.object({ op: z.literal('defaultIfEmpty'), value: z.string().max(255) }),\n z.object({ op: z.literal('number') }),\n z.object({ op: z.literal('europeanNumber') }),\n])\n\n/** One declarative cleaning operation. */\nexport type ColumnTransform = z.infer<typeof columnTransformSchema>\n\n/** The discriminator literal of a {@link ColumnTransform}. */\nexport type ColumnTransformOp = ColumnTransform['op']\n\n/** UI descriptor for one transform op: its key, a label, and whether it takes a `value` param. */\nexport interface ColumnTransformOpDescriptor {\n op: ColumnTransformOp\n label: string\n /** `true` → the op carries a `value` string param (`stripPrefix`/`stripSuffix`/`defaultIfEmpty`). */\n hasValue: boolean\n}\n\n/**\n * Static catalogue of the available per-column transform ops, for the PR3 wizard's\n * transform picker. The ORDER is the suggested UI order; `number`/`europeanNumber`\n * are terminal (see {@link applyColumnTransforms}) and labelled accordingly.\n */\nexport const COLUMN_TRANSFORM_OPS: readonly ColumnTransformOpDescriptor[] = [\n { op: 'trim', label: 'Trim whitespace', hasValue: false },\n { op: 'uppercase', label: 'Uppercase', hasValue: false },\n { op: 'lowercase', label: 'Lowercase', hasValue: false },\n { op: 'stripPrefix', label: 'Strip prefix', hasValue: true },\n { op: 'stripSuffix', label: 'Strip suffix', hasValue: true },\n { op: 'defaultIfEmpty', label: 'Default if empty', hasValue: true },\n { op: 'number', label: 'Parse number (uses number format)', hasValue: false },\n { op: 'europeanNumber', label: 'Parse European number (1.234,56)', hasValue: false },\n] as const\n\n/** The value a column carries as it moves through the transform pipeline. */\nexport type CellValue = string | number | null\n\n/**\n * Apply a transform pipeline to a raw cell, left-to-right. String ops act on the\n * string value; the `number` / `europeanNumber` ops are terminal (convert to\n * `number | null`, with `NaN` for a malformed numeric — the schema's `buildRow`\n * surfaces that as a row error). Once the value is non-string, later string ops\n * pass through unchanged.\n */\nexport function applyColumnTransforms(\n raw: string,\n transforms: readonly ColumnTransform[],\n format: NumberFormat,\n): CellValue {\n let value: CellValue = raw\n for (const t of transforms) {\n if (typeof value !== 'string') {\n // A prior number op already produced number|null — string ops are no-ops\n // past this point (number ops are terminal by convention).\n continue\n }\n switch (t.op) {\n case 'trim':\n value = value.trim()\n break\n case 'uppercase':\n value = value.toUpperCase()\n break\n case 'lowercase':\n value = value.toLowerCase()\n break\n case 'stripPrefix':\n value = value.startsWith(t.value) ? value.slice(t.value.length) : value\n break\n case 'stripSuffix':\n value = value.endsWith(t.value) ? value.slice(0, value.length - t.value.length) : value\n break\n case 'defaultIfEmpty':\n value = value.trim() === '' ? t.value : value\n break\n case 'number':\n value = parseNumber(value, format)\n break\n case 'europeanNumber':\n value = parseEuropeanNumber(value)\n break\n }\n }\n return value\n}\n","/**\n * Mapping engine (commerce-demo phase 02, D005).\n *\n * Turns a declarative `MappingConfig` (which source column → which whitelisted\n * target field, plus per-column transforms + a number format) into a\n * `RowTransform` the existing runner drives. The pipeline per row is:\n *\n * raw cells → applyColumnTransforms(per column) → MappedRow → schema.buildRow\n *\n * Two whitelists guard the boundary (R02 injection model): the `target_schema`\n * name resolves to a registered `TargetSchema`, and every `column_map[].target`\n * must be one of that schema's fields — enforced here at build time (throws),\n * before any row is processed.\n *\n * `resolveMapping` implements the D002 most-specific ladder used at enqueue time\n * to pick the `(supplier, brand)` mapping over the `(supplier, NULL)` default.\n */\n\nimport { z } from 'zod'\nimport type { RowTransform } from '../transform.js'\nimport { applyColumnTransforms, columnTransformSchema } from './column-transforms.js'\nimport type { MappedRow, TargetSchema } from './target-schema.js'\n\n/** One column-map entry. `target` is checked against the schema whitelist separately. */\nexport const mappingColumnSchema = z.object({\n /** Source header name (or positional index string for headerless feeds). */\n source: z.string().min(1).max(256),\n /** Target field key — must be in the chosen schema's whitelist (enforced in `buildMappingTransform`). */\n target: z.string().min(1).max(64),\n /** Ordered per-column cleaning ops. Defaults to none. */\n transforms: z.array(columnTransformSchema).default([]),\n})\nexport type MappingColumn = z.infer<typeof mappingColumnSchema>\n\n/**\n * Decimal + thousands separators for the `number` transform.\n *\n * The two MUST differ: `parseNumber` strips every thousands separator FIRST, so a\n * config where `decimal === thousands` (e.g. both `,`) would strip the decimal\n * separator too — `326,88` → `32688`, a silent 100× error rather than a row error.\n * Rejected at save (`assertValidMapping`) AND at run (`mappingConfigSchema`).\n */\nexport const numberFormatSchema = z\n .object({\n decimal: z.enum(['.', ',']),\n thousands: z.enum(['', '.', ',', ' ']),\n })\n .refine((f) => f.decimal !== f.thousands, {\n message: 'decimal and thousands separators must differ',\n path: ['thousands'],\n })\n\n/**\n * The portable mapping config — the subset of an `import_mapping` row the engine\n * needs. The enqueuer SNAPSHOTS this into `import_run.params.mapping` so a run\n * records the exact mapping it used (reproducible) and the transform stays a pure\n * function of `(row, ctx)` with no per-row DB I/O.\n */\nexport const mappingConfigSchema = z.object({\n targetSchema: z.string().min(1).max(64),\n numberFormat: numberFormatSchema,\n columnMap: z.array(mappingColumnSchema),\n})\nexport type MappingConfig = z.infer<typeof mappingConfigSchema>\n\n/**\n * Build a `RowTransform` from a validated mapping config + its target schema.\n *\n * Throws at BUILD time if any `column_map` target is outside the schema's\n * whitelist (deny-by-default, R02) — so a misconfigured mapping fails loudly\n * before a single row is written, rather than silently dropping a column.\n *\n * Generic over the schema's output type, so passing the concrete `partsSchema`\n * yields a `RowTransform<PartsDocument>` (the carmaker path keeps its typing); the\n * registry-resolved `TargetSchema<unknown>` yields `RowTransform<unknown>` (the\n * `commerce:mapping` path, where the output type isn't known until runtime).\n */\nexport function buildMappingTransform<TOut>(\n config: MappingConfig,\n schema: TargetSchema<TOut>,\n): RowTransform<TOut> {\n const allowed = new Set(schema.fields.map((f) => f.key))\n for (const col of config.columnMap) {\n if (!allowed.has(col.target)) {\n throw new Error(\n `Mapping target \"${col.target}\" is not a field of schema \"${schema.name}\" ` +\n `(allowed: ${[...allowed].join(', ')})`,\n )\n }\n }\n\n const format = config.numberFormat\n const columnMap = config.columnMap\n\n return async (row, ctx) => {\n const mapped: MappedRow = {}\n for (const col of columnMap) {\n // `source` is read as data from the parsed row map — never used as a key\n // into framework objects (R02). A missing source column yields `''`.\n const raw = row[col.source] ?? ''\n mapped[col.target] = applyColumnTransforms(raw, col.transforms, format)\n }\n return schema.buildRow(mapped, {\n importRunId: ctx.importRunId,\n rowNumber: ctx.rowNumber,\n params: ctx.params,\n })\n }\n}\n\n/** The minimal shape `resolveMapping` needs — satisfied by an `import_mapping` row or a test fixture. */\nexport interface ResolvableMapping {\n /** `null` = supplier default (applies to all brands). */\n brandId: string | null\n active: boolean\n}\n\n/**\n * Most-specific resolution (D002 ladder): given the candidate mappings for one\n * `(supplier, target_schema)` and the requested brand, return the brand-specific\n * mapping if one exists, else the supplier-default (`brandId === null`). Inactive\n * mappings are skipped. Returns `undefined` when nothing matches.\n */\nexport function resolveMapping<M extends ResolvableMapping>(\n mappings: readonly M[],\n brandId: string | null,\n): M | undefined {\n const active = mappings.filter((m) => m.active)\n if (brandId !== null) {\n const specific = active.find((m) => m.brandId === brandId)\n if (specific) return specific\n }\n return active.find((m) => m.brandId === null)\n}\n","/**\n * TargetSchema — the contract a mapping writes INTO (commerce-demo phase 02, D005).\n *\n * A mapping's `column_map` points source columns at a schema's whitelisted target\n * fields; the engine applies the per-column transforms to build a `MappedRow`, then\n * hands it to `schema.buildRow` to produce the typed output document (an ES\n * `PartsDocument` for `parts`, a discount row for `discount_table`).\n *\n * The `fields` list is the SECOND injection-boundary whitelist (the first is the\n * registered schema name): a `column_map` entry whose `target` is not a field key\n * here is rejected at validation time (R02). Two schemas ship in PR2 — `parts` and\n * `discount_table`.\n */\n\nimport type { RowResult } from '../transform.js'\nimport type { CellValue } from './column-transforms.js'\n\n/** One whitelisted target field on a schema. */\nexport interface TargetField {\n /** Stable target key referenced by `column_map[].target`. */\n key: string\n /** Human label for the wizard dropdown (PR3). */\n label: string\n /** Whether a mapping for this schema MUST include this target. */\n required: boolean\n /** Coarse type — drives the wizard's transform suggestions; `buildRow` does the real coercion. */\n type: 'string' | 'number'\n}\n\n/** The per-row product of the engine: target field key → cleaned value. */\nexport type MappedRow = Record<string, CellValue>\n\n/** Per-row context handed to `buildRow`. `params` is the opaque per-run params object. */\nexport interface BuildRowContext {\n importRunId: string\n rowNumber: number\n /**\n * Opaque per-run params (the parts schema reads brand/supplier identity from\n * it; `discount_table` ignores it). The schema validates the shape it needs.\n */\n params: Record<string, unknown>\n}\n\n/** A registered target schema: a whitelist + a row builder. */\nexport interface TargetSchema<TOut> {\n /** Registered name — also the `import_mapping.target_schema` value. */\n name: string\n /** The target-field whitelist (injection boundary). */\n fields: readonly TargetField[]\n /** Build a typed output (success) / skip / error from a mapped row. */\n buildRow(mapped: MappedRow, ctx: BuildRowContext): RowResult<TOut>\n}\n\n// ---------------------------------------------------------------------------\n// Shared `buildRow` helpers (used by every concrete schema).\n// ---------------------------------------------------------------------------\n\n/**\n * A nullable text cell → trimmed string, or `null` for empty / non-string. Accepts\n * `undefined` (an UNMAPPED target — indexed access under `noUncheckedIndexedAccess`)\n * and treats it as absent.\n */\nexport function textOrNull(v: CellValue | undefined): string | null {\n if (typeof v !== 'string') return null\n const t = v.trim()\n return t === '' ? null : t\n}\n\n/** The three outcomes a `number` / `europeanNumber` op produces for a cell. */\nexport type NumberOutcome =\n | { kind: 'value'; value: number }\n | { kind: 'absent' }\n | { kind: 'invalid' }\n\n/**\n * Classify a cell that should hold a number. `undefined` (unmapped) / `null` →\n * absent; a real number → value (`NaN` → invalid); a non-empty raw string (no\n * number op declared) → invalid; a blank raw string → absent.\n */\nexport function numberCell(v: CellValue | undefined): NumberOutcome {\n if (v === undefined || v === null) return { kind: 'absent' }\n if (typeof v === 'number') {\n return Number.isNaN(v) ? { kind: 'invalid' } : { kind: 'value', value: v }\n }\n return v.trim() === '' ? { kind: 'absent' } : { kind: 'invalid' }\n}\n\n/** Build a `RowResult` error variant (assignable to any `TargetSchema<TOut>`). */\nexport function rowError(errorType: string, message: string, field: string) {\n return { kind: 'error' as const, error: { errorType, message, field } }\n}\n","/**\n * `parts` target schema — maps a cleaned feed row into the ES `PartsDocument`\n * (commerce-demo phase 02, D005). This is the SHARED assembly + validation logic\n * the reimplemented carmaker transform and every operator-built `parts` mapping\n * run through, so the row taxonomy (code normalization, currency gate, price\n * validation) is uniform across all supplier feeds.\n *\n * **Field coverage.** This schema populates the direct passthroughs (the existing\n * `PartsDocument` columns plus `list_price_minor`, `discount_group`,\n * `superseding_code`) AND the DERIVED `purchase_price_minor` (PR4 — direct net, or\n * two-file `list × (1 − groupDiscount)` resolved from `params.discountGroups`). All\n * `core_charge_*` fields stay null until PR7; `core_charge` is deliberately NOT in\n * the whitelist (deny-by-default — PR7 adds it with its `*S`-row + column ingestion).\n * Replacement-pair SIDE EFFECTS (separate entity) are PR6.\n *\n * **Money.** Every money field is integer minor units via `toMinorUnits` (F003), and\n * the two-file derivation stays in minor units end-to-end (no major↔minor\n * round-trip). The legacy `net_price_eur` stays a major-unit float for back-compat:\n * it holds the direct net price, or — for a two-file LIST feed with no direct net —\n * the list price as a display placeholder (the true cost is `purchase_price_minor`).\n *\n * **Cost-pending warning.** A list-only row whose `discount_group` doesn't resolve\n * against `params.discountGroups` is still indexed (list set, `purchase_price_minor`\n * null) and carries a `no_purchase_price` WARNING — not a row error (phase 03 §5).\n */\n\nimport type { PartsDocument } from '@murumets-ee/search-elasticsearch'\nimport { z } from 'zod'\nimport type { RowWarning } from '../transform.js'\nimport { toMinorUnits } from './number-format.js'\nimport {\n type BuildRowContext,\n type MappedRow,\n numberCell,\n rowError,\n type TargetField,\n type TargetSchema,\n textOrNull,\n} from './target-schema.js'\n\n/**\n * Params the `parts` schema reads from the run — the per-(supplier, brand)\n * identity resolved at enqueue time. Identical to the carmaker transform's params\n * (the carmaker path re-exports this), so brand/supplier injection + the optional\n * supplier code-prefix strip are validated the same way everywhere.\n */\nexport const partsParamsSchema = z.object({\n brandId: z.string().uuid(),\n brandSlug: z.string().min(1).max(64),\n supplierId: z.string().uuid(),\n supplierDisplayName: z.string().min(1).max(255),\n importBatchId: z.string().uuid(),\n /** Supplier code prefix to strip before normalization (e.g. `'ME-'`). Optional. */\n supplierCodePrefix: z\n .string()\n .max(8)\n .regex(/^[A-Z0-9-]+$/, 'must be uppercase alphanumeric or \"-\"')\n .optional(),\n /**\n * Purchase-discount map for this supplier (commerce-demo PR4, phase 03 §2):\n * `{ groupCode → discountPct }`, resolved from `supplier_discount_group` at\n * enqueue time and snapshotted onto the run so `buildRow` stays a pure function\n * (no per-row DB I/O; `@murumets-ee/imports` stays a leaf). Absent for direct-net\n * suppliers + the carmaker path. Percentages are 0–100 (the entity constrains them).\n */\n discountGroups: z.record(z.string(), z.number().min(0).max(100)).optional(),\n})\nexport type PartsParams = z.infer<typeof partsParamsSchema>\n\n/** Per-run memoization of the parsed params (keyed on the `ctx.params` reference). */\nconst paramsCache = new WeakMap<object, PartsParams>()\n\nfunction resolveParams(raw: Record<string, unknown>): PartsParams {\n const cached = paramsCache.get(raw)\n if (cached) return cached\n const parsed = partsParamsSchema.parse(raw)\n paramsCache.set(raw, parsed)\n return parsed\n}\n\n/** Per-locale display-name target fields (8 locales). */\nconst NAME_LOCALES = ['de', 'en', 'es', 'fr', 'it', 'nl', 'pt', 'ro'] as const\ntype NameLocale = (typeof NAME_LOCALES)[number]\nconst nameKey = (l: NameLocale): `name_${NameLocale}` => `name_${l}`\n\n/** The `parts` target-field whitelist (PR2 subset — `core_charge`/`purchase_price` are later PRs). */\nconst PARTS_FIELDS: readonly TargetField[] = [\n { key: 'code', label: 'Code', required: true, type: 'string' },\n { key: 'net_price', label: 'Net price', required: false, type: 'number' },\n { key: 'gross_price', label: 'Gross price', required: false, type: 'number' },\n { key: 'list_price', label: 'List price', required: false, type: 'number' },\n { key: 'currency', label: 'Currency', required: false, type: 'string' },\n { key: 'barcode', label: 'Barcode', required: false, type: 'string' },\n { key: 'description1', label: 'Description 1', required: false, type: 'string' },\n { key: 'description2', label: 'Description 2', required: false, type: 'string' },\n { key: 'discount_group', label: 'Discount group', required: false, type: 'string' },\n { key: 'superseding_code', label: 'Superseding code', required: false, type: 'string' },\n ...NAME_LOCALES.map(\n (l): TargetField => ({\n key: nameKey(l),\n label: `Name (${l})`,\n required: false,\n type: 'string',\n }),\n ),\n] as const\n\nconst NORMALIZE_CODE_RE = /[\\s-]/g\n/** Uppercase + dash/whitespace-stripped (D2). `*`/other chars are preserved (e.g. `*S` markers). */\nfunction normalizeCode(s: string): string {\n return s.toUpperCase().replace(NORMALIZE_CODE_RE, '')\n}\n\n/**\n * Two-file purchase price (phase 03 §2): `list × (1 − discountPct/100)`, computed\n * ENTIRELY in integer minor units so no major↔minor round-trip can introduce drift\n * (F003). `listMinor` is already minor units; the result is rounded to the nearest\n * minor unit. Worked (R03 A+B): `5502 × 0.74 = 4071.48 → 4071` (€40.71).\n */\nfunction deriveTwoFilePurchaseMinor(listMinor: number, discountPct: number): number {\n return Math.round(listMinor * (1 - discountPct / 100))\n}\n\n/**\n * The `parts` schema. `buildRow` THROWS on invalid run params (so a misconfigured\n * run surfaces as a transform error, matching the carmaker contract) and returns a\n * typed `RowResult` per row otherwise.\n */\nexport const partsSchema: TargetSchema<PartsDocument> = {\n name: 'parts',\n fields: PARTS_FIELDS,\n buildRow(mapped: MappedRow, ctx: BuildRowContext) {\n const params = resolveParams(ctx.params)\n\n // 1. Code (required) → raw `code` + normalized `code_normalized` + doc id.\n const codeRaw = typeof mapped.code === 'string' ? mapped.code.trim() : ''\n if (!codeRaw) return rowError('missing_code', 'code is empty', 'code')\n const stripped =\n params.supplierCodePrefix && codeRaw.startsWith(params.supplierCodePrefix)\n ? codeRaw.slice(params.supplierCodePrefix.length)\n : codeRaw\n const codeNormalized = normalizeCode(stripped)\n if (!codeNormalized) {\n return rowError('empty_normalized_code', `Code \"${codeRaw}\" is empty after normalization`, 'code')\n }\n\n // 2. Currency gate (D9 EUR-only). Unmapped/blank defaults to EUR.\n const currencyText = textOrNull(mapped.currency)\n const currency = currencyText ?? 'EUR'\n if (currency !== 'EUR') {\n return rowError('unsupported_currency', `Currency \"${currency}\" not supported (EUR-only)`, 'currency')\n }\n\n // 3. Prices. net/gross/list each: absent → null, invalid → error, value → number.\n const net = numberCell(mapped.net_price)\n if (net.kind === 'invalid') return rowError('invalid_net_price', 'net_price is not a number', 'net_price')\n const gross = numberCell(mapped.gross_price)\n if (gross.kind === 'invalid') {\n return rowError('invalid_gross_price', 'gross_price is not a number', 'gross_price')\n }\n const list = numberCell(mapped.list_price)\n if (list.kind === 'invalid') return rowError('invalid_list_price', 'list_price is not a number', 'list_price')\n\n const netValue = net.kind === 'value' ? net.value : null\n const listValue = list.kind === 'value' ? list.value : null\n // At least one price must resolve. `net_price_eur` is non-nullable on the doc,\n // so it takes the direct net or — for a list-only two-file feed — the list\n // price as a display placeholder (the true cost is `purchase_price_minor`).\n const displayPrice = netValue ?? listValue\n if (displayPrice === null) {\n return rowError('missing_price', 'no net_price or list_price provided', 'price')\n }\n\n // 4. Purchase-price derivation (phase 03 §2, PR4). All money stays in integer\n // minor units (F003 — never round-trip back to major).\n // - direct net column (R03 C/D) → purchase = net\n // - two-file list × (1 − groupDiscount) → purchase = round(list × (1 − d/100))\n // - neither resolvable → cost pending: index list-only + WARN\n // (a `no_purchase_price` warning, NOT a row error — the part is still indexed).\n const discountGroup = textOrNull(mapped.discount_group)\n const listMinor = listValue === null ? null : toMinorUnits(listValue)\n let purchasePriceMinor: number | null = null\n const warnings: RowWarning[] = []\n if (netValue !== null) {\n purchasePriceMinor = toMinorUnits(netValue)\n } else if (listMinor !== null) {\n // List-only row: resolve the supplier's discount group → derive cost.\n const discountPct = discountGroup !== null ? params.discountGroups?.[discountGroup] : undefined\n if (discountPct !== undefined) {\n purchasePriceMinor = deriveTwoFilePurchaseMinor(listMinor, discountPct)\n } else {\n warnings.push({\n warningType: 'no_purchase_price',\n message:\n discountGroup === null\n ? 'list price present but no discount group on the row — cost pending'\n : `no discount group \"${discountGroup}\" configured for this supplier — cost pending`,\n field: 'discount_group',\n })\n }\n }\n\n const docId = `${codeNormalized}__${params.supplierId}`\n const doc: PartsDocument = {\n doc_id: docId,\n code: codeRaw,\n code_normalized: codeNormalized,\n brand_id: params.brandId,\n brand_slug: params.brandSlug,\n supplier_id: params.supplierId,\n supplier_display_name: params.supplierDisplayName,\n net_price_eur: displayPrice,\n gross_price_eur: gross.kind === 'value' ? gross.value : null,\n currency: 'EUR',\n barcode: textOrNull(mapped.barcode),\n name_de: textOrNull(mapped.name_de),\n name_en: textOrNull(mapped.name_en),\n name_es: textOrNull(mapped.name_es),\n name_fr: textOrNull(mapped.name_fr),\n name_it: textOrNull(mapped.name_it),\n name_nl: textOrNull(mapped.name_nl),\n name_pt: textOrNull(mapped.name_pt),\n name_ro: textOrNull(mapped.name_ro),\n description1: textOrNull(mapped.description1),\n description2: textOrNull(mapped.description2),\n // commerce-demo: list price → minor units; group + code passthroughs.\n list_price_minor: listMinor,\n discount_group: discountGroup,\n superseding_code: textOrNull(mapped.superseding_code),\n // Derived purchase cost (PR4 — direct net or two-file list × (1 − d)); null = cost pending.\n purchase_price_minor: purchasePriceMinor,\n // core_charge fields stay null until PR7.\n core_charge_minor: null,\n core_charge_currency: null,\n core_charge_refundable: null,\n core_charge_tax_class: null,\n core_charge_tax_inclusive: null,\n import_batch_id: params.importBatchId,\n imported_at: new Date().toISOString(),\n }\n\n return warnings.length > 0\n ? { kind: 'success', id: docId, doc, warnings }\n : { kind: 'success', id: docId, doc }\n },\n}\n"],"mappings":"wBA6BA,MAAM,EAAmB,kBAkBzB,SAAgB,EAAY,EAAa,EAAqC,CAC5E,IAAM,EAAU,EAAI,KAAK,EACzB,GAAI,IAAY,IAAM,EAAQ,YAAY,IAAM,KAAM,OAAO,KAE7D,GAAM,CAAE,UAAS,aAAc,EAC3B,EAAI,EAOR,GALI,IAAc,KAAI,EAAI,EAAE,MAAM,CAAS,CAAC,CAAC,KAAK,EAAE,GAKhD,IAAY,IAAK,CACnB,GAAI,EAAE,SAAS,GAAG,EAAG,MAAO,KAC5B,EAAI,EAAE,MAAM,GAAG,CAAC,CAAC,KAAK,GAAG,CAC3B,MAEE,GAAI,EAAE,SAAS,GAAG,EAAG,MAAO,KAI9B,OADK,EAAiB,KAAK,CAAC,EACrB,OAAO,WAAW,CAAC,EADY,GAExC,CASA,SAAgB,EAAa,EAAuB,CAClD,OAAO,KAAK,MAAM,EAAQ,GAAG,CAC/B,CAwBA,SAAgB,EAAoB,EAA0B,CAC5D,IAAM,EAAU,EAAE,KAAK,EACvB,GAAI,IAAY,IAAM,EAAQ,YAAY,IAAM,KAAM,OAAO,KAC7D,IAAM,EAAW,EAAQ,SAAS,GAAG,EAC/B,EAAS,EAAQ,SAAS,GAAG,EAC/B,EACJ,GAAI,GAAY,EAGd,EAAU,EAAQ,QAAQ,MAAO,EAAE,CAAC,CAAC,QAAQ,IAAK,GAAG,OAChD,GAAI,EACT,EAAU,EAAQ,QAAQ,IAAK,GAAG,OAC7B,GAAI,EACT,MAAO,SAEP,GAAU,EAGZ,OADK,EAAiB,KAAK,CAAO,EAC3B,OAAO,WAAW,CAAO,EADY,GAE9C,CCxGA,MAOa,EAAwB,EAAE,mBAAmB,KAAM,CAC9D,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,MAAM,CAAE,CAAC,EAClC,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,WAAW,CAAE,CAAC,EACvC,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,WAAW,CAAE,CAAC,EACvC,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,aAAa,EAAG,MAAO,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAa,CAAE,CAAC,EACtF,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,aAAa,EAAG,MAAO,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAa,CAAE,CAAC,EACtF,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,gBAAgB,EAAG,MAAO,EAAE,OAAO,CAAC,CAAC,IAAI,GAAG,CAAE,CAAC,EACxE,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,QAAQ,CAAE,CAAC,EACpC,EAAE,OAAO,CAAE,GAAI,EAAE,QAAQ,gBAAgB,CAAE,CAAC,CAC9C,CAAC,EAqBY,EAA+D,CAC1E,CAAE,GAAI,OAAQ,MAAO,kBAAmB,SAAU,EAAM,EACxD,CAAE,GAAI,YAAa,MAAO,YAAa,SAAU,EAAM,EACvD,CAAE,GAAI,YAAa,MAAO,YAAa,SAAU,EAAM,EACvD,CAAE,GAAI,cAAe,MAAO,eAAgB,SAAU,EAAK,EAC3D,CAAE,GAAI,cAAe,MAAO,eAAgB,SAAU,EAAK,EAC3D,CAAE,GAAI,iBAAkB,MAAO,mBAAoB,SAAU,EAAK,EAClE,CAAE,GAAI,SAAU,MAAO,oCAAqC,SAAU,EAAM,EAC5E,CAAE,GAAI,iBAAkB,MAAO,mCAAoC,SAAU,EAAM,CACrF,EAYA,SAAgB,EACd,EACA,EACA,EACW,CACX,IAAI,EAAmB,EACvB,IAAK,IAAM,KAAK,EACV,UAAO,GAAU,SAKrB,OAAQ,EAAE,GAAV,CACE,IAAK,OACH,EAAQ,EAAM,KAAK,EACnB,MACF,IAAK,YACH,EAAQ,EAAM,YAAY,EAC1B,MACF,IAAK,YACH,EAAQ,EAAM,YAAY,EAC1B,MACF,IAAK,cACH,EAAQ,EAAM,WAAW,EAAE,KAAK,EAAI,EAAM,MAAM,EAAE,MAAM,MAAM,EAAI,EAClE,MACF,IAAK,cACH,EAAQ,EAAM,SAAS,EAAE,KAAK,EAAI,EAAM,MAAM,EAAG,EAAM,OAAS,EAAE,MAAM,MAAM,EAAI,EAClF,MACF,IAAK,iBACH,EAAQ,EAAM,KAAK,IAAM,GAAK,EAAE,MAAQ,EACxC,MACF,IAAK,SACH,EAAQ,EAAY,EAAO,CAAM,EACjC,MACF,IAAK,iBACH,EAAQ,EAAoB,CAAK,EACjC,KACJ,CAEF,OAAO,CACT,CC7FA,MAAa,EAAsB,EAAE,OAAO,CAE1C,OAAQ,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,GAAG,EAEjC,OAAQ,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,EAEhC,WAAY,EAAE,MAAM,CAAqB,CAAC,CAAC,QAAQ,CAAC,CAAC,CACvD,CAAC,EAWY,EAAqB,EAC/B,OAAO,CACN,QAAS,EAAE,KAAK,CAAC,IAAK,GAAG,CAAC,EAC1B,UAAW,EAAE,KAAK,CAAC,GAAI,IAAK,IAAK,GAAG,CAAC,CACvC,CAAC,CAAC,CACD,OAAQ,GAAM,EAAE,UAAY,EAAE,UAAW,CACxC,QAAS,+CACT,KAAM,CAAC,WAAW,CACpB,CAAC,EAQU,EAAsB,EAAE,OAAO,CAC1C,aAAc,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,EACtC,aAAc,EACd,UAAW,EAAE,MAAM,CAAmB,CACxC,CAAC,EAeD,SAAgB,EACd,EACA,EACoB,CACpB,IAAM,EAAU,IAAI,IAAI,EAAO,OAAO,IAAK,GAAM,EAAE,GAAG,CAAC,EACvD,IAAK,IAAM,KAAO,EAAO,UACvB,GAAI,CAAC,EAAQ,IAAI,EAAI,MAAM,EACzB,MAAU,MACR,mBAAmB,EAAI,OAAO,8BAA8B,EAAO,KAAK,cACzD,CAAC,GAAG,CAAO,CAAC,CAAC,KAAK,IAAI,EAAE,EACzC,EAIJ,IAAM,EAAS,EAAO,aAChB,EAAY,EAAO,UAEzB,OAAO,MAAO,EAAK,IAAQ,CACzB,IAAM,EAAoB,CAAC,EAC3B,IAAK,IAAM,KAAO,EAAW,CAG3B,IAAM,EAAM,EAAI,EAAI,SAAW,GAC/B,EAAO,EAAI,QAAU,EAAsB,EAAK,EAAI,WAAY,CAAM,CACxE,CACA,OAAO,EAAO,SAAS,EAAQ,CAC7B,YAAa,EAAI,YACjB,UAAW,EAAI,UACf,OAAQ,EAAI,MACd,CAAC,CACH,CACF,CAeA,SAAgB,EACd,EACA,EACe,CACf,IAAM,EAAS,EAAS,OAAQ,GAAM,EAAE,MAAM,EAC9C,GAAI,IAAY,KAAM,CACpB,IAAM,EAAW,EAAO,KAAM,GAAM,EAAE,UAAY,CAAO,EACzD,GAAI,EAAU,OAAO,CACvB,CACA,OAAO,EAAO,KAAM,GAAM,EAAE,UAAY,IAAI,CAC9C,CCvEA,SAAgB,EAAW,EAAyC,CAClE,GAAI,OAAO,GAAM,SAAU,OAAO,KAClC,IAAM,EAAI,EAAE,KAAK,EACjB,OAAO,IAAM,GAAK,KAAO,CAC3B,CAaA,SAAgB,EAAW,EAAyC,CAKlE,OAJI,GAAyB,KAAa,CAAE,KAAM,QAAS,EACvD,OAAO,GAAM,SACR,OAAO,MAAM,CAAC,EAAI,CAAE,KAAM,SAAU,EAAI,CAAE,KAAM,QAAS,MAAO,CAAE,EAEpE,EAAE,KAAK,IAAM,GAAK,CAAE,KAAM,QAAS,EAAI,CAAE,KAAM,SAAU,CAClE,CAGA,SAAgB,EAAS,EAAmB,EAAiB,EAAe,CAC1E,MAAO,CAAE,KAAM,QAAkB,MAAO,CAAE,YAAW,UAAS,OAAM,CAAE,CACxE,CC5CA,MAAa,EAAoB,EAAE,OAAO,CACxC,QAAS,EAAE,OAAO,CAAC,CAAC,KAAK,EACzB,UAAW,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,EACnC,WAAY,EAAE,OAAO,CAAC,CAAC,KAAK,EAC5B,oBAAqB,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,GAAG,EAC9C,cAAe,EAAE,OAAO,CAAC,CAAC,KAAK,EAE/B,mBAAoB,EACjB,OAAO,CAAC,CACR,IAAI,CAAC,CAAC,CACN,MAAM,eAAgB,uCAAuC,CAAC,CAC9D,SAAS,EAQZ,eAAgB,EAAE,OAAO,EAAE,OAAO,EAAG,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,SAAS,CAC5E,CAAC,EAIK,EAAc,IAAI,QAExB,SAAS,EAAc,EAA2C,CAChE,IAAM,EAAS,EAAY,IAAI,CAAG,EAClC,GAAI,EAAQ,OAAO,EACnB,IAAM,EAAS,EAAkB,MAAM,CAAG,EAE1C,OADA,EAAY,IAAI,EAAK,CAAM,EACpB,CACT,CAGA,MAAM,EAAe,CAAC,KAAM,KAAM,KAAM,KAAM,KAAM,KAAM,KAAM,IAAI,EAE9D,EAAW,GAAwC,QAAQ,IAG3D,EAAuC,CAC3C,CAAE,IAAK,OAAQ,MAAO,OAAQ,SAAU,GAAM,KAAM,QAAS,EAC7D,CAAE,IAAK,YAAa,MAAO,YAAa,SAAU,GAAO,KAAM,QAAS,EACxE,CAAE,IAAK,cAAe,MAAO,cAAe,SAAU,GAAO,KAAM,QAAS,EAC5E,CAAE,IAAK,aAAc,MAAO,aAAc,SAAU,GAAO,KAAM,QAAS,EAC1E,CAAE,IAAK,WAAY,MAAO,WAAY,SAAU,GAAO,KAAM,QAAS,EACtE,CAAE,IAAK,UAAW,MAAO,UAAW,SAAU,GAAO,KAAM,QAAS,EACpE,CAAE,IAAK,eAAgB,MAAO,gBAAiB,SAAU,GAAO,KAAM,QAAS,EAC/E,CAAE,IAAK,eAAgB,MAAO,gBAAiB,SAAU,GAAO,KAAM,QAAS,EAC/E,CAAE,IAAK,iBAAkB,MAAO,iBAAkB,SAAU,GAAO,KAAM,QAAS,EAClF,CAAE,IAAK,mBAAoB,MAAO,mBAAoB,SAAU,GAAO,KAAM,QAAS,EACtF,GAAG,EAAa,IACb,IAAoB,CACnB,IAAK,EAAQ,CAAC,EACd,MAAO,SAAS,EAAE,GAClB,SAAU,GACV,KAAM,QACR,EACF,CACF,EAEM,EAAoB,SAE1B,SAAS,EAAc,EAAmB,CACxC,OAAO,EAAE,YAAY,CAAC,CAAC,QAAQ,EAAmB,EAAE,CACtD,CAQA,SAAS,EAA2B,EAAmB,EAA6B,CAClF,OAAO,KAAK,MAAM,GAAa,EAAI,EAAc,IAAI,CACvD,CAOA,MAAa,EAA2C,CACtD,KAAM,QACN,OAAQ,EACR,SAAS,EAAmB,EAAsB,CAChD,IAAM,EAAS,EAAc,EAAI,MAAM,EAGjC,EAAU,OAAO,EAAO,MAAS,SAAW,EAAO,KAAK,KAAK,EAAI,GACvE,GAAI,CAAC,EAAS,OAAO,EAAS,eAAgB,gBAAiB,MAAM,EAKrE,IAAM,EAAiB,EAHrB,EAAO,oBAAsB,EAAQ,WAAW,EAAO,kBAAkB,EACrE,EAAQ,MAAM,EAAO,mBAAmB,MAAM,EAC9C,CACuC,EAC7C,GAAI,CAAC,EACH,OAAO,EAAS,wBAAyB,SAAS,EAAQ,gCAAiC,MAAM,EAKnG,IAAM,EADe,EAAW,EAAO,QACX,GAAK,MACjC,GAAI,IAAa,MACf,OAAO,EAAS,uBAAwB,aAAa,EAAS,4BAA6B,UAAU,EAIvG,IAAM,EAAM,EAAW,EAAO,SAAS,EACvC,GAAI,EAAI,OAAS,UAAW,OAAO,EAAS,oBAAqB,4BAA6B,WAAW,EACzG,IAAM,EAAQ,EAAW,EAAO,WAAW,EAC3C,GAAI,EAAM,OAAS,UACjB,OAAO,EAAS,sBAAuB,8BAA+B,aAAa,EAErF,IAAM,EAAO,EAAW,EAAO,UAAU,EACzC,GAAI,EAAK,OAAS,UAAW,OAAO,EAAS,qBAAsB,6BAA8B,YAAY,EAE7G,IAAM,EAAW,EAAI,OAAS,QAAU,EAAI,MAAQ,KAC9C,EAAY,EAAK,OAAS,QAAU,EAAK,MAAQ,KAIjD,EAAe,GAAY,EACjC,GAAI,IAAiB,KACnB,OAAO,EAAS,gBAAiB,sCAAuC,OAAO,EASjF,IAAM,EAAgB,EAAW,EAAO,cAAc,EAChD,EAAY,IAAc,KAAO,KAAO,EAAa,CAAS,EAChE,EAAoC,KAClC,EAAyB,CAAC,EAChC,GAAI,IAAa,KACf,EAAqB,EAAa,CAAQ,OACrC,GAAI,IAAc,KAAM,CAE7B,IAAM,EAAc,IAAkB,KAAgD,IAAA,GAAzC,EAAO,iBAAiB,GACjE,IAAgB,IAAA,GAGlB,EAAS,KAAK,CACZ,YAAa,oBACb,QACE,IAAkB,KACd,qEACA,sBAAsB,EAAc,+CAC1C,MAAO,gBACT,CAAC,EATD,EAAqB,EAA2B,EAAW,CAAW,CAW1E,CAEA,IAAM,EAAQ,GAAG,EAAe,IAAI,EAAO,aACrC,EAAqB,CACzB,OAAQ,EACR,KAAM,EACN,gBAAiB,EACjB,SAAU,EAAO,QACjB,WAAY,EAAO,UACnB,YAAa,EAAO,WACpB,sBAAuB,EAAO,oBAC9B,cAAe,EACf,gBAAiB,EAAM,OAAS,QAAU,EAAM,MAAQ,KACxD,SAAU,MACV,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,QAAS,EAAW,EAAO,OAAO,EAClC,aAAc,EAAW,EAAO,YAAY,EAC5C,aAAc,EAAW,EAAO,YAAY,EAE5C,iBAAkB,EAClB,eAAgB,EAChB,iBAAkB,EAAW,EAAO,gBAAgB,EAEpD,qBAAsB,EAEtB,kBAAmB,KACnB,qBAAsB,KACtB,uBAAwB,KACxB,sBAAuB,KACvB,0BAA2B,KAC3B,gBAAiB,EAAO,cACxB,YAAa,IAAI,KAAK,CAAA,CAAE,YAAY,CACtC,EAEA,OAAO,EAAS,OAAS,EACrB,CAAE,KAAM,UAAW,GAAI,EAAO,MAAK,UAAS,EAC5C,CAAE,KAAM,UAAW,GAAI,EAAO,KAAI,CACxC,CACF"}
|
package/dist/plugin.d.mts
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { n as EsClientResolver, r as FilePathResolver, t as EnsureIndexResolver } from "./worker-
|
|
1
|
+
import { n as EsClientResolver, r as FilePathResolver, t as EnsureIndexResolver } from "./worker-Ck67AOnh.mjs";
|
|
2
2
|
import { ImportsStorageFactory } from "./storage.mjs";
|
|
3
3
|
|
|
4
4
|
//#region src/plugin.d.ts
|
package/dist/plugin.mjs
CHANGED
|
@@ -1,2 +1,2 @@
|
|
|
1
|
-
import{registerImportsStorageFactory as e}from"./storage.mjs";import{a as t,l as n,n as r}from"./transform-
|
|
1
|
+
import{registerImportsStorageFactory as e}from"./storage.mjs";import{a as t,l as n,n as r}from"./transform-BzacGXbo.mjs";import{createImportsRunDeadListener as i,createRunImportHandler as a,importsRunJob as o}from"./worker.mjs";import{definePlugin as s}from"@murumets-ee/core";const c=Symbol.for(`@murumets-ee/imports:dead-listener-unsubscribe`);function l(l){let u=l.esIndex??`parts`,d=l.storageBucket===null?null:{bucket:l.storageBucket??`lumi-imports`,orphanTtlHours:l.orphanUploadTtlHours??2,processedRetentionDays:l.processedRetentionDays??7};return l.storage&&e(l.storage),s({name:`@murumets-ee/imports`,server:{entities:[t,n],init:async e=>{if(!e.plugins.all().some(e=>e.name===`@murumets-ee/queue`)){e.logger.warn(`imports: queue() plugin not in plugins array — imports:run jobs will not be processed`);return}let n=await import(`@murumets-ee/queue/client`),s=e.getClient(t),f=r();n.registerJob(o,a({importRuns:s,transforms:f,esClient:l.esClient,esIndex:u,...l.ensureIndex!==void 0&&{ensureIndex:l.ensureIndex},...l.resolveFilePath!==void 0&&{resolveFilePath:l.resolveFilePath},logger:e.logger.child({pkg:`imports`})}));let p=globalThis[c];p&&p();let m=n.addQueueTerminalListener(i({importRuns:s,logger:e.logger.child({pkg:`imports`})}));if(globalThis[c]=m,d!==null){let{cleanupOrphanedImports:t,cleanupProcessedImports:r}=await import(`./cleanup-BzqyX9Jo.mjs`).then(e=>e.r),{runAsCli:i}=await import(`@murumets-ee/core`),a=e.logger.child({pkg:`imports`,job:`cleanup`});n.registerJob(n.defineJob({name:`imports:cleanup-orphaned-uploads`,description:`Delete toolkit_files rows in the imports bucket with no matching import_run reference and createdAt older than the configured TTL.`,schedule:`@every 1h`}),async()=>{await i(()=>t({app:e,bucket:d.bucket,ttlHours:d.orphanTtlHours,logger:a}))}),n.registerJob(n.defineJob({name:`imports:cleanup-processed-uploads`,description:`Delete toolkit_files rows for import_runs in terminal status (succeeded/failed/cancelled) whose finishedAt is older than the configured retention window. The import_run row itself is preserved.`,schedule:`@every 1h`}),async()=>{await i(()=>r({app:e,bucket:d.bucket,retentionDays:d.processedRetentionDays,logger:a}))})}e.logger.info({esIndex:u,cleanup:d===null?{enabled:!1}:{enabled:!0,...d}},`Imports plugin initialized`)}}})}export{l as imports};
|
|
2
2
|
//# sourceMappingURL=plugin.mjs.map
|
|
@@ -0,0 +1,2 @@
|
|
|
1
|
+
import{z as e}from"zod";import{createReadStream as t}from"node:fs";import{parse as n}from"csv-parse";import{bulkUpsert as r}from"@murumets-ee/search-elasticsearch";const i=50,a=5;var o=class{patterns=new Map;maxPatterns;maxSamplesPerPattern;droppedSignatures=0;constructor(e={}){this.maxPatterns=e.maxPatterns??50,this.maxSamplesPerPattern=e.maxSamplesPerPattern??5}addError(e,t,n,r,i){let a=`${t}:${r??`GENERAL`}:${n}`,o=this.patterns.get(a);if(!o){if(this.patterns.size>=this.maxPatterns){this.droppedSignatures+=1;return}o={errorType:t,field:r??null,message:n,count:0,firstOccurrence:e,lastOccurrence:e,samples:[]},this.patterns.set(a,o)}o.count+=1,o.lastOccurrence=e,o.samples.length<this.maxSamplesPerPattern&&o.samples.push({rowNumber:e,rowData:i})}getTotalErrorCount(){let e=0;for(let t of this.patterns.values())e+=t.count;return e}getDistinctPatternCount(){return this.patterns.size}getDroppedSignatureCount(){return this.droppedSignatures}getTopPatterns(){let e=Array.from(this.patterns.values()).sort((e,t)=>t.count-e.count),t=e.reduce((e,t)=>e+t.count,0);return e.map(e=>({errorType:e.errorType,field:e.field,message:e.message,count:e.count,firstOccurrence:e.firstOccurrence,lastOccurrence:e.lastOccurrence,samples:e.samples.slice(),percentage:t>0?e.count/t*100:0}))}snapshot(){return{totalErrors:this.getTotalErrorCount(),distinctPatterns:this.getDistinctPatternCount(),droppedSignatures:this.droppedSignatures,patterns:this.getTopPatterns()}}};function s(e){let t=new Set;for(let n of e)if(n!==``){if(t.has(n))throw Error(`Duplicate header "${n}" after trimming — column names must be unique`);t.add(n)}}async function c(e,n={}){let{hasHeader:r=!0}=n,i=t(e),a=0,o=0;for await(let e of i){for(let t=0;t<e.length;t+=1)e[t]===10&&(a+=1);e.length>0&&(o=e[e.length-1]??0)}return o!==0&&o!==10&&(a+=1),r?Math.max(0,a-1):a}async function*l(e){let{filePath:r,delimiter:i=` `,hasHeader:a=!0,columns:o,relaxColumnCount:c=!1,quote:l=!1}=e,u=o?Array.from(o):a?e=>{let t=e.map(e=>e.trim());return s(t),t}:!1,d=t(r),f=d.pipe(n({delimiter:i,columns:u,bom:!0,skip_empty_lines:!0,relax_column_count:c,quote:l})),p=0;try{for await(let e of f){p+=1;let t={};if(Array.isArray(e)){let n=e;for(let e=0;e<n.length;e+=1)t[String(e)]=n[e]??``}else for(let[n,r]of Object.entries(e))t[n]=r??``;yield{rowNumber:p,row:t}}}finally{d.destroy()}}const u=e.object({format:e.enum([`csv`,`tsv`,`xlsx`]).optional(),delimiter:e.string().min(1).max(4).optional(),hasHeader:e.boolean().optional(),headerRowIndex:e.number().int().min(0).optional(),sheet:e.string().min(1).max(128).optional()}).strict();function d(e){let t=e.delimiter??(e.format===`csv`?`,`:` `);return{filePath:e.filePath,delimiter:t,...e.hasHeader!==void 0&&{hasHeader:e.hasHeader},...e.columns!==void 0&&{columns:e.columns},...e.relaxColumnCount!==void 0&&{relaxColumnCount:e.relaxColumnCount},...e.quote!==void 0&&{quote:e.quote}}}async function*f(e){if(e.format===`xlsx`){let{streamXlsx:t}=await import(`./xlsx-reader-oApgZRJd.mjs`);yield*t({filePath:e.filePath,...e.hasHeader!==void 0&&{hasHeader:e.hasHeader},...e.headerRowIndex!==void 0&&{headerRowIndex:e.headerRowIndex},...e.sheet!==void 0&&{sheet:e.sheet}});return}yield*l(d(e))}const p=1e3;async function m(e){let{importRunId:t,runLabel:n,params:i,transform:a,feed:s,esClient:c,esIndex:l,batchSize:u=p,onProgress:d,rowLimit:m,signal:h,errorTracker:g=new o,warningTracker:_=new o,logger:v}=e;if(u<1)throw Error(`batchSize must be >= 1 (got ${u})`);let y=Date.now();v?.info({importRunId:t,runLabel:n,batchSize:u,esIndex:l},`imports.runner: started`);let b=0,x=0,S=0,C=0,w=0,T=0,E=[],D=async()=>{if(E.length===0)return;let e=E;E=[];let n=Date.now(),i;try{i=await r(c,{index:l,docs:e.map(({id:e,doc:t})=>({id:e,doc:t})),...h!==void 0&&{signal:h}})}catch(t){let n=h?.aborted??(t instanceof Error&&(t.name===`AbortError`||/abort/i.test(t.message)))?`aborted`:`bulk_request_failed`,r=t instanceof Error?t.message:String(t);for(let{rowNumber:t}of e)g.addError(t,n,r,void 0,null);throw S+=e.length,t}x+=i.succeeded,S+=i.failures.length,T+=1;let a=Date.now()-n,o=(Date.now()-y)/1e3;if(v?.info({importRunId:t,batch:T,flushMs:a,submitted:e.length,succeeded:i.succeeded,failed:i.failures.length,rowsRead:b,rowsPerSecond:o>0?Math.round(b/o):0},`imports.runner: batch flushed`),i.failures.length>0){let t=new Map(e.map(e=>[e.id,e.rowNumber]));for(let e of i.failures){let n=t.get(e.id)??-1;g.addError(n,e.type,e.reason,void 0,{id:e.id})}}let s=new Set(i.failures.map(e=>e.id));for(let t of e)if(!(!t.warnings||t.warnings.length===0)&&!s.has(t.id)){w+=1;for(let e of t.warnings)_.addError(t.rowNumber,e.warningType,e.message,e.field,t.rowData)}if(d){let e=(Date.now()-y)/1e3;d({rowsRead:b,rowsSucceeded:x,rowsFailed:S,rowsSkipped:C,rowsWarned:w,batchesCompleted:T,elapsedSeconds:e,rowsPerSecond:e>0?b/e:0,distinctErrorPatterns:g.getDistinctPatternCount()})}},O=u,k=new Map;for await(let{rowNumber:e,row:r}of f(s)){if(h?.aborted||m!==void 0&&b>=m)break;if(b+=1,b===1&&v?.info({importRunId:t,rowNumber:e,sampleKeys:Object.keys(r)},`imports.runner: STREAM_FIRST_ROW`),b%O===0){let e=(Date.now()-y)/1e3;v?.info({importRunId:t,rowsRead:b,rowsSucceeded:x,rowsFailed:S,rowsSkipped:C,pending:E.length,elapsedSec:Math.round(e*10)/10,rowsPerSecond:e>0?Math.round(b/e):0,distinctFirstErrorTypes:k.size},`imports.runner: STREAM_TICK`)}let o={importRunId:t,params:i,runLabel:n,rowNumber:e},s;try{s=await a(r,o)}catch(n){let i=n instanceof Error?n.message:String(n);g.addError(e,`transform_threw`,i,void 0,r),k.has(`transform_threw`)||(k.set(`transform_threw`,{rowNumber:e,message:i}),v?.warn({importRunId:t,rowNumber:e,errorType:`transform_threw`,message:i},`imports.runner: FIRST_ERROR transform_threw`)),S+=1;continue}if(s.kind===`skip`){C+=1;continue}if(s.kind===`error`){g.addError(e,s.error.errorType,s.error.message,s.error.field,r),k.has(s.error.errorType)||(k.set(s.error.errorType,{rowNumber:e,message:s.error.message}),v?.warn({importRunId:t,rowNumber:e,errorType:s.error.errorType,field:s.error.field,message:s.error.message},`imports.runner: FIRST_ERROR ${s.error.errorType}`)),S+=1;continue}E.push({id:s.id,doc:s.doc,rowNumber:e,rowData:r,...s.warnings&&s.warnings.length>0&&{warnings:s.warnings}}),E.length>=u&&await D()}await D();let A=(Date.now()-y)/1e3;return v?.info({importRunId:t,rowsRead:b,rowsSucceeded:x,rowsFailed:S,rowsSkipped:C,batchesCompleted:T,totalSec:Math.round(A*10)/10,rowsPerSecond:A>0?Math.round(b/A):0},`imports.runner: finished`),{rowsRead:b,rowsSucceeded:x,rowsFailed:S,rowsSkipped:C,rowsWarned:w,batchesCompleted:T,errors:g.snapshot(),warnings:_.snapshot()}}export{s as a,i as c,f as i,a as l,m as n,c as o,u as r,l as s,p as t,o as u};
|
|
2
|
+
//# sourceMappingURL=runner-0r6UICjo.mjs.map
|