@murumets-ee/imports 0.37.0 → 0.38.0

package/dist/xlsx-reader-CUbaPt3-.mjs.map

@@ -1 +0,0 @@
-{"version":3,"file":"xlsx-reader-CUbaPt3-.mjs","names":[],"sources":["../src/xlsx-reader.ts"],"sourcesContent":["/**\n * Streaming XLSX reader (commerce-demo phase 02, D005) — the exceljs branch of\n * the feed reader. Yields the SAME `AsyncIterable<{ rowNumber, row }>` contract as\n * the csv-parse `streamFeed`, so the runner / transform layer downstream is\n * format-agnostic.\n *\n * **exceljs over SheetJS** (R02): we parse operator-uploaded spreadsheets — the\n * exact threat model of the unfixed npm `xlsx` prototype-pollution CVE\n * (CVE-2023-30533). exceljs is MIT, ships its own types, has a real streaming\n * `WorkbookReader`, and carries no equivalent advisory.\n *\n * **exceljs is `await import()`-ed**, never statically imported, so it stays out\n * of this package's static module graph — heavy + Node-only, it must not be\n * inlinable into any `*-ui` server-pages bundle (CLAUDE.md bundle budgets). The\n * `read-feed` dispatcher additionally only loads THIS module for `format: 'xlsx'`.\n */\n\nimport type { CellValue } from 'exceljs'\nimport { assertUniqueHeaders, type StreamFeedRow } from './streaming.js'\n\nexport interface XlsxReadOptions {\n  /** Path to the .xlsx file on local disk. */\n  filePath: string\n  /** First row is the header (column names). `false` → positional `{ \"0\": … }` keys. Default `true`. */\n  hasHeader?: boolean\n  /** 0-based index of the header row (rows before it are skipped). Default `0`. */\n  headerRowIndex?: number\n  /** Worksheet to read, as a 1-based index string. Omit → the first worksheet. */\n  sheet?: string\n}\n\n/** Convert one exceljs cell value to a plain string (feeds are text/number cells). */\nfunction cellToString(v: CellValue): string {\n  if (v === null || v === undefined) return ''\n  if (typeof v === 'string') return v\n  if (typeof v === 'number' || typeof v === 'boolean') return String(v)\n  if (v instanceof Date) return v.toISOString()\n  // Object cell values: rich text, hyperlink, formula result, error.\n  if ('richText' in v && Array.isArray(v.richText)) {\n    return v.richText.map((r) => r.text).join('')\n  }\n  if ('text' in v && typeof v.text === 'string') return v.text\n  if ('result' in v && v.result !== undefined && v.result !== null) {\n    const r = v.result\n    if (r instanceof Date) return r.toISOString()\n    // A nested error object (formula error) surfaces as empty; primitives stringify.\n    return typeof r === 'object' ? '' : String(r)\n  }\n  return ''\n}\n\n/** Normalise an exceljs `row.values` (1-indexed array OR object form) to a 0-indexed string[]. */\nfunction rowCells(values: CellValue[] | { [key: string]: CellValue }): string[] {\n  const out: string[] = []\n  if (Array.isArray(values)) {\n    // exceljs row arrays are 1-indexed (`values[0]` is empty); shift to 0-based.\n    for (let i = 1; i < values.length; i += 1) out[i - 1] = cellToString(values[i])\n  } else {\n    for (const [k, v] of Object.entries(values)) {\n      const idx = Number.parseInt(k, 10)\n      if (!Number.isNaN(idx) && idx >= 1) out[idx - 1] = cellToString(v)\n    }\n  }\n  // Fill any holes (sparse rows) with empty strings — the contract is \"missing\n  // cell = ''\", matching streamFeed.\n  for (let i = 0; i < out.length; i += 1) if (out[i] === undefined) out[i] = ''\n  return out\n}\n\n/**\n * Async-iterable over the data rows of an .xlsx file. `rowNumber` is the 1-based\n * DATA row index (header excluded). Cell values are coerced to strings so the\n * mapping engine's per-column transforms (which expect raw string cells) apply\n * uniformly across csv/tsv/xlsx.\n */\nexport async function* streamXlsx(options: XlsxReadOptions): AsyncIterable<StreamFeedRow> {\n  const { filePath, hasHeader = true, headerRowIndex = 0, sheet } = options\n  const ExcelJS = await import('exceljs')\n  const reader = new ExcelJS.stream.xlsx.WorkbookReader(filePath, {\n    worksheets: 'emit',\n    sharedStrings: 'cache',\n    styles: 'ignore',\n    hyperlinks: 'ignore',\n    entries: 'ignore',\n  })\n\n  // `WorksheetReader` doesn't expose its sheet id, so count emitted worksheets\n  // (1-based) to match the configured `sheet`. Default: the first worksheet.\n  // A provided-but-invalid sheet FAILS FAST rather than silently reading sheet 1.\n  let targetSheetIndex: number | undefined\n  if (sheet !== undefined) {\n    targetSheetIndex = Number.parseInt(sheet, 10)\n    if (Number.isNaN(targetSheetIndex) || targetSheetIndex < 1) {\n      throw new Error(`Invalid xlsx sheet \"${sheet}\" — expected a 1-based worksheet index`)\n    }\n  }\n  let dataRowNumber = 0\n  let sheetIndex = 0\n\n  for await (const worksheet of reader) {\n    sheetIndex += 1\n    const isTarget =\n      targetSheetIndex === undefined ? sheetIndex === 1 : sheetIndex === targetSheetIndex\n\n    if (!isTarget) {\n      // Drain a non-target worksheet so the underlying zip stream advances to the\n      // next; exceljs's streaming reader stalls if a worksheet is left unconsumed.\n      for await (const _skip of worksheet) {\n        void _skip\n      }\n      continue\n    }\n\n    let excelRowNum = 0\n    let header: string[] | null = null\n    for await (const row of worksheet) {\n      excelRowNum += 1\n      const cells = rowCells(row.values)\n\n      if (hasHeader && header === null) {\n        // Skip rows before the configured header row, then capture it (trimmed).\n        if (excelRowNum < headerRowIndex + 1) continue\n        header = cells.map((c) => c.trim())\n        assertUniqueHeaders(header)\n        continue\n      }\n\n      dataRowNumber += 1\n      const rec: Record<string, string> = {}\n      if (header) {\n        for (let i = 0; i < header.length; i += 1) {\n          const key = header[i]\n          if (key) rec[key] = cells[i] ?? ''\n        }\n      } else {\n        for (let i = 0; i < cells.length; i += 1) rec[String(i)] = cells[i] ?? ''\n      }\n      yield { rowNumber: dataRowNumber, row: rec }\n    }\n\n    // Target worksheet fully read — done (remaining sheets are not consumed).\n    return\n  }\n\n  // Reached only when a specific sheet was requested but never matched — the\n  // first-sheet default always returns inside the loop on a non-empty workbook.\n  if (targetSheetIndex !== undefined) {\n    throw new Error(`xlsx worksheet ${targetSheetIndex} not found`)\n  }\n}\n"],"mappings":"0CAgCA,SAAS,EAAa,EAAsB,CAC1C,GAAI,GAAM,KAAyB,MAAO,GAC1C,GAAI,OAAO,GAAM,SAAU,OAAO,EAClC,GAAI,OAAO,GAAM,UAAY,OAAO,GAAM,UAAW,OAAO,OAAO,CAAC,EACpE,GAAI,aAAa,KAAM,OAAO,EAAE,YAAY,EAE5C,GAAI,aAAc,GAAK,MAAM,QAAQ,EAAE,QAAQ,EAC7C,OAAO,EAAE,SAAS,IAAK,GAAM,EAAE,IAAI,CAAC,CAAC,KAAK,EAAE,EAE9C,GAAI,SAAU,GAAK,OAAO,EAAE,MAAS,SAAU,OAAO,EAAE,KACxD,GAAI,WAAY,GAAK,EAAE,SAAW,IAAA,IAAa,EAAE,SAAW,KAAM,CAChE,IAAM,EAAI,EAAE,OAGZ,OAFI,aAAa,KAAa,EAAE,YAAY,EAErC,OAAO,GAAM,SAAW,GAAK,OAAO,CAAC,CAC9C,CACA,MAAO,EACT,CAGA,SAAS,EAAS,EAA8D,CAC9E,IAAM,EAAgB,CAAC,EACvB,GAAI,MAAM,QAAQ,CAAM,EAEtB,IAAK,IAAI,EAAI,EAAG,EAAI,EAAO,OAAQ,GAAK,EAAG,EAAI,EAAI,GAAK,EAAa,EAAO,EAAE,OAE9E,IAAK,GAAM,CAAC,EAAG,KAAM,OAAO,QAAQ,CAAM,EAAG,CAC3C,IAAM,EAAM,OAAO,SAAS,EAAG,EAAE,EAC7B,CAAC,OAAO,MAAM,CAAG,GAAK,GAAO,IAAG,EAAI,EAAM,GAAK,EAAa,CAAC,EACnE,CAIF,IAAK,IAAI,EAAI,EAAG,EAAI,EAAI,OAAQ,GAAK,EAAO,EAAI,KAAO,IAAA,KAAW,EAAI,GAAK,IAC3E,OAAO,CACT,CAQA,eAAuB,EAAW,EAAwD,CACxF,GAAM,CAAE,WAAU,YAAY,GAAM,iBAAiB,EAAG,SAAU,EAE5D,EAAS,IAAI,MADG,OAAO,YAAA,CACF,OAAO,KAAK,eAAe,EAAU,CAC9D,WAAY,OACZ,cAAe,QACf,OAAQ,SACR,WAAY,SACZ,QAAS,QACX,CAAC,EAKG,EACJ,GAAI,IAAU,IAAA,KACZ,EAAmB,OAAO,SAAS,EAAO,EAAE,EACxC,OAAO,MAAM,CAAgB,GAAK,EAAmB,GACvD,MAAU,MAAM,uBAAuB,EAAM,uCAAuC,EAGxF,IAAI,EAAgB,EAChB,EAAa,EAEjB,UAAW,IAAM,KAAa,EAAQ,CAKpC,GAJA,GAAc,EAIV,EAFF,IAAqB,IAAA,GAAY,IAAe,EAAI,IAAe,GAEtD,CAGb,UAAW,IAAM,KAAS,GAG1B,QACF,CAEA,IAAI,EAAc,EACd,EAA0B,KAC9B,UAAW,IAAM,KAAO,EAAW,CACjC,GAAe,EACf,IAAM,EAAQ,EAAS,EAAI,MAAM,EAEjC,GAAI,GAAa,IAAW,KAAM,CAEhC,GAAI,EAAc,EAAiB,EAAG,SACtC,EAAS,EAAM,IAAK,GAAM,EAAE,KAAK,CAAC,EAClC,EAAoB,CAAM,EAC1B,QACF,CAEA,GAAiB,EACjB,IAAM,EAA8B,CAAC,EACrC,GAAI,EACF,IAAK,IAAI,EAAI,EAAG,EAAI,EAAO,OAAQ,GAAK,EAAG,CACzC,IAAM,EAAM,EAAO,GACf,IAAK,EAAI,GAAO,EAAM,IAAM,GAClC,MAEA,IAAK,IAAI,EAAI,EAAG,EAAI,EAAM,OAAQ,GAAK,EAAG,EAAI,OAAO,CAAC,GAAK,EAAM,IAAM,GAEzE,KAAM,CAAE,UAAW,EAAe,IAAK,CAAI,CAC7C,CAGA,MACF,CAIA,GAAI,IAAqB,IAAA,GACvB,MAAU,MAAM,kBAAkB,EAAiB,WAAW,CAElE"}