feat: unify pdf datasheet architecture and regenerate all products

scripts/pdf/model/build-datasheet-model.ts

@@ -0,0 +1,894 @@
+import * as fs from 'fs';
+import * as path from 'path';
+
+import type { DatasheetModel, DatasheetVoltageTable, KeyValueItem, ProductData } from './types';
+import type { ExcelMatch, MediumVoltageCrossSectionExcelMatch } from './excel-index';
+import { findExcelForProduct, findMediumVoltageCrossSectionExcelForProduct } from './excel-index';
+import { getLabels, getProductUrl, normalizeValue, stripHtml } from './utils';
+
+type ExcelRow = Record<string, unknown>;
+type VoltageTableModel = {
+  voltageLabel: string;
+  metaItems: KeyValueItem[];
+  crossSections: string[];
+  columns: Array<{ key: string; label: string; get: (rowIndex: number) => string }>;
+};
+type BuildExcelModelResult = {
+  ok: boolean;
+  technicalItems: KeyValueItem[];
+  voltageTables: VoltageTableModel[];
+};
+type AssetMap = Record<string, string>;
+
+const ASSET_MAP_FILE = path.join(process.cwd(), 'data/processed/asset-map.json');
+
+function readAssetMap(): AssetMap {
+  try {
+    if (!fs.existsSync(ASSET_MAP_FILE)) return {};
+    return JSON.parse(fs.readFileSync(ASSET_MAP_FILE, 'utf8')) as AssetMap;
+  } catch {
+    return {};
+  }
+}
+const ASSET_MAP: AssetMap = readAssetMap();
+
+function normalizeUnit(unitRaw: string): string {
+  const u = normalizeValue(unitRaw);
+  if (!u) return '';
+  if (/^c$/i.test(u) || /^°c$/i.test(u)) return '°C';
+  return u.replace(/Ω/gi, 'Ohm').replace(/[\u00B5\u03BC]/g, 'u');
+}
+
+function formatExcelHeaderLabel(key: string, unit?: string): string {
+  const k = normalizeValue(key);
+  if (!k) return '';
+  const u = normalizeValue(unit || '');
+  const compact = k
+    .replace(/\s*\(approx\.?\)\s*/gi, ' (approx.) ')
+    .replace(/\s+/g, ' ')
+    .trim();
+  if (!u) return compact;
+  if (new RegExp(`\\(${u.replace(/[.*+?^${}()|[\\]\\]/g, '\\$&')}\\)`, 'i').test(compact))
+    return compact;
+  return `${compact} (${u})`;
+}
+
+function normalizeVoltageLabel(raw: string): string {
+  const v = normalizeValue(raw);
+  if (!v) return '';
+  const cleaned = v.replace(/\s+/g, ' ');
+  if (/\bkv\b/i.test(cleaned)) return cleaned.replace(/\bkv\b/i, 'kV');
+  const num = cleaned.match(/\d+(?:[.,]\d+)?(?:\s*\/\s*\d+(?:[.,]\d+)?)?/);
+  if (!num) return cleaned;
+  if (/[a-z]/i.test(cleaned)) return cleaned;
+  return `${cleaned} kV`;
+}
+
+function parseVoltageSortKey(voltageLabel: string): number {
+  const v = normalizeVoltageLabel(voltageLabel);
+  const nums = v
+    .replace(/,/g, '.')
+    .match(/\d+(?:\.\d+)?/g)
+    ?.map((n) => Number(n))
+    .filter((n) => Number.isFinite(n));
+  if (!nums || nums.length === 0) return Number.POSITIVE_INFINITY;
+  return nums[nums.length - 1];
+}
+
+function compactNumericForLocale(value: string, locale: 'en' | 'de'): string {
+  const v = normalizeValue(value);
+  if (!v) return '';
+
+  // Compact common bending-radius style: "15xD (Single core); 12xD (Multi core)" -> "15/12xD".
+  // Keep semantics, reduce width. Never truncate with ellipses.
+  if (/\d+xD/i.test(v)) {
+    const nums = Array.from(v.matchAll(/(\d+)xD/gi))
+      .map((m) => m[1])
+      .filter(Boolean);
+    const unique: string[] = [];
+    for (const n of nums) {
+      if (!unique.includes(n)) unique.push(n);
+    }
+    if (unique.length) return `${unique.join('/')}xD`;
+  }
+
+  const hasDigit = /\d/.test(v);
+  if (!hasDigit) return v;
+  const trimmed = v.replace(/\s+/g, ' ').trim();
+  const parts = trimmed.split(/(–|-)/);
+  const out = parts.map((p) => {
+    if (p === '–' || p === '-') return p;
+    const s = p.trim();
+    if (!/^-?\d+(?:[.,]\d+)?$/.test(s)) return p;
+    const n = s.replace(/,/g, '.');
+    const compact = n
+      .replace(/\.0+$/, '')
+      .replace(/(\.\d*?)0+$/, '$1')
+      .replace(/\.$/, '');
+    const hadPlus = /^\+/.test(s);
+    const withPlus = hadPlus && !/^\+/.test(compact) ? `+${compact}` : compact;
+    return locale === 'de' ? withPlus.replace(/\./g, ',') : withPlus;
+  });
+  return out.join('');
+}
+
+function compactCellForDenseTable(
+  value: string,
+  unit: string | undefined,
+  locale: 'en' | 'de',
+): string {
+  let v = normalizeValue(value);
+  if (!v) return '';
+  const u = normalizeValue(unit || '');
+  if (u) {
+    const esc = u.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+    v = v.replace(new RegExp(`\\s*${esc}\\b`, 'ig'), '').trim();
+    v = v
+      .replace(/\bkg\s*\/\s*km\b/gi, '')
+      .replace(/\bohm\s*\/\s*km\b/gi, '')
+      .replace(/\bΩ\s*\/\s*km\b/gi, '')
+      .replace(/\bu\s*f\s*\/\s*km\b/gi, '')
+      .replace(/\bmh\s*\/\s*km\b/gi, '')
+      .replace(/\bkA\b/gi, '')
+      .replace(/\bmm\b/gi, '')
+      .replace(/\bkv\b/gi, '')
+      .replace(/\b°?c\b/gi, '')
+      .replace(/\s+/g, ' ')
+      .trim();
+  }
+  v = v
+    .replace(/\s*–\s*/g, '-')
+    .replace(/\s*-\s*/g, '-')
+    .replace(/\s*\/\s*/g, '/')
+    .replace(/\s+/g, ' ')
+    .trim();
+  return compactNumericForLocale(v, locale);
+}
+
+function resolveMediaToLocalPath(urlOrPath: string | null | undefined): string | null {
+  if (!urlOrPath) return null;
+  if (urlOrPath.startsWith('/')) return urlOrPath;
+  if (/^media\//i.test(urlOrPath)) return `/${urlOrPath}`;
+  const mapped = ASSET_MAP[urlOrPath];
+  if (mapped) {
+    if (mapped.startsWith('/')) return mapped;
+    if (/^public\//i.test(mapped)) return `/${mapped.replace(/^public\//i, '')}`;
+    if (/^media\//i.test(mapped)) return `/${mapped}`;
+    return mapped;
+  }
+  return urlOrPath;
+}
+
+function guessColumnKey(row: ExcelRow, patterns: RegExp[]): string | null {
+  const keys = Object.keys(row || {});
+  for (const re of patterns) {
+    const k = keys.find((x) => {
+      const key = String(x);
+      if (re.test('conductor') && /ross section conductor/i.test(key)) return false;
+      if (re.test('insulation thickness') && /Diameter over insulation/i.test(key)) return false;
+      if (re.test('conductor') && !/^conductor$/i.test(key)) return false;
+      if (re.test('insulation') && !/^insulation$/i.test(key)) return false;
+      if (re.test('sheath') && !/^sheath$/i.test(key)) return false;
+      if (re.test('norm') && !/^norm$/i.test(key)) return false;
+      return re.test(key);
+    });
+    if (k) return k;
+  }
+  return null;
+}
+
+function technicalFullLabel(args: { key: string; excelKey: string; locale: 'en' | 'de' }): string {
+  if (args.locale === 'en') return normalizeValue(args.excelKey);
+  const raw = normalizeValue(args.excelKey);
+  if (!raw) return '';
+  return raw
+    .replace(/\(approx\.?\)/gi, '(ca.)')
+    .replace(/\bcapacitance\b/gi, 'Kapazität')
+    .replace(/\binductance\b/gi, 'Induktivität')
+    .replace(/\breactance\b/gi, 'Reaktanz')
+    .replace(/\btest voltage\b/gi, 'Prüfspannung')
+    .replace(/\brated voltage\b/gi, 'Nennspannung')
+    .replace(/\boperating temperature range\b/gi, 'Temperaturbereich')
+    .replace(/\bminimum sheath thickness\b/gi, 'Manteldicke (min.)')
+    .replace(/\bsheath thickness\b/gi, 'Manteldicke')
+    .replace(/\bnominal insulation thickness\b/gi, 'Isolationsdicke (nom.)')
+    .replace(/\binsulation thickness\b/gi, 'Isolationsdicke')
+    .replace(/\bdc resistance at 20\s*°?c\b/gi, 'DC-Leiterwiderstand (20 °C)')
+    .replace(/\bouter diameter(?: of cable)?\b/gi, 'Außen-Ø')
+    .replace(/\bbending radius\b/gi, 'Biegeradius')
+    .replace(/\bpackaging\b/gi, 'Verpackung')
+    .replace(/\bce\s*-?conformity\b/gi, 'CE-Konformität');
+}
+
+function metaFullLabel(args: { key: string; excelKey: string; locale: 'en' | 'de' }): string {
+  const key = normalizeValue(args.key);
+  if (args.locale === 'de') {
+    switch (key) {
+      case 'test_volt':
+        return 'Prüfspannung';
+      case 'temp_range':
+        return 'Temperaturbereich';
+      case 'max_op_temp':
+        return 'Leitertemperatur (max.)';
+      case 'max_sc_temp':
+        return 'Kurzschlusstemperatur (max.)';
+      case 'min_lay_temp':
+        return 'Minimale Verlegetemperatur';
+      case 'min_store_temp':
+        return 'Minimale Lagertemperatur';
+      case 'cpr':
+        return 'CPR-Klasse';
+      case 'flame':
+        return 'Flammhemmend';
+      default:
+        return formatExcelHeaderLabel(args.excelKey);
+    }
+  }
+
+  switch (key) {
+    case 'test_volt':
+      return 'Test voltage';
+    case 'temp_range':
+      return 'Operating temperature range';
+    case 'max_op_temp':
+      return 'Conductor temperature (max.)';
+    case 'max_sc_temp':
+      return 'Short-circuit temperature (max.)';
+    case 'min_lay_temp':
+      return 'Minimum laying temperature';
+    case 'min_store_temp':
+      return 'Minimum storage temperature';
+    case 'cpr':
+      return 'CPR class';
+    case 'flame':
+      return 'Flame retardant';
+    default:
+      return formatExcelHeaderLabel(args.excelKey);
+  }
+}
+
+function denseAbbrevLabel(args: { key: string; locale: 'en' | 'de'; unit?: string }): string {
+  const u = normalizeUnit(args.unit || '');
+  const unitSafe = u.replace(/Ω/gi, 'Ohm').replace(/[\u00B5\u03BC]/g, 'u');
+  const suffix = unitSafe ? ` [${unitSafe}]` : '';
+
+  switch (args.key) {
+    case 'DI':
+    case 'RI':
+    case 'Wi':
+    case 'Ibl':
+    case 'Ibe':
+    case 'Wm':
+    case 'Rbv':
+    case 'Fzv':
+    case 'G':
+      return `${args.key}${suffix}`;
+    case 'Ik_cond':
+      return `Ik${suffix}`;
+    case 'Ik_screen':
+      return `Ik_s${suffix}`;
+    case 'Ø':
+      return `Ø${suffix}`;
+    case 'Cond':
+      return args.locale === 'de' ? 'Leiter' : 'Cond.';
+    case 'shape':
+      return args.locale === 'de' ? 'Form' : 'Shape';
+    // Electrical
+    case 'cap':
+      // Capacitance. Use a clear label; lowercase "cap" looks like an internal key.
+      return `Cap${suffix}`;
+    case 'X':
+      return `X${suffix}`;
+    case 'test_volt':
+      return `U_test${suffix}`;
+    case 'rated_volt':
+      return `U0/U${suffix}`;
+    case 'temp_range':
+      return `T${suffix}`;
+    case 'max_op_temp':
+      return `T_op${suffix}`;
+    case 'max_sc_temp':
+      return `T_sc${suffix}`;
+    case 'min_store_temp':
+      return `T_st${suffix}`;
+    case 'min_lay_temp':
+      return `T_lay${suffix}`;
+    case 'cpr':
+      return `CPR${suffix}`;
+    case 'flame':
+      return `FR${suffix}`;
+    default:
+      return args.key || '';
+  }
+}
+
+function summarizeOptions(options: string[] | undefined): string {
+  const vals = (options || []).map(normalizeValue).filter(Boolean);
+  if (vals.length === 0) return '';
+  const uniq = Array.from(new Set(vals));
+  if (uniq.length === 1) return uniq[0];
+  // Never use ellipsis truncation in datasheets. Prefer full value list.
+  // (Long values should be handled by layout; if needed we can later add wrapping rules.)
+  return uniq.join(' / ');
+}
+
+function parseNumericOption(value: string): number | null {
+  const v = normalizeValue(value).replace(/,/g, '.');
+  const m = v.match(/-?\d+(?:\.\d+)?/);
+  if (!m) return null;
+  const n = Number(m[0]);
+  return Number.isFinite(n) ? n : null;
+}
+
+function summarizeNumericRange(options: string[] | undefined): { ok: boolean; text: string } {
+  const vals = (options || []).map(parseNumericOption).filter((n): n is number => n !== null);
+  if (vals.length < 3) return { ok: false, text: '' };
+  const uniq = Array.from(new Set(vals));
+  if (uniq.length < 4) return { ok: false, text: '' };
+  uniq.sort((a, b) => a - b);
+  const min = uniq[0];
+  const max = uniq[uniq.length - 1];
+  const fmt = (n: number) => (Number.isInteger(n) ? String(n) : String(n)).replace(/\.0+$/, '');
+  return { ok: true, text: `${fmt(min)}–${fmt(max)}` };
+}
+
+function summarizeSmartOptions(_label: string, options: string[] | undefined): string {
+  const range = summarizeNumericRange(options);
+  if (range.ok) return range.text;
+  return summarizeOptions(options);
+}
+
+function normalizeDesignation(value: string): string {
+  return String(value || '')
+    .toUpperCase()
+    .replace(/-\d+$/g, '')
+    .replace(/[^A-Z0-9]+/g, '');
+}
+
+function buildExcelModel(args: {
+  product: ProductData;
+  locale: 'en' | 'de';
+}): BuildExcelModelResult {
+  const match = findExcelForProduct(args.product) as ExcelMatch | null;
+  if (!match || match.rows.length === 0)
+    return { ok: false, technicalItems: [], voltageTables: [] };
+
+  const units = match.units || {};
+  const rows = match.rows;
+  let sample = rows.find((r) => r && Object.keys(r).length > 0) || {};
+  let maxColumns = Object.keys(sample).filter(
+    (k) => k && k !== 'Part Number' && k !== 'Units',
+  ).length;
+  for (const r of rows) {
+    const cols = Object.keys(r).filter((k) => k && k !== 'Part Number' && k !== 'Units').length;
+    if (cols > maxColumns) {
+      sample = r;
+      maxColumns = cols;
+    }
+  }
+
+  const columnMapping: Record<string, { header: string; unit: string; key: string }> = {
+    'number of cores and cross-section': {
+      header: 'Cross-section',
+      unit: '',
+      key: 'cross_section',
+    },
+    'ross section conductor': { header: 'Cross-section', unit: '', key: 'cross_section' },
+
+    'diameter over insulation': { header: 'DI', unit: 'mm', key: 'DI' },
+    'diameter over insulation (approx.)': { header: 'DI', unit: 'mm', key: 'DI' },
+    'dc resistance at 20 °C': { header: 'RI', unit: 'Ohm/km', key: 'RI' },
+    'dc resistance at 20°C': { header: 'RI', unit: 'Ohm/km', key: 'RI' },
+    'resistance conductor': { header: 'RI', unit: 'Ohm/km', key: 'RI' },
+    'maximum resistance of conductor': { header: 'RI', unit: 'Ohm/km', key: 'RI' },
+    'insulation thickness': { header: 'Wi', unit: 'mm', key: 'Wi' },
+    'nominal insulation thickness': { header: 'Wi', unit: 'mm', key: 'Wi' },
+    'current ratings in air, trefoil': { header: 'Ibl', unit: 'A', key: 'Ibl' },
+    'current ratings in air, trefoil*': { header: 'Ibl', unit: 'A', key: 'Ibl' },
+    'current ratings in ground, trefoil': { header: 'Ibe', unit: 'A', key: 'Ibe' },
+    'current ratings in ground, trefoil*': { header: 'Ibe', unit: 'A', key: 'Ibe' },
+    'conductor shortcircuit current': { header: 'Ik', unit: 'kA', key: 'Ik_cond' },
+    'screen shortcircuit current': { header: 'Ik', unit: 'kA', key: 'Ik_screen' },
+    'sheath thickness': { header: 'Wm', unit: 'mm', key: 'Wm' },
+    'minimum sheath thickness': { header: 'Wm', unit: 'mm', key: 'Wm' },
+    'nominal sheath thickness': { header: 'Wm', unit: 'mm', key: 'Wm' },
+    'bending radius': { header: 'Rbv', unit: 'mm', key: 'Rbv' },
+    'bending radius (min.)': { header: 'Rbv', unit: 'mm', key: 'Rbv' },
+    'outer diameter': { header: 'Ø', unit: 'mm', key: 'Ø' },
+    'outer diameter (approx.)': { header: 'Ø', unit: 'mm', key: 'Ø' },
+    'outer diameter of cable': { header: 'Ø', unit: 'mm', key: 'Ø' },
+    'pulling force': { header: 'Fzv', unit: 'N', key: 'Fzv' },
+    'max. pulling force': { header: 'Fzv', unit: 'N', key: 'Fzv' },
+    'conductor aluminum': { header: 'Cond.', unit: '', key: 'Cond' },
+    'conductor copper': { header: 'Cond.', unit: '', key: 'Cond' },
+    weight: { header: 'G', unit: 'kg/km', key: 'G' },
+    'weight (approx.)': { header: 'G', unit: 'kg/km', key: 'G' },
+    'cable weight': { header: 'G', unit: 'kg/km', key: 'G' },
+
+    'shape of conductor': { header: 'Conductor shape', unit: '', key: 'shape' },
+    'operating temperature range': {
+      header: 'Operating temp range',
+      unit: '°C',
+      key: 'temp_range',
+    },
+    'maximal operating conductor temperature': {
+      header: 'Max operating temp',
+      unit: '°C',
+      key: 'max_op_temp',
+    },
+    'maximal short-circuit temperature': {
+      header: 'Max short-circuit temp',
+      unit: '°C',
+      key: 'max_sc_temp',
+    },
+    'minimal storage temperature': {
+      header: 'Min storage temp',
+      unit: '°C',
+      key: 'min_store_temp',
+    },
+    'minimal temperature for laying': {
+      header: 'Min laying temp',
+      unit: '°C',
+      key: 'min_lay_temp',
+    },
+    'test voltage': { header: 'Test voltage', unit: 'kV', key: 'test_volt' },
+    'rated voltage': { header: 'Rated voltage', unit: 'kV', key: 'rated_volt' },
+    'cpr class': { header: 'CPR class', unit: '', key: 'cpr' },
+    'flame retardant': { header: 'Flame retardant', unit: '', key: 'flame' },
+    'self-extinguishing of single cable': { header: 'Flame retardant', unit: '', key: 'flame' },
+
+    // High-value electrical/screen columns
+    'capacitance (approx.)': { header: 'Capacitance', unit: 'uF/km', key: 'cap' },
+    capacitance: { header: 'Capacitance', unit: 'uF/km', key: 'cap' },
+    reactance: { header: 'Reactance', unit: 'Ohm/km', key: 'X' },
+    'diameter over screen': { header: 'Diameter over screen', unit: 'mm', key: 'D_screen' },
+    'metallic screen mm2': { header: 'Metallic screen', unit: 'mm2', key: 'S_screen' },
+    'metallic screen': { header: 'Metallic screen', unit: 'mm2', key: 'S_screen' },
+  };
+
+  const excelKeys = Object.keys(sample).filter((k) => k && k !== 'Part Number' && k !== 'Units');
+  const matchedColumns: Array<{
+    excelKey: string;
+    mapping: { header: string; unit: string; key: string };
+  }> = [];
+  for (const excelKey of excelKeys) {
+    const normalized = normalizeValue(excelKey).toLowerCase();
+    for (const [pattern, mapping] of Object.entries(columnMapping)) {
+      if (normalized === pattern.toLowerCase() || new RegExp(pattern, 'i').test(normalized)) {
+        matchedColumns.push({ excelKey, mapping });
+        break;
+      }
+    }
+  }
+
+  const seenKeys = new Set<string>();
+  const deduplicated: typeof matchedColumns = [];
+  for (const item of matchedColumns) {
+    if (!seenKeys.has(item.mapping.key)) {
+      seenKeys.add(item.mapping.key);
+      deduplicated.push(item);
+    }
+  }
+
+  const sampleKeys = Object.keys(sample)
+    .filter((k) => k && k !== 'Part Number' && k !== 'Units')
+    .sort();
+  const compatibleRows = rows.filter((r) => {
+    const rKeys = Object.keys(r)
+      .filter((k) => k && k !== 'Part Number' && k !== 'Units')
+      .sort();
+    return JSON.stringify(rKeys) === JSON.stringify(sampleKeys);
+  });
+  if (compatibleRows.length === 0) return { ok: false, technicalItems: [], voltageTables: [] };
+
+  const csKey =
+    guessColumnKey(sample, [
+      /number of cores and cross-section/i,
+      /cross.?section/i,
+      /ross section conductor/i,
+    ]) || null;
+  const voltageKey =
+    guessColumnKey(sample, [/rated voltage/i, /voltage rating/i, /nennspannung/i, /spannungs/i]) ||
+    null;
+  if (!csKey) return { ok: false, technicalItems: [], voltageTables: [] };
+
+  const byVoltage = new Map<string, number[]>();
+  for (let i = 0; i < compatibleRows.length; i++) {
+    const cs = normalizeValue(String(compatibleRows[i]?.[csKey] ?? ''));
+    if (!cs) continue;
+    const rawV = voltageKey ? normalizeValue(String(compatibleRows[i]?.[voltageKey] ?? '')) : '';
+    const voltageLabel = normalizeVoltageLabel(rawV || '');
+    const key = voltageLabel || (args.locale === 'de' ? 'Spannung unbekannt' : 'Voltage unknown');
+    const arr = byVoltage.get(key) ?? [];
+    arr.push(i);
+    byVoltage.set(key, arr);
+  }
+
+  const voltageKeysSorted = Array.from(byVoltage.keys()).sort((a, b) => {
+    const na = parseVoltageSortKey(a);
+    const nb = parseVoltageSortKey(b);
+    if (na !== nb) return na - nb;
+    return a.localeCompare(b);
+  });
+
+  const technicalItems: KeyValueItem[] = [];
+  const globalConstantColumns = new Set<string>();
+
+  for (const { excelKey, mapping } of deduplicated) {
+    const values = compatibleRows
+      .map((r) => normalizeValue(String(r?.[excelKey] ?? '')))
+      .filter(Boolean);
+    const unique = Array.from(new Set(values.map((v) => v.toLowerCase())));
+    if (unique.length === 1 && values.length > 0) {
+      globalConstantColumns.add(excelKey);
+      const unit = normalizeUnit(units[excelKey] || mapping.unit || '');
+      const labelBase = technicalFullLabel({ key: mapping.key, excelKey, locale: args.locale });
+      const label = formatExcelHeaderLabel(labelBase, unit);
+      const value = compactCellForDenseTable(values[0], unit, args.locale);
+      if (!technicalItems.find((t) => t.label === label))
+        technicalItems.push({ label, value, unit });
+    }
+  }
+  technicalItems.sort((a, b) => a.label.localeCompare(b.label));
+
+  const voltageTables: VoltageTableModel[] = [];
+  for (const vKey of voltageKeysSorted) {
+    const indices = byVoltage.get(vKey) || [];
+    if (!indices.length) continue;
+    const crossSections = indices.map((idx) =>
+      normalizeValue(String(compatibleRows[idx]?.[csKey] ?? '')),
+    );
+
+    const metaItems: KeyValueItem[] = [];
+    const metaCandidates = new Map<string, KeyValueItem>();
+
+    if (voltageKey) {
+      const rawV = normalizeValue(String(compatibleRows[indices[0]]?.[voltageKey] ?? ''));
+      metaItems.push({
+        label: args.locale === 'de' ? 'Spannung' : 'Voltage',
+        value: normalizeVoltageLabel(rawV || ''),
+      });
+    }
+
+    const metaKeyPriority = [
+      'test_volt',
+      'temp_range',
+      'max_op_temp',
+      'max_sc_temp',
+      'min_lay_temp',
+      'min_store_temp',
+      'cpr',
+      'flame',
+    ];
+    const metaKeyPrioritySet = new Set(metaKeyPriority);
+
+    const denseTableKeyOrder = [
+      'Cond',
+      'shape',
+      // Electrical properties (when present)
+      'cap',
+      'X',
+      // Dimensions and ratings
+      'DI',
+      'RI',
+      'Wi',
+      'Ibl',
+      'Ibe',
+      'Ik_cond',
+      'Wm',
+      'Rbv',
+      'Ø',
+      // Screen data (when present)
+      'D_screen',
+      'S_screen',
+      'Fzv',
+      'G',
+    ] as const;
+    const denseTableKeys = new Set<string>(denseTableKeyOrder);
+
+    const tableColumns: Array<{
+      excelKey: string;
+      mapping: { header: string; unit: string; key: string };
+    }> = [];
+    for (const { excelKey, mapping } of deduplicated) {
+      if (excelKey === csKey || excelKey === voltageKey) continue;
+      const values = indices
+        .map((idx) => normalizeValue(String(compatibleRows[idx]?.[excelKey] ?? '')))
+        .filter(Boolean);
+      if (!values.length) continue;
+      const unique = Array.from(new Set(values.map((v) => v.toLowerCase())));
+      const unit = normalizeUnit(units[excelKey] || mapping.unit || '');
+
+      if (denseTableKeys.has(mapping.key)) {
+        tableColumns.push({ excelKey, mapping });
+        continue;
+      }
+
+      if (globalConstantColumns.has(excelKey) && !metaKeyPrioritySet.has(mapping.key)) {
+        continue;
+      }
+
+      const value =
+        unique.length === 1
+          ? compactCellForDenseTable(values[0], unit, args.locale)
+          : summarizeSmartOptions(excelKey, values);
+      const label = metaFullLabel({ key: mapping.key, excelKey, locale: args.locale });
+      metaCandidates.set(mapping.key, { label, value, unit });
+    }
+
+    for (const k of metaKeyPriority) {
+      const item = metaCandidates.get(k);
+      if (item && item.label && item.value) metaItems.push(item);
+    }
+
+    const mappedByKey = new Map<
+      string,
+      { excelKey: string; mapping: { header: string; unit: string; key: string } }
+    >();
+    for (const c of tableColumns) {
+      if (!mappedByKey.has(c.mapping.key)) mappedByKey.set(c.mapping.key, c);
+    }
+
+    // If conductor material is missing in Excel, derive it from designation.
+    // NA... => Al, N... => Cu (common for this dataset).
+    if (!mappedByKey.has('Cond')) {
+      mappedByKey.set('Cond', {
+        excelKey: '',
+        mapping: { header: 'Cond.', unit: '', key: 'Cond' },
+      });
+    }
+
+    const orderedTableColumns = denseTableKeyOrder
+      .filter((k) => mappedByKey.has(k))
+      .map((k) => mappedByKey.get(k)!)
+      .map(({ excelKey, mapping }) => {
+        const unit = normalizeUnit((excelKey ? units[excelKey] : '') || mapping.unit || '');
+        return {
+          key: mapping.key,
+          label:
+            denseAbbrevLabel({ key: mapping.key, locale: args.locale, unit }) ||
+            formatExcelHeaderLabel(excelKey, unit),
+          get: (rowIndex: number) => {
+            const srcRowIndex = indices[rowIndex];
+
+            if (mapping.key === 'Cond' && !excelKey) {
+              const pn = normalizeDesignation(
+                args.product.name || args.product.slug || args.product.sku || '',
+              );
+              if (/^NA/.test(pn)) return 'Al';
+              if (/^N/.test(pn)) return 'Cu';
+              return '';
+            }
+
+            const raw = excelKey
+              ? normalizeValue(String(compatibleRows[srcRowIndex]?.[excelKey] ?? ''))
+              : '';
+            return compactCellForDenseTable(raw, unit, args.locale);
+          },
+        };
+      });
+
+    voltageTables.push({
+      voltageLabel: vKey,
+      metaItems,
+      crossSections,
+      columns: orderedTableColumns,
+    });
+  }
+
+  return { ok: true, technicalItems, voltageTables };
+}
+
+function isMediumVoltageProduct(product: ProductData): boolean {
+  const hay = [
+    product.slug,
+    product.path,
+    product.translationKey,
+    ...(product.categories || []).map((c) => c.name),
+  ]
+    .filter(Boolean)
+    .join(' ');
+  return /medium[-\s]?voltage|mittelspannung/i.test(hay);
+}
+
+type AbbrevColumn = { colKey: string; unit: string };
+
+function isAbbreviatedHeaderKey(key: string): boolean {
+  const k = normalizeValue(key);
+  if (!k) return false;
+  if (/^__EMPTY/i.test(k)) return false;
+
+  // Examples from the MV sheet: "LD mm", "RI Ohm", "G kg", "SBL 30", "SBE 20", "BK", "BR", "LF".
+  // Keep this permissive but focused on compact, non-sentence identifiers.
+  if (k.length > 12) return false;
+  if (/[a-z]{4,}/.test(k)) return false;
+  if (!/[A-ZØ]/.test(k)) return false;
+  return true;
+}
+
+function extractAbbrevColumnsFromMediumVoltageHeader(args: {
+  headerRow: Record<string, unknown>;
+  units: Record<string, string>;
+  partNumberKey: string;
+  crossSectionKey: string;
+  ratedVoltageKey: string | null;
+}): AbbrevColumn[] {
+  const out: AbbrevColumn[] = [];
+
+  for (const colKey of Object.keys(args.headerRow || {})) {
+    if (!colKey) continue;
+    if (colKey === args.partNumberKey) continue;
+    if (colKey === args.crossSectionKey) continue;
+    if (args.ratedVoltageKey && colKey === args.ratedVoltageKey) continue;
+
+    if (!isAbbreviatedHeaderKey(colKey)) continue;
+
+    const unit = normalizeUnit(args.units[colKey] || '');
+    out.push({ colKey, unit });
+  }
+
+  return out;
+}
+
+function buildMediumVoltageCrossSectionTableFromNewExcel(args: {
+  product: ProductData;
+  locale: 'en' | 'de';
+}): BuildExcelModelResult & { legendItems: KeyValueItem[] } {
+  const mv = findMediumVoltageCrossSectionExcelForProduct(
+    args.product,
+  ) as MediumVoltageCrossSectionExcelMatch | null;
+  if (!mv || !mv.rows.length)
+    return { ok: false, technicalItems: [], voltageTables: [], legendItems: [] };
+  if (!mv.crossSectionKey)
+    return { ok: false, technicalItems: [], voltageTables: [], legendItems: [] };
+
+  const abbrevCols = extractAbbrevColumnsFromMediumVoltageHeader({
+    headerRow: mv.headerRow,
+    units: mv.units,
+    partNumberKey: mv.partNumberKey,
+    crossSectionKey: mv.crossSectionKey,
+    ratedVoltageKey: mv.ratedVoltageKey,
+  });
+  if (!abbrevCols.length)
+    return { ok: false, technicalItems: [], voltageTables: [], legendItems: [] };
+
+  // Collect legend items: abbreviation -> description from header row
+  const legendItems: KeyValueItem[] = [];
+  for (const col of abbrevCols) {
+    const description = normalizeValue(String(mv.headerRow[col.colKey] || ''));
+    if (description && description !== col.colKey) {
+      legendItems.push({
+        label: col.colKey,
+        value: description,
+      });
+    }
+  }
+
+  const byVoltage = new Map<string, number[]>();
+  for (let i = 0; i < mv.rows.length; i++) {
+    const cs = normalizeValue(
+      String((mv.rows[i] as Record<string, unknown>)?.[mv.crossSectionKey] ?? ''),
+    );
+    if (!cs) continue;
+
+    const rawV = mv.ratedVoltageKey
+      ? normalizeValue(String((mv.rows[i] as Record<string, unknown>)?.[mv.ratedVoltageKey] ?? ''))
+      : '';
+
+    const voltageLabel = normalizeVoltageLabel(rawV || '');
+    const key = voltageLabel || (args.locale === 'de' ? 'Spannung unbekannt' : 'Voltage unknown');
+    const arr = byVoltage.get(key) ?? [];
+    arr.push(i);
+    byVoltage.set(key, arr);
+  }
+
+  const voltageKeysSorted = Array.from(byVoltage.keys()).sort((a, b) => {
+    const na = parseVoltageSortKey(a);
+    const nb = parseVoltageSortKey(b);
+    if (na !== nb) return na - nb;
+    return a.localeCompare(b);
+  });
+
+  const voltageTables: VoltageTableModel[] = [];
+  for (const vKey of voltageKeysSorted) {
+    const indices = byVoltage.get(vKey) || [];
+    if (!indices.length) continue;
+
+    const crossSections = indices.map((idx) =>
+      normalizeValue(String((mv.rows[idx] as Record<string, unknown>)?.[mv.crossSectionKey] ?? '')),
+    );
+
+    const metaItems: KeyValueItem[] = [];
+    if (mv.ratedVoltageKey) {
+      const rawV = normalizeValue(
+        String((mv.rows[indices[0]] as Record<string, unknown>)?.[mv.ratedVoltageKey] ?? ''),
+      );
+      metaItems.push({
+        label: args.locale === 'de' ? 'Spannung' : 'Voltage',
+        value: normalizeVoltageLabel(rawV || ''),
+      });
+    }
+
+    const columns = abbrevCols.map((col) => {
+      return {
+        key: col.colKey,
+        // Use the abbreviated title from the first row as the table header.
+        label: normalizeValue(col.colKey),
+        get: (rowIndex: number) => {
+          const srcRowIndex = indices[rowIndex];
+          const raw = normalizeValue(
+            String((mv.rows[srcRowIndex] as Record<string, unknown>)?.[col.colKey] ?? ''),
+          );
+          return compactCellForDenseTable(raw, col.unit, args.locale);
+        },
+      };
+    });
+
+    voltageTables.push({ voltageLabel: vKey, metaItems, crossSections, columns });
+  }
+
+  return { ok: true, technicalItems: [], voltageTables, legendItems };
+}
+
+export function buildDatasheetModel(args: {
+  product: ProductData;
+  locale: 'en' | 'de';
+}): DatasheetModel {
+  const labels = getLabels(args.locale);
+  const categoriesLine = (args.product.categories || []).map((c) => stripHtml(c.name)).join(' • ');
+  const descriptionText = stripHtml(
+    args.product.shortDescriptionHtml || args.product.descriptionHtml || '',
+  );
+  const heroSrc = resolveMediaToLocalPath(
+    args.product.featuredImage || args.product.images?.[0] || null,
+  );
+  const productUrl = getProductUrl(args.product);
+
+  // Technical data MUST stay sourced from the existing Excel index (legacy sheets).
+  const excelModel = buildExcelModel({ product: args.product, locale: args.locale });
+
+  // Cross-section tables: for medium voltage only, prefer the new MV sheet (abbrev columns in header row).
+  const crossSectionModel = isMediumVoltageProduct(args.product)
+    ? buildMediumVoltageCrossSectionTableFromNewExcel({
+        product: args.product,
+        locale: args.locale,
+      })
+    : { ok: false, technicalItems: [], voltageTables: [], legendItems: [] };
+
+  const voltageTablesSrc = crossSectionModel.ok
+    ? crossSectionModel.voltageTables
+    : excelModel.ok
+      ? excelModel.voltageTables
+      : [];
+
+  const voltageTables: DatasheetVoltageTable[] = voltageTablesSrc.map((t) => {
+    const columns = t.columns.map((c) => ({ key: c.key, label: c.label }));
+    const rows = t.crossSections.map((configuration, rowIndex) => ({
+      configuration,
+      cells: t.columns.map((c) => compactNumericForLocale(c.get(rowIndex), args.locale)),
+    }));
+    return {
+      voltageLabel: t.voltageLabel,
+      metaItems: t.metaItems,
+      columns,
+      rows,
+    };
+  });
+
+  return {
+    locale: args.locale,
+    product: {
+      id: args.product.id,
+      name: stripHtml(args.product.name),
+      sku: args.product.sku,
+      categoriesLine,
+      descriptionText,
+      heroSrc,
+      productUrl,
+    },
+    labels,
+    technicalItems: excelModel.ok ? excelModel.technicalItems : [],
+    voltageTables,
+    legendItems: crossSectionModel.legendItems || [],
+  };
+}

scripts/pdf/model/excel-index.ts

@@ -0,0 +1,204 @@
+import * as fs from 'fs';
+import * as path from 'path';
+import { execSync } from 'child_process';
+
+import type { ProductData } from './types';
+import { normalizeValue } from './utils';
+
+type ExcelRow = Record<string, unknown>;
+export type ExcelMatch = { rows: ExcelRow[]; units: Record<string, string> };
+
+export type MediumVoltageCrossSectionExcelMatch = {
+  headerRow: ExcelRow;
+  rows: ExcelRow[];
+  units: Record<string, string>;
+  partNumberKey: string;
+  crossSectionKey: string;
+  ratedVoltageKey: string | null;
+};
+
+const EXCEL_SOURCE_FILES = [
+  path.join(process.cwd(), 'data/excel/high-voltage.xlsx'),
+  path.join(process.cwd(), 'data/excel/medium-voltage-KM.xlsx'),
+  path.join(process.cwd(), 'data/excel/low-voltage-KM.xlsx'),
+  path.join(process.cwd(), 'data/excel/solar-cables.xlsx'),
+];
+
+// Medium-voltage cross-section table (new format with multi-row header).
+// IMPORTANT: this must NOT be used for the technical data table.
+const MV_CROSS_SECTION_FILE = path.join(process.cwd(), 'data/excel/medium-voltage-KM 170126.xlsx');
+
+type MediumVoltageCrossSectionIndex = {
+  headerRow: ExcelRow;
+  units: Record<string, string>;
+  partNumberKey: string;
+  crossSectionKey: string;
+  ratedVoltageKey: string | null;
+  rowsByDesignation: Map<string, ExcelRow[]>;
+};
+
+let EXCEL_INDEX: Map<string, ExcelMatch> | null = null;
+let MV_CROSS_SECTION_INDEX: MediumVoltageCrossSectionIndex | null = null;
+
+export function normalizeExcelKey(value: string): string {
+  return String(value || '')
+    .toUpperCase()
+    .replace(/-\d+$/g, '')
+    .replace(/[^A-Z0-9]+/g, '');
+}
+
+function loadExcelRows(filePath: string): ExcelRow[] {
+  const out = execSync(`npx -y xlsx-cli -j "${filePath}"`, {
+    encoding: 'utf8',
+    stdio: ['ignore', 'pipe', 'ignore'],
+  });
+  const trimmed = out.trim();
+  const jsonStart = trimmed.indexOf('[');
+  if (jsonStart < 0) return [];
+  const jsonText = trimmed.slice(jsonStart);
+  try {
+    return JSON.parse(jsonText) as ExcelRow[];
+  } catch {
+    return [];
+  }
+}
+
+function findKeyByHeaderValue(headerRow: ExcelRow, pattern: RegExp): string | null {
+  for (const [k, v] of Object.entries(headerRow || {})) {
+    const text = normalizeValue(String(v ?? ''));
+    if (!text) continue;
+    if (pattern.test(text)) return k;
+  }
+  return null;
+}
+
+function getMediumVoltageCrossSectionIndex(): MediumVoltageCrossSectionIndex {
+  if (MV_CROSS_SECTION_INDEX) return MV_CROSS_SECTION_INDEX;
+
+  const rows = fs.existsSync(MV_CROSS_SECTION_FILE) ? loadExcelRows(MV_CROSS_SECTION_FILE) : [];
+  const headerRow = (rows[0] || {}) as ExcelRow;
+
+  const partNumberKey = findKeyByHeaderValue(headerRow, /^part\s*number$/i) || '__EMPTY';
+  const crossSectionKey = findKeyByHeaderValue(headerRow, /querschnitt|cross.?section/i) || '';
+  const ratedVoltageKey =
+    findKeyByHeaderValue(headerRow, /rated voltage|voltage rating|nennspannung/i) || null;
+
+  const unitsRow =
+    rows.find((r) => normalizeValue(String((r as ExcelRow)?.[partNumberKey] ?? '')) === 'Units') ||
+    null;
+  const units: Record<string, string> = {};
+  if (unitsRow) {
+    for (const [k, v] of Object.entries(unitsRow)) {
+      if (k === partNumberKey) continue;
+      const unit = normalizeValue(String(v ?? ''));
+      if (unit) units[k] = unit;
+    }
+  }
+
+  const rowsByDesignation = new Map<string, ExcelRow[]>();
+  for (const r of rows) {
+    if (r === headerRow) continue;
+    const pn = normalizeValue(String((r as ExcelRow)?.[partNumberKey] ?? ''));
+    if (!pn || pn === 'Units' || pn === 'Part Number') continue;
+
+    const key = normalizeExcelKey(pn);
+    if (!key) continue;
+
+    const cur = rowsByDesignation.get(key) || [];
+    cur.push(r);
+    rowsByDesignation.set(key, cur);
+  }
+
+  MV_CROSS_SECTION_INDEX = {
+    headerRow,
+    units,
+    partNumberKey,
+    crossSectionKey,
+    ratedVoltageKey,
+    rowsByDesignation,
+  };
+  return MV_CROSS_SECTION_INDEX;
+}
+
+export function getExcelIndex(): Map<string, ExcelMatch> {
+  if (EXCEL_INDEX) return EXCEL_INDEX;
+  const idx = new Map<string, ExcelMatch>();
+
+  for (const file of EXCEL_SOURCE_FILES) {
+    if (!fs.existsSync(file)) continue;
+    const rows = loadExcelRows(file);
+
+    const unitsRow = rows.find((r) => r && r['Part Number'] === 'Units') || null;
+    const units: Record<string, string> = {};
+    if (unitsRow) {
+      for (const [k, v] of Object.entries(unitsRow)) {
+        if (k === 'Part Number') continue;
+        const unit = normalizeValue(String(v ?? ''));
+        if (unit) units[k] = unit;
+      }
+    }
+
+    for (const r of rows) {
+      const pn = r?.['Part Number'];
+      if (!pn || pn === 'Units') continue;
+      const key = normalizeExcelKey(String(pn));
+      if (!key) continue;
+      const cur = idx.get(key);
+      if (!cur) {
+        idx.set(key, { rows: [r], units });
+      } else {
+        cur.rows.push(r);
+        if (Object.keys(cur.units).length < Object.keys(units).length) cur.units = units;
+      }
+    }
+  }
+
+  EXCEL_INDEX = idx;
+  return idx;
+}
+
+export function findExcelForProduct(product: ProductData): ExcelMatch | null {
+  const idx = getExcelIndex();
+  const candidates = [
+    product.name,
+    product.slug ? product.slug.replace(/-\d+$/g, '') : '',
+    product.sku,
+    product.translationKey,
+  ].filter(Boolean) as string[];
+
+  for (const c of candidates) {
+    const key = normalizeExcelKey(c);
+    const match = idx.get(key);
+    if (match && match.rows.length) return match;
+  }
+  return null;
+}
+
+export function findMediumVoltageCrossSectionExcelForProduct(
+  product: ProductData,
+): MediumVoltageCrossSectionExcelMatch | null {
+  const idx = getMediumVoltageCrossSectionIndex();
+  const candidates = [
+    product.name,
+    product.slug ? product.slug.replace(/-\d+$/g, '') : '',
+    product.sku,
+    product.translationKey,
+  ].filter(Boolean) as string[];
+
+  for (const c of candidates) {
+    const key = normalizeExcelKey(c);
+    const rows = idx.rowsByDesignation.get(key) || [];
+    if (rows.length) {
+      return {
+        headerRow: idx.headerRow,
+        rows,
+        units: idx.units,
+        partNumberKey: idx.partNumberKey,
+        crossSectionKey: idx.crossSectionKey,
+        ratedVoltageKey: idx.ratedVoltageKey,
+      };
+    }
+  }
+
+  return null;
+}

scripts/pdf/model/types.ts

@@ -0,0 +1,51 @@
+export interface ProductData {
+  id: number;
+  name: string;
+  shortDescriptionHtml: string;
+  descriptionHtml: string;
+  images: string[];
+  featuredImage: string | null;
+  sku: string;
+  slug?: string;
+  path?: string;
+  translationKey?: string;
+  locale?: 'en' | 'de';
+  categories: Array<{ name: string }>;
+  attributes: Array<{
+    name: string;
+    options: string[];
+  }>;
+}
+
+export type KeyValueItem = { label: string; value: string; unit?: string };
+
+export type DatasheetVoltageTable = {
+  voltageLabel: string;
+  metaItems: KeyValueItem[];
+  columns: Array<{ key: string; label: string }>;
+  rows: Array<{ configuration: string; cells: string[] }>;
+};
+
+export type DatasheetModel = {
+  locale: 'en' | 'de';
+  product: {
+    id: number;
+    name: string;
+    sku: string;
+    categoriesLine: string;
+    descriptionText: string;
+    heroSrc: string | null;
+    productUrl: string;
+  };
+  labels: {
+    datasheet: string;
+    description: string;
+    technicalData: string;
+    crossSection: string;
+    sku: string;
+    noImage: string;
+  };
+  technicalItems: KeyValueItem[];
+  voltageTables: DatasheetVoltageTable[];
+  legendItems: KeyValueItem[];
+};

scripts/pdf/model/utils.ts

@@ -0,0 +1,74 @@
+import * as path from 'path';
+
+import type { ProductData } from './types';
+
+export const CONFIG = {
+  siteUrl: 'https://klz-cables.com',
+  publicDir: path.join(process.cwd(), 'public'),
+  assetMapFile: path.join(process.cwd(), 'data/processed/asset-map.json'),
+} as const;
+
+export function stripHtml(html: string): string {
+  if (!html) return '';
+  let text = String(html)
+    .replace(/<[^>]*>/g, '')
+    .normalize('NFC');
+  text = text
+    .replace(/[\u00A0\u202F]/g, ' ')
+    .replace(/[\u2013\u2014]/g, '-')
+    .replace(/[\u2018\u2019]/g, "'")
+    .replace(/[\u201C\u201D]/g, '"')
+    .replace(/\u2026/g, '...')
+    .replace(/[\u2022]/g, '·')
+    .replace(/[\u2264]/g, '<=')
+    .replace(/[\u2265]/g, '>=')
+    .replace(/[\u2248]/g, '~')
+    .replace(/[\u03A9\u2126]/g, 'Ohm')
+    .replace(/[\u00B5\u03BC]/g, 'u')
+    .replace(/[\u2193]/g, 'v')
+    .replace(/[\u2191]/g, '^')
+    .replace(/[\u00B0]/g, '°');
+  // eslint-disable-next-line no-control-regex
+  text = text.replace(/[\u0000-\u001F\u007F]/g, '');
+  return text.replace(/\s+/g, ' ').trim();
+}
+
+export function normalizeValue(value: string): string {
+  return stripHtml(value).replace(/\s+/g, ' ').trim();
+}
+
+export function getProductUrl(product: ProductData): string {
+  if (product.path) return `${CONFIG.siteUrl}${product.path}`;
+  return CONFIG.siteUrl;
+}
+
+export function generateFileName(product: ProductData, locale: 'en' | 'de'): string {
+  const baseName = product.slug || product.translationKey || `product-${product.id}`;
+  const cleanSlug = baseName
+    .toLowerCase()
+    .replace(/[^a-z0-9-]/g, '-')
+    .replace(/-+/g, '-')
+    .replace(/^-|-$/g, '');
+  return `${cleanSlug}-${locale}.pdf`;
+}
+
+export function getLabels(locale: 'en' | 'de') {
+  return {
+    en: {
+      datasheet: 'PRODUCT DATASHEET',
+      description: 'DESCRIPTION',
+      technicalData: 'TECHNICAL DATA',
+      crossSection: 'CROSS-SECTION DATA',
+      sku: 'SKU',
+      noImage: 'No image available',
+    },
+    de: {
+      datasheet: 'PRODUKTDATENBLATT',
+      description: 'BESCHREIBUNG',
+      technicalData: 'TECHNISCHE DATEN',
+      crossSection: 'QUERSCHNITTSDATEN',
+      sku: 'ARTIKELNUMMER',
+      noImage: 'Kein Bild verfügbar',
+    },
+  }[locale];
+}