feat(lieferung/neu): PDF upload with SSE streaming and result dialog

- Add PDF-Upload button to ManifestHeader (delivery forms only)
- New ManifestPdfDialog: live SSE event log, structured basedata,
  and line-items table streamed from POST /api/delivery/parsePDF
- ApiClient.delivery.parsePdfSse: multipart/form-data upload,
  async generator yielding typed SSE events (status, llm_call,
  tool_call, usage, progress, result, error)
- ManifestForm.handlePdfUpload: drives SSE generator, updates
  dialog state in real-time; dialog opens immediately on upload
- Fix MUI icon imports (CheckCircle, ErrorOutlined)

src/utils/importArtikelCsv.js

@@ -0,0 +1,184 @@
+/**
+ * Client-side bulk import of the shop `artikeltabelle.csv` export format.
+ *
+ * The CSV is semicolon-separated (often wrapped in `="…"` from Excel export)
+ * with columns:
+ *   ArtNr; Name; Preis; USt; Pfand; Hersteller; Kategorie; Barcode; V/K
+ *
+ * All work is done through the existing dashback API:
+ *   - categories  → POST /api/category
+ *   - brands      → PUT /api/brand
+ *   - articles    → POST /api/article (+ price / brand / ean / categories PUTs)
+ */
+
+/** Parse one `="value"` / `"value"` / plain Excel-style CSV cell. */
+export function stripCell(value) {
+  if (value == null) return '';
+  let text = String(value).trim();
+  if (text.startsWith('=')) text = text.slice(1);
+  text = text.trim();
+  if (
+    text.length >= 2 &&
+    text.startsWith('"') &&
+    text.endsWith('"')
+  ) {
+    text = text.slice(1, -1).replace(/""/g, '"');
+  }
+  return text.trim();
+}
+
+/**
+ * Split a semicolon CSV line into cells, honoring double-quoted fields.
+ * (Cell values are still raw; use `stripCell` for the Excel `="…"` wrapper.)
+ */
+export function splitCsvLine(line) {
+  const cells = [];
+  let cur = '';
+  let inQuotes = false;
+  for (let i = 0; i < line.length; i++) {
+    const ch = line[i];
+    if (inQuotes) {
+      if (ch === '"') {
+        if (line[i + 1] === '"') {
+          cur += '"';
+          i++;
+        } else {
+          inQuotes = false;
+        }
+      } else {
+        cur += ch;
+      }
+    } else if (ch === '"') {
+      inQuotes = true;
+    } else if (ch === ';') {
+      cells.push(cur);
+      cur = '';
+    } else {
+      cur += ch;
+    }
+  }
+  cells.push(cur);
+  return cells.map((c) => stripCell(c));
+}
+
+/**
+ * Parse a CSV string into row objects keyed by the header names.
+ * Skips blank lines and the optional `sep=;` hint line.
+ */
+export function parseArtikelCsv(csvText) {
+  const lines = String(csvText ?? '')
+    .replace(/\r\n/g, '\n')
+    .split('\n')
+    .map((l) => l.trim())
+    .filter((l) => l !== '' && l !== 'sep=;');
+
+  if (lines.length === 0) return [];
+
+  const headers = splitCsvLine(lines[0]);
+  const rows = [];
+  for (let i = 1; i < lines.length; i++) {
+    const cells = splitCsvLine(lines[i]);
+    const row = {};
+    headers.forEach((header, idx) => {
+      row[header] = cells[idx] ?? '';
+    });
+    if (row.Name) rows.push(row);
+  }
+  return rows;
+}
+
+/**
+ * Resolve a de-DE (or plain) number from a CSV field.
+ * Accepts `2,00`, `1.234,56`, and plain `2.00`.
+ */
+export function parseCsvNumber(value) {
+  const text = String(value ?? '').trim();
+  if (text === '') return null;
+  let normalized = text.replace(/\s/g, '');
+  if (normalized.includes(',')) {
+    normalized = normalized.replace(/\./g, '').replace(',', '.');
+  }
+  const parsed = Number(normalized);
+  return Number.isFinite(parsed) && parsed >= 0 ? parsed : null;
+}
+
+/** EAN from a CSV cell (digits only), or null. */
+export function parseCsvEan(value) {
+  const digits = String(value ?? '').replace(/\D/g, '');
+  return digits || null;
+}
+
+/**
+ * Split a `Kategorie` path (`Haupt | Grow | Erde`) into trimmed segments.
+ * Empty / falsy path → [].
+ */
+export function parseCategoryPath(value) {
+  return String(value ?? '')
+    .split('|')
+    .map((s) => s.trim())
+    .filter(Boolean);
+}
+
+/**
+ * Column index → name. Uses a forgiving match so the exact header text
+ * (with or without `="` wrapper, case) doesn't matter.
+ */
+function headerName(name) {
+  return stripCell(name).toLowerCase().trim();
+}
+
+export const COLUMNS = {
+  artNr: 'artnr',
+  name: 'name',
+  price: 'preis',
+  tax: 'ust',
+  deposit: 'pfand',
+  brand: 'hersteller',
+  category: 'kategorie',
+  ean: 'barcode',
+  vk: 'v/k',
+};
+
+/**
+ * Normalize one raw CSV row object into the typed import record.
+ * Unknown/missing columns become null/[] so downstream code can skip them.
+ */
+export function normalizeRow(row, headers = []) {
+  const byName = new Map();
+  (headers.length ? headers : Object.keys(row)).forEach((h) => {
+    byName.set(headerName(h), row[h] ?? row[stripCell(h)] ?? '');
+  });
+  const get = (key) => byName.get(COLUMNS[key]) ?? byName.get(key) ?? '';
+
+  const artNr = get('artNr');
+  const name = get('name');
+  const price = parseCsvNumber(get('price'));
+  const tax = parseCsvNumber(get('tax'));
+  const brand = get('brand');
+  const categoryPath = parseCategoryPath(get('category'));
+  const ean = parseCsvEan(get('ean'));
+  const vk = get('vk');
+
+  return {
+    artNr,
+    name,
+    price,
+    tax,
+    deposit: parseCsvNumber(get('deposit')),
+    brand,
+    categoryPath,
+    ean,
+    vk,
+  };
+}
+
+/**
+ * Parse a full CSV string into typed import records.
+ * @param {string} csvText
+ * @returns {Array<ReturnType<typeof normalizeRow>>}
+ */
+export function parseImportCsv(csvText) {
+  const rawRows = parseArtikelCsv(csvText);
+  const headers = rawRows.length ? Object.keys(rawRows[0]) : [];
+  return rawRows.map((row) => normalizeRow(row, headers));
+}