Nomad changes

e2e/node/tests/format_specific.test.ts

@@ -0,0 +1,55 @@
+// This file is auto-generated by alef — DO NOT EDIT.
+// alef:hash:4e15143f4af1ae8bafbdb1506ef057da924484c66a19483966333558ad437e75
+// To regenerate: alef generate
+// To verify freshness: alef verify --exit-code
+// Issues & docs: https://github.com/kreuzberg-dev/alef
+
+import { describe, expect, it } from 'vitest';import { extractFile, extractBytesSync, extractFileSync } from '@kreuzberg/node';
+function _alefE2eText(value: unknown): string {
+  return value == null ? "" : String(value);
+}
+
+function _alefE2eItemTexts(item: unknown): string[] {
+  if (item == null || typeof item !== "object") {
+    return [_alefE2eText(item)];
+  }
+  const record = item as Record<string, unknown>;
+  const itemsText = Array.isArray(record.items) ? record.items.map(_alefE2eText).join(" ") : "";
+  return [_alefE2eText(item), _alefE2eText(record.kind), _alefE2eText(record.name), _alefE2eText(record.source), _alefE2eText(record.alias), _alefE2eText(record.text), _alefE2eText(record.signature), itemsText];
+}
+
+function _alefE2eFormatMetadataDisplay(fm: unknown): string {
+  if (fm == null) return "";
+  if (typeof fm !== "object") return String(fm);
+  const record = fm as Record<string, unknown>;
+  const formatType = record.format_type;
+
+  // FormatMetadata is a tagged union: { format_type: 'image', image: { format: 'PNG', ... }, ... }
+  // Extract the display string based on the variant type
+  if (formatType === "image" && typeof record.image === "object") {
+    const imageData = record.image as Record<string, unknown>;
+    if (typeof imageData.format === "string") return imageData.format;
+  }
+
+  // Fallback: return format_type variant name
+  if (typeof record.format_type === "string") return record.format_type;
+  return "";
+}
+
+
+describe('format_specific', () => {  it('format_docx_standalone: Standalone DOCX extraction using extract_bytes_sync', async () => {    const _content_content = await (await import('node:fs/promises')).readFile('docx/fake.docx');    const result = extractBytesSync(_content_content, "application/vnd.openxmlformats-officedocument.wordprocessingml.document", undefined);    expect(result.content.length).toBeGreaterThanOrEqual(20);
+
+  }, 30000);
+  it('format_hwpx_standalone: Standalone HWPX extraction using extract_bytes_sync', async () => {    const _content_content = await (await import('node:fs/promises')).readFile('hwpx/simple.hwpx');    const result = extractBytesSync(_content_content, "application/haansofthwpx", undefined);    expect(result.content.length).toBeGreaterThanOrEqual(20);
+    expect(result.content).toContain("Hello from HWPX");
+
+  }, 30000);
+  it('format_pdf_text: Standalone PDF text extraction using extract_bytes_sync', async () => {    const _content_content = await (await import('node:fs/promises')).readFile('pdf/fake_memo.pdf');    const result = extractBytesSync(_content_content, "application/pdf", undefined);    expect(result.content.length).toBeGreaterThanOrEqual(50);
+    expect(["Mallori", "May"].some((v) => result.content.includes(v))).toBe(true);
+
+  }, 30000);
+  it('format_pptx: PPTX presentation extraction using extract_file_sync', () => {    extractFileSync("pptx/simple.pptx", "application/vnd.openxmlformats-officedocument.presentationml.presentation", undefined);
+  }, 30000);
+  it('format_xlsx: XLSX spreadsheet extraction using extract_file_sync', () => {    extractFileSync("xlsx/stanley_cups.xlsx", "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", undefined);
+  }, 30000);
+});