import { describe, it, expect, beforeEach } from "bun:test";
import { PDFProcessor } from "../src/pdf-processor";
const JAPANESE_PDF_URL = "https://www.stat.go.jp/data/nenkan/74nenkan/zenbun/jp74/data/676/src/jp74.pdf";
describe("Japanese PDF Handling", () => {
let processor: PDFProcessor;
let pdfId: string;
beforeEach(() => {
processor = new PDFProcessor();
});
it("should properly extract pages from Japanese statistical yearbook", async () => {
// Load the PDF
const result = await processor.loadPDF(JAPANESE_PDF_URL);
pdfId = result.id;
expect(result.pageCount).toBe(831);
// Test single page extraction
const page1 = await processor.extractPage(pdfId, 1);
// Handle potential spacing in Japanese text
expect(page1.replace(/\s/g, '')).toContain("第七十四回");
expect(page1).toContain("JAPAN");
expect(page1).toContain("STATISTICAL YEARBOOK");
expect(page1.length).toBeLessThan(500); // Should be a single page, not the entire document
// Test range extraction
const range = await processor.extractRange(pdfId, 1, 3);
expect(range).toContain("--- Page 1 ---");
expect(range).toContain("--- Page 2 ---");
expect(range).toContain("--- Page 3 ---");
// Handle spacing in Japanese text
expect(range.replace(/\s/g, '')).toContain("まえがき"); // Preface on page 3
// Verify each page is properly separated
const pages = range.split(/--- Page \d+ ---/);
expect(pages.length).toBe(4); // Empty first element + 3 pages
// Check that pages have reasonable content
const page1Content = pages[1].trim();
const page2Content = pages[2].trim();
const page3Content = pages[3].trim();
expect(page1Content.length).toBeGreaterThan(0);
expect(page1Content.length).toBeLessThan(1000);
// Page 2 might be empty in some PDFs
expect(page2Content.length).toBeGreaterThanOrEqual(0);
expect(page2Content.length).toBeLessThan(1000);
expect(page3Content.length).toBeGreaterThan(0);
expect(page3Content.length).toBeLessThan(2000);
}, 60000); // 60 second timeout for large PDF
it("should handle page extraction for middle pages", async () => {
// Always load fresh for this test
const result = await processor.loadPDF(JAPANESE_PDF_URL);
const localPdfId = result.id;
// Test extraction of pages in the middle
const range = await processor.extractRange(localPdfId, 100, 102);
const pages = range.split(/--- Page \d+ ---/).filter(p => p.trim());
expect(pages.length).toBe(3);
// Each page should have content
pages.forEach(page => {
expect(page.trim().length).toBeGreaterThan(0);
});
}, 60000);
});
