Read complete PDFs and surface per-page OCR failures

src/services/pdfDocumentReader.js

@@ -0,0 +1,23 @@
+/** Read every requested page in order, retaining text/OCR provenance per page. */
+export async function readPdfPages(doc, {maxPages=Infinity, ocrPage, forceOcr=false, onProgress}={}) {
+  if (!(maxPages === Infinity || (Number.isInteger(maxPages) && maxPages > 0))) throw new Error('Geçersiz sayfa sınırı.');
+  const pages = Math.min(doc.numPages,maxPages);
+  const results=[];
+  for(let number=1;number<=pages;number++) {
+    onProgress?.({sayfa:number,toplam:pages});
+    const page=await doc.getPage(number);
+    try {
+      const content=await page.getTextContent();
+      let text=content.items.map(item=>item.str + (item.hasEOL ? '\n' : ' ')).join('').trim();
+      let method='text';
+      if(ocrPage && (forceOcr || text.replace(/\s/g,'').length < 100)) {
+        try { text=await ocrPage(page); method='ocr'; }
+        catch { throw new Error(`${number}. sayfada OCR tamamlanamadı. Okunabilir belgeyle yeniden deneyin.`); }
+      }
+      results.push({page:number,text,method});
+    } finally { page.cleanup?.(); }
+  }
+  return {text:results.map(p=>p.text).join('\n\n').replace(/[ \t]+/g,' ').trim(),
+    pageCount:doc.numPages,usedPages:pages,pages:results,ocr:results.some(p=>p.method==='ocr'),
+    complete:pages===doc.numPages};
+}