From 6dd34fe70354891a57b717f422d10b233ce27b3f Mon Sep 17 00:00:00 2001
From: mstfyldz <mstfyldz@mstfyldz-MacBook-Air.local>
Date: Fri, 14 Aug 2026 13:13:37 +0300
Subject: [PATCH] feat: enhance UDF file extraction pipeline to unzip container
 and handle content.xml or embedded PDF files

---
 src/controllers/document.controller.ts  | 49 ++++++++++++++++++++-----
 src/controllers/precedent.controller.ts |  8 ++--
 2 files changed, 44 insertions(+), 13 deletions(-)

diff --git a/src/controllers/document.controller.ts b/src/controllers/document.controller.ts
index bb65744..ddb2b74 100644
--- a/src/controllers/document.controller.ts
+++ b/src/controllers/document.controller.ts
@@ -407,17 +407,48 @@ export async function extractTextFromUploadedFile(file: { buffer: Buffer; mimety
     try {
       const zip = new AdmZip(file.buffer);
       const zipEntries = zip.getEntries();
-      const contentEntry = zipEntries.find((entry: any) => entry.entryName === "content.xml" || entry.entryName.endsWith(".xml"));
-      if (!contentEntry) {
-        throw new Error("No XML content found inside UDF/ZIP");
+      
+      // 1. Önce XML dosyasını kontrol et (content.xml veya *.xml)
+      const xmlEntry = zipEntries.find((entry: any) => 
+        !entry.isDirectory && (entry.entryName === "content.xml" || entry.entryName.toLowerCase().endsWith(".xml"))
+      );
+
+      // 2. İçerisinde gömülü PDF dosyası var mı kontrol et (*.pdf)
+      const pdfEntry = zipEntries.find((entry: any) => 
+        !entry.isDirectory && entry.entryName.toLowerCase().endsWith(".pdf")
+      );
+
+      // Eğer XML bulunmuşsa XML üzerinden devam et
+      if (xmlEntry) {
+        const xmlData = zip.readAsText(xmlEntry);
+        if (xmlData && xmlData.trim()) {
+          const plainText = xmlData.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
+          if (plainText.length > 20) {
+            return plainText;
+          }
+        }
       }
-      const xmlData = zip.readAsText(contentEntry);
-      // Remove XML tags and replace with space, then replace multiple spaces with single space
-      const plainText = xmlData.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
-      return plainText;
-    } catch (err) {
+
+      // Eğer UDF içerisinden PDF çıktıysa PDF ayrıştırıcı (officeParser) üzerinden devam et
+      if (pdfEntry) {
+        const pdfBuffer = pdfEntry.getData();
+        const pdfResult: any = await officeParser.parseOffice(pdfBuffer);
+        const pdfText = typeof pdfResult === 'string' ? pdfResult : (pdfResult?.toText ? pdfResult.toText() : String(pdfResult || ''));
+        if (pdfText && pdfText.trim()) {
+          return pdfText.trim();
+        }
+      }
+
+      // Yedek durum: Son çare XML metnini döndür
+      if (xmlEntry) {
+        const xmlData = zip.readAsText(xmlEntry);
+        return xmlData.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
+      }
+
+      throw new Error("UDF/ZIP arşivinde okunabilir XML veya PDF içeriği bulunamadı.");
+    } catch (err: any) {
       console.error("UDF parse error:", err);
-      throw new Error("UDF dosyası okunamadı veya bozuk.");
+      throw new Error(`UDF dosyası ayrıştırılamadı: ${err.message || 'Bozuk arşiv'}`);
     }
   }
 
diff --git a/src/controllers/precedent.controller.ts b/src/controllers/precedent.controller.ts
index 827dc66..196cf05 100644
--- a/src/controllers/precedent.controller.ts
+++ b/src/controllers/precedent.controller.ts
@@ -60,15 +60,15 @@ SADECE anahtar kelimeleri aralarına boşluk koyarak yaz. Noktalama işareti, a
       return res.status(mcpResponse.status).json({ error: 'Arama servisi hata döndürdü', details: errorData });
     }
 
-    const data = await mcpResponse.json();
+    let data = await mcpResponse.json();
 
     // Hukuki niyet ve sonuç (örn. beraat, bozma) odaklı akıllı yeniden sıralama ve puanlama
     let rawItems: any[] = data.decisions || data.results || data.data || data.hits || (Array.isArray(data) ? data : []);
 
     if (Array.isArray(rawItems) && rawItems.length > 0) {
-      const searchTerms = phrase.toLowerCase().split(/\s+/).filter(Boolean);
-      const isBeraatSearch = searchTerms.some(t => t.includes('beraat'));
-      const isBozmaSearch = searchTerms.some(t => t.includes('bozma') || t.includes('bozulması'));
+      const searchTerms = phrase.toLowerCase().split(/\s+/).filter((Boolean) as (value: string) => boolean);
+      const isBeraatSearch = searchTerms.some((t: string) => t.includes('beraat'));
+      const isBozmaSearch = searchTerms.some((t: string) => t.includes('bozma') || t.includes('bozulması'));
 
       const scoredItems = rawItems.map((item: any) => {
         const fullText = (