original code

app.py

@@ -10,11 +10,25 @@ from multilingual_pdf2text.models.document_model.document import Document
 
 import pdfplumber
 import docx2txt
+import shutil
+import subprocess
+
+print("🔍 Checking if Tesseract OCR is installed...")
+
+tess_path = shutil.which("tesseract")
+if tess_path:
+    print(f"✅ Tesseract found at: {tess_path}")
+    try:
+        version = subprocess.check_output(["tesseract", "--version"]).decode()
+        print(f"🧠 Tesseract version:\n{version}")
+    except Exception as e:
+        print(f"⚠️ Tesseract found but failed to run: {e}")
+else:
+    print("❌ Tesseract not found. Arabic PDF extraction will likely fail.")
+
 
 # --- Config ---
 FLOWISE_URL = "https://siraflowappl.happywave-0e1819cd.uaenorth.azurecontainerapps.io/api/v1/prediction/5f2aa074-ae2e-4b06-a4a2-6f8dccbe59bb"
-
-#
 COUNTRIES = ["USA", "Germany", "UAE", "Turkey", "Jordan"]
 LANGUAGES = ["English", "Arabic"]
 
@@ -78,7 +92,7 @@ def send_to_flowise(file, country: str, language: str) -> str:
             return "⚠️ No content extracted from the uploaded file."
 
         prompt = f"""Please audit the following employment contract in light of {country} labor law.
-Only refer to the provided law document (do not use external knowledge). Do not mention the law NO. in your response.
+Only refer to the provided law document (do not use external knowledge), but you don't have to mention the exact source of info while answering.
 Is the contract compliant? Provide reasoning and highlight any issues, and suggest improvements.
 Be precise and don't miss any details in the contract.
 Please respond in {language}.