input_text with PDF

app.py

@@ -4,6 +4,7 @@ import torch
 from transformers import AutoProcessor, MllamaForConditionalGeneration
 from PIL import Image
 import spaces
+import tempfile
 
 # Check if we're running in a Hugging Face Space and if SPACES_ZERO_GPU is enabled
 IS_SPACES_ZERO = os.environ.get("SPACES_ZERO_GPU", "0") == "1"
@@ -54,17 +55,38 @@ processor = AutoProcessor.from_pretrained(model_name, use_auth_token=HF_TOKEN)
 #     # Decode the output to return the final response
 #     response = processor.decode(outputs[0], skip_special_tokens=True)
 #     return response
+
+def extract_text_from_pdf(pdf_url):
+    try:
+        response = requests.get(pdf_url)
+        response.raise_for_status()
+        with tempfile.NamedTemporaryFile(delete=False) as temp_pdf:
+            temp_pdf.write(response.content)
+            temp_pdf_path = temp_pdf.name
+        
+        reader = PdfReader(temp_pdf_path)
+        text = ""
+        for page in reader.pages:
+            text += page.extract_text()
+        
+        os.remove(temp_pdf_path)
+        return text
+    except Exception as e:
+        raise HTTPException(status_code=400, detail=f"Error extracting text from PDF: {str(e)}")
+
 @spaces.GPU
-def predict_text(text):
+def predict_text(text, url = 'https://arinsight.co/2024_FA_AEC_1200_GR1_GR2.pdf'):
+    pdf_text = extract_text_from_pdf(url)
+     text_combined = text + "\n\nExtracted Text from PDF:\n" + pdf_text
     # Prepare the input messages
-    messages = [{"role": "user", "content": [{"type": "text", "text": text}]}]
+    messages = [{"role": "user", "content": [{"type": "text", "text": text_combined}]}]
     
     # Create the input text using the processor's chat template
     input_text = processor.apply_chat_template(messages, add_generation_prompt=True)
     
     # Process the inputs and move to the appropriate device
     # inputs = processor(image, input_text, return_tensors="pt").to(device)
-    inputs = processor(text=text, return_tensors="pt").to("cuda")
+    inputs = processor(text=input_text, return_tensors="pt").to("cuda")
     # Generate a response from the model
     outputs = model.generate(**inputs, max_new_tokens=1024)