doc upload option added

app.py

@@ -4,13 +4,15 @@ from typing import Dict
 import os
 import shutil
 import logging
-
+ 
 import torch
 from transformers import LayoutLMv3Processor, LayoutLMv3ForTokenClassification
 
 from dotenv import load_dotenv
 import os
 
+from utils import doc_processing
+
 # Load .env file
 load_dotenv()
 
@@ -33,6 +35,8 @@ aadhar_model = LayoutLMv3ForTokenClassification.from_pretrained(
     aadhar_model,
     use_auth_token=HUGGINGFACE_AUTH_TOKEN
 )
+
+
 aadhar_model = aadhar_model.to(device)
 
 # pan model
@@ -40,6 +44,8 @@ pan_model = "AuditEdge/doc_ocr_p"  # Replace with your fine-tuned model if appli
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 print(f"Using device: {device}")
 
+
+
 # Load the processor (tokenizer + image processor)
 processor_pan = LayoutLMv3Processor.from_pretrained(
     pan_model,
@@ -113,7 +119,9 @@ app.add_middleware(
 
 # Configure directories
 UPLOAD_FOLDER = './uploads/'
+processing_folder = "./processed_images"
 os.makedirs(UPLOAD_FOLDER, exist_ok=True)  # Ensure the main upload folder exists
+os.makedirs(processing_folder,exist_ok=True)
 
 UPLOAD_DIRS = {
     "aadhar_file": "uploads/aadhar/",
@@ -122,9 +130,22 @@ UPLOAD_DIRS = {
     "gst_file": "uploads/gst/",
 }
 
+process_dirs = {
+    "aadhar_file": "processed_images/aadhar/",
+    "pan_file": "processed_images/pan/",
+    "cheque_file": "processed_images/cheque/",
+    "gst_file": "processed_images/gst/",
+
+}
+
 # Ensure individual directories exist
 for dir_path in UPLOAD_DIRS.values():
     os.makedirs(dir_path, exist_ok=True)
+    
+for dir_path in process_dirs.values():
+    os.makedirs(dir_path, exist_ok=True)
+    
+    
 
 # Logger configuration
 logging.basicConfig(level=logging.INFO)
@@ -217,14 +238,27 @@ async def aadhar_ocr(
         # Log received files
         logging.info(f"Received files: {list(file_paths.keys())}")
         print("file_paths",file_paths)
-        import sys
-        # sys.exit()
-
+        
+        files = {}
+        for key, value in file_paths.items():
+            name = value.split("/")[-1].split(".")[0]
+            id_type = key.split("_")[0]
+            doc_type = value.split("/")[-1].split(".")[1]
+            f_path = value
+            preprocessing = doc_processing(name,id_type,doc_type,f_path)
+            response = preprocessing.process()
+            files[key] = response["output_p"]
+            print("response",response)
+
+        
         # Perform inference
-        result = perform_inference(file_paths)
+        result = perform_inference(files)
 
         return {"status": "success", "result": result}
 
     except Exception as e:
         logging.error(f"Error processing files: {e}")
-        raise HTTPException(status_code=500, detail="Internal Server Error")
+        # raise HTTPException(status_code=500, detail="Internal Server Error")
+        return {"status":400}
+    
+    

utils.py

@@ -0,0 +1,110 @@
+import fitz
+from PIL import Image
+
+class doc_processing:
+
+    def __init__(self, name, id_type, doc_type, f_path):
+ 
+        self.name = name
+        self.id_type = id_type
+        self.doc_type = doc_type
+        self.f_path = f_path
+        # self.o_path = o_path
+        
+        
+    def pdf_to_image_scale(self):
+        pdf_document = fitz.open(self.f_path)
+        if self.id_type == "gst":
+            page_num = 2
+        else:
+            page_num = 0
+        
+        page = pdf_document.load_page(page_num)
+        pix = page.get_pixmap()  # Render page as a pixmap (image)
+        
+        # Convert pixmap to PIL Image
+        image = Image.frombytes("RGB", [pix.width, pix.height], pix.samples)
+        
+        original_width, original_height = image.size
+        
+        print("original_width",original_width)
+        print("original_height",original_height)
+
+
+        new_width = (1000 / original_width) * original_width
+        new_height = (1000 / original_height) * original_height
+        
+        print("new_width",new_width)
+        print("new_height",new_height)
+        # new_width = 
+        # new_height =  
+        image.resize((int(new_width), int(new_height)), Image.Resampling.LANCZOS)
+        output_path = "processed_images/{}/{}.jpeg".format(self.id_type,self.name)
+        image.save(output_path)
+        return  {"success":200,"output_p":output_path}
+    
+
+    def scale_img(self):
+        
+        image = Image.open(self.f_path).convert("RGB")
+        original_width, original_height = image.size
+        
+        print("original_width",original_width)
+        print("original_height",original_height)
+
+
+        new_width = (1000 / original_width) * original_width
+        new_height = (1000 / original_height) * original_height
+        
+        print("new_width",new_width)
+        print("new_height",new_height)
+        # new_width = 
+        # new_height =  
+        image.resize((int(new_width), int(new_height)), Image.Resampling.LANCZOS)
+        output_path = "processed_images/{}/{}.jpeg".format(self.id_type,self.name)
+        image.save(output_path)
+        return {"success":200,"output_p":output_path}
+
+    def process(self):
+        if self.doc_type == "pdf":
+            response = self.pdf_to_image_scale()
+        else:
+            response = self.scale_img()
+
+        return response
+
+
+
+
+        
+# files = {
+#     "aadhar_file": "/home/javmulla/model_one/test_images_aadhar/test_two.jpg",
+#     "pan_file": "/home/javmulla/model_one/test_images_pan/6ea33087.jpeg", 
+#     "cheque_file": "/home/javmulla/model_one/test_images_cheque/0f81678a.jpeg",
+#     "gst_file": "/home/javmulla/model_one/test_images_gst/0a52fbcb_page3_image_0.jpg"
+# }
+
+
+# files = {
+#     "aadhar_file": "/home/javmulla/model_one/test_images_aadhar/test_two.jpg",
+#     "pan_file": "/home/javmulla/model_one/test_images_pan/6ea33087.jpeg", 
+#     "cheque_file": "/home/javmulla/model_one/test_images_cheque/0f81678a.jpeg",
+#     "gst_file": "test_Images_folder/gst/e.pdf"
+# }
+
+# for key, value in files.items():
+#     name = value.split("/")[-1].split(".")[0]
+#     id_type = key.split("_")[0]
+#     doc_type = value.split("/")[-1].split(".")[1]
+#     f_path = value
+#     preprocessing = doc_processing(name,id_type,doc_type,f_path)
+#     response = preprocessing.process()
+#     print("response",response)
+    
+    
+    
+    
+    
+    # id_type, doc_type, f_path
+    
+