Deploy Django API with Docker Build Caching

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+local_ai_model/* filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -7,16 +7,20 @@ RUN useradd -m -u 1000 user
 WORKDIR /home/user/app
 
 # Copy requirements first (better cache)
-COPY requirements.txt .
+COPY --chown=user:user requirements.txt .
 
 # Install dependencies
 RUN pip install --no-cache-dir -r requirements.txt
 
-# Copy project files
-COPY . .
-
-# Switch to non-root user
+# --- THE FIX: CACHE THE MODEL DURING BUILD ---
+# This command downloads the model into the Docker image's cache directory.
+# This means your app will start instantly without downloading it on boot!
 USER user
+RUN python -c "from transformers import pipeline; pipeline('zero-shot-classification', model='valhalla/distilbart-mnli-12-3')"
+# ---------------------------------------------
+
+# Copy project files
+COPY --chown=user:user . .
 
 # Expose HF port
 EXPOSE 7860

api/admin.py

@@ -1,6 +1,6 @@
 from django.contrib import admin
 from .models import UserData, AIKey, AppVersion
-from .models import Contact
+from .models import Contact , ExpenseLog
 
 @admin.register(Contact)
 class ContactAdmin(admin.ModelAdmin):
@@ -41,3 +41,11 @@ class UserDataAdmin(admin.ModelAdmin):
             return "Not set"
 
     get_user_name.short_description = "User Name"
+
+
+@admin.register(ExpenseLog)
+class ExpenseLogAdmin(admin.ModelAdmin):
+    list_display = ("text", "predicted_category", "confidence_score", "created_at")
+    list_filter = ("predicted_category", "created_at")
+    search_fields = ("text", "predicted_category")
+    ordering = ("-created_at",)

api/apps.py

@@ -1,6 +1,49 @@
 from django.apps import AppConfig
+import torch
+from transformers import pipeline
+import os
 
 
 class ApiConfig(AppConfig):
-    default_auto_field = 'django.db.models.BigAutoField'
-    name = 'api'
+    default_auto_field = "django.db.models.BigAutoField"
+    name = "api"
+
+    def ready(self):
+        # Prevent loading twice (Django sometimes runs ready() twice in dev mode)
+        if os.environ.get("RUN_MAIN", None) != "true" and not os.environ.get(
+            "KUBERNETES_PORT"
+        ):
+            pass  # Keep going if running via Gunicorn or in production
+
+        print("Loading Expense Categorizer AI from cache...")
+
+        device = 0 if torch.cuda.is_available() else -1
+
+        # This will automatically use the cached version built by Docker
+        self.classifier = pipeline(
+            "zero-shot-classification",
+            model="valhalla/distilbart-mnli-12-3",
+            device=device,
+        )
+
+        self.categories = [
+            "Food & Drinks",
+            "Groceries",
+            "Shopping",
+            "Bills & Utilities",
+            "Entertainment",
+            "Health",
+            "Education",
+            "Subscriptions",
+            "Travel",
+            "Rent",
+            "Family & Friends",
+            "Miscellaneous",
+            "Gifts",
+            "Party",
+            "Personal Care",
+            "Home & Hygiene",
+            "Others",
+            "Recharge",
+        ]
+        print("AI Model loaded successfully!")

api/migrations/0002_expenselog.py

@@ -0,0 +1,23 @@
+# Generated by Django 5.2.4 on 2026-03-06 16:28
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('api', '0001_initial'),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='ExpenseLog',
+            fields=[
+                ('id', models.BigAutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('text', models.CharField(max_length=500)),
+                ('predicted_category', models.CharField(max_length=100)),
+                ('confidence_score', models.FloatField()),
+                ('created_at', models.DateTimeField(auto_now_add=True)),
+            ],
+        ),
+    ]

api/migrations/0003_expenselog_amount.py

@@ -0,0 +1,18 @@
+# Generated by Django 5.2.4 on 2026-03-06 16:43
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('api', '0002_expenselog'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='expenselog',
+            name='amount',
+            field=models.FloatField(blank=True, null=True),
+        ),
+    ]

api/models.py

@@ -2,6 +2,7 @@ from django.db import models
 from django.contrib.postgres.fields import JSONField
 import uuid
 
+
 class UserData(models.Model):
     user_id = models.CharField(max_length=100, unique=True)
     expenses = models.JSONField(default=dict, blank=True)
@@ -11,7 +12,7 @@ class UserData(models.Model):
     has_music_url_access = models.BooleanField(default=False)
     has_ai_access = models.BooleanField(default=False)
     data_backup_key = models.CharField(max_length=100, unique=True, blank=True)
-    
+
     def __str__(self):
         return f"Data for User {self.user_id}"
 
@@ -38,7 +39,9 @@ class AppVersion(models.Model):
     def save(self, *args, **kwargs):
         # If this version is active → deactivate others
         if self.isActive:
-            AppVersion.objects.filter(isActive=True).exclude(id=self.id).update(isActive=False)
+            AppVersion.objects.filter(isActive=True).exclude(id=self.id).update(
+                isActive=False
+            )
 
         super().save(*args, **kwargs)
 
@@ -50,3 +53,17 @@ class Contact(models.Model):
     name = models.CharField(max_length=20)
     email = models.EmailField(max_length=50)
     message = models.TextField(max_length=500)
+
+    def __str__(self):
+        return self.name
+
+
+class ExpenseLog(models.Model):
+    text = models.CharField(max_length=500)
+    amount = models.FloatField(null=True, blank=True)
+    predicted_category = models.CharField(max_length=100)
+    confidence_score = models.FloatField()
+    created_at = models.DateTimeField(auto_now_add=True)
+
+    def __str__(self):
+        return f"{self.predicted_category} (${self.amount}): {self.text[:30]}"

api/serializers.py

@@ -15,3 +15,10 @@ class ContactSerializer(serializers.ModelSerializer):
     class Meta:
         model = Contact
         fields = ["id", "name", "email", "message"]
+
+
+class ExpenseRequestSerializer(serializers.Serializer):
+    text = serializers.CharField(
+        max_length=500,
+        help_text="Enter the expense description to categorize (e.g., 'Paid 1500 for electricity')",
+    )

api/urls.py

@@ -1,8 +1,11 @@
 from django.urls import path
-from .views import UserDataPostView, GetFieldView, ContactList1
+from .views import UserDataPostView, GetFieldView, ContactList1, CategorizeExpenseView
 
 urlpatterns = [
     path("post/", UserDataPostView.as_view(), name="post_user_data"),
-    path("<str:field>/<str:identifier>/", GetFieldView.as_view(), name="get_field_data"),
-    path('contact/', ContactList1.as_view()),
-]
+    path(
+        "<str:field>/<str:identifier>/", GetFieldView.as_view(), name="get_field_data"
+    ),
+    path("contact/", ContactList1.as_view()),
+    path("categorize/", CategorizeExpenseView.as_view(), name="categorize_expense"),
+]

api/views.py

@@ -3,9 +3,13 @@ from rest_framework.response import Response
 from rest_framework import status
 from .models import UserData, AIKey, AppVersion, Contact
 from django.shortcuts import get_object_or_404
-from .serializers import ContactSerializer
-from rest_framework.generics import CreateAPIView
+from .serializers import ContactSerializer, ExpenseRequestSerializer
+from rest_framework.generics import CreateAPIView, GenericAPIView
 import uuid
+from .models import ExpenseLog
+from django.apps import apps
+import re
+
 
 class UserDataPostView(APIView):
     def post(self, request):
@@ -27,7 +31,7 @@ class UserDataPostView(APIView):
                 setattr(obj, key, request.data[key])
                 updated = True
 
-        obj.save()   # Save once after changes
+        obj.save()  # Save once after changes
 
         # AI key logic
         if obj.has_ai_access:
@@ -40,13 +44,9 @@ class UserDataPostView(APIView):
                 "message": (
                     "Data created successfully"
                     if created
-                    else "Data updated successfully"
-                    if updated
-                    else "No changes made"
+                    else "Data updated successfully" if updated else "No changes made"
                 ),
-                "app_version": get_object_or_404(
-                    AppVersion, isActive=True
-                ).version,
+                "app_version": get_object_or_404(AppVersion, isActive=True).version,
                 "has_music_url_access": obj.has_music_url_access,
                 "has_ai_access": obj.has_ai_access,
                 "ai_key": AI_KEY,
@@ -59,8 +59,10 @@ class UserDataPostView(APIView):
 class GetFieldView(APIView):
     def get(self, request, field, identifier):
         # identifier can be user_id OR data_backup_key
-        obj = UserData.objects.filter(user_id=identifier).first() \
-              or UserData.objects.filter(data_backup_key=identifier).first()
+        obj = (
+            UserData.objects.filter(user_id=identifier).first()
+            or UserData.objects.filter(data_backup_key=identifier).first()
+        )
 
         if not obj:
             return Response({"error": "User not found"}, status=404)
@@ -89,10 +91,59 @@ class ContactList1(CreateAPIView):
             serializer.save()
             return Response(
                 {"message": "Contact saved successfully"},
-                status=status.HTTP_201_CREATED
+                status=status.HTTP_201_CREATED,
             )
 
         return Response(
             {"message": "Validation failed", "errors": serializer.errors},
-            status=status.HTTP_400_BAD_REQUEST
-        )
+            status=status.HTTP_400_BAD_REQUEST,
+        )
+
+
+class CategorizeExpenseView(GenericAPIView):
+    # 2. Assign the serializer class to automatically populate the Swagger schema
+    serializer_class = ExpenseRequestSerializer
+
+    def post(self, request):
+        # 3. Validate input using the serializer
+        serializer = self.get_serializer(data=request.data)
+        if not serializer.is_valid():
+            return Response(serializer.errors, status=400)
+
+        expense_text = serializer.validated_data["text"]
+
+        # 4. Extract Amount using Regex
+        # Matches formats like 150, 1,500, 150.50, 1,500.50
+        amount_match = re.search(r"\d+(?:,\d+)*(?:\.\d+)?", expense_text)
+        extracted_amount = (
+            float(amount_match.group().replace(",", "")) if amount_match else None
+        )
+
+        # 5. Use the safe Django app registry to fetch the pre-loaded AI model
+        api_config = apps.get_app_config("api")
+
+        result = api_config.classifier(
+            expense_text,
+            candidate_labels=api_config.categories,
+            hypothesis_template="This expense is for {}.",
+        )
+
+        top_category = result["labels"][0]
+        confidence_score = result["scores"][0]
+
+        # 6. Save to database
+        ExpenseLog.objects.create(
+            text=expense_text,
+            amount=extracted_amount,
+            predicted_category=top_category,
+            confidence_score=confidence_score,
+        )
+
+        return Response(
+            {
+                "expense": expense_text,
+                "amount": extracted_amount,
+                "category": top_category,
+                "confidence": round(confidence_score * 100, 2),
+            }
+        )

dev.py

@@ -0,0 +1,49 @@
+import requests
+import json
+
+# Replace this URL with your actual Hugging Face Space URL once deployed
+# For local testing (e.g., python manage.py runserver), use: "[http://127.0.0.1:8000/api/categorize/](http://127.0.0.1:8000/api/categorize/)"
+API_URL = "http://127.0.0.1:8000/api/categorize/"
+
+
+def test_categorize_expense(expense_text):
+    # The JSON body that your Django API expects
+    payload = {"text": expense_text}
+
+    headers = {"Content-Type": "application/json"}
+
+    print(f"Sending request to: {API_URL}")
+    print(f"Expense: '{expense_text}'\n")
+
+    try:
+        # Send the POST request to your API
+        response = requests.post(API_URL, json=payload, headers=headers)
+
+        # Check if the request was successful
+        if response.status_code == 200:
+            print("--- API Response ---")
+            # Parse and print the formatted JSON response
+            print(json.dumps(response.json(), indent=4))
+        else:
+            print(f"Error: API returned status code {response.status_code}")
+            print(response.text)
+
+    except requests.exceptions.RequestException as e:
+        print(f"Connection Error: Could not connect to the API. Details: {e}")
+
+
+if __name__ == "__main__":
+    # Test cases to check your API
+    print("Starting API tests...\n")
+
+    test_categorize_expense("I spent 150 rupees on a burger today")
+
+    print("-" * 40)
+
+    test_categorize_expense("Paid 1,500.50 for the electricity and water bill")
+
+    print("-" * 40)
+
+    test_categorize_expense(
+        "Bought paracetamol and cough syrup"
+    )  # Will return amount: null