removed the async tasks to reduce complexity

Author: Priyanshu-Batham

Algolyzer/Algolyzer/__init__.py

@@ -1,3 +0,0 @@
-from .celery_app import celery_app
-
-__all__ = ("celery_app",)

Algolyzer/Algolyzer/celery_app.py

@@ -97,15 +97,5 @@
 STATICFILES_DIRS = [BASE_DIR / "static"]
 STATIC_ROOT = BASE_DIR / "staticfiles"
 
-# Celery Configurations
-# settings.py
-
-# Celery Settings
-CELERY_BROKER_URL = os.getenv(
-    "CELERY_BROKER_URL", "redis://localhost:6379/0"
-)  # URL for Redis
-CELERY_ACCEPT_CONTENT = ["json"]  # Accept JSON formatted tasks
-CELERY_TASK_SERIALIZER = "json"  # Use JSON serialization
-
 # AIML downloaded models directory
 MODEL_DIR = "/aiml_models"

Algolyzer/Algolyzer/settings/base.py

@@ -7,5 +7,5 @@
 class PlaygroundTaskAdmin(admin.ModelAdmin):
     list_display = ("id", "user", "model_name", "status", "created_at", "completed_at")
     list_filter = ("status", "model_name", "created_at")
-    search_fields = ("user__username", "celery_task_id", "model_name", "input_data")
-    readonly_fields = ("created_at", "completed_at", "celery_task_id")
+    search_fields = ("user__username", "model_name", "input_data")
+    readonly_fields = ("created_at", "completed_at")

Algolyzer/playground/admin.py

@@ -0,0 +1,25 @@
+# Generated by Django 5.1.4 on 2025-04-03 11:06
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("playground", "0002_playgroundtask_input_image"),
+    ]
+
+    operations = [
+        migrations.RemoveField(
+            model_name="playgroundtask",
+            name="celery_task_id",
+        ),
+        migrations.AlterField(
+            model_name="playgroundtask",
+            name="status",
+            field=models.CharField(
+                choices=[("COMPLETED", "Completed"), ("FAILED", "Failed")],
+                default="PENDING",
+                max_length=20,
+            ),
+        ),
+    ]

Algolyzer/playground/migrations/0003_remove_playgroundtask_celery_task_id_and_more.py

@@ -4,14 +4,10 @@
 
 class PlaygroundTask(models.Model):
     STATUS_CHOICES = [
-        ("PENDING", "Pending"),
-        ("PROCESSING", "Processing"),
         ("COMPLETED", "Completed"),
         ("FAILED", "Failed"),
     ]
-
     user = models.ForeignKey(User, on_delete=models.CASCADE)
-    celery_task_id = models.CharField(max_length=255, unique=True)  # Celery's Task ID
     model_name = models.CharField(max_length=100)  # Selected ML model
     input_data = models.TextField()  # Store user input (can be JSON)
     input_image = models.ImageField(upload_to="doodles/", blank=True, null=True)
@@ -21,4 +17,4 @@ class PlaygroundTask(models.Model):
     completed_at = models.DateTimeField(blank=True, null=True)
 
     def __str__(self):
-        return f"Task {self.celery_task_id} - {self.status}"
+        return f"Task {self.model_name} - {self.status}"

Algolyzer/playground/models.py

@@ -38,7 +38,7 @@ <h2 class="text-xl font-semibold mb-3 px-4">Previous Results</h2>
                             <div class="stat place-items-center">
                               <div class="stat-title">Status</div>
                               <div class="stat-value text-primary">{{ result.status }}</div>
-                              <div class="stat-desc">Pending | Processing | Completed</div>
+                              <div class="stat-desc">Failed | Completed</div>
                             </div>
 
                             <div class="stat place-items-center">

Algolyzer/playground/templates/playground/doodle_classifier.html

@@ -32,7 +32,7 @@ <h2 class="text-xl font-semibold mb-3 px-4">Previous Results</h2>
                             <div class="stat place-items-center">
                               <div class="stat-title">Status</div>
                               <div class="stat-value text-primary">{{ result.status }}</div>
-                              <div class="stat-desc">Pending | Processing | Completed</div>
+                              <div class="stat-desc">Failed | Completed</div>
                             </div>
 
                             <div class="stat place-items-center">

Algolyzer/playground/templates/playground/sentiment_analysis.html

@@ -1,11 +1,15 @@
 import base64
 import json
+import os
 
+from django.conf import settings
 from django.contrib.auth.decorators import login_required
 from django.core.files.base import ContentFile
 from django.shortcuts import redirect, render
+from django.utils.timezone import now
 from home.decorators import profile_required
-from playground.tasks import doodle_classifier_task, sentiment_analysis_task
+from PIL import Image
+from transformers import AutoModelForSequenceClassification, AutoTokenizer, pipeline
 
 from .models import PlaygroundTask
 
@@ -28,17 +32,47 @@ def sentiment_analysis(request):
     if request.method == "POST":
         input_data = request.POST.get("input_data")
 
-        # Create a task entry in the database
-        task = PlaygroundTask.objects.create(
-            user=user,
-            input_data=input_data,
-            model_name="sentiment_analysis",
-            status="PENDING",
-        )
-
         # Call Celery and attach the Celery task ID to the database entry
-        celery_task = sentiment_analysis_task.apply_async(args=[task.id, input_data])
-        task.celery_task_id = celery_task.id
+        try:
+            MODEL_DIR = os.path.join(
+                settings.BASE_DIR,
+                "playground",
+                "aiml_models",
+                "finiteautomata_bertweet-base-sentiment-analysis",
+            )
+
+            # Load model from saved directory
+            tokenizer = AutoTokenizer.from_pretrained(MODEL_DIR)
+            model = AutoModelForSequenceClassification.from_pretrained(MODEL_DIR)
+            sentiment_pipeline = pipeline(
+                "sentiment-analysis", model=model, tokenizer=tokenizer
+            )
+
+            if not model:
+                raise ValueError(
+                    "Model 'finiteautomata_bertweet-base-sentiment-analysis' not found."
+                )
+
+            # Process the input with the model
+            result = sentiment_pipeline(input_data)[0]
+
+            # Create a task entry in the database
+            task = PlaygroundTask.objects.create(
+                user=user,
+                input_data=input_data,
+                model_name="sentiment_analysis",
+                result=str(result),
+                completed_at=now(),
+                status="COMPLETED",
+            )
+
+            task.save()
+
+        except Exception as e:
+            task.status = "FAILED"
+            task.result = str(e)
+            task.save()
+            # raise e
         task.save()
 
         return redirect("sentiment_analysis")
@@ -51,19 +85,22 @@ def sentiment_analysis(request):
         ).order_by("-created_at")
 
         for task in previous_results:
-            if task.result:
-                try:
-                    # Replace single quotes with double quotes to make it valid JSON
-                    formatted_result = task.result.replace("'", '"')
-                    task.result = json.loads(formatted_result)
-                except json.JSONDecodeError:
-                    task.result = (
-                        {}
-                    )  # If JSON is invalid, set it as an empty dictionary
+            try:
+                # Replace single quotes with double quotes to make it valid JSON
+                formatted_result = task.result.replace("'", '"')
+                task.result = json.loads(formatted_result)
+            except json.JSONDecodeError:
+                task.result = {}  # If JSON is invalid, set it as an empty dictionary
         context = {"previous_results": previous_results}
         return render(request, "playground/sentiment_analysis.html", context=context)
 
 
+# Load the pre-trained MNIST model from Hugging Face
+mnist_classifier = pipeline(
+    "image-classification", model="farleyknight/mnist-digit-classification-2022-09-04"
+)
+
+
 @login_required
 def doodle_classifier(request):
     user = request.user
@@ -82,16 +119,34 @@ def doodle_classifier(request):
             user=user,
             input_data=image_data,  # Store the base64 data for reference
             model_name="doodle_classifier",
-            status="PENDING",
         )
 
         # Save the image file to the task's input_image field
         task.input_image.save(f"doodle_{task.id}.png", image_file)
         task.save()
 
         # Call Celery task
-        celery_task = doodle_classifier_task.apply_async(args=[task.id])
-        task.celery_task_id = celery_task.id
+        try:
+            # Open the saved PNG image from the input_image field
+            image = Image.open(task.input_image.path)
+            print("image being loaded: ", image.filename)
+
+            # Run inference using the pre-trained model
+            result = mnist_classifier(image)
+            print(str(result))
+            best_prediction = max(result, key=lambda x: x["score"])
+            best_label = best_prediction["label"]
+            # print("Label: ", best_label, "Score: ", best_score)
+
+            # Update the task with the result
+            task.result = best_label  # Convert the result to a string for storage
+            task.status = "COMPLETED"
+            task.save()
+
+        except Exception as e:
+            # Handle any errors that occur during processing
+            task.result = str(e)  # Store the error message
+            task.status = "FAILED"
         task.save()
 
         return redirect("doodle_classifier")