FastAPI setup and Docker configuration

Author: bk-ghauri

Dockerfile

@@ -11,14 +11,14 @@ COPY requirements.txt /app/
 RUN pip install --no-cache-dir --upgrade pip \
     && pip install --no-cache-dir -r requirements.txt
 
-RUN pip install --user -U nltk
+RUN pip install -U nltk
 
 # Then copy the rest of the application files
 COPY . /app
 
-# Expose the port Flask runs on
-EXPOSE 5000
+# Expose the port FastAPI runs on
+EXPOSE 8000
 
 # Define the entry point command
-CMD ["python", "app/main.py"]
+CMD ["uvicorn", "main:app", "--reload"]
 

app/classifier_model.py

@@ -7,18 +7,22 @@
 import nltk
 from nltk.corpus import stopwords
 import gdown
+import os
+
+file_path = "bert_model.pth"  # Change to your file's actual path
 
 def download_model(file_id, output_name):
-    url = f"https://drive.google.com/uc?id={file_id}"
-    gdown.download(url, output_name, quiet=False)
+        url = f"https://drive.google.com/uc?id={file_id}"
+        gdown.download(url, output_name, quiet=False)
 
-# Model file IDs from Google Drive
-bert_model_id = "14OBJIgUtGLujlCzEaBb2Mxc5eUsAMZk5"
-cascade_bert_model_id = "14Shk7Yt6ilSrzFvppjSsqqZBv2RM1qwt"
+if not os.path.exists(file_path):
+    # Model file IDs from Google Drive
+    bert_model_id = "14OBJIgUtGLujlCzEaBb2Mxc5eUsAMZk5"
+    cascade_bert_model_id = "14Shk7Yt6ilSrzFvppjSsqqZBv2RM1qwt"
 
-# Download models
-download_model(bert_model_id, "bert_model.pth")
-download_model(cascade_bert_model_id, "cascade_bert_model.pth")
+    # Download models
+    download_model(bert_model_id, "bert_model.pth")
+    download_model(cascade_bert_model_id, "cascade_bert_model.pth")
 
 
 nltk.download('stopwords')
@@ -102,10 +106,4 @@ def classify_text(text, model_ai_hum, model_llm, tokenizer):
         else:
             predicted_label_llm = "Gemini"
 
-    return {"type": predicted_label, "llm": predicted_label_llm}
-
-    
-# Example usage:
-# abstract = df["Abstract"].values[39000]
-# result = classify_abstract(abstract, model_ai_hum, model_llm, tokenizer)
-# print(result)
+    return {"type": predicted_label, "llm": predicted_label_llm}

app/main.py

@@ -1,31 +1,33 @@
-print("svc_texpose - starting")
-
+from fastapi import FastAPI, Request, Form
+from fastapi.templating import Jinja2Templates
+from fastapi.responses import HTMLResponse
 from classifier_model import load_models, classify_text
-from flask import Flask, render_template, request
 
+print("svc_texpose - starting")
+
+# Load models
 tokenizer, model_ai_hum, model_llm = load_models()
 
-# Flask App
-app = Flask(__name__)
-
-@app.route("/", methods=["GET", "POST"])
-def index():
-    prediction = None
-    if request.method == "GET":
-        return {"svc_Texpose is alive :) "}
-    
-    if request.method == "POST":
-        input_text = request.form.get("text", "")
-        if input_text.strip():  #  input if not empty then proceed
-            result = classify_text(input_text, model_ai_hum, model_llm, tokenizer)
-            prediction = result["type"]
-            if len(result["llm"]) > 1:
-                prediction += " Using " + result["llm"]
-        
-            print(prediction)
-    return render_template("index.html", prediction=prediction if prediction is not None else "")
-
-
-if __name__ == "__main__":
-    app.run(debug=True)
+# Initialize FastAPI app
+app = FastAPI()
+
+# Set up Jinja2 templates
+templates = Jinja2Templates(directory="templates")
+
+@app.get("/", response_class=HTMLResponse)
+async def root(request: Request):
+    return templates.TemplateResponse("index.html", {"request": request, "prediction": None, "input_text": ""})
+
+@app.post("/classify", response_class=HTMLResponse)
+async def classify(request: Request, input_text: str = Form(...)):
+    if input_text.strip():  # Ensure input is not empty
+        result = classify_text(input_text, model_ai_hum, model_llm, tokenizer)
+        prediction = result["type"]
+
+        # Ensure "llm" exists and is not empty
+        if result.get("llm"):
+            prediction += f" Using {result['llm']}"
+
+        return templates.TemplateResponse("index.html", {"request": request, "prediction": prediction, "input_text": input_text})
 
+    return templates.TemplateResponse("index.html", {"request": request, "prediction": "No text provided.", "input_text": ""})

app/templates/index.html

@@ -83,9 +83,10 @@
 <body>
     <div class="container">
         <h2><i class="fas fa-brain"></i> AI vs Human Text Detector</h2>
-        <form method="POST">
-            <textarea name="text" rows="5" placeholder="Type here..." 
-            class="{% if prediction == 'Human' %}human-text{% elif prediction %}ai-text{% endif %}">{{ request.form.get('text', '') }}</textarea>
+        <form method="POST" action="/classify">
+            <textarea name="input_text" rows="5" placeholder="Type here..." 
+            class="{% if prediction == 'Human' %}human-text{% elif prediction %}ai-text{% endif %}">{{ input_text }}</textarea>
+
             <button type="submit"><i class="fas fa-search"></i> Analyze</button>
         </form>
 

requirements.txt

@@ -5,4 +5,5 @@ torch
 transformers
 numpy
 flask
-gdown
+gdown
+jinja2