app.py

from fastapi import FastAPI, HTTPException
from pydantic import BaseModel
from fastapi.responses import JSONResponse
import torch
from transformers import AutoModelForSequenceClassification, AutoTokenizer
import os
import re
import logging
 
app = FastAPI()
 
# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
 
# Set the cache directory for Hugging Face
os.environ['TRANSFORMERS_CACHE'] = os.getenv('TRANSFORMERS_CACHE', '/app/cache')
 
# Load model and tokenizer
model_name = "BIJOY087/Bangla_barta_shurkha_mobilebert"
try:
    model = AutoModelForSequenceClassification.from_pretrained(model_name)
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    logger.info("Model and tokenizer loaded successfully")
except Exception as e:
    logger.error(f"Failed to load model or tokenizer: {e}")
    raise RuntimeError(f"Failed to load model or tokenizer: {e}")
 
class TextRequest(BaseModel):
    text: str
 
class BatchTextRequest(BaseModel):
    texts: list[str]
 
# Regular expression to detect Bangla characters
bangla_regex = re.compile('[\u0980-\u09FF]')
 
def contains_bangla(text):
    return bool(bangla_regex.search(text))
 
@app.post("/batch_predict/")
async def batch_predict(request: BatchTextRequest):
    try:
        model.eval()
 
        # Prepare the batch results
        results = []
 
        for idx, text in enumerate(request.texts):
            
            # Check if text contains Bangla characters
            if not contains_bangla(text):
                results.append({"id": idx + 1, "text": text, "prediction": "other"})
                continue
 
            # Encode and predict for texts containing Bangla characters
            inputs = tokenizer.encode_plus(
                text,
                add_special_tokens=True,
                max_length=64,
                truncation=True,
                padding='max_length',
                return_attention_mask=True,
                return_tensors='pt'
            )
 
            with torch.no_grad():
                logits = model(inputs['input_ids'], attention_mask=inputs['attention_mask']).logits
                prediction = torch.argmax(logits, dim=1).item()
                label = "Spam" if prediction == 1 else "Ham"
                results.append({"id": idx + 1, "text": text, "prediction": label})
 
        logger.info(f"Batch prediction results: {results}")
        return JSONResponse(content={"results": results}, media_type="application/json; charset=utf-8")
 
    except Exception as e:
        logger.error(f"Batch prediction failed: {e}")
        raise HTTPException(status_code=500, detail="Batch prediction failed. Please try again.")
 
@app.get("/")
async def root():
    return {"message": "Welcome to the MobileBERT API"}