app.py

from flask import Flask, render_template, request
from joblib import load
import pandas as pd
import re
from customFunctions import *
import json
import datetime
import numpy as np
from huggingface_hub import hf_hub_download
import torch
import os
import logging

pd.set_option('display.max_colwidth', 1000)

# Patch torch.load to always load on CPU
original_torch_load = torch.load
def cpu_load(*args, **kwargs):
    return original_torch_load(*args, map_location=torch.device('cpu'), **kwargs)
torch.load = cpu_load

app = Flask(__name__)

# Logging setup
LOG_DIR = "/tmp/logs"  # Use a universally writable directory
LOG_FILE = os.path.join(LOG_DIR, "usage_log.jsonl")
os.makedirs(LOG_DIR, exist_ok=True)
logging.basicConfig(
    filename=LOG_FILE,
    level=logging.INFO,
    format='%(asctime)s [%(levelname)s] %(message)s'
)

PIPELINES = [
    {'id': 8, 'name': 'Embedded using BioWordVec', 'filename': "pipeline_ex3_s4.joblib"},
    {'id': 1, 'name': 'Baseline', 'filename': "pipeline_ex1_s1.joblib"},
    {'id': 2, 'name': 'Trained on a FeedForward NN', 'filename': "pipeline_ex1_s2.joblib"},
    {'id': 3, 'name': 'Trained on a CRF', 'filename': "pipeline_ex1_s3.joblib"},
    {'id': 4, 'name': 'Trained on a small dataset', 'filename': "pipeline_ex2_s3.joblib"},
    {'id': 5, 'name': 'Trained on a large dataset', 'filename': "pipeline_ex2_s2.joblib"},
    {'id': 6, 'name': 'Embedded using TFIDF', 'filename': "pipeline_ex3_s2.joblib"},
    {'id': 7, 'name': 'Embedded using GloVe', 'filename': "pipeline_ex3_s3.joblib"},
]

pipeline_metadata = [{'id': p['id'], 'name': p['name']} for p in PIPELINES]

def load_pipeline_from_hub(filename):
    cache_dir = "/tmp/hf_cache"
    os.environ["HF_HUB_CACHE"] = cache_dir
    repo_id = 'hw01558/nlp-coursework-pipelines'
    local_path = hf_hub_download(repo_id=repo_id, filename=filename, cache_dir=cache_dir)
    return load(local_path)

def get_pipeline_by_id(pipelines, pipeline_id):
    return next((p['filename'] for p in pipelines if p['id'] == pipeline_id), None)

def get_name_by_id(pipelines, pipeline_id):
    return next((p['name'] for p in pipelines if p['id'] == pipeline_id), None)

def requestResults(text, pipeline):
    labels = pipeline.predict(text)
    if isinstance(labels, np.ndarray):
        labels = labels.tolist()
    return labels[0]

def log_interaction(user_input, model_name, predictions):
    log_entry = {
        "timestamp": datetime.datetime.utcnow().isoformat(),
        "model": model_name,
        "user_input": user_input,
        "predictions": predictions
    }
    try:
        logging.info(json.dumps(log_entry))
        print("[INFO] Logged interaction successfully.")
    except Exception as e:
        print(f"[ERROR] Could not write log entry: {e}")


@app.route('/')
def index():
    return render_template('index.html', pipelines=pipeline_metadata)

@app.route('/', methods=['POST'])
def get_data():
    if request.method == 'POST':
        text = request.form['search']
        tokens = re.findall(r"\w+|[^\w\s]", text)
        tokens_formatted = pd.Series([pd.Series(tokens)])

        pipeline_id = int(request.form['pipeline_select'])
        pipeline = load_pipeline_from_hub(get_pipeline_by_id(PIPELINES, pipeline_id))
        name = get_name_by_id(PIPELINES, pipeline_id)

        labels = requestResults(tokens_formatted, pipeline)
        results = dict(zip(tokens, labels))

        log_interaction(text, name, results)

        return render_template('index.html', results=results, name=name, pipelines=pipeline_metadata)


if __name__ == '__main__':
    app.run(host="0.0.0.0", port=7860)