Rowan Martnishn
Update app.py
037b7f7 verified
raw
history blame
9.87 kB
import gradio as gr
import schedule
import time
import datetime
import praw
import joblib
import torch
import scipy.sparse as sp
import torch.nn as nn
import pandas as pd
import re
import numpy as np
import matplotlib.pyplot as plt
from scipy.interpolate import make_interp_spline
from transformers import AutoTokenizer
import matplotlib.font_manager as fm
# Load models and data (your existing code)
autovectorizer = joblib.load('AutoVectorizer.pkl')
autoclassifier = joblib.load('AutoClassifier.pkl')
MODEL = "cardiffnlp/xlm-twitter-politics-sentiment"
tokenizer = AutoTokenizer.from_pretrained(MODEL)
class ScorePredictor(nn.Module):
# ... (Your ScorePredictor class)
def __init__(self, vocab_size, embedding_dim=128, hidden_dim=256, output_dim=1):
super(ScorePredictor, self).__init__()
self.embedding = nn.Embedding(vocab_size, embedding_dim, padding_idx=0)
self.lstm = nn.LSTM(embedding_dim, hidden_dim, batch_first=True)
self.fc = nn.Linear(hidden_dim, output_dim)
self.sigmoid = nn.Sigmoid()
def forward(self, input_ids, attention_mask):
embedded = self.embedding(input_ids)
lstm_out, _ = self.lstm(embedded)
final_hidden_state = lstm_out[:, -1, :]
output = self.fc(final_hidden_state)
return self.sigmoid(output)
score_model = ScorePredictor(tokenizer.vocab_size)
score_model.load_state_dict(torch.load("score_predictor.pth"))
score_model.eval()
sentiment_model = joblib.load('sentiment_forecast_model.pkl')
reddit = praw.Reddit(
client_id="PH99oWZjM43GimMtYigFvA",
client_secret="3tJsXQKEtFFYInxzLEDqRZ0s_w5z0g",
user_agent='MyAPI/0.0.1',
check_for_async=False)
subreddits = [
"centrist",
"libertarian",
"southpark",
"truechristian",
"conservatives"
]
# Global variables for data
global prediction_plot_base64
prediction_plot_base64 = None
def process_data():
"""Fetches data, performs analysis, and generates the plot."""
global prediction_plot_base64
end_date = datetime.datetime.utcnow()
start_date = end_date - datetime.timedelta(days=14)
def fetch_all_recent_posts(subreddit_name, start_time, limit=500):
# ... (Your fetch_all_recent_posts function)
subreddit = reddit.subreddit(subreddit_name)
posts = []
try:
for post in subreddit.top(limit=limit): # Fetch recent posts
post_time = datetime.datetime.utcfromtimestamp(post.created_utc)
if post_time >= start_time: # Filter only within last 14 days
posts.append({
"subreddit": subreddit_name,
"timestamp": post.created_utc,
"date": post_time.strftime('%Y-%m-%d %H:%M:%S'),
"post_text": post.title
})
except Exception as e:
print(f"Error fetching posts from r/{subreddit_name}: {e}")
return posts
def preprocess_text(text):
# ... (Your preprocess_text function)
text = text.lower()
text = re.sub(r'http\S+', '', text)
text = re.sub(r'[^a-zA-Z0-9\s.,!?]', '', text)
text = re.sub(r'\s+', ' ', text).strip()
return text
def predict_score(text):
# ... (Your predict_score function)
if not text:
return 0.0
max_length = 512
encoded_input = tokenizer(
text.split(),
return_tensors='pt',
padding=True,
truncation=True,
max_length=max_length
)
input_ids, attention_mask = encoded_input["input_ids"], encoded_input["attention_mask"]
with torch.no_grad():
score = score_model(input_ids, attention_mask)[0].item()
return score
start_time = datetime.datetime.utcnow() - datetime.timedelta(days=14)
all_posts = []
for sub in subreddits:
print(f"Fetching posts from r/{sub}")
posts = fetch_all_recent_posts(sub, start_time)
all_posts.extend(posts)
print(f"Fetched {len(posts)} posts from r/{sub}")
filtered_posts = []
for post in all_posts:
vector = autovectorizer.transform([post['post_text']])
prediction = autoclassifier.predict(vector)
if prediction[0] == 1:
filtered_posts.append(post)
all_posts = filtered_posts
df = pd.DataFrame(all_posts)
df['date'] = pd.to_datetime(df['date'])
df['date_only'] = df['date'].dt.date
df = df.sort_values(by=['date_only'])
df['sentiment_score'] = df['post_text'].apply(predict_score)
last_14_dates = df['date_only'].unique()
num_dates = min(len(last_14_dates), 14)
last_14_dates = sorted(last_14_dates, reverse=True)[:num_dates]
filtered_df = df[df['date_only'].isin(last_14_dates)]
daily_sentiment = filtered_df.groupby('date_only')['sentiment_score'].median()
if len(daily_sentiment) < 14:
mean_sentiment = daily_sentiment.mean()
padding = [mean_sentiment] * (14 - len(daily_sentiment))
daily_sentiment = np.concatenate([daily_sentiment.values, padding])
daily_sentiment = pd.Series(daily_sentiment)
sentiment_scores_np = daily_sentiment.values.reshape(1, -1)
prediction = sentiment_model.predict(sentiment_scores_np)
pred = (prediction[0])
font_path = "AfacadFlux-VariableFont_slnt,wght[1].ttf"
custom_font = fm.FontProperties(fname=font_path)
today = datetime.date.today()
days = [today + datetime.timedelta(days=i) for i in range(7)]
days_str = [day.strftime('%a %m/%d') for day in days]
xnew = np.linspace(0, 6, 300)
spline = make_interp_spline(np.arange(7), pred, k=3)
pred_smooth = spline(xnew)
fig, ax = plt.subplots(figsize=(12, 7))
ax.fill_between(xnew, pred_smooth, color='#244B48', alpha=0.4)
ax.plot(xnew, pred_smooth, color='#244B48', lw=3, label='Forecast')
ax.scatter(np.arange(7), pred, color='#244B48', s=100, zorder=5)
ax.set_title(f"7-Day Political Sentiment Forecast - {datetime.datetime.now()}", fontsize=22, fontweight='bold', pad=20, fontproperties=custom_font)
ax.set_xlabel("Day", fontsize=16, fontproperties=custom_font)
ax.set_ylabel("Negative Sentiment (0-1)", fontsize=16, fontproperties=custom_font)
ax.set_xticks(np.arange(7))
ax.set_xticklabels(days_str, fontsize=14, fontproperties=custom_font)
# Continue from previous app.py code
ax.set_yticklabels([f"{tick:.2f}" for tick in ax.get_yticks()], fontsize=14, fontproperties=custom_font)
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
ax.spines['left'].set_visible(False)
ax.spines['bottom'].set_visible(False)
ax.legend(fontsize=14, loc='upper right', prop=custom_font)
plt.tight_layout()
import io
import base64
buffer = io.BytesIO()
plt.savefig(buffer, format='png')
buffer.seek(0)
prediction_plot_base64 = base64.b64encode(buffer.getvalue()).decode()
plt.close(fig)
def display_plot():
"""Displays the plot in the Gradio interface."""
global prediction_plot_base64
if prediction_plot_base64:
return f'<img src="data:image/png;base64,{prediction_plot_base64}" alt="Prediction Plot">'
else:
return "Processing data..."
# process_data()
# # Schedule daily refresh
# def run_daily():
# process_data()
# print("Data refreshed at:", datetime.datetime.now())
# schedule.every().day.at("00:00").do(run_daily)
# def run_schedule():
# while True:
# schedule.run_pending()
# time.sleep(60)
# import threading
# thread = threading.Thread(target=run_schedule)
# thread.daemon = True
# thread.start()
# # Gradio Interface
# if prediction_plot_base64:
# html_content = f'<img src="data:image/png;base64,{prediction_plot_base64}" alt="Prediction Plot">'
# else:
# html_content = "Processing data..."
# iface = gr.Interface(fn=None, inputs=None, outputs=gr.HTML(value=html_content))
# iface.launch()
# Initial data processing
process_data()
# Schedule daily refresh
def run_daily():
process_data()
print("Data refreshed at:", datetime.datetime.now())
#schedule.every().day.at("00:00").do(run_daily)
schedule.every(10).seconds.do(run_daily)
def run_schedule():
while True:
schedule.run_pending()
#time.sleep(60)
import threading
thread = threading.Thread(target=run_schedule)
thread.daemon = True
thread.start()
# # Gradio Interface
# if prediction_plot_base64:
# html_content = (
# f'<img src="data:image/png;base64,{prediction_plot_base64}" '
# 'alt="Prediction Plot" '
# 'style="width: 100vw; height: 100vh; object-fit: contain;">'
# )
# else:
# html_content = "Processing data..."
# custom_css = """
# body, .gradio-container {
# margin: 0;
# padding: 0;
# }
# """
# with gr.Blocks(css=custom_css) as demo:
# gr.HTML(value=html_content)
# print("Data refreshed at:", datetime.datetime.now())
# demo.launch()
custom_css = """
body, .gradio-container {
margin: 0;
padding: 0;
}
"""
with gr.Blocks(css=custom_css) as demo:
# Initialize the HTML output with a default message
html_output = gr.HTML("Processing data...")
# Define the refresh function
def refresh_html():
if prediction_plot_base64:
return (
f'<img src="data:image/png;base64,{prediction_plot_base64}" '
'alt="Prediction Plot" '
'style="width: 100vw; height: 100vh; object-fit: contain;">'
)
else:
return "Processing data..."
# Use the Timer component according to the documentation
timer = gr.Timer(3600, refresh_html, [], html_output)
# Initial call to set the HTML content when the page loads
demo.load(refresh_html, [], html_output)
# Launch the demo
demo.launch()