app.py

import os
import streamlit as st
import requests
import pandas as pd
import nltk
from wordcloud import WordCloud
import matplotlib.pyplot as plt
from sklearn.feature_extraction.text import CountVectorizer, ENGLISH_STOP_WORDS
from sklearn.decomposition import LatentDirichletAllocation
from datetime import datetime, timedelta

# Download VADER lexicon (if not already downloaded)
nltk.download('vader_lexicon')
from nltk.sentiment.vader import SentimentIntensityAnalyzer

# Global VADER instance
sia = SentimentIntensityAnalyzer()

# Environment variable for the API key
API_KEY = os.getenv("FMP_API_KEY")

# Maximum pages to fetch
MAX_PAGES = 3

# Store stateful data
if "all_run" not in st.session_state:
    st.session_state.all_run = False
if "symbol_run" not in st.session_state:
    st.session_state.symbol_run = False
if "selected_symbol" not in st.session_state:
    st.session_state.selected_symbol = "AAPL"
if "selected_date" not in st.session_state:
    st.session_state.selected_date = datetime.now().date() - timedelta(days=30)
if "selected_topics_all" not in st.session_state:
    st.session_state.selected_topics_all = 10
if "selected_topics_symbol" not in st.session_state:
    st.session_state.selected_topics_symbol = 10

#############################
# Utility Functions
#############################

def process_press_releases_df(df: pd.DataFrame) -> pd.DataFrame:
    """
    Add a sentiment score using VADER for each press release row.
    Returns the updated DataFrame.
    """
    if df.empty:
        return df
    df["sentiment"] = df["text"].apply(lambda x: sia.polarity_scores(x)["compound"])
    return df

def generate_wordcloud(df: pd.DataFrame):
    """
    Generate and display a word cloud from the 'text' column.
    """
    all_text = " ".join(df["text"].dropna().tolist())
    if not all_text:
        st.write("No text found for generating a word cloud.")
        return
    wc = WordCloud(width=800, height=400, background_color="white").generate(all_text)
    fig, ax = plt.subplots(figsize=(10, 5))
    ax.imshow(wc, interpolation="bilinear")
    ax.axis("off")
    st.pyplot(fig)

def run_topic_modeling(df: pd.DataFrame, n_topics=10, n_top_words=10):
    """
    Perform topic modeling using LDA. Display top words for each topic.
    """
    texts = df["text"].dropna().tolist()
    if not texts:
        st.write("No text available for topic modeling.")
        return
    # Extend default English stop words with common press release terms.
    custom_stop_words = list(ENGLISH_STOP_WORDS.union({
        "said", "reuters", "inc", "llc", "corp", "co", "company", "news", "press", "release"
    }))
    vectorizer = CountVectorizer(stop_words=custom_stop_words)
    X = vectorizer.fit_transform(texts)
    lda = LatentDirichletAllocation(n_components=n_topics, random_state=42)
    lda.fit(X)

    # Build a dictionary of topic names -> top words
    topics = {}
    for topic_idx, topic in enumerate(lda.components_):
        top_features_ind = topic.argsort()[:-n_top_words - 1:-1]
        top_features = [vectorizer.get_feature_names_out()[i] for i in top_features_ind]
        topics[f"Topic {topic_idx+1}"] = top_features

    st.write("### Topic Modeling Results")
    for topic_label, words in topics.items():
        st.write(f"**{topic_label}:** {', '.join(words)}")

#############################
# PAGE 1: Press Releases Live Feed
#############################

@st.cache_data(show_spinner=False)
def fetch_press_releases_all() -> pd.DataFrame:
    """
    Fetch recent press releases from multiple companies across several pages.
    Returns a combined DataFrame.
    """
    frames = []
    for page in range(MAX_PAGES):
        url = f"https://financialmodelingprep.com/api/v3/press-releases?page={page}&apikey={API_KEY}"
        try:
            response = requests.get(url)
            response.raise_for_status()
            data = response.json()
            if not data:
                break
            frames.append(pd.DataFrame(data))
        except Exception:
            # Fail gracefully without naming the data source
            return pd.DataFrame()

    if frames:
        df = pd.concat(frames, ignore_index=True)
        if "date" in df.columns:
            df["date"] = pd.to_datetime(df["date"])
        return df
    return pd.DataFrame()

def run_all_press_releases():
    st.write("**Press Releases Live Feed**")
    st.write(
        "Here, you will see the latest press releases aggregated from various companies. "
        "Explore the table for publication dates, text content, and automated sentiment. "
        "Use the Word Cloud and Topic Modeling below to uncover common themes."
    )

    df = fetch_press_releases_all()
    if df.empty:
        st.error("No press releases found.")
        return

    # Process text for sentiment
    df = process_press_releases_df(df)
    st.dataframe(df, use_container_width=True)

    st.subheader("Word Cloud")
    generate_wordcloud(df)

    st.subheader("Topic Modeling")
    run_topic_modeling(df, n_topics=st.session_state.selected_topics_all)

#############################
# PAGE 2: Press Releases by Company
#############################

@st.cache_data(show_spinner=False)
def fetch_press_releases_by_symbol(symbol: str) -> pd.DataFrame:
    """
    Fetch recent press releases for a single company symbol across several pages.
    Returns a combined DataFrame.
    """
    frames = []
    for page in range(MAX_PAGES):
        url = f"https://financialmodelingprep.com/api/v3/press-releases/{symbol}?page={page}&apikey={API_KEY}"
        try:
            response = requests.get(url)
            response.raise_for_status()
            data = response.json()
            if not data:
                break
            frames.append(pd.DataFrame(data))
        except Exception:
            # Fail gracefully without naming the data source
            return pd.DataFrame()

    if frames:
        df = pd.concat(frames, ignore_index=True)
        if "date" in df.columns:
            df["date"] = pd.to_datetime(df["date"])
        return df
    return pd.DataFrame()

def run_symbol_press_releases(symbol: str, start_date, n_topics):
    st.write("**Press Releases by Company**")
    st.write(
        f"Browse recent press releases for **{symbol}**, starting from {start_date}. "
        "View release text, publication dates, and sentiment analysis. "
        "Below, discover prevalent words and recurring topics for these press releases."
    )

    df = fetch_press_releases_by_symbol(symbol)
    if df.empty:
        st.error(f"No press releases found for {symbol}.")
        return

    # Filter by user-chosen date
    if "date" in df.columns:
        df = df[df["date"].dt.date >= start_date]

    # Process text for sentiment
    df = process_press_releases_df(df)
    st.dataframe(df, use_container_width=True)

    st.subheader("Word Cloud")
    generate_wordcloud(df)

    st.subheader("Topic Modeling")
    run_topic_modeling(df, n_topics=n_topics)

#############################
# MAIN APP
#############################

def main():
    st.set_page_config(page_title="Press Releases", layout="wide")
    st.title("Press Releases Analysis")
    st.write(
        "Explore recent press releases from multiple companies or focus on a single company. "
        "Each page provides a table of press releases, sentiment analysis, a word cloud, and topic modeling."
    )

    # Sidebar navigation
    with st.sidebar.expander("Navigation and Options", expanded=True):
        page = st.radio(
            "Select Page",
            ("Press Releases Live Feed", "Press Releases by Company"),
            help="Choose between a broad overview or a single company's releases."
        )

        if page == "Press Releases Live Feed":
            st.session_state.selected_topics_all = st.number_input(
                "Number of Topics for Live Feed",
                value=st.session_state.selected_topics_all,
                min_value=1,
                max_value=20,
                help="Choose how many topics you want to see in the topic model."
            )
            if st.button("Run"):
                st.session_state.all_run = True

        elif page == "Press Releases by Company":
            symbol = st.text_input(
                "Ticker Symbol",
                value=st.session_state.selected_symbol,
                help="Type the company's ticker symbol."
            )
            st.session_state.selected_symbol = symbol

            start_date = st.date_input(
                "Start Date",
                value=st.session_state.selected_date,
                help="Only press releases on or after this date will appear."
            )
            st.session_state.selected_date = start_date

            st.session_state.selected_topics_symbol = st.number_input(
                "Number of Topics for Company",
                value=st.session_state.selected_topics_symbol,
                min_value=1,
                max_value=20,
                help="Choose how many topics you want to see in the topic model."
            )
            if st.button("Run"):
                st.session_state.symbol_run = True

    # Main body content
    if page == "Press Releases Live Feed":
        st.header("Press Releases Live Feed")
        if st.session_state.all_run:
            run_all_press_releases()
        else:
            st.info("Pick how many topics to show, then click 'Run Press Releases Live Feed'.")

    elif page == "Press Releases by Company":
        st.header("Press Releases by Company")
        if st.session_state.symbol_run:
            run_symbol_press_releases(
                st.session_state.selected_symbol,
                st.session_state.selected_date,
                st.session_state.selected_topics_symbol
            )
        else:
            st.info("Enter a ticker symbol, date, and number of topics, then click 'Run Press Releases by Company'.")

if __name__ == "__main__":
    main()

hide_streamlit_style = """
<style>
#MainMenu {visibility: hidden;}
footer {visibility: hidden;}
</style>
"""
st.markdown(hide_streamlit_style, unsafe_allow_html=True)