app.py

import streamlit as st
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from wordcloud import WordCloud
import os

# Import the data loading and processing function from analysis.py
# Assuming analysis.py is in the same directory
from analysis import load_and_process_data

# --- Streamlit Application ---
st.set_page_config(layout="wide", page_title="CORD-19 Research Paper Analysis")

st.title("CORD-19 Research Paper Analysis")
st.markdown("""
This application provides an interactive analysis of the CORD-19 (COVID-19 Open Research Dataset) metadata.
Explore trends in publication years, top journals, and common keywords in paper titles.
""")

@st.cache_data # Cache data to avoid reloading on every rerun
def get_data():
    return load_and_process_data()

df = get_data()

if df.empty:
    st.error("Could not load data. Please ensure 'metadata.csv' is in the project directory and try again.")
else:
    st.sidebar.header("Filter Options")

    # Year Range Slider
    min_year = int(df['year'].min()) if not df.empty else 2000
    max_year = int(df['year'].max()) if not df.empty else 2022
    year_range = st.sidebar.slider(
        "Select Year Range",
        min_value=min_year,
        max_value=max_year,
        value=(min_year, max_year)
    )

    # Journal Dropdown
    all_journals = ["All"] + sorted(df['journal'].unique().tolist())
    selected_journal = st.sidebar.selectbox("Select Journal", all_journals)

    # Apply filters
    filtered_df = df[(df['year'] >= year_range[0]) & (df['year'] <= year_range[1])]
    if selected_journal != "All":
        filtered_df = filtered_df[filtered_df['journal'] == selected_journal]

    st.header("Analysis Results")

    if filtered_df.empty:
        st.warning("No data available for the selected filters.")
    else:
        # --- Visualizations ---
        st.subheader("Publications Over Time")
        papers_by_year = filtered_df['year'].value_counts().sort_index()
        fig1, ax1 = plt.subplots(figsize=(10, 5))
        sns.lineplot(x=papers_by_year.index, y=papers_by_year.values, ax=ax1)
        ax1.set_title('Number of Publications Over Time')
        ax1.set_xlabel('Year')
        ax1.set_ylabel('Number of Papers')
        ax1.grid(True)
        st.pyplot(fig1)
        plt.close(fig1)

        st.subheader("Top Publishing Journals")
        top_journals = filtered_df['journal'].value_counts().head(10)
        if not top_journals.empty:
            fig2, ax2 = plt.subplots(figsize=(10, 6))
            sns.barplot(x=top_journals.values, y=top_journals.index, palette='viridis', ax=ax2)
            ax2.set_title('Top 10 Publishing Journals')
            ax2.set_xlabel('Number of Papers')
            ax2.set_ylabel('Journal')
            st.pyplot(fig2)
            plt.close(fig2)
        else:
            st.info("No journals found for the selected filters.")

        st.subheader("Word Cloud of Paper Titles")
        all_titles = ' '.join(filtered_df['title'].dropna().tolist())
        if all_titles:
            wordcloud = WordCloud(width=800, height=400, background_color='white').generate(all_titles)
            fig3, ax3 = plt.subplots(figsize=(10, 5))
            ax3.imshow(wordcloud, interpolation='bilinear')
            ax3.axis('off')
            ax3.set_title('Word Cloud of Paper Titles')
            st.pyplot(fig3)
            plt.close(fig3)
        else:
            st.info("No titles available for word cloud generation with the selected filters.")

        st.subheader("Distribution of Paper Counts by Source")
        if 'source_x' in filtered_df.columns:
            paper_sources = filtered_df['source_x'].value_counts().head(10)
            source_col = 'source_x'
        else:
            paper_sources = filtered_df['journal'].value_counts().head(10) # Fallback
            source_col = 'journal'

        if not paper_sources.empty:
            fig4, ax4 = plt.subplots(figsize=(10, 6))
            sns.barplot(x=paper_sources.values, y=paper_sources.index, palette='plasma', ax=ax4)
            ax4.set_title(f'Top 10 Paper Sources ({source_col})')
            ax4.set_xlabel('Number of Papers')
            ax4.set_ylabel('Source')
            st.pyplot(fig4)
            plt.close(fig4)
        else:
            st.info("No sources found for the selected filters.")

        st.subheader("Data Sample")
        st.write(filtered_df.head(10))