Hariommm/libraries-scanner

1import streamlit as st2import requests3import pandas as pd4from datetime import datetime5import tempfile6import time7import plotly.express as px8import base649 10# --- 1. UI SETUP ---11st.set_page_config(page_title="Libraries Scanner", layout="wide", page_icon="📦")12st.title("📦 Libraries Scanner")13st.markdown("Live Vulnerability (CVE) & Health Scanning Engine")14 15# --- 2. SECURITY ENGINE ---16def check_vulnerabilities(package, version, ecosystem):17    url = "https://api.osv.dev/v1/query"18    payload = {"version": version, "package": {"name": package, "ecosystem": ecosystem}}19    for _ in range(2): 20        try:21            r = requests.post(url, json=payload, timeout=7)22            if r.status_code == 200:23                data = r.json()24                count = len(data.get("vulns", []))25                return f"🚨 {count} CVEs" if count > 0 else "✅ Secure"26        except:27            time.sleep(1)28    return "✅ Secure"29 30def classify_owner(name):31    if not name or name == "N/A": return "Unknown"32    org_k = ['team', 'foundation', 'project', 'org', 'inc', 'llc', 'group', 'maintainers']33    if any(k in str(name).lower() for k in org_k): return f"{name} (Organization)"34    return f"{name} (Individual)"35 36def get_health_status(date_str):37    if date_str in ["N/A", "Unknown", None]: return "❌ Not Found"38    try:39        days = (datetime.now() - datetime.strptime(date_str, '%Y-%m-%d')).days40        if days <= 180: return "✅ Healthy"41        elif days <= 365: return "⚠️ Warning"42        else: return "❌ Outdated"43    except: return "❌ Error"44 45# --- 3. PDF GENERATOR ---46def create_pdf(df):47    try: 48        from fpdf import FPDF49    except ImportError: 50        return None51    def clean_text(text):52        text = str(text).replace("✅", "").replace("❌", "").replace("⚠️", "").replace("🚨", "")53        return text.encode('latin-1', 'ignore').decode('latin-1').strip()54    pdf = FPDF()55    pdf.add_page()56    pdf.set_font("Arial", 'B', 16)57    pdf.cell(190, 10, "Libraries Scanner - Audit Report", ln=True, align='C')58    pdf.ln(5)59    pdf.set_font("Arial", 'B', 10)60    cols, w = ["Library", "Health", "Version", "Security", "Owner"], [45, 30, 20, 25, 70]61    for c, width in zip(cols, w): pdf.cell(width, 10, c, 1, 0, 'C')62    pdf.ln()63    pdf.set_font("Arial", '', 9)64    for _, row in df.iterrows():65        pdf.cell(w[0], 10, clean_text(row['Library'])[:25], 1)66        pdf.cell(w[1], 10, clean_text(row['Health Status'])[:15], 1)67        pdf.cell(w[2], 10, clean_text(row['Version'])[:10], 1, 0, 'C')68        pdf.cell(w[3], 10, clean_text(row['Vulnerabilities'])[:15], 1, 0, 'C')69        pdf.cell(w[4], 10, clean_text(row['Owner'])[:38], 1); pdf.ln()70    with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp:71        pdf.output(tmp.name)72        with open(tmp.name, "rb") as f: return f.read()73 74# --- 4. DATA FETCHING ---75@st.cache_data(ttl=60)76def get_pypi_data(package):77    pkg = str(package).strip().lower()78    try:79        r = requests.get(f"https://pypi.org/pypi/{pkg}/json", timeout=7)80        if r.status_code == 200:81            data = r.json()82            ver = data['info']['version']83            upd = data['releases'].get(ver, [{}])[0].get('upload_time', 'Unknown').split('T')[0]84            owner = classify_owner(data['info'].get('author') or "Community")85            urls = data['info'].get('project_urls') or {}86            src = urls.get('Source') or urls.get('Repository') or urls.get('Homepage')87            cve = check_vulnerabilities(pkg, ver, "PyPI")88            return {89                "Library": pkg, 90                "Health Status": get_health_status(upd), 91                "Vulnerabilities": cve, 92                "Version": ver, 93                "Last Updated": upd, 94                "Owner": owner, 95                "Source": src, 96                "Registry": "PyPI"97            }98    except: return None99    return None100 101@st.cache_data(ttl=60)102def get_npm_data(package):103    pkg = str(package).strip().lower()104    try:105        r = requests.get(f"https://registry.npmjs.org/{pkg}", timeout=7)106        if r.status_code == 200:107            data = r.json()108            ver = data['dist-tags']['latest']109            upd = data['time'].get(ver, "").split('T')[0]110            raw_auth = data.get('author')111            auth = raw_auth.get('name') if isinstance(raw_auth, dict) else raw_auth112            raw_repo = data.get('repository', {})113            src = raw_repo.get('url', '') if isinstance(raw_repo, dict) else raw_repo114            cve = check_vulnerabilities(pkg, ver, "npm")115            return {116                "Library": pkg, 117                "Health Status": get_health_status(upd), 118                "Vulnerabilities": cve, 119                "Version": ver, 120                "Last Updated": upd, 121                "Owner": classify_owner(auth or "Community"), 122                "Source": str(src).replace('git+', ''), 123                "Registry": "NPM"124            }125    except: return None126    return None127 128# --- 5. INTERFACE ---129with st.sidebar:130    # --- CUSTOM DRAWN CYBER LOGO ---131    svg_logo = """132    <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 200 200">133      <defs>134        <filter id="neonGlow" x="-50%" y="-50%" width="200%" height="200%">135          <feGaussianBlur stdDeviation="3" result="blur"/>136          <feMerge>137            <feMergeNode in="blur"/>138            <feMergeNode in="SourceGraphic"/>139          </feMerge>140        </filter>141      </defs>142      <circle cx="100" cy="100" r="85" fill="none" stroke="#38bdf8" stroke-width="2" stroke-dasharray="15 10" opacity="0.7"/>143      <circle cx="100" cy="100" r="65" fill="none" stroke="#00FF41" stroke-width="4" stroke-dasharray="40 10 5 10" filter="url(#neonGlow)"/>144      <polygon points="85,65 135,100 85,135" fill="none" stroke="#E0E0E0" stroke-width="4" filter="url(#neonGlow)"/>145      <circle cx="95" cy="100" r="12" fill="#00FF41" filter="url(#neonGlow)"/>146      <line x1="15" y1="100" x2="35" y2="100" stroke="#38bdf8" stroke-width="3"/>147      <line x1="185" y1="100" x2="165" y2="100" stroke="#38bdf8" stroke-width="3"/>148      <line x1="100" y1="15" x2="100" y2="35" stroke="#38bdf8" stroke-width="3"/>149      <line x1="100" y1="185" x2="100" y2="165" stroke="#38bdf8" stroke-width="3"/>150    </svg>151    """152    b64_svg = base64.b64encode(svg_logo.encode('utf-8')).decode('utf-8')153    st.markdown(f'<img src="data:image/svg+xml;base64,{b64_svg}" width="120" style="display:block; margin:auto; margin-bottom: 20px;">', unsafe_allow_html=True)154    155    st.title("System Controls")156    with st.form("audit_form"):157        libs_input = st.text_area("Target Assets (e.g. axios, pandas):", "axios, requests, pandas", height=150)158        run_btn = st.form_submit_button("🚀 Execute Scan", use_container_width=True)159 160if run_btn:161    lib_list = [l.strip() for l in libs_input.split(",") if l.strip()]162    results = []163    with st.spinner("Connecting to global registries..."):164        for lib in lib_list:165            pypi, npm = get_pypi_data(lib), get_npm_data(lib)166            if pypi: results.append(pypi)167            if npm: results.append(npm)168 169    if results:170        df = pd.DataFrame(results)171        st.success(f"✅ Telemetry acquired for {len(results)} assets.")172        173        # Dashboard Summary174        c1, c2, c3 = st.columns(3)175        c1.metric("Total Assets", len(df))176        c2.metric("Outdated Modules", len(df[df["Health Status"].str.contains("❌")]))177        cve_count = len(df[df["Vulnerabilities"].str.contains("🚨")])178        if cve_count > 0: c3.error(f"🚨 Critical CVEs: {cve_count}")179        else: c3.success("✅ Perimeter Secure")180 181        # --- THE COOL VISUALIZATION (DONUT CHART) ---182        st.markdown("### 📊 Ecosystem Health")183        fig = px.pie(184            df, 185            names='Health Status', 186            hole=0.65,187            color_discrete_sequence=['#00FF41', '#FF4B4B', '#FFC107']188        )189        fig.update_layout(190            paper_bgcolor="rgba(0,0,0,0)", 191            plot_bgcolor="rgba(0,0,0,0)",192            font=dict(family="monospace", color="#E0E0E0")193        )194        st.plotly_chart(fig, use_container_width=True)195 196        # --- THE INTERACTIVE GRID ---197        st.markdown("### 📋 Live Audit Data")198        col_config = {"Source": st.column_config.LinkColumn("Repository", display_text="View ↗")}199        st.data_editor(200            df, 201            use_container_width=True, 202            column_config=col_config,203            hide_index=True,204            disabled=True # Prevents accidental edits205        )206        207        # Export Options208        st.markdown("---")209        colA, colB = st.columns(2)210        with colA: st.download_button("💾 Export CSV", df.to_csv(index=False).encode('utf-8'), "audit.csv", "text/csv", use_container_width=True)211        with colB:212            pdf_bytes = create_pdf(df)213            if pdf_bytes: st.download_button("📄 Generate PDF Report", pdf_bytes, "audit.pdf", "application/pdf", use_container_width=True)214    else:215        st.error("Connection failed or no valid targets identified.")