app.py

import streamlit as st
import pandas as pd
import plotly.graph_objects as go
import plotly.express as px
import time

from utils.data_processor import detect_columns, clean_and_tier, get_tier_summary, TIER_META
from utils.embeddings import find_merge_candidates, load_embedder
from utils.llm_client import MODEL_OPTIONS, test_connection, enrich_merge_pairs
from utils.excel_builder import build_excel

# ── Page config ───────────────────────────────────────────────────────────────
st.set_page_config(
    page_title="Blog Audit Tool",
    page_icon="📊",
    layout="wide",
    initial_sidebar_state="expanded",
)

# ── Global CSS ────────────────────────────────────────────────────────────────
st.markdown("""
<style>
  .block-container { padding-top: 1.5rem; }
  .metric-card {
    background: #F8F9FA;
    border: 1px solid #E0E0E0;
    border-radius: 10px;
    padding: 16px 20px;
    text-align: center;
  }
  .metric-card .value { font-size: 2rem; font-weight: 700; }
  .metric-card .label { font-size: 0.82rem; color: #666; margin-top: 2px; }
  .section-header {
    font-size: 1.15rem; font-weight: 700;
    color: #1A1A2E; margin: 1.2rem 0 0.4rem 0;
    padding-bottom: 4px; border-bottom: 2px solid #4F46E5;
    display: inline-block;
  }
  .status-ok  { color: #2ECC71; font-weight: 600; }
  .status-err { color: #E74C3C; font-weight: 600; }
  div[data-testid="stDataFrame"] { border-radius: 8px; }
  .stButton > button {
    border-radius: 8px; font-weight: 600;
  }
  .approve-row { background: #E8F5E9 !important; }
  .reject-row  { background: #FFEBEE !important; }
</style>
""", unsafe_allow_html=True)


# ── Session state init ────────────────────────────────────────────────────────
for key in ["df_clean", "col_map", "merge_pairs", "analysis_done",
            "api_key", "model_id", "model_label", "site_name"]:
    if key not in st.session_state:
        st.session_state[key] = None

if "analysis_done" not in st.session_state or st.session_state["analysis_done"] is None:
    st.session_state["analysis_done"] = False


# ── Sidebar ───────────────────────────────────────────────────────────────────
with st.sidebar:
    st.image("https://via.placeholder.com/200x50/4F46E5/FFFFFF?text=Blog+Audit", width=200)
    st.markdown("---")
    st.markdown("### 📋 Steps")
    st.markdown("""
1. 📁 Upload your CSV  
2. 🔑 Enter API key & pick model  
3. 🚀 Run analysis  
4. 📊 Review results  
5. 💾 Download report  
""")
    st.markdown("---")
    st.markdown("### ℹ️ CSV Format")
    st.markdown("""
Your CSV must have:
- A **URL** column (`Blogs` or similar)
- A **title** column (`H1` or `Title`)
- **4 monthly click** columns (any names)
- A **Total Clicks** column
""")
    st.markdown("---")
    st.caption("Blog Audit Tool — works for any website")


# ══════════════════════════════════════════════════════════════════════════════
# HEADER
# ══════════════════════════════════════════════════════════════════════════════
st.title("📊 Blog Performance Audit Tool")
st.markdown("Upload any website's blog click data CSV to get a full performance audit with AI-powered merge recommendations.")
st.markdown("---")


# ══════════════════════════════════════════════════════════════════════════════
# STEP 1 — UPLOAD CSV
# ══════════════════════════════════════════════════════════════════════════════
st.markdown('<p class="section-header">Step 1 — Upload Your CSV</p>', unsafe_allow_html=True)

col_up1, col_up2 = st.columns([2, 1])
with col_up1:
    uploaded_file = st.file_uploader(
        "Upload your blog clicks CSV",
        type=["csv"],
        help="CSV with URL column, H1/title column, 4 monthly click columns, and Total Clicks column.",
        label_visibility="collapsed",
    )

with col_up2:
    site_name_input = st.text_input(
        "Website / Brand name (for report)",
        value="My Website",
        placeholder="e.g. Edstellar, TechBlog...",
    )

if uploaded_file:
    try:
        df_raw = pd.read_csv(uploaded_file)
        col_map = detect_columns(df_raw)

        # Validate detection
        if not col_map["url_col"]:
            st.error("❌ Could not detect a URL column. Make sure your CSV has a column with URLs (e.g. 'Blogs').")
            st.stop()
        if not col_map["month_cols"]:
            st.error("❌ Could not detect monthly click columns. Ensure your numeric click columns are present.")
            st.stop()

        df_clean = clean_and_tier(df_raw, col_map)
        st.session_state["df_clean"]  = df_clean
        st.session_state["col_map"]   = col_map
        st.session_state["site_name"] = site_name_input

        # Preview
        with st.expander("📋 CSV Preview & Column Detection", expanded=True):
            c1, c2, c3, c4 = st.columns(4)
            c1.metric("Total Blogs", len(df_clean))
            c2.metric("URL Column",   col_map["url_col"])
            c3.metric("Title Column", col_map["title_col"] or "—")
            c4.metric("Month Columns", len(col_map["month_cols"]))

            st.markdown(f"**Detected month columns:** `{'` · `'.join(col_map['month_cols'])}`")
            st.dataframe(df_clean[["url", "title"] + col_map["month_cols"] + ["total_clicks", "tier", "trend"]].head(8),
                         use_container_width=True, height=220)

        st.success(f"✅ File loaded: **{len(df_clean)} blogs** detected.")

    except Exception as e:
        st.error(f"❌ Error reading CSV: {e}")
        st.stop()

st.markdown("---")


# ══════════════════════════════════════════════════════════════════════════════
# STEP 2 — AI CONFIGURATION
# ══════════════════════════════════════════════════════════════════════════════
st.markdown('<p class="section-header">Step 2 — AI Configuration (OpenRouter)</p>', unsafe_allow_html=True)
st.caption("Get a free API key at [openrouter.ai/keys](https://openrouter.ai/keys)")

ai_col1, ai_col2, ai_col3 = st.columns([2, 2, 1])

with ai_col1:
    api_key_input = st.text_input(
        "🔑 OpenRouter API Key",
        type="password",
        placeholder="sk-or-v1-xxxxxxxxxxxxxxxxxxxxxxxx",
        value=st.secrets.get("OPENROUTER_API_KEY", "") if hasattr(st, "secrets") else "",
    )

with ai_col2:
    model_label = st.selectbox(
        "🧠 Select LLM Model",
        options=list(MODEL_OPTIONS.keys()),
        index=0,
    )

with ai_col3:
    st.markdown("<br>", unsafe_allow_html=True)
    test_btn = st.button("🔌 Test Connection", use_container_width=True)

if test_btn:
    if not api_key_input:
        st.warning("⚠️ Please enter your API key first.")
    else:
        with st.spinner("Testing connection..."):
            ok, msg = test_connection(api_key_input, MODEL_OPTIONS[model_label])
        if ok:
            st.session_state["api_key"]     = api_key_input
            st.session_state["model_id"]    = MODEL_OPTIONS[model_label]
            st.session_state["model_label"] = model_label
            st.success(f"✅ {msg} Model: **{model_label}**")
        else:
            st.error(f"❌ {msg}")

# Auto-save key if already entered without test
if api_key_input and not test_btn:
    st.session_state["api_key"]     = api_key_input
    st.session_state["model_id"]    = MODEL_OPTIONS[model_label]
    st.session_state["model_label"] = model_label

st.markdown("---")


# ══════════════════════════════════════════════════════════════════════════════
# STEP 3 — RUN ANALYSIS
# ══════════════════════════════════════════════════════════════════════════════
st.markdown('<p class="section-header">Step 3 — Run Analysis</p>', unsafe_allow_html=True)

run_ready = (
    st.session_state.get("df_clean") is not None
    and st.session_state.get("api_key")
)

if not run_ready:
    st.info("⬆️ Complete Steps 1 and 2 above before running the analysis.")

run_btn = st.button(
    "🚀 Run Full Analysis",
    disabled=not run_ready,
    use_container_width=False,
    type="primary",
)

if run_btn:
    df_clean   = st.session_state["df_clean"]
    col_map    = st.session_state["col_map"]
    api_key    = st.session_state["api_key"]
    model_id   = st.session_state["model_id"]
    model_label = st.session_state["model_label"]

    progress_bar = st.progress(0, text="Starting analysis...")
    status_box   = st.empty()

    # Step A — Embeddings
    status_box.info("🧠 Step 1/3 — Loading embedding model and generating title vectors...")
    progress_bar.progress(10, text="Loading embedding model...")
    with st.spinner("Generating embeddings (this may take 30–60s on first run)..."):
        load_embedder()   # warm up cache
        merge_candidates = find_merge_candidates(df_clean, threshold=0.72, max_weak_clicks=200)
    progress_bar.progress(45, text=f"Found {len(merge_candidates)} merge candidate pairs.")
    status_box.success(f"✅ Embeddings done — {len(merge_candidates)} potential merge pairs found.")
    time.sleep(0.4)

    # Step B — LLM enrichment
    if merge_candidates:
        status_box.info(f"🤖 Step 2/3 — Sending {len(merge_candidates)} pairs to {model_label} for merge reasoning...")
        progress_bar.progress(50, text="LLM analysis in progress...")

        llm_progress = st.empty()
        llm_prog_bar = st.progress(0)

        def llm_cb(done, total):
            pct = int(done / total * 100)
            llm_prog_bar.progress(pct, text=f"LLM: {done}/{total} pairs processed")

        enriched_pairs = enrich_merge_pairs(
            merge_candidates, api_key, model_id,
            batch_size=10, progress_callback=llm_cb,
        )
        llm_prog_bar.empty()
        llm_progress.empty()
    else:
        enriched_pairs = []

    st.session_state["merge_pairs"] = enriched_pairs
    progress_bar.progress(90, text="Building report...")

    # Step C — Done
    status_box.success(f"✅ Analysis complete! {len(enriched_pairs)} merge recommendations generated.")
    progress_bar.progress(100, text="✅ Analysis complete!")
    st.session_state["analysis_done"] = True
    time.sleep(0.5)
    st.rerun()

st.markdown("---")


# ══════════════════════════════════════════════════════════════════════════════
# STEP 4 — RESULTS DASHBOARD
# ══════════════════════════════════════════════════════════════════════════════
if st.session_state["analysis_done"] and st.session_state["df_clean"] is not None:
    df_clean    = st.session_state["df_clean"]
    col_map     = st.session_state["col_map"]
    month_cols  = col_map["month_cols"]
    merge_pairs = st.session_state.get("merge_pairs", [])
    site_name   = st.session_state.get("site_name", "Website")

    st.markdown('<p class="section-header">Step 4 — Results Dashboard</p>', unsafe_allow_html=True)

    counts = get_tier_summary(df_clean)

    # ── KPI metric row ──────────────────────────────────────────────────────
    m_cols = st.columns(7)
    kpis = [
        ("Total Blogs",       len(df_clean),              "#4F46E5"),
        ("🔴 Take Down",      counts["TAKEDOWN_ZERO"] + counts["TAKEDOWN_LOW"], "#E74C3C"),
        ("🔵 Merge",          len(merge_pairs),            "#2980B9"),
        ("🟡 Monitor",        counts["MONITOR"],           "#F39C12"),
        ("✅ Performing",     counts["PERFORMING"],         "#27AE60"),
        ("💪 Strong",         counts["STRONG"],            "#1ABC9C"),
        ("🏆 Top",            counts["TOP"],               "#9B59B6"),
    ]
    for col_w, (label, val, color) in zip(m_cols, kpis):
        col_w.markdown(
            f'<div class="metric-card">'
            f'<div class="value" style="color:{color}">{val}</div>'
            f'<div class="label">{label}</div>'
            f'</div>',
            unsafe_allow_html=True,
        )

    st.markdown("<br>", unsafe_allow_html=True)

    # ── Charts ──────────────────────────────────────────────────────────────
    chart_col1, chart_col2 = st.columns(2)

    with chart_col1:
        st.markdown("**Click Distribution by Tier**")
        tier_labels = [TIER_META[t]["label"] for t in counts]
        tier_values = list(counts.values())
        tier_colors = [TIER_META[t]["color"] for t in counts]
        fig_pie = go.Figure(go.Pie(
            labels=tier_labels, values=tier_values,
            marker_colors=tier_colors,
            hole=0.45, textinfo="label+percent",
        ))
        fig_pie.update_layout(
            showlegend=False, margin=dict(t=10, b=10, l=10, r=10), height=280
        )
        st.plotly_chart(fig_pie, use_container_width=True)

    with chart_col2:
        st.markdown("**Monthly Click Trend (Top 5 Blogs)**")
        top5 = df_clean.nlargest(5, "total_clicks")
        fig_line = go.Figure()
        for _, row in top5.iterrows():
            short_title = row["title"][:40] + "…" if len(row["title"]) > 40 else row["title"]
            fig_line.add_trace(go.Scatter(
                x=month_cols,
                y=[row[m] for m in month_cols],
                mode="lines+markers",
                name=short_title,
            ))
        fig_line.update_layout(
            margin=dict(t=10, b=10, l=10, r=10), height=280,
            legend=dict(font=dict(size=9)),
            xaxis_title="Month", yaxis_title="Clicks",
        )
        st.plotly_chart(fig_line, use_container_width=True)

    st.markdown("---")

    # ── Result Tabs ─────────────────────────────────────────────────────────
    tab_td, tab_merge, tab_mon, tab_perf, tab_str, tab_top = st.tabs([
        f"🔴 Take Down ({counts['TAKEDOWN_ZERO'] + counts['TAKEDOWN_LOW']})",
        f"🔵 Merge ({len(merge_pairs)})",
        f"🟡 Monitor ({counts['MONITOR']})",
        f"✅ Performing ({counts['PERFORMING']})",
        f"💪 Strong ({counts['STRONG']})",
        f"🏆 Top ({counts['TOP']})",
    ])

    display_cols_base = ["url", "title"] + month_cols + ["total_clicks", "trend", "action"]

    # ── Take Down tab ────────────────────────────────────────────────────────
    with tab_td:
        st.markdown("Blogs with zero or negligible traffic. Recommended for removal or merge.")
        td_df = df_clean[df_clean["tier"].isin(["TAKEDOWN_ZERO", "TAKEDOWN_LOW"])].sort_values("total_clicks")
        if len(td_df):
            # Search filter
            search_td = st.text_input("🔍 Filter by URL or title", key="search_td", placeholder="Type to filter...")
            if search_td:
                mask = td_df["url"].str.contains(search_td, case=False, na=False) | \
                       td_df["title"].str.contains(search_td, case=False, na=False)
                td_df = td_df[mask]
            st.dataframe(td_df[display_cols_base], use_container_width=True, height=400)
            st.caption(f"{len(td_df)} blogs")
        else:
            st.success("🎉 No take-down candidates found!")

    # ── Merge tab ────────────────────────────────────────────────────────────
    with tab_merge:
        st.markdown("AI-detected topically overlapping blogs. Review and approve/reject each suggestion.")

        if merge_pairs:
            # Review interface
            st.markdown("**✅ = Keep in report &nbsp;&nbsp; ❌ = Reject suggestion**")

            # Init approval state
            if "merge_approvals" not in st.session_state:
                st.session_state["merge_approvals"] = {i: True for i in range(len(merge_pairs))}

            col_filter, col_actions = st.columns([3, 1])
            with col_filter:
                search_merge = st.text_input("🔍 Filter merges", key="search_merge", placeholder="Filter by title or cluster...")
            with col_actions:
                st.markdown("<br>", unsafe_allow_html=True)
                if st.button("✅ Approve All", key="approve_all"):
                    for i in range(len(merge_pairs)):
                        st.session_state["merge_approvals"][i] = True
                    st.rerun()

            # Display table
            merge_display = []
            for i, p in enumerate(merge_pairs):
                if search_merge and search_merge.lower() not in p.get("weak_title","").lower() \
                        and search_merge.lower() not in p.get("strong_title","").lower() \
                        and search_merge.lower() not in p.get("topic_cluster","").lower():
                    continue
                approved = st.session_state["merge_approvals"].get(i, True)
                merge_display.append({
                    "#": i + 1,
                    "Weak Blog (Merge FROM)":  p.get("weak_title", "")[:60],
                    "Weak Clicks":             p.get("weak_clicks", 0),
                    "Strong Blog (Merge INTO)":p.get("strong_title", "")[:60],
                    "Strong Clicks":           p.get("strong_clicks", 0),
                    "Similarity":              p.get("similarity", ""),
                    "Topic Cluster":           p.get("topic_cluster", ""),
                    "AI Merge Reason":         p.get("merge_reason", "")[:100],
                    "Approved":                "✅ Yes" if approved else "❌ Rejected",
                })

            merge_df_display = pd.DataFrame(merge_display)
            st.dataframe(merge_df_display, use_container_width=True, height=380)

            # Approve/Reject individual rows
            st.markdown("**Toggle individual suggestions:**")
            toggle_cols = st.columns(5)
            for i, p in enumerate(merge_pairs[:20]):  # show first 20 toggles
                col_idx = i % 5
                with toggle_cols[col_idx]:
                    current = st.session_state["merge_approvals"].get(i, True)
                    label = f"#{i+1}: {p.get('weak_title','')[:25]}…"
                    new_val = st.checkbox(label, value=current, key=f"merge_toggle_{i}")
                    st.session_state["merge_approvals"][i] = new_val

            approved_count = sum(st.session_state["merge_approvals"].values())
            st.caption(f"✅ {approved_count} approved · ❌ {len(merge_pairs) - approved_count} rejected")
        else:
            st.info("No merge candidates detected above the similarity threshold.")

    # ── Generic tier tab helper ──────────────────────────────────────────────
    def render_tier_tab(tier_key, tab_obj, search_key):
        with tab_obj:
            subset = df_clean[df_clean["tier"] == tier_key].sort_values("total_clicks", ascending=False)
            search = st.text_input("🔍 Filter", key=search_key, placeholder="Filter by URL or title...")
            if search:
                mask = subset["url"].str.contains(search, case=False, na=False) | \
                       subset["title"].str.contains(search, case=False, na=False)
                subset = subset[mask]
            trend_filter = st.selectbox(
                "Trend filter", ["All", "📈 Growing", "📉 Declining", "➡️ Stable"],
                key=f"{search_key}_trend"
            )
            if trend_filter != "All":
                subset = subset[subset["trend"] == trend_filter]
            st.dataframe(subset[display_cols_base], use_container_width=True, height=400)
            st.caption(f"{len(subset)} blogs")

    render_tier_tab("MONITOR",    tab_mon,  "search_mon")
    render_tier_tab("PERFORMING", tab_perf, "search_perf")
    render_tier_tab("STRONG",     tab_str,  "search_str")
    render_tier_tab("TOP",        tab_top,  "search_top")

    st.markdown("---")

    # ══════════════════════════════════════════════════════════════════════════
    # STEP 5 — DOWNLOAD REPORT
    # ══════════════════════════════════════════════════════════════════════════
    st.markdown('<p class="section-header">Step 5 — Download Excel Report</p>', unsafe_allow_html=True)

    # Apply approvals to merge pairs before export
    approvals = st.session_state.get("merge_approvals", {})
    approved_pairs = [p for i, p in enumerate(merge_pairs) if approvals.get(i, True)]

    dl_col1, dl_col2 = st.columns([2, 3])
    with dl_col1:
        with st.spinner("Building Excel report..."):
            excel_bytes = build_excel(
                df_clean, approved_pairs, month_cols,
                site_name=st.session_state.get("site_name", "Website"),
            )
        filename = f"{site_name.replace(' ', '_').lower()}_blog_audit_report.xlsx"
        st.download_button(
            label="⬇️ Download 7-Tab Excel Report",
            data=excel_bytes,
            file_name=filename,
            mime="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
            type="primary",
            use_container_width=True,
        )

    with dl_col2:
        st.markdown(f"""
**Report includes 7 tabs:**
- 📊 Summary Dashboard
- 🔴 Take Down ({counts['TAKEDOWN_ZERO'] + counts['TAKEDOWN_LOW']} blogs)
- 🔵 Merge Recommendations ({len(approved_pairs)} approved pairs)
- 🟡 Monitor · ✅ Performing · 💪 Strong · 🏆 Top Performers
        """)
        st.caption(f"Model used: {st.session_state.get('model_label','—')}")

elif not st.session_state["analysis_done"]:
    st.markdown('<p class="section-header">Step 4 — Results Dashboard</p>', unsafe_allow_html=True)
    st.info("⬆️ Complete Steps 1–3 above to see results here.")