"""VynFi × pm4py: Interactive Process Mining Demo"""

import streamlit as st
import pandas as pd
from collections import Counter

st.set_page_config(page_title="VynFi Process Mining", page_icon="📊", layout="wide")
st.title("📊 VynFi × pm4py: Process Mining Demo")
st.caption("Synthetic supply-chain event log from [VynFi](https://vynfi.com)")


@st.cache_data
def load_data():
    from datasets import load_dataset
    ds = load_dataset("VynFi/vynfi-supply-chain-ocel", "events", split="train")
    df = ds.to_pandas()
    df["timestamp"] = pd.to_datetime(df["timestamp"], errors="coerce")
    # Drop rows with NaT timestamps (pm4py can't handle them)
    df = df.dropna(subset=["timestamp"])
    # Rename for pm4py — use safe names without colons for display
    df = df.rename(columns={
        "case_id": "case_id_pm",
        "activity_name": "activity",
        "timestamp": "ts",
    })
    return df


df = load_data()

st.sidebar.header("Dataset")
st.sidebar.metric("Events", f"{len(df):,}")
st.sidebar.metric("Activities", df["activity"].nunique())
st.sidebar.metric("Cases", df["case_id_pm"].nunique())

tab1, tab2, tab3, tab4 = st.tabs(["Process Model", "Variants", "Statistics", "Raw Data"])

with tab1:
    st.subheader("Directly-Follows Graph")
    try:
        import pm4py
        # Convert to pm4py format
        pm_df = df.rename(columns={
            "case_id_pm": "case:concept:name",
            "activity": "concept:name",
            "ts": "time:timestamp",
        })
        event_log = pm4py.convert_to_event_log(pm_df)
        dfg, sa, ea = pm4py.discover_dfg(event_log)
        from pm4py.visualization.dfg import visualizer as dfg_vis
        gviz = dfg_vis.apply(dfg, log=event_log, variant=dfg_vis.Variants.FREQUENCY,
            parameters={
                dfg_vis.Variants.FREQUENCY.value.Parameters.START_ACTIVITIES: sa,
                dfg_vis.Variants.FREQUENCY.value.Parameters.END_ACTIVITIES: ea,
                dfg_vis.Variants.FREQUENCY.value.Parameters.FORMAT: "svg",
            })
        st.image(dfg_vis.serialize(gviz).decode("utf-8"), use_container_width=True)
    except Exception as e:
        st.warning(f"Could not render DFG: {e}")
        st.info("Try the Variants or Statistics tabs instead.")

with tab2:
    st.subheader("Process Variants")
    variants = {}
    for cid, grp in df.sort_values("ts").groupby("case_id_pm"):
        variants[cid] = tuple(grp["activity"].tolist())
    vc = Counter(variants.values())
    total = len(variants)
    st.metric("Unique Variants", len(vc))
    rows = [{"Trace": " → ".join(t), "Count": c, "Frequency": f"{c/total*100:.1f}%"}
            for t, c in vc.most_common(20)]
    st.dataframe(pd.DataFrame(rows), use_container_width=True, hide_index=True)
    if vc:
        hp = vc.most_common(1)[0]
        st.info(f"**Happy path**: {' → '.join(hp[0])} ({hp[1]} cases, {hp[1]/total*100:.1f}%)")

with tab3:
    st.subheader("Activity Frequency")
    ac = df["activity"].value_counts().reset_index()
    ac.columns = ["Activity", "Count"]
    st.bar_chart(ac, x="Activity", y="Count")

    st.subheader("Events Over Time")
    if "ts" in df.columns:
        weekly = df.set_index("ts").resample("W").size().reset_index()
        weekly.columns = ["Week", "Events"]
        st.line_chart(weekly, x="Week", y="Events")

with tab4:
    st.subheader("Raw Event Data")
    st.dataframe(df.head(200), use_container_width=True)

st.divider()
st.caption("[VynFi](https://vynfi.com) · [pm4py](https://pm4py.fit.fraunhofer.de/) · [Dataset](https://huggingface.co/datasets/VynFi/vynfi-supply-chain-ocel)")