app.py

import streamlit as st
import pandas as pd
import plotly.express as px
import glob
import re
import os

# ======================
# PAGE CONFIG
# ======================
st.set_page_config(
    page_title="HF KPI Dashboard",
    layout="wide"
)

# ======================
# TAB NAVIGATION
# ======================
tab1, tab2 = st.tabs([
    "📊 Dashboard KPI Monitoring Daily",
    "📈 Dashboard Capacity & Utilization"
])

# ======================
# LOAD DATA (SHARED)
# ======================
@st.cache_data
def load_data():
    files = (
        glob.glob("src/LTE_CELL_DAILY *.csv") +
        glob.glob("src/LTE_CELL_DAILY *.csv.gz")
    )
    if not files:
        return None

    dfs = []
    for f in files:
        tmp = pd.read_csv(f, compression="infer")

        if "DATE_ID" in tmp.columns:
            tmp["DATE_ID"] = pd.to_datetime(
                tmp["DATE_ID"].astype(str).str.strip(),
                format="%m/%d/%Y",
                errors="coerce"
            )

        if tmp["DATE_ID"].isna().all():
            m = re.search(r"(\d{8})", os.path.basename(f))
            if m:
                tmp["DATE_ID"] = pd.to_datetime(m.group(1), format="%Y%m%d")

        dfs.append(tmp)

    df = pd.concat(dfs, ignore_index=True)
    df = df.dropna(subset=["DATE_ID"])
    df["DATE_ID"] = df["DATE_ID"].dt.normalize()
    return df


df_raw = load_data()
if df_raw is None or df_raw.empty:
    st.error("Data CSV / CSV.GZ tidak ditemukan")
    st.stop()

# ======================
# KPI ALIAS
# ======================
COLUMN_ALIAS = {
    "Intra-Frequency Handover Out Success Rate":
        "Intra_Frequency_HO_Out_Success_Rate"
}
for src, dst in COLUMN_ALIAS.items():
    if src in df_raw.columns:
        df_raw[dst] = df_raw[src]

# ======================
# FILTER (GLOBAL)
# ======================
with tab1:
    st.title("DASHBOARDS KPI MONITORING DAILY")
    st.caption("📊 Dashboard By Muhammad Defri")

    st.subheader("Filter")

    selected_sites = st.multiselect(
        "Site ID",
        sorted(df_raw["ERBS"].dropna().unique())
    )

    if not selected_sites:
        st.warning("⚠️ Input Site ID")
        st.stop()

    df = df_raw[df_raw["ERBS"].isin(selected_sites)]

    col_start, col_end = st.columns(2)

    with col_start:
        start = st.date_input(
            "Start Date",
            df["DATE_ID"].min().date(),
            min_value=df_raw["DATE_ID"].min().date(),
            max_value=df_raw["DATE_ID"].max().date()
        )

    with col_end:
        end = st.date_input(
            "End Date",
            df_raw["DATE_ID"].max().date(),
            min_value=start,
            max_value=df_raw["DATE_ID"].max().date()
        )

    df = df[
        (df["DATE_ID"] >= pd.to_datetime(start)) &
        (df["DATE_ID"] < pd.to_datetime(end) + pd.Timedelta(days=1))
    ]

    # ======================
    # MAP BAND & SECTOR (UNCHANGED)
    # ======================
    def map_band(cell):
        if pd.isna(cell): 
            return "OTHER"
        cell = str(cell).upper()
        if re.search(r"(MT|IT)", cell): return "L900"
        if re.search(r"(ML|IL|RL)", cell): return "L1800"
        if re.search(r"(MR|RR|IR)", cell): return "L2100"
        if re.search(r"(ME|MF|MV|IE|IF|IV|VE|VF|VV)", cell): return "L2300"
        return "OTHER"

    def map_sector(cell):
        if pd.isna(cell):
            return "SEC_OTHER"
        cell = str(cell).upper().strip()
        m = re.search(r"(RR|RL|IR|IL)(\d{2})$", cell)
        if m:
            sec = int(m.group(2)) // 10
            return f"SEC{sec}" if sec in (1, 2, 3) else "SEC_OTHER"
        m = re.search(r"(\d{1,2})$", cell)
        if m:
            sec = ((int(m.group(1)) - 1) % 3) + 1
            return f"SEC{sec}"
        return "SEC_OTHER"

    df["BAND"] = df["EUTRANCELLFDD"].apply(map_band)
    df["SECTOR"] = df["EUTRANCELLFDD"].apply(map_sector)

    available_bands = ["L900", "L1800", "L2100", "L2300"]
    selected_bands = st.multiselect("Band", available_bands, default=available_bands)
    df = df[df["BAND"].isin(selected_bands)]

    # ======================
    # GLOBAL DATE AXIS
    # ======================
    X_MIN = pd.to_datetime(start)
    X_MAX = pd.to_datetime(end) + pd.Timedelta(days=1) - pd.Timedelta(seconds=1)
    FULL_DATES = pd.DataFrame({
        "DATE_ID": pd.date_range(start=X_MIN, end=pd.to_datetime(end), freq="D")
    })
    DTICK = "D1"

    # ======================
    # AUTO UNIT
    # ======================
    def auto_unit(d, kpi):
        d = d.copy()
        max_val = d[kpi].max()
        if pd.isna(max_val):
            return d.round(2), "GB"
        if max_val >= 1024:
            d[kpi] = (d[kpi] / 1024).round(2)
            return d, "TB"
        d[kpi] = d[kpi].round(2)
        return d, "GB"

    # ======================
    # SECTOR CHART (UNCHANGED)
    # ======================
    def sector_chart(df, sector, kpi, title, is_pct=False):
        base = df[df["SECTOR"] == sector]
        if base.empty or kpi not in base.columns:
            return None

        frames = []
        for cell in sorted(base["EUTRANCELLFDD"].unique()):
            g = base[base["EUTRANCELLFDD"] == cell][["DATE_ID", kpi]]
            g = FULL_DATES.merge(g, on="DATE_ID", how="left")
            g["EUTRANCELLFDD"] = cell
            frames.append(g)

        d = pd.concat(frames, ignore_index=True)

        if ("TRAFFIC" in kpi.upper()) or ("PAYLOAD" in kpi.upper()):
            d, unit = auto_unit(d, kpi)
            fig = px.area(d, x="DATE_ID", y=kpi, color="EUTRANCELLFDD",
                          title=f"{title} ({unit})")
        else:
            fig = px.line(d, x="DATE_ID", y=kpi,
                          color="EUTRANCELLFDD",
                          title=title, markers=True)

        fig.update_layout(
            height=420,
            dragmode=False,
            xaxis=dict(tickformat="%d-%b", tickangle=-45),
            legend=dict(orientation="h", y=-0.35, x=0.5, xanchor="center")
        )

        fig.update_xaxes(
            range=[X_MIN, X_MAX],
            tickmode="linear",
            dtick=DTICK,
            fixedrange=True
        )

        if is_pct:
            fig.update_yaxes(range=[0, 100])

        return fig

    KPI_LIST = [
        ("Radio_Network_Availability_Rate","Availability",True),
        ("Session_Abnormal_Release","Session Abnormal Release",False),
        ("RRC_Setup_Success_Rate_Service","RRC Setup Success Rate",True),
        ("ERAB_Setup_Success_Rate_All","ERAB Setup Success Rate",True),
        ("Session_Setup_Success_Rate","Session Setup Success Rate",True),
        ("Intra_Frequency_HO_Out_Success_Rate","Intra-Frequency HO Out",True),
        ("inter_freq_HO","Inter Frequency HO",False),
        ("UL_RSSI_dbm","UL RSSI (dBm)",False),
        ("Average_CQI_nonHOME","Average CQI",False),
        ("SE_2","Spectral Efficiency",False),
        ("Total_Traffic_Volume_new","Total Traffic Volume New",False),
        ("CA_PAYLOAD_GB","CA Payload (GB)",False),
    ]

    for kpi, title, is_pct in KPI_LIST:
        st.markdown("---")
        c1, c2, c3 = st.columns(3)
        for col, sec in zip([c1, c2, c3], ["SEC1","SEC2","SEC3"]):
            with col:
                fig = sector_chart(df, sec, kpi, f"{title} - {sec}", is_pct)
                if fig:
                    st.plotly_chart(
                        fig,
                        use_container_width=True,
                        config={"displayModeBar": False, "scrollZoom": False}
                    )

# ======================
# DASHBOARD 2
# ======================
with tab2:
    st.title("DASHBOARD CAPACITY & UTILIZATION")
    st.caption("📈 Clean • Enterprise • Management Ready")

    KPI_LIST_2 = [
        "DL_Resource_Block_Utilizing_Rate",
        "UL_Resource_Block_Utilizing_Rate",
        "User_Uplink_Average_Throughput",
        "DL_PDCP_User_Throughput",
        "Cell_Downlink_Average_Throughput",
        "Cell_Uplink_Average_Throughput",
        "LTE_CSFB_SR",
        "Maximum_User_Number_RrcConn",
        "PRB_MAX",
        "AU_MAX",
        "Max DL Cell Downlink Throughput",
        "Max UL Cell Downlink Throughput"
    ]

    def simple_kpi_chart(df, kpi):
        if kpi not in df.columns:
            return None

        d = df.groupby("DATE_ID")[kpi].mean().reset_index()
        d = FULL_DATES.merge(d, on="DATE_ID", how="left")

        fig = px.area(d, x="DATE_ID", y=kpi, title=kpi)
        fig.update_layout(
            height=420,
            dragmode=False,
            xaxis=dict(tickformat="%d-%b"),
        )
        fig.update_xaxes(
            range=[X_MIN, X_MAX],
            tickmode="linear",
            dtick=DTICK,
            fixedrange=True
        )
        return fig

    for i in range(0, len(KPI_LIST_2), 3):
        cols = st.columns(3)
        for col, kpi in zip(cols, KPI_LIST_2[i:i+3]):
            with col:
                fig = simple_kpi_chart(df, kpi)
                if fig:
                    st.plotly_chart(
                        fig,
                        use_container_width=True,
                        config={"displayModeBar": False, "scrollZoom": False}
                    )