app.py

from fastapi import FastAPI, HTTPException, Query
from functools import lru_cache
import pandas as pd
import os
from typing import Dict, Any

app = FastAPI(title="Dashboard Akademik API")

# Lokasi CSV dapat dioverride di HuggingFace: Settings → Variables → CSV_PATH
CSV_PATH = os.getenv("CSV_PATH", "generated_dummy_data.csv")

# Konversi nilai huruf ke skala 4.0
MAP_NILAI = {
    "A": 4.0,
    "AB": 3.5,
    "B": 3.0,
    "BC": 2.5,
    "C": 2.0,
    "D": 1.0,
    "E": 0.0
}

# Pastikan CSV memiliki kolom wajib
def _ensure_columns(df: pd.DataFrame):
    required = {
        "kode_mhs", "nama_prodi", "id_smt", "kode_mk", "nama_mk",
        "RMK", "sks", "nilai_akhir", "nilai_huruf", "Tahun angkatan",
        "Semester_sekarang", "Deskripsi Matkul"
    }
    missing = required - set(df.columns)
    if missing:
        raise ValueError(f"CSV missing required columns: {missing}")

# Helper untuk normalisasi nilai_huruf sel yang mungkin berisi list/Series/None
def _to_simple_string(val):
    try:
        # pandas NA
        if pd.isna(val):
            return ""
    except Exception:
        # if val is unhashable, continue
        pass

    # If the cell contains a pandas Series or ndarray-like
    if isinstance(val, pd.Series):
        # take first non-null element if possible
        non_null = val.dropna()
        if not non_null.empty:
            return str(non_null.iloc[0])
        if not val.empty:
            return str(val.iloc[0])
        return ""
    if isinstance(val, (list, tuple)):
        if len(val) == 0:
            return ""
        return str(val[0])
    # dict or other object: convert to string
    try:
        return str(val)
    except Exception:
        return ""

# Loader CSV dengan cache
@lru_cache(maxsize=1)
def load_data_cached() -> pd.DataFrame:
    if not os.path.exists(CSV_PATH):
        raise FileNotFoundError(f"CSV not found at: {CSV_PATH}")

    # Read CSV (let pandas infer types). If encoding issues happen, set encoding='utf-8'
    df = pd.read_csv(CSV_PATH)
    # Normalize column names: strip whitespace
    df.columns = [c.strip() for c in df.columns]

    _ensure_columns(df)

    # Normalize numeric columns
    df["sks"] = pd.to_numeric(df["sks"], errors="coerce").fillna(0).astype(int)
    df["id_smt"] = pd.to_numeric(df["id_smt"], errors="coerce").fillna(0).astype(int)

    # SAFELY normalize nilai_huruf (handle lists, Series, NaN, ints, etc.)
    df["nilai_huruf"] = df["nilai_huruf"].apply(_to_simple_string).astype(str).str.strip()
    # Map A, AB, B, ... -> numeric
    df["nilai_numerik"] = df["nilai_huruf"].map(MAP_NILAI)

    # Jika huruf invalid → fallback ke nilai_akhir (0–100 -> skala 0-4)
    def fallback_numeric(row):
        # jika sudah mapped, kembalikan
        if pd.notna(row["nilai_numerik"]):
            return row["nilai_numerik"]
        try:
            v = float(row.get("nilai_akhir", 0))
            if v >= 86: return 4.0
            if v >= 76: return 3.5
            if v >= 66: return 3.0
            if v >= 61: return 2.5
            if v >= 56: return 2.0
            if v >= 41: return 1.0
            return 0.0
        except Exception:
            return 0.0

    df["nilai_numerik"] = df.apply(fallback_numeric, axis=1)

    return df

# Pakai nilai terakhir per (mhs, matkul)
def get_final_records(df: pd.DataFrame) -> pd.DataFrame:
    df_sorted = df.sort_values(["kode_mhs", "kode_mk", "id_smt"])
    return df_sorted.groupby(["kode_mhs", "kode_mk"], as_index=False).last()

# ----------------------------------------------------
# ENDPOINT 1 — TOTAL MAHASISWA
# ----------------------------------------------------
@app.get("/jumlah_mahasiswa")
def jumlah_mahasiswa(reload: bool = Query(False, description="reload CSV cache")):
    if reload:
        load_data_cached.cache_clear()
    try:
        df = load_data_cached()
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

    total = int(df["kode_mhs"].nunique())
    return {"total_mahasiswa": total}

# ----------------------------------------------------
# ENDPOINT 2 — JUMLAH PER ANGKATAN
# ----------------------------------------------------
@app.get("/jumlah_per_angkatan")
def jumlah_per_angkatan(reload: bool = Query(False, description="reload CSV cache")):
    if reload:
        load_data_cached.cache_clear()
    try:
        df = load_data_cached()
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

    per_ang = df.groupby("Tahun angkatan")["kode_mhs"].nunique().sort_index().to_dict()
    per_ang = {str(k): int(v) for k, v in per_ang.items()}
    return {"mahasiswa_per_angkatan": per_ang}

# ----------------------------------------------------
# ENDPOINT 3 — MAHASISWA ELIGIBLE TA
# ----------------------------------------------------
@app.get("/eligible_ta")
def eligible_ta(reload: bool = Query(False, description="reload CSV cache"),
                min_sks: int = Query(110, description="threshold SKS untuk eligible (default:110)")):
    if reload:
        load_data_cached.cache_clear()
    try:
        df = load_data_cached()
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

    final = get_final_records(df)
    sks_per_mhs = final.groupby("kode_mhs")["sks"].sum()
    eligible = sks_per_mhs[sks_per_mhs > min_sks]

    data = [{"kode_mhs": m, "total_sks": int(sks_per_mhs[m])} for m in eligible.index]
    return {"jumlah_eligible": len(data), "daftar": data}

# ----------------------------------------------------
# ENDPOINT 4 — IPK RATA-RATA
# ----------------------------------------------------
@app.get("/ipk_rata_rata")
def ipk_rata_rata(reload: bool = Query(False, description="reload CSV cache")):
    if reload:
        load_data_cached.cache_clear()
    try:
        df = load_data_cached()
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

    final = get_final_records(df)
    final["total_bobot"] = final["sks"] * final["nilai_numerik"]

    grp = final.groupby(["kode_mhs", "id_smt"]).agg(
        total_bobot=("total_bobot", "sum"),
        total_sks=("sks", "sum")
    ).reset_index()

    grp["ips"] = grp.apply(lambda r: (r["total_bobot"] / r["total_sks"]) if r["total_sks"] > 0 else 0.0, axis=1)
    ipk_series = grp.groupby("kode_mhs")["ips"].mean()

    if ipk_series.empty:
        return {"ipk_rata_rata": 0.0, "ipk_quartiles": {}}

    mean_ipk = float(round(ipk_series.mean(), 3))
    q = ipk_series.quantile([0.25, 0.5, 0.75]).to_dict()
    q = {str(k): float(v) for k, v in q.items()}

    return {"ipk_rata_rata": mean_ipk, "ipk_quartiles": q}

# ----------------------------------------------------
# ENDPOINT 5 — DASHBOARD SUMMARY (SEMUA RINGKASAN)
# ----------------------------------------------------
@app.get("/dashboard_summary")
def dashboard_summary(reload: bool = Query(False, description="reload CSV cache")):
    if reload:
        load_data_cached.cache_clear()
    try:
        df = load_data_cached()
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

    final = get_final_records(df)
    total_mhs = int(df["kode_mhs"].nunique())

    per_ang = df.groupby("Tahun angkatan")["kode_mhs"].nunique().to_dict()
    per_ang = {str(k): int(v) for k, v in per_ang.items()}

    sks_per_mhs = final.groupby("kode_mhs")["sks"].sum()
    eligible = sks_per_mhs[sks_per_mhs > 110]
    eligible_list = [{"kode_mhs": m, "total_sks": int(sks_per_mhs[m])} for m in eligible.index]

    # Hitung IPK
    final["total_bobot"] = final["sks"] * final["nilai_numerik"]
    grp = final.groupby(["kode_mhs", "id_smt"]).agg(
        total_bobot=("total_bobot", "sum"),
        total_sks=("sks", "sum")
    ).reset_index()
    grp["ips"] = grp.apply(lambda r: (r["total_bobot"] / r["total_sks"]) if r["total_sks"] > 0 else 0.0, axis=1)
    ipk_series = grp.groupby("kode_mhs")["ips"].mean()
    mean_ipk = float(round(ipk_series.mean(), 3)) if not ipk_series.empty else 0.0

    return {
        "total_mahasiswa": total_mhs,
        "mahasiswa_per_angkatan": per_ang,
        "eligible_ta": {
            "jumlah": len(eligible_list),
            "daftar": eligible_list
        },
        "ipk": {
            "rata_rata_ipk": mean_ipk
        }
    }

# ----------------------------------------------------
# ADDITIONAL ENDPOINT — RATA-RATA SKS
# ----------------------------------------------------
@app.get("/rata_sks")
def rata_sks(reload: bool = Query(False, description="reload CSV cache")):
    if reload:
        load_data_cached.cache_clear()
    try:
        df = load_data_cached()
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

    final = get_final_records(df)
    sks_per_mhs = final.groupby("kode_mhs")["sks"].sum()
    if sks_per_mhs.empty:
        return {"rata_rata_sks": 0.0}
    rata2 = float(round(sks_per_mhs.mean(), 2))
    return {"rata_rata_sks": rata2}

# ----------------------------------------------------
# ADDITIONAL ENDPOINT — IPS TREND (per angkatan per semester)
# ----------------------------------------------------
@app.get("/ips_trend")
def ips_trend(reload: bool = Query(False, description="reload CSV cache")):
    if reload:
        load_data_cached.cache_clear()
    try:
        df = load_data_cached()
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

    final = get_final_records(df)
    final["total_bobot"] = final["sks"] * final["nilai_numerik"]

    grp = final.groupby(["kode_mhs", "id_smt", "Tahun angkatan"]).agg(
        total_bobot=("total_bobot", "sum"),
        total_sks=("sks", "sum")
    ).reset_index()

    grp["ips"] = grp.apply(lambda r: (r["total_bobot"] / r["total_sks"]) if r["total_sks"] > 0 else 0.0, axis=1)

    result = grp.groupby(["Tahun angkatan", "id_smt"])["ips"].mean().round(3).reset_index()

    output = {}
    for _, row in result.iterrows():
        angkatan = str(int(row["Tahun angkatan"]))
        semester = str(int(row["id_smt"]))
        output.setdefault(angkatan, {})[semester] = float(row["ips"])
    return output

# ----------------------------------------------------
# ADDITIONAL ENDPOINT — POPULASI (jumlah mahasiswa per angkatan)
# ----------------------------------------------------
@app.get("/populasi")
def populasi(reload: bool = Query(False, description="reload CSV cache")):
    if reload:
        load_data_cached.cache_clear()
    try:
        df = load_data_cached()
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

    per_ang = df.groupby("Tahun angkatan")["kode_mhs"].nunique().sort_index().to_dict()
    per_ang = {str(k): int(v) for k, v in per_ang.items()}
    return {"populasi": per_ang}

# ----------------------------------------------------
# ENDPOINT — RELOAD CSV
# ----------------------------------------------------
@app.post("/reload_data")
def reload_data():
    load_data_cached.cache_clear()
    try:
        _ = load_data_cached()
    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Reload failed: {e}")
    return {"status": "reloaded"}



# ----------------------------------------------------
# ENDPOINT 7 — TREN IPS PER ANGKATAN / SEMESTER
# ----------------------------------------------------
@app.get("/tren_ips")
def tren_ips(reload: bool = False):
    if reload:
        load_data_cached.cache_clear()

    try:
        df = load_data_cached()
    except Exception as e:
        raise HTTPException(500, str(e))

    final = get_final_records(df)
    final["total_bobot"] = final["sks"] * final["nilai_numerik"]

    # hitung IPS per (angkatan, semester)
    grouped = final.groupby(["Tahun angkatan", "id_smt"]).agg(
        total_bobot=("total_bobot", "sum"),
        total_sks=("sks", "sum")
    ).reset_index()

    grouped["ips"] = grouped.apply(
        lambda r: (r["total_bobot"] / r["total_sks"]) if r["total_sks"] > 0 else 0.0,
        axis=1
    )

    result = {}
    for angkatan, d in grouped.groupby("Tahun angkatan"):
        d_sorted = d.sort_values("id_smt")
        result[str(angkatan)] = d_sorted["ips"].round(3).tolist()

    return result


# ----------------------------------------------------
# ENDPOINT 8 — DISTRIBUSI POPULASI PER ANGKATAN
# ----------------------------------------------------
@app.get("/distribusi_populasi")
def distribusi_populasi(reload: bool = False, min_sks: int = 110):
    if reload:
        load_data_cached.cache_clear()

    try:
        df = load_data_cached()
    except Exception as e:
        raise HTTPException(500, str(e))

    final = get_final_records(df)

    # total mahasiswa per angkatan
    total_mhs = df.groupby("Tahun angkatan")["kode_mhs"].nunique()

    # eligible per angkatan
    sks_per_mhs = final.groupby("kode_mhs")["sks"].sum()
    eligible = sks_per_mhs[sks_per_mhs > min_sks]

    angkatan_map = df.set_index("kode_mhs")["Tahun angkatan"].to_dict()

    eligible_count = {}
    for mhs in eligible.index:
        ang = angkatan_map.get(mhs)
        if ang is not None:
            eligible_count.setdefault(ang, 0)
            eligible_count[ang] += 1

    # bentuk response
    final_result = {}
    for angkatan in sorted(total_mhs.index):
        final_result[str(angkatan)] = {
            "total": int(total_mhs[angkatan]),
            "eligible": int(eligible_count.get(angkatan, 0))
        }

    return final_result