Update app.py

app.py

@@ -1,41 +1,23 @@
 # -*- coding: utf-8 -*-
 """
 IPLM 2025 — Final (Target Sampel 33.88% per Jenis) — TANPA Kinerja Relatif / Percentile
-
-KONSEP / DOKUMENTASI
-
-A. Skor ABSOLUT (untuk akuntabilitas)
-1) Indeks_Dasar_0_100 (level entitas):
-   Yeo-Johnson (per indikator) -> MinMax global (0–1) -> sub-indeks -> dimensi -> indeks
-   dim_kepatuhan = mean(sub_koleksi, sub_sdm)
-   dim_kinerja   = mean(sub_pelayanan, sub_pengelolaan)
-   Indeks_Dasar_0_100 = 100 * (W_KEPATUHAN*dim_kepatuhan + W_KINERJA*dim_kinerja)
-
-2) Penyesuaian kecukupan sampel (TARGET 33.88% per jenis) pada level wilayah×jenis:
-   target_total_33_88_jenis = pop_total_jenis * TARGET_RATIO
-   faktor_penyesuaian_jenis = min(n_jenis / target_total_33_88_jenis, 1.0)
-   Indeks_Final_Agregat_0_100 = Indeks_Dasar_Agregat_0_100 * faktor_penyesuaian_jenis
-
-3) Agregat wilayah keseluruhan = rata-rata 3 jenis (FIX, missing dianggap 0 dan tetap dibagi 3):
-   Indeks_Dasar_Agregat_0_100(keseluruhan) = (dasar_sekolah + dasar_umum + dasar_khusus)/3
-   Indeks_Final_Wilayah_0_100(keseluruhan) = (final_sekolah + final_umum + final_khusus)/3
-
-B. UI
-- KPI Dashboard: hanya 2 kartu (Indeks Final & Indeks Dasar)
-- Tanpa kartu Coverage
-- Bell curve: menampilkan Indeks_Dasar_0_100 per entitas per jenis, hover menampilkan nama perpustakaan
-
-C. UPDATE PERMINTAAN ANDA (LLM -> WORD tabel seperti gambar)
-- LLM tidak lagi menulis narasi panjang.
-- LLM mengisi kolom "Interpretasi" dan "Rekomendasi" pada tabel Word:
-  No | Dimensi | Nilai | Interpretasi | Rekomendasi
-- Nilai diisi dari hasil hitung (angka 0–100).
+UPDATE UTAMA (sesuai instruksi Anda):
+- LLM TIDAK lagi menulis narasi 3 paragraf.
+- LLM sekarang mengisi kolom "Interpretasi" dan "Rekomendasi" untuk tabel:
+  (Kepatuhan, Koleksi, Tenaga, Kinerja, Pelayanan, Penyelenggaraan/Pengelolaan, Nilai IPLM)
+- Output tabel tersebut dibuat dalam format MS Word (.docx) dan bisa diunduh dari aplikasi.
+- Nilai (kolom "Nilai") diambil APA ADANYA dari hasil perhitungan aplikasi (bukan dari LLM).
+
+Catatan:
+- Script ini tetap mempertahankan seluruh pipeline perhitungan Anda (Yeo-Johnson + MinMax + agregasi + penyesuaian 33.88%).
+- Saya hanya "mengganti fungsi LLM + Word report" agar menghasilkan tabel interpretasi & rekomendasi seperti contoh.
 """
 
 import os
 import re
 import time
 import json
+import math
 import tempfile
 from pathlib import Path
 
@@ -45,15 +27,18 @@ import pandas as pd
 import plotly.graph_objects as go
 from sklearn.preprocessing import PowerTransformer
 
-# python-docx opsional
+# python-docx (wajib kalau mau Word)
 DOCX_AVAILABLE = True
 try:
     from docx import Document
+    from docx.shared import Pt, Inches
+    from docx.oxml import OxmlElement
+    from docx.oxml.ns import qn
 except Exception:
     DOCX_AVAILABLE = False
     Document = None
 
-# huggingface client opsional
+# huggingface client (opsional)
 HF_AVAILABLE = True
 try:
     from huggingface_hub import InferenceClient
@@ -368,10 +353,20 @@ def _parse_pop_khusus(path_xlsx: str) -> pd.DataFrame:
         if mm.startswith("PROVINSI "):
             prov_name = mm.replace("PROVINSI", "").strip()
             current_prov = prov_name
-            rows.append({"LEVEL": "PROV", "Provinsi_Label": f"PROVINSI {prov_name}", "Kab_Kota_Label": None, "Pop_Total_Jenis": pval})
+            rows.append({
+                "LEVEL": "PROV",
+                "Provinsi_Label": f"PROVINSI {prov_name}",
+                "Kab_Kota_Label": None,
+                "Pop_Total_Jenis": pval,
+            })
             continue
 
-        rows.append({"LEVEL": "KAB", "Provinsi_Label": f"PROVINSI {current_prov}" if current_prov else None, "Kab_Kota_Label": mm, "Pop_Total_Jenis": pval})
+        rows.append({
+            "LEVEL": "KAB",
+            "Provinsi_Label": f"PROVINSI {current_prov}" if current_prov else None,
+            "Kab_Kota_Label": mm,
+            "Pop_Total_Jenis": pval,
+        })
 
     pop = pd.DataFrame(rows)
     if pop.empty:
@@ -383,13 +378,17 @@ def _parse_pop_khusus(path_xlsx: str) -> pd.DataFrame:
     return pop
 
 def load_default_files(force=False):
-    key = (DATA_FILE, POP_KAB, POP_PROV, POP_KHUSUS, _mtime(DATA_FILE), _mtime(POP_KAB), _mtime(POP_PROV), _mtime(POP_KHUSUS))
+    key = (
+        DATA_FILE, POP_KAB, POP_PROV, POP_KHUSUS,
+        _mtime(DATA_FILE), _mtime(POP_KAB), _mtime(POP_PROV), _mtime(POP_KHUSUS)
+    )
+
     if (not force) and _CACHE["key"] == key and _CACHE["df_all"] is not None:
         return _CACHE["df_all"], _CACHE["df_raw"], _CACHE["pop_kab"], _CACHE["pop_prov"], _CACHE["pop_khusus"], _CACHE["meta"], _CACHE["info"]
 
     for p, label in [(DATA_FILE, "DM"), (POP_KAB, "POP_KAB"), (POP_PROV, "POP_PROV"), (POP_KHUSUS, "POP_KHUSUS")]:
         if not Path(p).exists():
-            info = f"File {label} tidak ditemukan: `{p}`"
+            info = f"File tidak ditemukan ({label}): {p}"
             _CACHE.update({"key": key, "df_all": None, "df_raw": None, "pop_kab": None, "pop_prov": None, "pop_khusus": None, "meta": {}, "info": info})
             return None, None, None, None, None, {}, info
 
@@ -438,6 +437,7 @@ def load_default_files(force=False):
     df_raw = df_raw.drop_duplicates(subset=["_row_key"], keep="first").copy()
     after = len(df_raw)
 
+    # POP KAB
     pk = pd.read_excel(POP_KAB)
     c_kab  = pick_col(pk, ["KABUPATEN_KOTA","Kab/Kota","Kabupaten/Kota","KAB/KOTA","Kabupaten_Kota","kab_kota","kabupaten_kota"])
     c_prov = pick_col(pk, ["PROVINSI","Provinsi","provinsi"])
@@ -452,6 +452,7 @@ def load_default_files(force=False):
     pop_kab["kab_key"] = pop_kab["Kab_Kota_Label"].apply(norm_kab_label)
     pop_kab = pop_kab.groupby("kab_key", as_index=False).first()
 
+    # POP PROV
     pp = pd.read_excel(POP_PROV)
     c_pr = pick_col(pp, ["Provinsi","PROVINSI","provinsi","Propinsi","PROPINSI","propinsi"])
     if c_pr is None:
@@ -464,6 +465,7 @@ def load_default_files(force=False):
     pop_prov["prov_key"] = pop_prov["Provinsi_Label"].apply(norm_prov_label)
     pop_prov = pop_prov.groupby("prov_key", as_index=False).first()
 
+    # POP KHUSUS
     try:
         pop_khusus = _parse_pop_khusus(POP_KHUSUS)
     except Exception as e:
@@ -479,12 +481,21 @@ def load_default_files(force=False):
         f"DM: {fp.name} | Baris: {before} -> dedup: {after}\n"
         f"POP_KAB: {Path(POP_KAB).name} (n={len(pop_kab)})\n"
         f"POP_PROV: {Path(POP_PROV).name} (n={len(pop_prov)})\n"
-        f"POP_KHUSUS: {Path(POP_KHUSUS).name} (n={len(pop_khusus)}) (termasuk baris PROV)\n"
+        f"POP_KHUSUS: {Path(POP_KHUSUS).name} (n={len(pop_khusus)})\n"
         f"TARGET sampel per jenis: {TARGET_RATIO*100:.2f}%\n"
         f"mtime: DM={time.ctime(_mtime(DATA_FILE))} | Kab={time.ctime(_mtime(POP_KAB))} | Prov={time.ctime(_mtime(POP_PROV))} | Khusus={time.ctime(_mtime(POP_KHUSUS))}"
-    ).replace("\n", "<br>")
+    )
 
-    _CACHE.update({"key": key, "df_all": df_all, "df_raw": df_raw, "pop_kab": pop_kab, "pop_prov": pop_prov, "pop_khusus": pop_khusus, "meta": meta, "info": info})
+    _CACHE.update({
+        "key": key,
+        "df_all": df_all,
+        "df_raw": df_raw,
+        "pop_kab": pop_kab,
+        "pop_prov": pop_prov,
+        "pop_khusus": pop_khusus,
+        "meta": meta,
+        "info": info
+    })
     return df_all, df_raw, pop_kab, pop_prov, pop_khusus, meta, info
 
 
@@ -492,7 +503,7 @@ def load_default_files(force=False):
 # 6) FAKTOR WILAYAH — PER JENIS (TARGET 33.88%)
 # ============================================================
 
-def build_faktor_wilayah_jenis(df_filtered: pd.DataFrame, pop_kab: pd.DataFrame, pop_prov: pd.DataFrame, pop_khusus: pd.DataFrame, kew_value: str):
+def build_faktor_wilayah_jenis(df_filtered, pop_kab, pop_prov, pop_khusus, kew_value):
     if df_filtered is None or df_filtered.empty:
         return pd.DataFrame()
 
@@ -508,13 +519,13 @@ def build_faktor_wilayah_jenis(df_filtered: pd.DataFrame, pop_kab: pd.DataFrame,
         key_col, label_col, label_name, mode = "prov_key", "PROV_DISP", "Provinsi", "PROV"
         base_pop = pop_prov.copy() if (pop_prov is not None and not pop_prov.empty) else pd.DataFrame()
         if not base_pop.empty and "prov_key" not in base_pop.columns:
-            base_pop["prov_key"] = base_pop["Provinsi_Label"].apply(norm_prov_label) if "Provinsi_Label" in base_pop.columns else base_pop.iloc[:, 0].apply(norm_prov_label)
+            base_pop["prov_key"] = base_pop["Provinsi_Label"].apply(norm_prov_label)
         base_pop = base_pop.set_index("prov_key") if (not base_pop.empty and "prov_key" in base_pop.columns) else pd.DataFrame().set_index(pd.Index([]))
     else:
         key_col, label_col, label_name, mode = "kab_key", "KAB_DISP", "Kab/Kota", "KAB"
         base_pop = pop_kab.copy() if (pop_kab is not None and not pop_kab.empty) else pd.DataFrame()
         if not base_pop.empty and "kab_key" not in base_pop.columns:
-            base_pop["kab_key"] = base_pop["Kab_Kota_Label"].apply(norm_kab_label) if "Kab_Kota_Label" in base_pop.columns else base_pop.iloc[:, 0].apply(norm_kab_label)
+            base_pop["kab_key"] = base_pop["Kab_Kota_Label"].apply(norm_kab_label)
         base_pop = base_pop.set_index("kab_key") if (not base_pop.empty and "kab_key" in base_pop.columns) else pd.DataFrame().set_index(pd.Index([]))
 
     base_keys = df[[key_col, label_col]].drop_duplicates().rename(columns={key_col: "group_key", label_col: label_name})
@@ -619,7 +630,7 @@ def build_faktor_wilayah_jenis(df_filtered: pd.DataFrame, pop_kab: pd.DataFrame,
 # 7) AGREGAT WILAYAH × JENIS
 # ============================================================
 
-def build_agg_wilayah_jenis(df_filtered: pd.DataFrame, faktor_wilayah_jenis: pd.DataFrame, kew_value: str):
+def build_agg_wilayah_jenis(df_filtered, faktor_wilayah_jenis, kew_value):
     if df_filtered is None or df_filtered.empty:
         return pd.DataFrame()
 
@@ -652,8 +663,8 @@ def build_agg_wilayah_jenis(df_filtered: pd.DataFrame, faktor_wilayah_jenis: pd.
     ).reset_index().rename(columns={key_col: "group_key", label_col: label_name, "_dataset": "Jenis"})
 
     agg_real["Jenis"] = agg_real["Jenis"].astype(str).str.lower().str.strip()
-    agg = full.merge(agg_real, on=["group_key", label_name, "Jenis"], how="left")
 
+    agg = full.merge(agg_real, on=["group_key", label_name, "Jenis"], how="left")
     for c in ["Jumlah","Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_sub_pelayanan","Rata2_sub_pengelolaan",
               "Rata2_dim_kepatuhan","Rata2_dim_kinerja","Indeks_Dasar_Agregat_0_100"]:
         if c in agg.columns:
@@ -662,11 +673,6 @@ def build_agg_wilayah_jenis(df_filtered: pd.DataFrame, faktor_wilayah_jenis: pd.
 
     if faktor_wilayah_jenis is None or faktor_wilayah_jenis.empty:
         agg["faktor_penyesuaian_jenis"] = 1.0
-        agg["target_total_33_88_jenis"] = 0
-        agg["pop_total_jenis"] = 0
-        agg["coverage_jenis_%"] = 0.0
-        agg["gap_target33_88_jenis"] = 0
-        agg["n_jenis"] = 0
     else:
         fw = faktor_wilayah_jenis.copy()
         fw["Jenis"] = fw["Jenis"].astype(str).str.lower().str.strip()
@@ -674,37 +680,33 @@ def build_agg_wilayah_jenis(df_filtered: pd.DataFrame, faktor_wilayah_jenis: pd.
                 "faktor_penyesuaian_jenis", "target_total_33_88_jenis", "pop_total_jenis",
                 "coverage_jenis_%", "gap_target33_88_jenis", "n_jenis"]
         fw = fw[[c for c in keep if c in fw.columns]].copy()
-
         agg = agg.merge(fw, on=["group_key", label_name, "Jenis"], how="left")
         agg["faktor_penyesuaian_jenis"] = pd.to_numeric(agg["faktor_penyesuaian_jenis"], errors="coerce").fillna(1.0)
 
-        for c in ["target_total_33_88_jenis","pop_total_jenis","gap_target33_88_jenis","n_jenis"]:
-            if c in agg.columns:
-                agg[c] = pd.to_numeric(agg[c], errors="coerce").fillna(0).round(0).astype(int)
-
-        if "coverage_jenis_%" in agg.columns:
-            agg["coverage_jenis_%"] = pd.to_numeric(agg["coverage_jenis_%"], errors="coerce").fillna(0.0).round(2)
-
     agg["Indeks_Final_Agregat_0_100"] = (
         pd.to_numeric(agg["Indeks_Dasar_Agregat_0_100"], errors="coerce").fillna(0.0)
         * pd.to_numeric(agg["faktor_penyesuaian_jenis"], errors="coerce").fillna(1.0)
     )
 
-    for c in ["Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_sub_pelayanan","Rata2_sub_pengelolaan","Rata2_dim_kepatuhan","Rata2_dim_kinerja"]:
+    for c in [
+        "Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_sub_pelayanan","Rata2_sub_pengelolaan",
+        "Rata2_dim_kepatuhan","Rata2_dim_kinerja"
+    ]:
         if c in agg.columns:
             agg[c] = pd.to_numeric(agg[c], errors="coerce").fillna(0.0).round(3)
     for c in ["Indeks_Dasar_Agregat_0_100","Indeks_Final_Agregat_0_100"]:
         if c in agg.columns:
             agg[c] = pd.to_numeric(agg[c], errors="coerce").fillna(0.0).round(2)
+
     agg["faktor_penyesuaian_jenis"] = pd.to_numeric(agg["faktor_penyesuaian_jenis"], errors="coerce").fillna(1.0).round(3)
     return agg
 
 
 # ============================================================
-# 8) AGREGAT WILAYAH (KESELURUHAN) — avg3 FIX
+# 8) AGREGAT WILAYAH (KESELURUHAN) — avg3 dari 3 jenis
 # ============================================================
 
-def build_agg_wilayah_total_from_jenis(agg_jenis: pd.DataFrame, faktor_wilayah_jenis: pd.DataFrame, kew_value: str):
+def build_agg_wilayah_total_from_jenis(agg_jenis, faktor_wilayah_jenis, kew_value):
     if agg_jenis is None or agg_jenis.empty:
         return pd.DataFrame()
 
@@ -718,12 +720,16 @@ def build_agg_wilayah_total_from_jenis(agg_jenis: pd.DataFrame, faktor_wilayah_j
     base_keys = a[["group_key", label_name]].drop_duplicates()
     full = base_keys.assign(_tmp=1).merge(pd.DataFrame({"Jenis": jenis_list, "_tmp": 1}), on="_tmp").drop(columns="_tmp")
 
-    cols_need = ["Jumlah","Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_sub_pelayanan","Rata2_sub_pengelolaan","Rata2_dim_kepatuhan","Rata2_dim_kinerja",
-                 "Indeks_Dasar_Agregat_0_100","Indeks_Final_Agregat_0_100"]
-    cols_present = [c for c in cols_need if c in a.columns]
-
-    full = full.merge(a[["group_key", label_name, "Jenis"] + cols_present], on=["group_key", label_name, "Jenis"], how="left")
+    cols_present = [c for c in [
+        "Jumlah",
+        "Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_sub_pelayanan","Rata2_sub_pengelolaan",
+        "Rata2_dim_kepatuhan","Rata2_dim_kinerja",
+        "Indeks_Dasar_Agregat_0_100",
+        "Indeks_Final_Agregat_0_100",
+    ] if c in a.columns]
 
+    full = full.merge(a[["group_key", label_name, "Jenis"] + cols_present],
+                      on=["group_key", label_name, "Jenis"], how="left")
     for c in cols_present:
         full[c] = pd.to_numeric(full[c], errors="coerce").fillna(0.0)
 
@@ -739,47 +745,12 @@ def build_agg_wilayah_total_from_jenis(agg_jenis: pd.DataFrame, faktor_wilayah_j
         Indeks_Final_Wilayah_0_100=("Indeks_Final_Agregat_0_100", "mean"),
     )
 
-    if faktor_wilayah_jenis is not None and not faktor_wilayah_jenis.empty:
-        fw = faktor_wilayah_jenis.copy()
-        fw["Jenis"] = fw["Jenis"].astype(str).str.lower().str.strip()
-
-        piv = fw.pivot_table(
-            index=["group_key", label_name],
-            columns="Jenis",
-            values=["pop_total_jenis", "target_total_33_88_jenis", "n_jenis", "gap_target33_88_jenis", "faktor_penyesuaian_jenis"],
-            aggfunc="first"
-        )
-        piv.columns = [f"{v}_{k}" for v, k in piv.columns]
-        piv = piv.reset_index()
-        out = out.merge(piv, on=["group_key", label_name], how="left")
-
-        for j in ["sekolah", "umum", "khusus"]:
-            for basecol in ["pop_total_jenis", "target_total_33_88_jenis", "n_jenis", "gap_target33_88_jenis"]:
-                c = f"{basecol}_{j}"
-                if c in out.columns:
-                    out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0).round(0).astype(int)
-            cfac = f"faktor_penyesuaian_jenis_{j}"
-            if cfac in out.columns:
-                out[cfac] = pd.to_numeric(out[cfac], errors="coerce").fillna(1.0).round(3)
-
-        out["pop_total_all"] = (out.get("pop_total_jenis_sekolah", 0) + out.get("pop_total_jenis_umum", 0) + out.get("pop_total_jenis_khusus", 0)).astype(int)
-        out["target_total_33_88_all"] = (out.get("target_total_33_88_jenis_sekolah", 0) + out.get("target_total_33_88_jenis_umum", 0) + out.get("target_total_33_88_jenis_khusus", 0)).astype(int)
-        out["terkumpul_all"] = (out.get("n_jenis_sekolah", 0) + out.get("n_jenis_umum", 0) + out.get("n_jenis_khusus", 0)).astype(int)
-        out["coverage_target33_88_all_%"] = np.where(
-            pd.to_numeric(out["target_total_33_88_all"], errors="coerce").fillna(0).values > 0,
-            (pd.to_numeric(out["terkumpul_all"], errors="coerce").fillna(0).values / pd.to_numeric(out["target_total_33_88_all"], errors="coerce").fillna(0).values) * 100.0,
-            0.0
-        )
-        out["coverage_target33_88_all_%"] = pd.to_numeric(out["coverage_target33_88_all_%"], errors="coerce").fillna(0.0).round(2)
-
     for c in ["Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_sub_pelayanan","Rata2_sub_pengelolaan","Rata2_dim_kepatuhan","Rata2_dim_kinerja"]:
-        if c in out.columns:
-            out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0.0).round(3)
+        out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0.0).round(3)
     for c in ["Indeks_Dasar_Agregat_0_100","Indeks_Final_Wilayah_0_100"]:
-        if c in out.columns:
-            out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0.0).round(2)
-
+        out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0.0).round(2)
     out["n_total"] = pd.to_numeric(out["n_total"], errors="coerce").fillna(0).round(0).astype(int)
+
     return out
 
 
@@ -787,7 +758,7 @@ def build_agg_wilayah_total_from_jenis(agg_jenis: pd.DataFrame, faktor_wilayah_j
 # 9) SUMMARY (PER JENIS) + KESELURUHAN
 # ============================================================
 
-def build_summary_per_jenis(agg_jenis: pd.DataFrame, agg_total: pd.DataFrame):
+def build_summary_per_jenis(agg_jenis, agg_total):
     jenis_list = ["sekolah", "umum", "khusus"]
 
     def _row_default(jenis):
@@ -809,7 +780,6 @@ def build_summary_per_jenis(agg_jenis: pd.DataFrame, agg_total: pd.DataFrame):
     if agg_jenis is not None and not agg_jenis.empty:
         a = agg_jenis.copy()
         a["Jenis"] = a["Jenis"].astype(str).str.lower().str.strip()
-
         for c in ["Jumlah","Indeks_Dasar_Agregat_0_100","Indeks_Final_Agregat_0_100","pop_total_jenis","target_total_33_88_jenis"]:
             if c in a.columns:
                 a[c] = pd.to_numeric(a[c], errors="coerce").fillna(0)
@@ -843,16 +813,32 @@ def build_summary_per_jenis(agg_jenis: pd.DataFrame, agg_total: pd.DataFrame):
 
     rows = [rows_by_jenis[j] for j in jenis_list]
 
-    dasar_all = (rows_by_jenis["sekolah"]["Indeks_Dasar_0_100"] + rows_by_jenis["umum"]["Indeks_Dasar_0_100"] + rows_by_jenis["khusus"]["Indeks_Dasar_0_100"]) / 3.0
-    final_all = (rows_by_jenis["sekolah"]["Indeks_Final_Disesuaikan_0_100"] + rows_by_jenis["umum"]["Indeks_Final_Disesuaikan_0_100"] + rows_by_jenis["khusus"]["Indeks_Final_Disesuaikan_0_100"]) / 3.0
+    dasar_all = (rows_by_jenis["sekolah"]["Indeks_Dasar_0_100"]
+                 + rows_by_jenis["umum"]["Indeks_Dasar_0_100"]
+                 + rows_by_jenis["khusus"]["Indeks_Dasar_0_100"]) / 3.0
+
+    final_all = (rows_by_jenis["sekolah"]["Indeks_Final_Disesuaikan_0_100"]
+                 + rows_by_jenis["umum"]["Indeks_Final_Disesuaikan_0_100"]
+                 + rows_by_jenis["khusus"]["Indeks_Final_Disesuaikan_0_100"]) / 3.0
+
+    pop_all = int(rows_by_jenis["sekolah"]["Pop_Total_Jenis"]
+                  + rows_by_jenis["umum"]["Pop_Total_Jenis"]
+                  + rows_by_jenis["khusus"]["Pop_Total_Jenis"])
+
+    target_all = int(rows_by_jenis["sekolah"]["Target33_88_Total_Jenis"]
+                     + rows_by_jenis["umum"]["Target33_88_Total_Jenis"]
+                     + rows_by_jenis["khusus"]["Target33_88_Total_Jenis"])
+
+    terkumpul_all = int(rows_by_jenis["sekolah"]["Terkumpul_Jenis"]
+                        + rows_by_jenis["umum"]["Terkumpul_Jenis"]
+                        + rows_by_jenis["khusus"]["Terkumpul_Jenis"])
 
-    pop_all = int(rows_by_jenis["sekolah"]["Pop_Total_Jenis"] + rows_by_jenis["umum"]["Pop_Total_Jenis"] + rows_by_jenis["khusus"]["Pop_Total_Jenis"])
-    target_all = int(rows_by_jenis["sekolah"]["Target33_88_Total_Jenis"] + rows_by_jenis["umum"]["Target33_88_Total_Jenis"] + rows_by_jenis["khusus"]["Target33_88_Total_Jenis"])
-    terkumpul_all = int(rows_by_jenis["sekolah"]["Terkumpul_Jenis"] + rows_by_jenis["umum"]["Terkumpul_Jenis"] + rows_by_jenis["khusus"]["Terkumpul_Jenis"])
     coverage_all = (terkumpul_all / target_all * 100.0) if target_all > 0 else 0.0
 
     jumlah_wilayah_all = int(agg_total.shape[0]) if (agg_total is not None and not agg_total.empty) else int(
-        max(rows_by_jenis["sekolah"]["Jumlah_Wilayah"], rows_by_jenis["umum"]["Jumlah_Wilayah"], rows_by_jenis["khusus"]["Jumlah_Wilayah"])
+        max(rows_by_jenis["sekolah"]["Jumlah_Wilayah"],
+            rows_by_jenis["umum"]["Jumlah_Wilayah"],
+            rows_by_jenis["khusus"]["Jumlah_Wilayah"])
     )
 
     rows.append({
@@ -869,22 +855,18 @@ def build_summary_per_jenis(agg_jenis: pd.DataFrame, agg_total: pd.DataFrame):
     })
 
     out = pd.DataFrame(rows)
-
     for c in ["Jumlah_Wilayah","Total_Perpus","Pop_Total_Jenis","Target33_88_Total_Jenis","Terkumpul_Jenis"]:
-        if c in out.columns:
-            out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0).round(0).astype(int)
+        out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0).round(0).astype(int)
     for c in ["Coverage_Target33_88_Jenis_%","Indeks_Dasar_0_100","Indeks_Final_Disesuaikan_0_100","Penyesuaian_Poin"]:
-        if c in out.columns:
-            out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0.0).round(2)
-
+        out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0.0).round(2)
     return out
 
 
 # ============================================================
-# 10) DETAIL ENTITAS: Final menempel dari agg_total (wilayah)
+# 10) DETAIL ENTITAS (Final menempel dari wilayah)
 # ============================================================
 
-def attach_final_to_detail(df_filtered: pd.DataFrame, agg_total: pd.DataFrame, meta: dict, kew_value: str):
+def attach_final_to_detail(df_filtered, agg_total, meta, kew_value):
     if df_filtered is None or df_filtered.empty:
         return pd.DataFrame()
 
@@ -928,15 +910,14 @@ def attach_final_to_detail(df_filtered: pd.DataFrame, agg_total: pd.DataFrame, m
     for c in ["Indeks_Dasar_0_100","Indeks_Final_0_100"]:
         if c in out.columns:
             out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0.0).round(2)
-
     return out
 
 
 # ============================================================
-# 11) VERIFIKASI PER JENIS (TARGET 33.88%)
+# 11) VERIF (kecukupan sampel)
 # ============================================================
 
-def build_verif_jenis(faktor_wilayah_jenis: pd.DataFrame, kew_value: str):
+def build_verif_jenis(faktor_wilayah_jenis, kew_value):
     if faktor_wilayah_jenis is None or faktor_wilayah_jenis.empty:
         return pd.DataFrame()
 
@@ -944,17 +925,19 @@ def build_verif_jenis(faktor_wilayah_jenis: pd.DataFrame, kew_value: str):
     label_col = "Provinsi" if "PROV" in kew_norm else "Kab/Kota"
 
     out = faktor_wilayah_jenis.copy()
-    keep = [c for c in [label_col, "Jenis", "pop_total_jenis", "target_total_33_88_jenis", "n_jenis",
-                       "coverage_jenis_%", "faktor_penyesuaian_jenis", "gap_target33_88_jenis"] if c in out.columns]
+    keep = [c for c in [
+        label_col, "Jenis",
+        "pop_total_jenis", "target_total_33_88_jenis", "n_jenis",
+        "coverage_jenis_%", "faktor_penyesuaian_jenis", "gap_target33_88_jenis"
+    ] if c in out.columns]
+
     out = out[keep].copy()
 
     for c in ["pop_total_jenis", "target_total_33_88_jenis", "n_jenis", "gap_target33_88_jenis"]:
         if c in out.columns:
             out[c] = pd.to_numeric(out[c], errors="coerce").fillna(0).round(0).astype(int)
-
     if "coverage_jenis_%" in out.columns:
         out["coverage_jenis_%"] = pd.to_numeric(out["coverage_jenis_%"], errors="coerce").fillna(0.0).round(2)
-
     if "faktor_penyesuaian_jenis" in out.columns:
         out["faktor_penyesuaian_jenis"] = pd.to_numeric(out["faktor_penyesuaian_jenis"], errors="coerce").fillna(1.0).round(3)
 
@@ -962,11 +945,10 @@ def build_verif_jenis(faktor_wilayah_jenis: pd.DataFrame, kew_value: str):
 
 
 # ============================================================
-# 12) BELL CURVE — Indeks Dasar per Entitas (per Jenis) + Hover Nama Perpus
+# 12) BELL CURVE — Indeks Dasar per Entitas (per Jenis) + Hover
 # ============================================================
 
-def _make_bell_curve_entitas(dfp: pd.DataFrame, title: str, xcol: str = "Indeks_Dasar_0_100",
-                            label_col: str = "nm_perpustakaan", hover_cols=None, min_points: int = 2):
+def _make_bell_curve_entitas(dfp, title, xcol="Indeks_Dasar_0_100", label_col="nm_perpustakaan", hover_cols=None, min_points=2):
     fig = go.Figure()
     fig.update_layout(
         title=title,
@@ -1021,7 +1003,12 @@ def _make_bell_curve_entitas(dfp: pd.DataFrame, title: str, xcol: str = "Indeks_
 
     if len(x) < min_points:
         x_single = float(x[0])
-        fig.add_trace(go.Scatter(x=[x_single], y=[0], mode="markers", showlegend=False, hovertext=[hover_text[0]] if hover_text else None, hoverinfo="text"))
+        fig.add_trace(go.Scatter(
+            x=[x_single], y=[0],
+            mode="markers", showlegend=False,
+            hovertext=[hover_text[0]] if hover_text else None,
+            hoverinfo="text"
+        ))
         fig.add_vline(x=x_single, line_width=1, line_dash="dash", annotation_text=f"Nilai: {x_single:.1f}", annotation_position="top")
         fig.update_xaxes(range=[0, 100])
         fig.update_yaxes(rangemode="tozero")
@@ -1037,7 +1024,12 @@ def _make_bell_curve_entitas(dfp: pd.DataFrame, title: str, xcol: str = "Indeks_
     pdf = (1.0 / (sigma * np.sqrt(2 * np.pi))) * np.exp(-0.5 * ((xs - mu) / sigma) ** 2)
 
     fig.add_trace(go.Scatter(x=xs, y=pdf, mode="lines", name="Kurva Normal (fit)"))
-    fig.add_trace(go.Scatter(x=x, y=np.zeros_like(x), mode="markers", showlegend=False, hovertext=hover_text if hover_text else None, hoverinfo="text"))
+    fig.add_trace(go.Scatter(
+        x=x, y=np.zeros_like(x),
+        mode="markers", showlegend=False,
+        hovertext=hover_text if hover_text else None,
+        hoverinfo="text"
+    ))
 
     q1, q2, q3 = np.percentile(x, [25, 50, 75])
     for xv, lab in [(q1, "Q1"), (q2, "Q2 (Median)"), (q3, "Q3"), (mu, "Mean")]:
@@ -1049,7 +1041,7 @@ def _make_bell_curve_entitas(dfp: pd.DataFrame, title: str, xcol: str = "Indeks_
 
 
 # ============================================================
-# 13) KPI DASHBOARD (HANYA 2 KARTU: FINAL + DASAR)
+# 13) KPI DASHBOARD (2 kartu: final + dasar)
 # ============================================================
 
 def _safe_first(df, col, default=0.0, where=None):
@@ -1062,16 +1054,11 @@ def _safe_first(df, col, default=0.0, where=None):
         return default
     return float(pd.to_numeric(sub[col], errors="coerce").fillna(default).iloc[0])
 
-def compute_dashboard_kpis(summary_jenis: pd.DataFrame):
-    final_all = _safe_first(summary_jenis, "Indeks_Final_Disesuaikan_0_100", 0.0, where=summary_jenis["Jenis"].astype(str).str.lower().eq("keseluruhan"))
-    dasar_all = _safe_first(summary_jenis, "Indeks_Dasar_0_100", 0.0, where=summary_jenis["Jenis"].astype(str).str.lower().eq("keseluruhan"))
-    return {"final_all": final_all, "dasar_all": dasar_all}
-
-def build_kpi_markdown(summary_jenis: pd.DataFrame) -> str:
+def build_kpi_markdown(summary_jenis):
     if summary_jenis is None or summary_jenis.empty:
         return ""
-
-    k = compute_dashboard_kpis(summary_jenis)
+    final_all = _safe_first(summary_jenis, "Indeks_Final_Disesuaikan_0_100", 0.0, where=summary_jenis["Jenis"].astype(str).str.lower().eq("keseluruhan"))
+    dasar_all = _safe_first(summary_jenis, "Indeks_Dasar_0_100", 0.0, where=summary_jenis["Jenis"].astype(str).str.lower().eq("keseluruhan"))
 
     def fmt(x, nd=2):
         return "NA" if pd.isna(x) else f"{x:.{nd}f}"
@@ -1080,13 +1067,13 @@ def build_kpi_markdown(summary_jenis: pd.DataFrame) -> str:
 <div style="display:flex; gap:12px; flex-wrap:wrap;">
   <div style="border:1px solid #333; border-radius:10px; padding:10px 12px; min-width:260px;">
     <div style="opacity:0.8;">Indeks IPLM FINAL (Disesuaikan 33.88%)</div>
-    <div style="font-size:26px; font-weight:700;">{fmt(k["final_all"],2)}</div>
-    <div style="opacity:0.7;">Skor absolut</div>
+    <div style="font-size:26px; font-weight:700;">{fmt(final_all,2)}</div>
+    <div style="opacity:0.7;">Skor absolut (untuk akuntabilitas)</div>
   </div>
 
   <div style="border:1px solid #333; border-radius:10px; padding:10px 12px; min-width:260px;">
     <div style="opacity:0.8;">Indeks Dasar (Tanpa Penyesuaian)</div>
-    <div style="font-size:26px; font-weight:700;">{fmt(k["dasar_all"],2)}</div>
+    <div style="font-size:26px; font-weight:700;">{fmt(dasar_all,2)}</div>
     <div style="opacity:0.7;">Sebelum faktor kecukupan sampel</div>
   </div>
 </div>
@@ -1094,7 +1081,7 @@ def build_kpi_markdown(summary_jenis: pd.DataFrame) -> str:
 
 
 # ============================================================
-# 14) LLM -> WORD TABEL (SESUAI GAMBAR) + WORD REPORT
+# 14) LLM: Isi Interpretasi & Rekomendasi (TABEL) + WORD
 # ============================================================
 
 _HF_CLIENT = None
@@ -1113,154 +1100,229 @@ def get_llm_client():
         _HF_CLIENT = None
         return None
 
-def _get_overall_numbers_for_llm(summary_jenis: pd.DataFrame, agg_total: pd.DataFrame) -> dict:
-    def _mean_0_1_to_100(df, col):
-        if df is None or df.empty or col not in df.columns:
+def _to_2dec(x):
+    try:
+        if x is None or (isinstance(x, float) and math.isnan(x)):
             return 0.0
-        v = pd.to_numeric(df[col], errors="coerce").fillna(0.0).mean()
-        return float(v) * 100.0
-
-    nilai_iplm = 0.0
-    if summary_jenis is not None and not summary_jenis.empty and "Jenis" in summary_jenis.columns:
-        mask = summary_jenis["Jenis"].astype(str).str.lower().eq("keseluruhan")
-        if mask.any() and "Indeks_Final_Disesuaikan_0_100" in summary_jenis.columns:
-            nilai_iplm = float(pd.to_numeric(summary_jenis.loc[mask, "Indeks_Final_Disesuaikan_0_100"], errors="coerce").fillna(0.0).iloc[0])
-
-    return {
-        "kepatuhan": round(_mean_0_1_to_100(agg_total, "Rata2_dim_kepatuhan"), 2),
-        "koleksi": round(_mean_0_1_to_100(agg_total, "Rata2_sub_koleksi"), 2),
-        "tenaga": round(_mean_0_1_to_100(agg_total, "Rata2_sub_sdm"), 2),
-        "kinerja": round(_mean_0_1_to_100(agg_total, "Rata2_dim_kinerja"), 2),
-        "pelayanan": round(_mean_0_1_to_100(agg_total, "Rata2_sub_pelayanan"), 2),
-        "pengelolaan": round(_mean_0_1_to_100(agg_total, "Rata2_sub_pengelolaan"), 2),
-        "iplm": round(nilai_iplm, 2),
-    }
+        return float(x)
+    except Exception:
+        return 0.0
+
+def build_interpretasi_table_values(agg_total, wilayah_label, target_ratio):
+    """
+    Mengambil NILAI apa adanya dari hasil aplikasi (agg_total):
+    - Kepatuhan = 100 * Rata2_dim_kepatuhan
+    - Koleksi = 100 * Rata2_sub_koleksi
+    - Tenaga = 100 * Rata2_sub_sdm
+    - Kinerja = 100 * Rata2_dim_kinerja
+    - Pelayanan = 100 * Rata2_sub_pelayanan
+    - Penyelenggaraan/Pengelolaan = 100 * Rata2_sub_pengelolaan
+    - Nilai IPLM = Indeks_Final_Wilayah_0_100
+
+    Jika agg_total punya lebih dari 1 baris (mis. Nasional),
+    diambil rata-rata kolom-kolom tersebut.
+    """
+    if agg_total is None or agg_total.empty:
+        base = {
+            "kepatuhan": 0.0, "koleksi": 0.0, "tenaga": 0.0,
+            "kinerja": 0.0, "pelayanan": 0.0, "pengelolaan": 0.0,
+            "iplm": 0.0
+        }
+    else:
+        a = agg_total.copy()
+        for c in ["Rata2_dim_kepatuhan","Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_dim_kinerja","Rata2_sub_pelayanan","Rata2_sub_pengelolaan","Indeks_Final_Wilayah_0_100"]:
+            if c in a.columns:
+                a[c] = pd.to_numeric(a[c], errors="coerce").fillna(0.0)
+            else:
+                a[c] = 0.0
+
+        base = {
+            "kepatuhan": 100.0 * float(a["Rata2_dim_kepatuhan"].mean()),
+            "koleksi": 100.0 * float(a["Rata2_sub_koleksi"].mean()),
+            "tenaga": 100.0 * float(a["Rata2_sub_sdm"].mean()),
+            "kinerja": 100.0 * float(a["Rata2_dim_kinerja"].mean()),
+            "pelayanan": 100.0 * float(a["Rata2_sub_pelayanan"].mean()),
+            "pengelolaan": 100.0 * float(a["Rata2_sub_pengelolaan"].mean()),
+            "iplm": float(a["Indeks_Final_Wilayah_0_100"].mean()),
+        }
 
-def generate_llm_table_rows(summary_jenis, agg_total, wilayah, kew):
-    base_rows = [
-        {"No": "1",   "Dimensi": "Kepatuhan", "Nilai": None, "Interpretasi": "", "Rekomendasi": ""},
-        {"No": "1.1", "Dimensi": "Variabel Koleksi", "Nilai": None, "Interpretasi": "", "Rekomendasi": ""},
-        {"No": "1.2", "Dimensi": "Variabel Tenaga Perpustakaan", "Nilai": None, "Interpretasi": "", "Rekomendasi": ""},
-        {"No": "2",   "Dimensi": "Kinerja", "Nilai": None, "Interpretasi": "", "Rekomendasi": ""},
-        {"No": "2.1", "Dimensi": "Variabel Pelayanan", "Nilai": None, "Interpretasi": "", "Rekomendasi": ""},
-        {"No": "2.2", "Dimensi": "Variabel Penyelenggaraan/Pengelolaan", "Nilai": None, "Interpretasi": "", "Rekomendasi": ""},
-        {"No": "4",   "Dimensi": "Nilai IPLM", "Nilai": None, "Interpretasi": "", "Rekomendasi": ""},
+    # pembulatan display (tetap "nilai aplikasi", hanya format tampilan)
+    for k in base:
+        base[k] = round(_to_2dec(base[k]), 2)
+
+    rows = [
+        {"No":"1",   "Dimensi":"Kepatuhan", "Nilai":base["kepatuhan"]},
+        {"No":"1.1", "Dimensi":"Variabel Koleksi", "Nilai":base["koleksi"]},
+        {"No":"1.2", "Dimensi":"Variabel Tenaga Perpustakaan", "Nilai":base["tenaga"]},
+        {"No":"2",   "Dimensi":"Kinerja", "Nilai":base["kinerja"]},
+        {"No":"2.1", "Dimensi":"Variabel Pelayanan", "Nilai":base["pelayanan"]},
+        {"No":"2.2", "Dimensi":"Variabel Penyelenggaraan/Pengelolaan", "Nilai":base["pengelolaan"]},
+        {"No":"4",   "Dimensi":"Nilai IPLM", "Nilai":base["iplm"]},
     ]
 
-    nums = _get_overall_numbers_for_llm(summary_jenis, agg_total)
-    mapping = {"1": "kepatuhan", "1.1": "koleksi", "1.2": "tenaga", "2": "kinerja", "2.1": "pelayanan", "2.2": "pengelolaan", "4": "iplm"}
-    for r in base_rows:
-        r["Nilai"] = nums.get(mapping.get(r["No"]), 0.0)
-
+    header = {
+        "judul": f"Interpretasi dan Rekomendasi IPLM — {wilayah_label}",
+        "target_sampel": f"{target_ratio*100:.2f}%"
+    }
+    return header, rows
+
+def llm_fill_interpretasi_rekomendasi(header, rows, wilayah_label, kew_label):
+    """
+    LLM diminta mengisi kolom Interpretasi dan Rekomendasi
+    dengan gaya netral-deskriptif (tanpa label tinggi/rendah/baik/buruk).
+    Output wajib JSON agar mudah diparse.
+    """
     client = get_llm_client()
     if client is None or (not USE_LLM):
-        return base_rows
-
-    ctx = f"Wilayah={wilayah} | Kewenangan={kew} | Target={TARGET_RATIO*100:.2f}%"
-    angka_ctx = (
-        f"- Kepatuhan: {nums['kepatuhan']}\n"
-        f"- Variabel Koleksi: {nums['koleksi']}\n"
-        f"- Variabel Tenaga Perpustakaan: {nums['tenaga']}\n"
-        f"- Kinerja: {nums['kinerja']}\n"
-        f"- Variabel Pelayanan: {nums['pelayanan']}\n"
-        f"- Variabel Penyelenggaraan/Pengelolaan: {nums['pengelolaan']}\n"
-        f"- Nilai IPLM: {nums['iplm']}\n"
+        # fallback kosong
+        out = []
+        for r in rows:
+            out.append({**r, "Interpretasi":"", "Rekomendasi":""})
+        return out, "LLM tidak digunakan / tidak tersedia."
+
+    payload = {
+        "wilayah": wilayah_label,
+        "kewenangan": kew_label,
+        "target_sampel_per_jenis": header["target_sampel"],
+        "baris": rows
+    }
+
+    system = (
+        "Anda adalah analis kebijakan perpustakaan di Indonesia.\n"
+        "Tugas: isi kolom Interpretasi dan Rekomendasi untuk tiap baris tabel.\n"
+        "Gaya wajib: netral dan deskriptif; dilarang menggunakan label normatif seperti baik/buruk, tinggi/sedang/rendah, maju/tertinggal.\n"
+        "Gunakan kalimat yang menjelaskan makna angka sebagai ringkasan kondisi berdasarkan data yang dilaporkan, tanpa menghakimi.\n"
+        "Rekomendasi: operasional, spesifik, dan dapat ditindaklanjuti (2-3 butir ringkas) tanpa menyebut kategori penilaian.\n"
+        "Dilarang mengubah NILAI. NILAI hanya dipakai sebagai konteks.\n"
+        "Output harus JSON valid, tanpa teks tambahan."
     )
 
-    prompt_user = f"""
-{ctx}
-
-Saya akan membuat tabel Word dengan kolom:
-No | Dimensi | Nilai | Interpretasi | Rekomendasi
-
-Nilai (0–100) sudah ditetapkan sebagai berikut:
-{angka_ctx}
-
-Tugas Anda:
-1) Isi "Interpretasi" dan "Rekomendasi" untuk tiap baris secara netral dan deskriptif.
-2) Jangan gunakan label normatif seperti: baik/buruk, tinggi/rendah, memuaskan/tidak, optimal/tidak.
-3) Interpretasi menjelaskan apa yang dicerminkan angka itu (tanpa menghakimi).
-4) Rekomendasi berisi langkah tindak lanjut yang operasional.
-5) Output harus berupa JSON array saja (tanpa teks lain), tiap elemen berisi:
-   - "No"
-   - "Interpretasi"
-   - "Rekomendasi"
-Gunakan No persis: ["1","1.1","1.2","2","2.1","2.2","4"].
-""".strip()
+    user = (
+        "Berikut input data tabel (JSON). Kembalikan JSON dengan struktur:\n"
+        "{\n"
+        '  "rows": [\n'
+        '    {"No":"...","Dimensi":"...","Nilai":12.34,"Interpretasi":"...","Rekomendasi":"..."}\n'
+        "  ]\n"
+        "}\n"
+        "Pastikan jumlah baris sama dan urutan sama.\n\n"
+        f"INPUT:\n{json.dumps(payload, ensure_ascii=False)}"
+    )
 
     try:
         resp = client.chat_completion(
             model=LLM_MODEL_NAME,
             messages=[
-                {"role":"system","content":"Anda adalah analis kebijakan perpustakaan di Indonesia. Gaya netral-deskriptif, berbasis data, tanpa label normatif."},
-                {"role":"user","content":prompt_user}
+                {"role": "system", "content": system},
+                {"role": "user", "content": user},
             ],
-            max_tokens=800,
+            max_tokens=900,
             temperature=0.2,
             top_p=0.9,
         )
-        raw = (resp.choices[0].message.content or "").strip()
-        if not raw:
-            return base_rows
-
-        data = json.loads(raw)
-        by_no = {}
-        if isinstance(data, list):
-            for it in data:
-                no = str(it.get("No", "")).strip()
-                if no:
-                    by_no[no] = {
-                        "Interpretasi": str(it.get("Interpretasi", "") or "").strip(),
-                        "Rekomendasi": str(it.get("Rekomendasi", "") or "").strip(),
-                    }
-
-        for r in base_rows:
-            if r["No"] in by_no:
-                r["Interpretasi"] = by_no[r["No"]]["Interpretasi"]
-                r["Rekomendasi"] = by_no[r["No"]]["Rekomendasi"]
-        return base_rows
-
-    except Exception:
-        return base_rows
-
-def generate_llm_analysis(summary_jenis, agg_total, verif_total, wilayah, kew):
-    client = get_llm_client()
-    if client is None or (not USE_LLM):
-        return "Analisis otomatis (LLM) tidak digunakan / tidak tersedia."
-    return "Analisis otomatis disajikan pada Laporan Word dalam bentuk tabel (Interpretasi & Rekomendasi per dimensi)."
-
-def generate_word_report(wilayah, summary_jenis, analysis_text, agg_total=None, kew="(Semua)"):
+        text = resp.choices[0].message.content.strip()
+
+        # parse JSON
+        data = json.loads(text)
+        rows_out = data.get("rows", [])
+        # fallback jika tidak sesuai
+        if not isinstance(rows_out, list) or len(rows_out) != len(rows):
+            raise ValueError("Format JSON rows tidak sesuai.")
+        return rows_out, "LLM mengisi Interpretasi & Rekomendasi."
+    except Exception as e:
+        out = []
+        for r in rows:
+            out.append({**r, "Interpretasi":"", "Rekomendasi":""})
+        return out, f"LLM error: {repr(e)}"
+
+
+def _set_cell_shading(cell, fill_hex="1F1F1F"):
+    """
+    Set shading background untuk cell (python-docx).
+    """
+    tcPr = cell._tc.get_or_add_tcPr()
+    shd = OxmlElement("w:shd")
+    shd.set(qn("w:val"), "clear")
+    shd.set(qn("w:color"), "auto")
+    shd.set(qn("w:fill"), fill_hex)
+    tcPr.append(shd)
+
+def _set_cell_text_color(cell, rgb_hex="FFFFFF"):
+    """
+    Set font color untuk semua run dalam cell.
+    """
+    for p in cell.paragraphs:
+        for run in p.runs:
+            rPr = run._r.get_or_add_rPr()
+            color = OxmlElement("w:color")
+            color.set(qn("w:val"), rgb_hex)
+            rPr.append(color)
+
+def _set_table_borders(table):
+    """
+    Tambah border sederhana.
+    """
+    tbl = table._tbl
+    tblPr = tbl.tblPr
+    if tblPr is None:
+        tblPr = OxmlElement('w:tblPr')
+        tbl.append(tblPr)
+    tblBorders = OxmlElement('w:tblBorders')
+    for edge in ("top", "left", "bottom", "right", "insideH", "insideV"):
+        elem = OxmlElement(f'w:{edge}')
+        elem.set(qn('w:val'), 'single')
+        elem.set(qn('w:sz'), '8')
+        elem.set(qn('w:space'), '0')
+        elem.set(qn('w:color'), 'FFFFFF')
+        tblBorders.append(elem)
+    tblPr.append(tblBorders)
+
+def generate_word_table_interpretasi(header, rows_filled, wilayah_label):
     if (not DOCX_AVAILABLE) or (Document is None):
         return None
 
     doc = Document()
-    doc.add_heading(f"Interpretasi dan Rekomendasi IPLM — {wilayah}", level=1)
-    doc.add_paragraph(f"Target sampel per jenis: {TARGET_RATIO*100:.2f}%")
-
-    rows = generate_llm_table_rows(summary_jenis, agg_total, wilayah, kew)
-
-    table = doc.add_table(rows=1, cols=5)
-    table.style = "Table Grid"
-    hdr = table.rows[0].cells
-    hdr[0].text = "No"
-    hdr[1].text = "Dimensi"
-    hdr[2].text = "Nilai"
-    hdr[3].text = "Interpretasi"
-    hdr[4].text = "Rekomendasi"
 
-    def _fmt_nilai(x):
+    # Title
+    title = doc.add_paragraph()
+    run = title.add_run(header["judul"])
+    run.bold = True
+    run.font.size = Pt(18)
+
+    doc.add_paragraph(f"Target sampel per jenis: {header['target_sampel']}")
+
+    # Table
+    cols = ["No", "Dimensi", "Nilai", "Interpretasi", "Rekomendasi"]
+    table = doc.add_table(rows=1, cols=len(cols))
+    table.autofit = True
+    _set_table_borders(table)
+
+    hdr_cells = table.rows[0].cells
+    for i, c in enumerate(cols):
+        hdr_cells[i].text = c
+        _set_cell_shading(hdr_cells[i], "1A1A1A")
+        _set_cell_text_color(hdr_cells[i], "FFFFFF")
+        for p in hdr_cells[i].paragraphs:
+            for r in p.runs:
+                r.bold = True
+
+    for r in rows_filled:
+        row_cells = table.add_row().cells
+        row_cells[0].text = str(r.get("No",""))
+        row_cells[1].text = str(r.get("Dimensi",""))
+        # nilai (apa adanya dari aplikasi, hanya format 2 desimal)
         try:
-            return f"{float(x):.2f}"
+            row_cells[2].text = f"{float(r.get('Nilai',0.0)):.2f}"
         except Exception:
-            return ""
+            row_cells[2].text = str(r.get("Nilai",""))
+        row_cells[3].text = str(r.get("Interpretasi","") or "")
+        row_cells[4].text = str(r.get("Rekomendasi","") or "")
+
+        # shading body (gelap) + teks putih agar mirip contoh
+        for c in row_cells:
+            _set_cell_shading(c, "262626")
+            _set_cell_text_color(c, "FFFFFF")
 
-    for r in rows:
-        cells = table.add_row().cells
-        cells[0].text = str(r.get("No", "") or "")
-        cells[1].text = str(r.get("Dimensi", "") or "")
-        cells[2].text = _fmt_nilai(r.get("Nilai", ""))
-        cells[3].text = str(r.get("Interpretasi", "") or "")
-        cells[4].text = str(r.get("Rekomendasi", "") or "")
+    doc.add_paragraph("")  # spacer
 
     outpath = tempfile.mktemp(suffix=".docx")
     doc.save(outpath)
@@ -1274,13 +1336,22 @@ def generate_word_report(wilayah, summary_jenis, analysis_text, agg_total=None,
 def _empty_outputs(msg="Data belum siap."):
     empty = pd.DataFrame()
     empty_fig = go.Figure()
-    return ("", empty, empty, empty, empty, empty, None, None, None, None, None, empty_fig, empty_fig, empty_fig, msg, "Analisis belum tersedia.")
+    return (
+        "",  # kpi_md
+        empty, empty, empty, empty, empty,
+        None, None, None, None, None,
+        empty_fig, empty_fig, empty_fig,
+        msg,  # msg
+        "LLM belum tersedia.",  # status llm
+        None  # word path
+    )
 
 def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov, pop_khusus, meta):
     try:
         if df_all is None or df_all.empty or df_raw is None or df_raw.empty:
-            return _empty_outputs("Data belum ter-load. Pastikan file tersedia di repo/server.")
+            return _empty_outputs("Data belum ter-load. Pastikan file tersedia.")
 
+        # Filter
         df = df_all.copy()
         if prov_value and prov_value != "(Semua)":
             df = df[df["PROV_DISP"] == prov_value]
@@ -1288,7 +1359,6 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
             df = df[df["KAB_DISP"] == kab_value]
         if kew_value and kew_value != "(Semua)":
             df = df[df["KEW_NORM"] == kew_value]
-
         if df.empty:
             return _empty_outputs("Tidak ada data untuk filter ini.")
 
@@ -1301,13 +1371,17 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
         verif_total = build_verif_jenis(faktor_wilayah_jenis, kew_norm)
         detail_view = attach_final_to_detail(df, agg_total, meta, kew_norm)
 
+        # agg_jenis view (UI hanya sampai indeks dasar)
         if agg_jenis_full is None or agg_jenis_full.empty:
             agg_jenis_view = agg_jenis_full
         else:
             kew_norm2 = str(kew_norm).upper()
             label_name = "Kab/Kota" if ("KAB" in kew_norm2 or "KOTA" in kew_norm2) else ("Provinsi" if "PROV" in kew_norm2 else "Kab/Kota")
             cols_upto = [
-                "group_key", label_name, "Jenis", "Jumlah",
+                "group_key",
+                label_name,
+                "Jenis",
+                "Jumlah",
                 "Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_sub_pelayanan","Rata2_sub_pengelolaan",
                 "Rata2_dim_kepatuhan","Rata2_dim_kinerja",
                 "Indeks_Dasar_Agregat_0_100",
@@ -1315,6 +1389,7 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
             cols_upto = [c for c in cols_upto if c in agg_jenis_full.columns]
             agg_jenis_view = agg_jenis_full[cols_upto].copy()
 
+        # RAW download (hasil filter)
         raw = df_raw.copy()
         if prov_value and prov_value != "(Semua)":
             raw = raw[raw["PROV_DISP"] == prov_value]
@@ -1323,15 +1398,13 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
         if kew_value and kew_value != "(Semua)":
             raw = raw[raw["KEW_NORM"] == kew_value]
 
+        # Bell curve per jenis
         if detail_view is None or detail_view.empty:
             fig_umum = _make_bell_curve_entitas(pd.DataFrame(), "Bell Curve — Jenis: Umum")
             fig_sekolah = _make_bell_curve_entitas(pd.DataFrame(), "Bell Curve — Jenis: Sekolah")
             fig_khusus = _make_bell_curve_entitas(pd.DataFrame(), "Bell Curve — Jenis: Khusus")
         else:
-            hover_cols = []
-            for hc in ["Provinsi", "Kab/Kota", "Jenis"]:
-                if hc in detail_view.columns:
-                    hover_cols.append(hc)
+            hover_cols = [hc for hc in ["Provinsi", "Kab/Kota", "Jenis"] if hc in detail_view.columns]
 
             def _fig(j):
                 d = detail_view[detail_view["Jenis"].astype(str).str.lower() == j].copy()
@@ -1350,6 +1423,7 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
 
         kpi_md = build_kpi_markdown(summary_jenis)
 
+        # Export xlsx
         tmpdir = tempfile.mkdtemp()
         prov_slug = (_canon(prov_value or "SEMUA").upper() or "SEMUA")
         kab_slug = (_canon(kab_value or "SEMUA").upper() or "SEMUA")
@@ -1367,22 +1441,26 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
         detail_view.to_excel(p_detail, index=False)
         verif_total.to_excel(p_verif, index=False)
 
+        # ====== NEW: Word tabel interpretasi & rekomendasi ======
         wilayah_txt = kab_value if (kab_value and kab_value != "(Semua)") else (prov_value if (prov_value and prov_value != "(Semua)") else "Nasional/All")
-        analysis_text = generate_llm_analysis(summary_jenis, agg_total, verif_total, wilayah_txt, kew_value or "(Semua)")
-        word_path = generate_word_report(wilayah_txt, summary_jenis, analysis_text, agg_total=agg_total, kew=(kew_value or "(Semua)"))
+        header, rows = build_interpretasi_table_values(agg_total, wilayah_txt, TARGET_RATIO)
+        rows_filled, llm_status = llm_fill_interpretasi_rekomendasi(header, rows, wilayah_txt, kew_value or "(Semua)")
+        word_path = generate_word_table_interpretasi(header, rows_filled, wilayah_txt)
 
         msg = (
             f"Selesai (TARGET {TARGET_RATIO*100:.2f}%): raw={len(raw)} | entitas={len(detail_view)} | "
             f"wilayah(keseluruhan)={len(agg_total)} | jenis(agregat)={len(agg_jenis_full)}"
-            + ("" if DOCX_AVAILABLE else "<br>python-docx tidak tersedia -> laporan Word dimatikan.")
+            + ("" if DOCX_AVAILABLE else " | python-docx tidak tersedia (Word OFF)")
         )
 
         return (
             kpi_md,
             summary_jenis, agg_total, agg_jenis_view, detail_view, verif_total,
-            p_summary, p_total, p_raw, p_detail, (word_path if word_path else None),
+            p_summary, p_total, p_raw, p_detail, p_verif,
             fig_umum, fig_sekolah, fig_khusus,
-            msg, analysis_text
+            msg,
+            llm_status,
+            (word_path if word_path else None)
         )
 
     except Exception as e:
@@ -1445,11 +1523,9 @@ Dashboard KPI:
 - Indeks IPLM FINAL (disesuaikan 33.88%)
 - Indeks Dasar (tanpa penyesuaian)
 
-Bell Curve:
-- Indeks_Dasar_0_100 per entitas (per jenis), hover menampilkan nama perpustakaan.
-
-Laporan Word (LLM):
-- Tabel: No | Dimensi | Nilai | Interpretasi | Rekomendasi
+UPDATE LLM:
+- LLM mengisi tabel "Interpretasi & Rekomendasi IPLM" dalam Word (.docx) yang bisa diunduh.
+- Nilai tetap dari aplikasi.
 """)
 
     state_df = gr.State(None)
@@ -1479,7 +1555,7 @@ Laporan Word (LLM):
     gr.Markdown("## Agregat Wilayah (Keseluruhan) — FIX avg3")
     out_agg_total = gr.DataFrame(interactive=False)
 
-    gr.Markdown("## Agregat Wilayah × Jenis — (ditampilkan sampai Indeks_Dasar_Agregat_0_100)")
+    gr.Markdown("## Agregat Wilayah x Jenis — (ditampilkan sampai Indeks Dasar)")
     out_agg_jenis = gr.DataFrame(interactive=False)
 
     gr.Markdown("## Detail Entitas (Final menempel dari wilayah)")
@@ -1498,15 +1574,16 @@ Laporan Word (LLM):
     gr.Markdown("### Perpustakaan Khusus")
     bell_khusus = gr.Plot(scale=1)
 
-    gr.Markdown("## Analisis Otomatis (opsional)")
-    analysis_out = gr.Markdown()
+    gr.Markdown("## Status LLM (Isi Interpretasi & Rekomendasi)")
+    llm_status_out = gr.Markdown()
 
     with gr.Row():
         dl_summary = gr.DownloadButton(label="Download Ringkasan (.xlsx)")
         dl_total = gr.DownloadButton(label="Download Agregat Wilayah (.xlsx)")
         dl_raw = gr.DownloadButton(label="Download Data Mentah (.xlsx)")
         dl_detail = gr.DownloadButton(label="Download Detail Entitas (.xlsx)")
-        dl_word = gr.DownloadButton(label="Download Laporan Word (.docx)" if DOCX_AVAILABLE else "Download Laporan Word (OFF)")
+        dl_verif = gr.DownloadButton(label="Download Kecukupan Sampel (.xlsx)")
+        dl_word = gr.DownloadButton(label="Download Word: Interpretasi & Rekomendasi (.docx)" if DOCX_AVAILABLE else "Download Word (OFF)")
 
     run_btn.click(
         fn=run_calc,
@@ -1514,9 +1591,11 @@ Laporan Word (LLM):
         outputs=[
             kpi_out,
             out_summary, out_agg_total, out_agg_jenis, out_detail, out_verif,
-            dl_summary, dl_total, dl_raw, dl_detail, dl_word,
+            dl_summary, dl_total, dl_raw, dl_detail, dl_verif,
             bell_umum, bell_sekolah, bell_khusus,
-            msg_out, analysis_out
+            msg_out,
+            llm_status_out,
+            dl_word
         ]
     )
 
@@ -1526,4 +1605,4 @@ Laporan Word (LLM):
         outputs=[state_df, state_raw, state_pop_kab, state_pop_prov, state_pop_khusus, state_meta, info_box, dd_prov, dd_kab, dd_kew]
     )
 
-demo.launch()
+demo.launch()