Update app.py

app.py

@@ -1,16 +1,18 @@
 # app.py
+# pip install keplergl pandas numpy geopandas shapely gradio requests openpyxl
+
 import os
 import io
 import time
+import json
+import tempfile
 import requests
 import pandas as pd
 import numpy as np
 import geopandas as gpd
-import matplotlib.pyplot as plt
 from shapely.geometry import Point
-import folium
 import gradio as gr
-from PIL import Image
+from keplergl import KeplerGl
 
 # ----------------------------
 # 設定
@@ -20,13 +22,15 @@ GSI_USER_AGENT = os.environ.get(
     "jp-gsi-geocoding-demo (contact: your_email@example.com)"  # 連絡先付き推奨
 )
 GSI_TIMEOUT_SEC = float(os.environ.get("GSI_TIMEOUT_SEC", "10"))
-GEOCODE_DELAY_SEC = float(os.environ.get("GSI_RATE_LIMIT_SEC", "0.5"))  # マナーとして少し待機
+
+# ★ sleep最小（0秒）
+GEOCODE_DELAY_SEC = float(os.environ.get("GSI_RATE_LIMIT_SEC", "0.0"))
+
 GSI_GEOCODE_URL = "https://msearch.gsi.go.jp/address-search/AddressSearch"
 
 CACHE_DIR = "data/cache"
 os.makedirs(CACHE_DIR, exist_ok=True)
 CACHE_PATH = os.path.join(CACHE_DIR, "geocode_cache.csv")
-DEFAULT_ZIP = "data/japan_ver85.zip"
 
 # ----------------------------
 # キャッシュ
@@ -47,18 +51,6 @@ def save_cache(df_cache):
     except Exception:
         pass
 
-# ----------------------------
-# Shapefile 読み込み
-# ----------------------------
-def load_gdf_from_zip(zip_path: str) -> gpd.GeoDataFrame:
-    gdf = gpd.read_file(f"zip://{zip_path}")  # , engine="pyogrio"
-    try:
-        if gdf.crs:
-            gdf = gdf.to_crs("EPSG:4326")
-    except Exception:
-        pass
-    return gdf
-
 # ----------------------------
 # 国土地理院 ジオコーダ
 # ----------------------------
@@ -70,26 +62,21 @@ def make_gsi_session() -> requests.Session:
 def gsi_geocode_once(address: str, session: requests.Session) -> tuple[float, float]:
     """
     国土地理院 住所検索APIを1回呼び出し、(lat, lon) を返す。失敗時は (nan, nan)。
-    返却座標は [lon, lat] なので順を入れ替えて返す。
+    APIは [lon, lat] を返すので順を入れ替える。
     """
     try:
-        # 空やnan文字列はスキップ
-        if not address or address.strip() == "" or address.strip().lower() in ("nan", "none"):
+        if not address or str(address).strip() == "" or str(address).strip().lower() in ("nan", "none"):
             return (np.nan, np.nan)
 
         resp = session.get(GSI_GEOCODE_URL, params={"q": address}, timeout=GSI_TIMEOUT_SEC)
         if not resp.ok:
             return (np.nan, np.nan)
         data = resp.json()
-        # 返り値は配列（候補リスト）。最上位候補を採用
         if isinstance(data, list) and len(data) > 0:
             feat = data[0]
             coords = (feat.get("geometry") or {}).get("coordinates") or []
             if isinstance(coords, (list, tuple)) and len(coords) >= 2:
-                lon, lat = coords[0], coords[1]
-                # 数値化チェック
-                lat = float(lat)
-                lon = float(lon)
+                lon, lat = float(coords[0]), float(coords[1])
                 return (lat, lon)
     except Exception:
         pass
@@ -117,8 +104,10 @@ def geocode_with_cache(addresses, CFs, use_internet=True):
             continue
 
         lat, lon = gsi_geocode_once(a, session)
-        # マナーとして小休止
-        time.sleep(GEOCODE_DELAY_SEC)
+
+        # ★ 最小スリープ（デフォルト0.0秒）
+        if GEOCODE_DELAY_SEC > 0:
+            time.sleep(GEOCODE_DELAY_SEC)
 
         # キャッシュ更新
         cache = cache[cache["address_input"] != a]
@@ -132,86 +121,100 @@ def geocode_with_cache(addresses, CFs, use_internet=True):
     return pd.DataFrame(results)
 
 # ----------------------------
-# 可視化（matplotlib）
+# Kepler.gl HTML 生成（ポイントのみ）
 # ----------------------------
-def plot_map_png(
-    gdf_pref: gpd.GeoDataFrame,
-    gdf_pts: gpd.GeoDataFrame,
-    line_width: float = 0.6,
-    marker_size: int = 24,
-    legend_shrink: float = 0.6,
-    legend_fontsize: int = 8,
-    figsize=(7, 7),
-) -> Image.Image:
-    fig, ax = plt.subplots(figsize=figsize)
-    gdf_pref.boundary.plot(ax=ax, linewidth=line_width, color="black")
-
-    gdf_pts_valid = gdf_pts[gdf_pts.geometry.notnull()]
-    if not gdf_pts_valid.empty:
-        cf_num = pd.to_numeric(
-            gdf_pts_valid.get("CF", pd.Series([np.nan]*len(gdf_pts_valid))),
-            errors="coerce"
-        )
-        gdf_pts_valid.assign(CF_num=cf_num).plot(
-            ax=ax,
-            column="CF_num",
-            cmap="OrRd",
-            markersize=max(2, int(marker_size)),
-            alpha=0.85,
-            legend=True,
-            legend_kwds={"shrink": legend_shrink},
-        )
-        try:
-            for _ax in fig.axes:
-                if _ax is not ax:
-                    _ax.tick_params(labelsize=legend_fontsize)
-        except Exception:
-            pass
-
-    ax.set_axis_off()
-    plt.tight_layout()
-    buf = io.BytesIO()
-    fig.savefig(buf, format="png", dpi=200)
-    plt.close(fig)
-    buf.seek(0)
-    return Image.open(buf)
-
-# ----------------------------
-# 可視化（folium）
-# ----------------------------
-def make_folium_html(gdf_pref: gpd.GeoDataFrame, gdf_pts: gpd.GeoDataFrame, marker_size: int = 24):
-    gdf_pts_valid = gdf_pts[gdf_pts.geometry.notnull()]
-    if not gdf_pts_valid.empty:
-        center_lat = gdf_pts_valid.geometry.y.median()
-        center_lon = gdf_pts_valid.geometry.x.median()
-        zoom = 6
-    else:
-        center_lat, center_lon, zoom = 35.6812, 139.7671, 5
-
-    m = folium.Map(location=[center_lat, center_lon], zoom_start=zoom)
-
+def make_kepler_html(df_points: pd.DataFrame, height: int = 640) -> str:
+    """
+    df_points は 'lat','lon','address_input','CF' を含む DataFrame を想定。
+    ポイントレイヤのみを Kepler.gl で描画し、HTMLを文字列で返す。
+    """
+    df_valid = df_points.dropna(subset=["lat", "lon"]).copy()
+    if df_valid.empty:
+        # 空のKeplerでもHTMLは返す
+        m = KeplerGl(height=height)
+        return m._repr_html_()
+
+    # ほどよい初期中心
+    center_lat = float(df_valid["lat"].median())
+    center_lon = float(df_valid["lon"].median())
+
+    # Kepler 設定（ポイントレイヤのみ）
+    config = {
+        "version": "v1",
+        "config": {
+            "visState": {
+                "filters": [],
+                "layers": [
+                    {
+                        "id": "point_layer",
+                        "type": "point",
+                        "config": {
+                            "dataId": "points",
+                            "label": "Points",
+                            "color": [18, 147, 154],
+                            "columns": {"lat": "lat", "lng": "lon"},
+                            "isVisible": True,
+                            "visConfig": {
+                                "radius": 10,            # 基本半径
+                                "opacity": 0.9,
+                                "outline": False
+                            }
+                        },
+                        "visualChannels": {
+                            # CF列が数値ならサイズに反映（なければ自動で固定半径）
+                            "sizeField": {"name": "CF", "type": "real"} if pd.to_numeric(df_valid.get("CF", pd.Series([])), errors="coerce").notna().any() else None,
+                            "sizeScale": "sqrt",
+                        },
+                    }
+                ],
+                "interactionConfig": {
+                    "tooltip": {
+                        "enabled": True,
+                        "fieldsToShow": {
+                            "points": [ {"name": "address_input", "format": None},
+                                        {"name": "CF", "format": None},
+                                        {"name": "lat", "format": None},
+                                        {"name": "lon", "format": None} ]
+                        },
+                        "compareMode": False,
+                        "compareType": "absolute"
+                    }
+                },
+                "layerBlending": "normal"
+            },
+            "mapState": {
+                "bearing": 0,
+                "pitch": 0,
+                "latitude": center_lat,
+                "longitude": center_lon,
+                "zoom": 6
+            },
+            "mapStyle": {
+                "styleType": "light",
+                "topLayerGroups": {},
+                "visibleLayerGroups": {"label": True, "road": True, "border": False, "building": False, "water": True, "land": True}
+            }
+        }
+    }
+
+    m = KeplerGl(height=height, config=config)
+    # Kepler は DataFrame の列名で自動解釈（lat/lon）
+    m.add_data(data=df_valid[["lat", "lon", "address_input", "CF"]], name="points")
+
+    # Gradioへは _repr_html_ をそのまま返すのが簡単
     try:
-        folium.GeoJson(gdf_pref.to_json(), name="prefecture").add_to(m)
+        return m._repr_html_()
     except Exception:
-        pass
-
-    circle_radius = max(3, int(marker_size // 3))
-
-    for _, r in gdf_pts_valid.iterrows():
-        lat, lon = r.geometry.y, r.geometry.x
-        popup = f"{r.get('address_input','(no addr)')}<br>CF:{r.get('CF','')}"
-        folium.CircleMarker(
-            location=(float(lat), float(lon)),
-            radius=circle_radius,
-            fill=True,
-            fill_opacity=0.9,
-            popup=popup,
-        ).add_to(m)
-
-    return m._repr_html_()
+        # 万一ノートブック外で不安定な場合はHTMLファイルを生成して読み戻す
+        with tempfile.NamedTemporaryFile(suffix=".html", delete=False) as f:
+            tmp = f.name
+        m.save_to_html(file_name=tmp, read_only=True)
+        with open(tmp, "r", encoding="utf-8") as fh:
+            html = fh.read()
+        return html
 
 # ----------------------------
-# 実行パイプライン
+# 実行パイプライン（ポイントのみ）
 # ----------------------------
 def _parse_indexer(x):
     try:
@@ -219,85 +222,53 @@ def _parse_indexer(x):
     except Exception:
         return x
 
-def run(zip_file, excel_file, sheet_name, header_row, address_col, power_col,
-        use_inet, line_width, marker_size, legend_shrink, legend_fontsize):
-    # Shapefile
-    if zip_file is not None and hasattr(zip_file, "name") and os.path.exists(zip_file.name):
-        zip_path = zip_file.name
-    elif os.path.exists(DEFAULT_ZIP):
-        zip_path = DEFAULT_ZIP
-    else:
-        empty_df = pd.DataFrame(columns=["address_input", "CF", "lat", "lon"])
-        return None, None, "", empty_df, "Shapefile の ZIP をアップロードするか、data/japan_ver85.zip を配置してください。"
+def run(excel_file, sheet_name, header_row, address_col, power_col, use_inet):
+    # Excel 読み込み
+    if excel_file is None or not hasattr(excel_file, "name"):
+        table_df = pd.DataFrame(columns=["address_input", "CF", "lat", "lon"])
+        return "", table_df, "Excelファイルを指定してください。"
 
     try:
-        gdf_pref = load_gdf_from_zip(zip_path)
+        df = pd.read_excel(excel_file.name, sheet_name=sheet_name, header=int(header_row))
     except Exception as e:
         empty_df = pd.DataFrame(columns=["address_input", "CF", "lat", "lon"])
-        return None, None, "", empty_df, f"行政界の読み込みに失敗しました: {e}"
+        return "", empty_df, f"Excel の読み込みに失敗しました: {e}"
 
-    # Excel→ジオコーディング
-    if excel_file is None or not hasattr(excel_file, "name"):
-        gdf_pts = gpd.GeoDataFrame(columns=["address_input", "CF", "lat", "lon"], geometry=[], crs="EPSG:4326")
-        table_df = pd.DataFrame(columns=["address_input", "CF", "lat", "lon"])
-    else:
-        try:
-            df = pd.read_excel(excel_file.name, sheet_name=sheet_name, header=int(header_row))
-        except Exception as e:
-            empty_df = pd.DataFrame(columns=["address_input", "CF", "lat", "lon"])
-            return None, None, "", empty_df, f"Excel の読み込みに失敗しました: {e}"
-
-        addr_series = df.iloc[:, address_col] if isinstance(address_col, int) else df[address_col]
-        cf_series = df.iloc[:, power_col] if isinstance(power_col, int) else df[power_col]
-
-        addresses = addr_series.astype(str).tolist()
-        cfs = cf_series.tolist()
+    addr_series = df.iloc[:, address_col] if isinstance(address_col, int) else df[address_col]
+    cf_series = df.iloc[:, power_col] if isinstance(power_col, int) else df[power_col]
 
-        geo_df = geocode_with_cache(addresses, cfs, use_internet=bool(use_inet))
-        table_df = geo_df[["address_input", "CF", "lat", "lon"]].copy()
+    addresses = addr_series.astype(str).tolist()
+    cfs = cf_series.tolist()
 
-        geometry = [
-            Point(lon, lat) if (pd.notna(lat) and pd.notna(lon)) else None
-            for lat, lon in zip(geo_df["lat"], geo_df["lon"])
-        ]
-        gdf_pts = gpd.GeoDataFrame(geo_df, geometry=geometry, crs="EPSG:4326")
+    geo_df = geocode_with_cache(addresses, cfs, use_internet=bool(use_inet))
+    table_df = geo_df[["address_input", "CF", "lat", "lon"]].copy()
 
-    # 図と地図
-    try:
-        img = plot_map_png(
-            gdf_pref, gdf_pts,
-            line_width=float(line_width),
-            marker_size=int(marker_size),
-            legend_shrink=float(legend_shrink),
-            legend_fontsize=int(legend_fontsize),
-        )
-    except Exception as e:
-        return None, None, "", table_df, f"静的描画に失敗しました: {e}"
+    # GeoDataFrame も一応整備（未使用だが将来の拡張用）
+    geometry = [
+        Point(lon, lat) if (pd.notna(lat) and pd.notna(lon)) else None
+        for lat, lon in zip(geo_df["lat"], geo_df["lon"])
+    ]
+    gdf_pts = gpd.GeoDataFrame(geo_df, geometry=geometry, crs="EPSG:4326")
 
+    # Kepler.gl（ポイントのみ）
     try:
-        html = make_folium_html(gdf_pref, gdf_pts, marker_size=int(marker_size))
+        html = make_kepler_html(table_df, height=640)
     except Exception as e:
-        html = f"<p>folium描画に失敗しました: {e}</p>"
+        html = f"<p>Kepler.gl描画に失敗しました: {e}</p>"
 
-    # 情報
+    # 情報（地物数のみ）
     info = []
-    info.append(f"都道府県レコード数: {len(gdf_pref)}")
-    if gdf_pref.crs:
-        info.append(f"PREF CRS: {gdf_pref.crs}")
     info.append(f"ポイント数（有効座標）: {int(gdf_pts.geometry.notnull().sum())} / {len(gdf_pts)}")
-    if not gdf_pts.empty and gdf_pts.crs:
-        info.append(f"PTS CRS: {gdf_pts.crs}")
 
-    return img, html, "\n".join(info), table_df, ""
+    return html, table_df, "\n".join(info)
 
 # ----------------------------
-# Gradio UI
+# Gradio UI（ポイントのみ）
 # ----------------------------
-with gr.Blocks(title="Japan Shapefile + Excel Geocoding Plotter (GSI)") as demo:
-    gr.Markdown("## japan_ver85.shp（ZIP） + Excel住所 → 日本地図にプロ���ト（凡例小・点大の調整可）")
+with gr.Blocks(title="Excel住所 → Kepler.gl（ポイントのみ）") as demo:
+    gr.Markdown("## Excelの住所を国土地理院APIでジオコーディング → Kepler.gl に **ポイントのみ** を描画")
 
     with gr.Row():
-        zip_in = gr.File(label="Shapefile (ZIP)", file_count="single", file_types=[".zip"])
         xlsx_in = gr.File(label="Excelファイル（住所付き）", file_count="single", file_types=[".xlsx", ".xls"])
 
     with gr.Row():
@@ -310,21 +281,12 @@ with gr.Blocks(title="Japan Shapefile + Excel Geocoding Plotter (GSI)") as demo:
 
     with gr.Row():
         use_inet = gr.Checkbox(label="国土地理院APIに問い合わせ（オフでキャッシュのみ使用）", value=True)
-        line_width = gr.Slider(0.2, 2.0, value=0.6, step=0.1, label="境界線の太さ")
-
-    # 見た目調整スライダ
-    with gr.Row():
-        marker_size = gr.Slider(4, 64, value=24, step=2, label="ポイントサイズ（matplotlib / folium）")
-        legend_shrink = gr.Slider(0.3, 1.0, value=0.6, step=0.05, label="凡例の縮小率（小さいほど小さく）")
-        legend_fontsize = gr.Slider(6, 16, value=8, step=1, label="凡例の目盛フォントサイズ")
 
     run_btn = gr.Button("描画")
 
-    out_img = gr.Image(label="静的地図（matplotlib）", type="pil")
-    out_html = gr.HTML(label="インタラクティブ地図（folium）")
-    out_info = gr.Textbox(label="メタ情報", lines=4)
+    out_html = gr.HTML(label="インタラクティブ地図（Kepler.gl：ポイントのみ）")
     out_table = gr.Dataframe(label="ジオコーディング結果（住所・緯度・経度・CF）", wrap=True)
-    out_err = gr.Markdown(label="エラー", visible=True)
+    out_info = gr.Textbox(label="メタ情報", lines=2)
 
     def _parse(x):
         try:
@@ -332,16 +294,15 @@ with gr.Blocks(title="Japan Shapefile + Excel Geocoding Plotter (GSI)") as demo:
         except Exception:
             return x
 
-    def app_run(zipf, xls, s, h, a, p, inet, lw, ms, lsh, lfs):
+    def app_run(xls, s, h, a, p, inet):
         return run(
-            zipf, xls, s, int(h), _parse(a), _parse(p),
-            inet, lw, ms, lsh, lfs
+            xls, s, int(h), _parse(a), _parse(p), inet
         )
 
     run_btn.click(
         fn=app_run,
-        inputs=[zip_in, xlsx_in, sheet, header_row, address_col, power_col, use_inet, line_width, marker_size, legend_shrink, legend_fontsize],
-        outputs=[out_img, out_html, out_info, out_table, out_err],
+        inputs=[xlsx_in, sheet, header_row, address_col, power_col, use_inet],
+        outputs=[out_html, out_table, out_info],
     )
 
 if __name__ == "__main__":