perf: add dilution to speed up.

cnmaps/drawing.py

@@ -237,6 +237,8 @@ def clip_clabels_by_map(
     if ax is None:
         ax = plt.gca()
     map_polygon = _transform_polygon(map_polygon, ccrs.PlateCarree(), ax.projection)
+    if not map_polygon.is_valid:
+        map_polygon = map_polygon.buffer(0)
 
     for cbt in clabel_text:
         point = sgeom.Point(cbt.get_position())

cnmaps/maps.py

@@ -198,17 +198,25 @@ def make_mask_array(self, lons: np.ndarray, lats: np.ndarray):
         return ~contains(self, lons, lats)
 
 
-def read_mapjson(fp, wgs84=True):
+def read_mapjson(fp, wgs84=True, dilution_interval=1):
     """
     读取geojson地图边界文件
 
     参数:
         fp (str, 可选): geojson文件名.
         wgs84 (bool, 可选): 是否使用 WGS84 坐标
+        dilution_interval (int, 可选): 稀疏间隔, 默认为1, 即不稀疏.
+                                        该值越大, 稀释得越强烈. 默认为1.
 
     返回值:
         MapPolygon: 地图边界对象
     """
+    if not isinstance(dilution_interval, int):
+        raise ValueError("dilution_interval必须为整数")
+
+    if dilution_interval < 1:
+        raise ValueError("dilution_interval必须大于等于1")
+
     with open(fp, encoding="utf-8") as f:
         map_json = orjson.loads(f.read())
 
@@ -221,11 +229,14 @@ def read_mapjson(fp, wgs84=True):
     if "Polygon" in geometry["type"]:
         for _coords in geometry["coordinates"]:
             for coords in _coords:
+                __coords = coords[::dilution_interval]
+                if len(__coords) < 3:
+                    __coords = coords
                 if wgs84:
-                    wgs84_coords = [gcj02_to_wgs84(*coord) for coord in coords]
+                    wgs84_coords = [gcj02_to_wgs84(*coord) for coord in __coords]
                     polygon_list.append(sgeom.Polygon(wgs84_coords))
                 else:
-                    polygon_list.append(sgeom.Polygon(coords))
+                    polygon_list.append(sgeom.Polygon(__coords))
 
         return MapPolygon(polygon_list)
 
@@ -295,6 +306,7 @@ def get_adm_maps(
     only_polygon: bool = False,
     wgs84=True,
     simplify=False,
+    dilution_interval=1,
     *args,
     **kwargs,
 ):
@@ -332,6 +344,8 @@ def get_adm_maps(
                                 Defaults to False.
         wgs84 (bool, 可选): 是否使用 WGS84 坐标系, 若为 True 则转为 WGS84 坐标,
                                 若为 False 则使用高德默认的 GCJ02 火星坐标。Defaults to True.
+        dilution_interval (int, 可选): 稀疏间隔, 默认为1, 即不稀疏.
+                                        该值越大, 稀释得越强烈. Defaults to 1.
         simplify  (bool, 可选): 是否对边界进行简化, 若为 True 则进行简化处理, 否则不做简化。Defaults to True.
 
     异常:
@@ -412,7 +426,9 @@ def get_adm_maps(
     elif level in ["区", "县", "区县", "区/县"]:
         level_sql = "level='区县'"
     else:
-        raise ValueError(f'无法识别level等级: {level}, level参数请从"国", "省", "市", "区县"中选择')
+        raise ValueError(
+            f'无法识别level等级: {level}, level参数请从"国", "省", "市", "区县"中选择'
+        )
 
     meta_sql = (
         "SELECT country, province, city, district, level, source, kind"
@@ -432,13 +448,16 @@ def get_adm_maps(
     map_polygons = []
     for path in gemo_rows:
         mapjson = read_mapjson(
-            os.path.join(DATA_DIR, "geojson.min/", path[0]), wgs84=wgs84
+            os.path.join(DATA_DIR, "geojson.min/", path[0]),
+            wgs84=wgs84,
+            dilution_interval=dilution_interval,
         )
 
         map_polygons.append(mapjson)
 
     gdf = gpd.GeoDataFrame(
-        data=meta_rows, columns=["国家", "省/直辖市", "市", "区/县", "级别", "来源", "类型"]
+        data=meta_rows,
+        columns=["国家", "省/直辖市", "市", "区/县", "级别", "来源", "类型"],
     )
     gdf.set_geometry(map_polygons, inplace=True)
 

tests/test_drawing.py

@@ -25,14 +25,19 @@
 sample_districts = [random.choice(districts) for _ in range(100)]
 
 map_args = [
-    {"only_polygon": True, "record": "first", "name": "中华人民共和国", "simplify": True}
+    {
+        "only_polygon": True,
+        "record": "first",
+        "name": "中华人民共和国",
+        "dilution_interval": 10,
+    }
 ] + [
     {
         "province": p,
         "only_polygon": True,
         "record": "first",
         "name": p,
-        "simplify": True,
+        "dilution_interval": 10,
     }
     for p in ["黑龙江省", "内蒙古自治区"]
 ]
@@ -41,15 +46,22 @@
 def test_draw_maps():
     """测试多地图绘制功能"""
     map_args = (
-        [{"level": "国", "name": "中华人民共和国", "simplify": True}]
-        + [{"level": "省", "name": "中华人民共和国-分省", "simplify": True}]
-        + [{"level": "国", "engine": "geopandas", "name": "中华人民共和国", "simplify": True}]
+        [{"level": "国", "name": "中华人民共和国", "dilution_interval": 10}]
+        + [{"level": "省", "name": "中华人民共和国-分省", "dilution_interval": 10}]
+        + [
+            {
+                "level": "国",
+                "engine": "geopandas",
+                "name": "中华人民共和国",
+                "dilution_interval": 10,
+            }
+        ]
         + [
             {
                 "level": "省",
                 "engine": "geopandas",
                 "name": "中华人民共和国-分省",
-                "simplify": True,
+                "dilution_interval": 10,
             }
         ]
     )
@@ -221,7 +233,7 @@ def test_clip_clabel():
 
     lons, lats, data = load_dem()
 
-    map_polygon = get_adm_maps(record="first", only_polygon=True)
+    map_polygon = get_adm_maps(record="first", only_polygon=True, dilution_interval=10)
     fig = plt.figure(figsize=(10, 10))
     ax = fig.add_subplot(111, projection=ccrs.PlateCarree())
     contours = ax.contour(
@@ -272,7 +284,9 @@ def test_projection():
     lons, lats, data = load_dem()
 
     for projection in PROJECTIONS:
-        map_polygon = get_adm_maps(province="河南省", record="first", only_polygon=True)
+        map_polygon = get_adm_maps(
+            record="first", only_polygon=True, dilution_interval=10
+        )
         fig = plt.figure(figsize=(10, 10))
         ax = fig.add_subplot(111, projection=projection)
         contours = ax.contourf(

tests/test_map.py

@@ -127,6 +127,16 @@ def test_get_map_by_fp():
         read_mapjson(fp)
 
 
+def test_read_mapjson_with_dilution():
+    pattern = os.path.join(
+        os.path.dirname(os.path.abspath(__file__)),
+        "../cnmaps/data/geojson.min/*/*/*/*.geojson",
+    )
+    fps = sorted(glob(pattern))
+    for fp in fps:
+        read_mapjson(fp, dilution_interval=10)
+
+
 def test_map_load():
     """测试各级地图数量是否完整，以及各种规则是否都能加载成功."""
     assert len(get_adm_maps(level="国")) == 2
@@ -153,7 +163,11 @@ def test_map_load():
     )
 
     beijing = get_adm_maps(city="北京市")[0]
-    assert beijing["市"] == "北京市" and beijing["区/县"] is None and beijing["级别"] == "市"
+    assert (
+        beijing["市"] == "北京市"
+        and beijing["区/县"] is None
+        and beijing["级别"] == "市"
+    )
 
     chaoyang = get_adm_maps(district="朝阳区")
     assert len(chaoyang) == 2
@@ -261,7 +275,7 @@ def test_province_orthogonality():
 
     couples = sorted([couple for couple in combinations(province_names, r=2)])
 
-    for (one, another) in couples:
+    for one, another in couples:
         assert (
             get_adm_maps(province=one)[0]["geometry"]
             & get_adm_maps(province=another)[0]["geometry"]
@@ -295,7 +309,7 @@ def test_city_orthogonality():
             sorted(("阿勒泰地区", "北屯市")),
         ]
 
-        for (one, another) in couples:
+        for one, another in couples:
             if sorted([one, another]) in problem_set:
                 continue
             area = (
@@ -312,7 +326,7 @@ def test_province_union():
 
     couples = sorted([couple for couple in combinations(province_names, r=2)])
 
-    for (one, another) in couples:
+    for one, another in couples:
         _ = (
             get_adm_maps(province=one)[0]["geometry"]
             + get_adm_maps(province=another)[0]["geometry"]
@@ -326,7 +340,7 @@ def test_province_difference():
 
     couples = sorted([couple for couple in product(province_names, repeat=2)])
 
-    for (one, another) in couples:
+    for one, another in couples:
         _ = (
             get_adm_maps(province=one)[0]["geometry"]
             - get_adm_maps(province=another)[0]["geometry"]
@@ -360,13 +374,17 @@ def test_get_extent():
 
 def test_only_polygon_and_record():
     """测试only_polygon参数和record参数功能."""
-    polygons = get_adm_maps(city="北京市", record="all", level="区县", only_polygon=True)
+    polygons = get_adm_maps(
+        city="北京市", record="all", level="区县", only_polygon=True
+    )
     assert isinstance(polygons, list)
     assert len(polygons) == 16
     for p in polygons:
         assert isinstance(p, MapPolygon)
 
-    polygon = get_adm_maps(city="北京市", record="first", level="区县", only_polygon=True)
+    polygon = get_adm_maps(
+        city="北京市", record="first", level="区县", only_polygon=True
+    )
     assert isinstance(polygon, MapPolygon)
 
     meta = get_adm_maps(city="北京市", record="first", level="市")

tests/test_perf.py

@@ -34,7 +34,7 @@
     "only_polygon": True,
     "record": "first",
     "name": "中华人民共和国",
-    "simplify": True,
+    "dilution_interval": 10,
 }
 
 
@@ -233,7 +233,9 @@ def test_clip_clabel(benchmark):
     def inner():
         lons, lats, data = load_dem()
 
-        map_polygon = get_adm_maps(record="first", only_polygon=True, simplify=True)
+        map_polygon = get_adm_maps(
+            record="first", only_polygon=True, dilution_interval=10
+        )
         fig = plt.figure(figsize=(10, 10))
         ax = fig.add_subplot(111, projection=ccrs.PlateCarree())
         contours = ax.contour(
@@ -269,7 +271,9 @@ def test_projection(benchmark):
     def inner():
         lons, lats, data = load_dem()
 
-        map_polygon = get_adm_maps(record="first", only_polygon=True, simplify=True)
+        map_polygon = get_adm_maps(
+            record="first", only_polygon=True, dilution_interval=10
+        )
         fig = plt.figure(figsize=(10, 10))
         ax = fig.add_subplot(111, projection=ccrs.Orthographic(central_longitude=100))
         contours = ax.contourf(
@@ -304,7 +308,10 @@ def inner():
         mask_array = np.load(casefp)
 
         map_polygon = get_adm_maps(
-            province="宁夏回族自治区", only_polygon=True, record="first", wgs84=False
+            province="宁夏回族自治区",
+            only_polygon=True,
+            record="first",
+            wgs84=False,
         )
 
         lons, lats, data = load_dem()
@@ -350,7 +357,12 @@ def inner():
         lat = np.linspace(0, 60, 1000)
         lons, lats = np.meshgrid(lon, lat)
 
-        china = get_adm_maps(level="国", record="first", only_polygon=True, wgs84=False)
+        china = get_adm_maps(
+            level="国",
+            record="first",
+            only_polygon=True,
+            wgs84=False,
+        )
         china_maskout_array = china.make_mask_array(lons, lats)
 
         assert (china_maskout_array == mask_array).all()
@@ -362,7 +374,12 @@ def inner():
         lat = np.linspace(0, 60, 1000)
         lons, lats = np.meshgrid(lon, lat)
 
-        china = get_adm_maps(level="国", record="first", only_polygon=True, wgs84=True)
+        china = get_adm_maps(
+            level="国",
+            record="first",
+            only_polygon=True,
+            wgs84=True,
+        )
         china_maskout_array = china.make_mask_array(lons, lats)
 
         assert (china_maskout_array == mask_array).all()