def _area_select(gdf: gpd.GeoDataFrame, min_area: float,
max_area: float) -> gpd.GeoDataFrame:
"""Select basins from lower to higher levels that are between min_area and max_area in area.
Start by working out if any basins at e.g. level 1 are selected.
Then move on to higher levels (smaller basins).
At each level, only add basins if the basin at the level below has not been added.
e.g. level 3 basins 411 to 419 will not be added if at level 2 basin 41 was added.
Can also be used as a method on a `gpd.GeoDataFrame`:
`gdf.area_select(min_area, max_area)`
:param gdf: hydrobasins geodataframe to traverse
:param min_area: minimum area of basin
:param max_area: maximum area of basin
:return: filtered geodataframe from any level (favouring lower levels) with area between min and max
"""
all_good_index: Set[int] = set()
good_index = set()
for level in range(gdf.LEVEL.min(), gdf.LEVEL.max() + 1):
gdf_lev = gdf[gdf.LEVEL == level]
gdf_too_large = gdf_lev[gdf.SUB_AREA > max_area]
gdf_just_right = gdf_lev[(gdf.SUB_AREA <= max_area)
& (gdf.SUB_AREA >= min_area)]
gdf_too_small = gdf_lev[gdf.SUB_AREA < min_area]
logger.debug(f'Level {level}')
logger.debug(f' too large: {len(gdf_too_large)}')
logger.debug(f' just right: {len(gdf_just_right)}')
logger.debug(f' too small: {len(gdf_too_small)}')
if len(gdf_just_right) == 0:
logger.debug(f'Skipping level: {level}')
continue
for irow_index in range(len(gdf_lev)):
row_index = gdf_lev.index[irow_index]
row = gdf_lev.iloc[irow_index]
if max_area > row.SUB_AREA > min_area:
larger_basin = gdf.find_next_level_larger(row.PFAF_ID)
larger_basin_index = larger_basin.index[0] if len(
larger_basin) else -1
if larger_basin_index not in all_good_index:
good_index.add(row_index)
all_good_index.add(row_index)
return gdf.loc[good_index]
