ssb-sgis 1.0.3__py3-none-any.whl → 1.0.5__py3-none-any.whl

sgis/geopandas_tools/polygons_as_rings.py

@@ -8,6 +8,7 @@ from geopandas import GeoSeries
 from geopandas.array import GeometryArray
 from numpy.typing import NDArray
 from pyproj import CRS
+from shapely import difference
 from shapely import get_coordinates
 from shapely import get_exterior_ring
 from shapely import get_interior_ring
@@ -320,14 +321,14 @@ class PolygonsAsRings:
 
         exterior = self.rings.loc[self.is_exterior].sort_index()
         assert exterior.shape == (len(self.gdf),)
+        nonempty_exteriors = exterior.loc[lambda x: x.notna()]
+        empty_exteriors = exterior.loc[lambda x: x.isna()]
 
         nonempty_interiors = self.rings.loc[self.is_interior]
 
         if not len(nonempty_interiors):
-            try:
-                return make_valid(polygons(exterior.values))
-            except Exception:
-                return _geoms_to_linearrings_fallback(exterior).values
+            nonempty_exteriors.loc[:] = make_valid(polygons(nonempty_exteriors.values))
+            return pd.concat([empty_exteriors, nonempty_exteriors]).sort_index().values
 
         empty_interiors = pd.Series(
             [None for _ in range(len(self.gdf) * self.max_rings)],
@@ -343,10 +344,41 @@ class PolygonsAsRings:
         )
         assert interiors.shape == (len(self.gdf), self.max_rings), interiors.shape
 
-        try:
-            return make_valid(polygons(exterior.values, interiors.values))
-        except Exception:
-            return _geoms_to_linearrings_fallback(exterior, interiors).values
+        interiors = interiors.loc[
+            interiors.index.get_level_values(1).isin(
+                nonempty_exteriors.index.get_level_values(1)
+            )
+        ]
+        assert interiors.index.get_level_values(1).equals(
+            nonempty_exteriors.index.get_level_values(1)
+        )
+
+        # nan gives TypeError in shapely.polygons. None does not.
+        for i, _ in enumerate(interiors.columns):
+            interiors.loc[interiors.iloc[:, i].isna(), i] = None
+        nonempty_exteriors.loc[nonempty_exteriors.isna()] = None
+
+        # construct polygons with holes
+        polys = make_valid(
+            polygons(
+                nonempty_exteriors.values,
+                interiors.values,
+            )
+        )
+
+        # interiors might have moved (e.g. snapped) so that they are not within the exterior
+        # these interiors will not be holes, so we need to erase them manually
+        interiors_as_polys = make_valid(polygons(interiors.values))
+        # merge interior polygons into 1d array
+        interiors_as_polys = np.array(
+            [
+                make_valid(unary_union(interiors_as_polys[i, :]))
+                for i in range(interiors_as_polys.shape[0])
+            ]
+        )
+        # erase rowwise
+        nonempty_exteriors.loc[:] = make_valid(difference(polys, interiors_as_polys))
+        return pd.concat([empty_exteriors, nonempty_exteriors]).sort_index().values
 
 
 def get_linearring_series(geoms: GeoDataFrame | GeoSeries) -> pd.Series:

sgis/geopandas_tools/sfilter.py

@@ -4,7 +4,9 @@ import numpy as np
 import pandas as pd
 from geopandas import GeoDataFrame
 from geopandas import GeoSeries
+from geopandas import __version__ as geopandas_version
 from shapely import Geometry
+from shapely import STRtree
 
 from .conversion import to_gdf
 
@@ -15,6 +17,7 @@ def sfilter(
     gdf: GeoDataFrame | GeoSeries,
     other: GeoDataFrame | GeoSeries | Geometry,
     predicate: str = "intersects",
+    distance: int | float | None = None,
 ) -> GeoDataFrame:
     """Filter a GeoDataFrame or GeoSeries by spatial predicate.
 
@@ -29,6 +32,7 @@ def sfilter(
         gdf: The GeoDataFrame.
         other: The geometry object to filter 'gdf' by.
         predicate: Spatial predicate to use. Defaults to 'intersects'.
+        distance: Max distance to allow if predicate=="dwithin".
 
     Returns:
         A copy of 'gdf' with only the rows matching the
@@ -66,7 +70,7 @@ def sfilter(
     Also equivelent to using the intersects method, which
     is often a lot slower since df2 must be dissolved:
 
-    >>> df1.loc[df1.intersects(df2.unary_union)]
+    >>> df1.loc[df1.intersects(df2.union_all())]
                       geometry
     0  POINT (0.00000 0.00000)
 
@@ -76,7 +80,7 @@ def sfilter(
 
     other = _sfilter_checks(other, crs=gdf.crs)
 
-    indices = _get_sfilter_indices(gdf, other, predicate)
+    indices = _get_sfilter_indices(gdf, other, predicate, distance)
 
     return gdf.iloc[indices]
 
@@ -85,6 +89,7 @@ def sfilter_split(
     gdf: GeoDataFrame | GeoSeries,
     other: GeoDataFrame | GeoSeries | Geometry,
     predicate: str = "intersects",
+    distance: int | float | None = None,
 ) -> tuple[GeoDataFrame, GeoDataFrame]:
     """Split a GeoDataFrame or GeoSeries by spatial predicate.
 
@@ -95,6 +100,7 @@ def sfilter_split(
         gdf: The GeoDataFrame.
         other: The geometry object to filter 'gdf' by.
         predicate: Spatial predicate to use. Defaults to 'intersects'.
+        distance: Max distance to allow if predicate=="dwithin".
 
     Returns:
         A tuple of GeoDataFrames, one with the rows that match the spatial predicate
@@ -135,7 +141,7 @@ def sfilter_split(
     Also equivelent to using the intersects method, which
     is often slower since df2 must be dissolved:
 
-    >>> filt = df1.intersects(df2.unary_union)
+    >>> filt = df1.intersects(df2.union_all())
     >>> intersecting = df1.loc[filt]
     >>> not_intersecting = df1.loc[~filt]
 
@@ -145,7 +151,7 @@ def sfilter_split(
 
     other = _sfilter_checks(other, crs=gdf.crs)
 
-    indices = _get_sfilter_indices(gdf, other, predicate)
+    indices = _get_sfilter_indices(gdf, other, predicate, distance)
 
     return (
         gdf.iloc[indices],
@@ -157,6 +163,7 @@ def sfilter_inverse(
     gdf: GeoDataFrame | GeoSeries,
     other: GeoDataFrame | GeoSeries | Geometry,
     predicate: str = "intersects",
+    distance: int | float | None = None,
 ) -> GeoDataFrame | GeoSeries:
     """Filter a GeoDataFrame or GeoSeries by inverse spatial predicate.
 
@@ -166,6 +173,7 @@ def sfilter_inverse(
         gdf: The GeoDataFrame or GeoSeries.
         other: The geometry object to filter 'gdf' by.
         predicate: Spatial predicate to use. Defaults to 'intersects'.
+        distance: Max distance to allow if predicate=="dwithin".
 
     Returns:
         A copy of 'gdf' with only the rows that do not match the
@@ -202,7 +210,7 @@ def sfilter_inverse(
     Also equivelent to using the intersects method, which
     is often slower since df2 must be dissolved:
 
-    >>> not_intersecting = df1.loc[~df1.intersects(df2.unary_union)]
+    >>> not_intersecting = df1.loc[~df1.intersects(df2.union_all())]
 
     """
     if not isinstance(gdf, (GeoDataFrame | GeoSeries)):
@@ -210,7 +218,7 @@ def sfilter_inverse(
 
     other = _sfilter_checks(other, crs=gdf.crs)
 
-    indices = _get_sfilter_indices(gdf, other, predicate)
+    indices = _get_sfilter_indices(gdf, other, predicate, distance)
 
     return gdf.iloc[pd.Index(range(len(gdf))).difference(pd.Index(indices))]
 
@@ -243,6 +251,7 @@ def _get_sfilter_indices(
     left: GeoDataFrame | GeoSeries,
     right: GeoDataFrame | GeoSeries | Geometry,
     predicate: str,
+    distance: int | float | None,
 ) -> np.ndarray:
     """Compute geometric comparisons and get matching indices.
 
@@ -276,17 +285,25 @@ def _get_sfilter_indices(
             # contains is a faster predicate
             # see discussion at https://github.com/geopandas/geopandas/pull/1421
             predicate = "contains"
-            sindex = left.sindex
+            sindex, kwargs = _get_spatial_tree(left)
             input_geoms = right.geometry if isinstance(right, GeoDataFrame) else right
         else:
             # all other predicates are symmetric
             # keep them the same
-            sindex = right.sindex
+            sindex, kwargs = _get_spatial_tree(right)
             input_geoms = left.geometry if isinstance(left, GeoDataFrame) else left
 
-    l_idx, r_idx = sindex.query(input_geoms, predicate=predicate, sort=False)
+    l_idx, r_idx = sindex.query(
+        input_geoms, predicate=predicate, distance=distance, **kwargs
+    )
 
     if original_predicate == "within":
         return np.sort(np.unique(r_idx))
 
     return np.sort(np.unique(l_idx))
+
+
+def _get_spatial_tree(df):
+    if int(geopandas_version[0]) >= 1:
+        return df.sindex, {"sort": False}
+    return STRtree(df.geometry.values), {}

sgis/io/dapla_functions.py

@@ -1,24 +1,37 @@
 """Functions for reading and writing GeoDataFrames in Statistics Norway's GCS Dapla."""
 
+import json
+import multiprocessing
+import os
+from collections.abc import Iterable
 from pathlib import Path
 
 import dapla as dp
 import geopandas as gpd
 import joblib
 import pandas as pd
+import pyarrow
+import shapely
 from geopandas import GeoDataFrame
+from geopandas import GeoSeries
 from geopandas.io.arrow import _geopandas_to_arrow
 from pandas import DataFrame
-from pyarrow import parquet
+from pyarrow import ArrowInvalid
+
+from ..geopandas_tools.sfilter import sfilter
+
+PANDAS_FALLBACK_INFO = " Set pandas_fallback=True to ignore this error."
 
 
 def read_geopandas(
-    gcs_path: str | Path | list[str | Path],
+    gcs_path: str | Path | list[str | Path] | tuple[str | Path] | GeoSeries,
     pandas_fallback: bool = False,
     file_system: dp.gcs.GCSFileSystem | None = None,
+    mask: GeoSeries | GeoDataFrame | shapely.Geometry | tuple | None = None,
+    threads: int | None = None,
     **kwargs,
 ) -> GeoDataFrame | DataFrame:
-    """Reads geoparquet or other geodata from a file on GCS.
+    """Reads geoparquet or other geodata from one or more files on GCS.
 
     If the file has 0 rows, the contents will be returned as a pandas.DataFrame,
     since geopandas does not read and write empty tables.
@@ -33,6 +46,11 @@ def read_geopandas(
             not be read with geopandas and the number of rows is more than 0. If True,
             the file will be read with pandas if geopandas fails.
         file_system: Optional file system.
+        mask: Optional geometry mask to keep only intersecting geometries.
+            If 'gcs_path' is an iterable of multiple paths, only the files
+            with a bbox that intersects the mask are read, then filtered by location.
+        threads: Number of threads to use if reading multiple files. Defaults to
+            the number of files to read or the number of available threads (if lower).
         **kwargs: Additional keyword arguments passed to geopandas' read_parquet
             or read_file, depending on the file type.
 
@@ -42,14 +60,52 @@ def read_geopandas(
     if file_system is None:
         file_system = dp.FileClient.get_gcs_file_system()
 
-    if isinstance(gcs_path, (list, tuple)):
+    if not isinstance(gcs_path, (str | Path | os.PathLike)):
         kwargs |= {"file_system": file_system, "pandas_fallback": pandas_fallback}
+
+        if mask is not None:
+            if not isinstance(gcs_path, GeoSeries):
+                bounds_series: GeoSeries = get_bounds_series(
+                    gcs_path,
+                    file_system,
+                    threads=threads,
+                    pandas_fallback=pandas_fallback,
+                )
+            else:
+                bounds_series = gcs_path
+            new_bounds_series = sfilter(bounds_series, mask)
+            if not len(new_bounds_series):
+                if isinstance(kwargs.get("columns"), Iterable):
+                    cols = {col: [] for col in kwargs["columns"]}
+                else:
+                    cols = {}
+                    for path in bounds_series.index:
+                        try:
+                            cols |= {col: [] for col in _get_columns(path, file_system)}
+                        except ArrowInvalid as e:
+                            if file_system.isfile(path):
+                                raise ArrowInvalid(e, path) from e
+
+                return GeoDataFrame(cols | {"geometry": []})
+            paths = list(new_bounds_series.index)
+        else:
+            if isinstance(gcs_path, GeoSeries):
+                paths = list(gcs_path.index)
+            else:
+                paths = list(gcs_path)
+
+        if threads is None:
+            threads = min(len(gcs_path), int(multiprocessing.cpu_count())) or 1
+
         # recursive read with threads
-        with joblib.Parallel(n_jobs=len(gcs_path), backend="threading") as parallel:
+        with joblib.Parallel(n_jobs=threads, backend="threading") as parallel:
             dfs: list[GeoDataFrame] = parallel(
-                joblib.delayed(read_geopandas)(x, **kwargs) for x in gcs_path
+                joblib.delayed(read_geopandas)(x, **kwargs) for x in paths
             )
-        return pd.concat(dfs)
+        df = pd.concat(dfs)
+        if mask is not None:
+            return sfilter(df, mask)
+        return df
 
     if not isinstance(gcs_path, str):
         try:
@@ -60,20 +116,26 @@ def read_geopandas(
     if "parquet" in gcs_path or "prqt" in gcs_path:
         with file_system.open(gcs_path, mode="rb") as file:
             try:
-                return gpd.read_parquet(file, **kwargs)
+                df = gpd.read_parquet(file, **kwargs)
             except ValueError as e:
                 if "Missing geo metadata" not in str(e) and "geometry" not in str(e):
-                    raise e
+                    raise e.__class__(
+                        f"{e.__class__.__name__}: {e} for {gcs_path}."
+                    ) from e
                 df = dp.read_pandas(gcs_path, **kwargs)
 
                 if pandas_fallback or not len(df):
                     return df
                 else:
-                    raise e
+                    more_txt = PANDAS_FALLBACK_INFO if not len(df) else ""
+                    raise e.__class__(
+                        f"{e.__class__.__name__}: {e} for {df}." + more_txt
+                    ) from e
+
     else:
         with file_system.open(gcs_path, mode="rb") as file:
             try:
-                return gpd.read_file(file, **kwargs)
+                df = gpd.read_file(file, **kwargs)
             except ValueError as e:
                 if "Missing geo metadata" not in str(e) and "geometry" not in str(e):
                     raise e
@@ -82,7 +144,144 @@ def read_geopandas(
                 if pandas_fallback or not len(df):
                     return df
                 else:
-                    raise e
+                    more_txt = PANDAS_FALLBACK_INFO if not len(df) else ""
+                    raise e.__class__(
+                        f"{e.__class__.__name__}: {e} for {df}. " + more_txt
+                    ) from e
+
+    if mask is not None:
+        return sfilter(df, mask)
+    return df
+
+
+def _get_bounds_parquet(
+    path: str | Path, file_system: dp.gcs.GCSFileSystem, pandas_fallback: bool = False
+) -> tuple[list[float], dict] | tuple[None, None]:
+    with file_system.open(path) as f:
+        try:
+            num_rows = pyarrow.parquet.read_metadata(f).num_rows
+        except ArrowInvalid as e:
+            if not file_system.isfile(f):
+                return None, None
+            raise ArrowInvalid(e, path) from e
+        if not num_rows:
+            return None, None
+        meta = pyarrow.parquet.read_schema(f).metadata
+    try:
+        meta = json.loads(meta[b"geo"])["columns"]["geometry"]
+    except KeyError as e:
+        if pandas_fallback:
+            return None, None
+        raise KeyError(
+            f"{e.__class__.__name__}: {e} for {path}." + PANDAS_FALLBACK_INFO,
+            # f"{num_rows=}",
+            # meta,
+        ) from e
+    return meta["bbox"], meta["crs"]
+
+
+def _get_columns(path: str | Path, file_system: dp.gcs.GCSFileSystem) -> pd.Index:
+    with file_system.open(path) as f:
+        schema = pyarrow.parquet.read_schema(f)
+        index_cols = _get_index_cols(schema)
+        return pd.Index(schema.names).difference(index_cols)
+
+
+def _get_index_cols(schema: pyarrow.Schema) -> list[str]:
+    cols = json.loads(schema.metadata[b"pandas"])["index_columns"]
+    return [x for x in cols if not isinstance(x, dict)]
+
+
+def get_bounds_series(
+    paths: list[str | Path] | tuple[str | Path],
+    file_system: dp.gcs.GCSFileSystem | None = None,
+    threads: int | None = None,
+    pandas_fallback: bool = False,
+) -> GeoSeries:
+    """Get a GeoSeries with file paths as indexes and the file's bounds as values.
+
+    The returned GeoSeries can be used as the first argument of 'read_geopandas'
+    along with the 'mask' keyword.
+
+    Args:
+        paths: Iterable of file paths in gcs.
+        file_system: Optional instance of dp.gcs.GCSFileSystem.
+            If None, an instance is created within the function.
+            Note that this is slower in long loops.
+        threads: Number of threads to use if reading multiple files. Defaults to
+            the number of files to read or the number of available threads (if lower).
+        pandas_fallback: If False (default), an exception is raised if the file has
+            no geo metadata. If True, the geometry value is set to None for this file.
+
+    Returns:
+        A geopandas.GeoSeries with file paths as indexes and bounds as values.
+
+    Examples:
+    ---------
+    >>> import sgis as sg
+    >>> import dapla as dp
+    >>> file_system = dp.FileClient.get_gcs_file_system()
+    >>> all_paths = file_system.ls("...")
+
+    Get the bounds of all your file paths, indexed by path.
+
+    >>> bounds_series = sg.get_bounds_series(all_paths, file_system)
+    >>> bounds_series
+    .../0301.parquet    POLYGON ((273514.334 6638380.233, 273514.334 6...
+    .../1101.parquet    POLYGON ((6464.463 6503547.192, 6464.463 65299...
+    .../1103.parquet    POLYGON ((-6282.301 6564097.347, -6282.301 660...
+    .../1106.parquet    POLYGON ((-46359.891 6622984.385, -46359.891 6...
+    .../1108.parquet    POLYGON ((30490.798 6551661.467, 30490.798 658...
+                                                                                                            ...
+    .../5628.parquet    POLYGON ((1019391.867 7809550.777, 1019391.867...
+    .../5630.parquet    POLYGON ((1017907.145 7893398.317, 1017907.145...
+    .../5632.parquet    POLYGON ((1075687.587 7887714.263, 1075687.587...
+    .../5634.parquet    POLYGON ((1103447.451 7874551.663, 1103447.451...
+    .../5636.parquet    POLYGON ((1024129.618 7838961.91, 1024129.618 ...
+    Length: 357, dtype: geometry
+
+    Make a grid around the total bounds of the files,
+    and read geometries intersecting with the mask in a loop.
+
+    >>> grid = sg.make_grid(bounds_series, 10_000)
+    >>> for mask in grid.geometry:
+    ...     df = sg.read_geopandas(
+    ...         bounds_series,
+    ...         mask=mask,
+    ...         file_system=file_system,
+    ...     )
+
+    """
+    if file_system is None:
+        file_system = dp.FileClient.get_gcs_file_system()
+
+    if threads is None:
+        threads = min(len(paths), int(multiprocessing.cpu_count())) or 1
+
+    with joblib.Parallel(n_jobs=threads, backend="threading") as parallel:
+        bounds: list[tuple[list[float], dict]] = parallel(
+            joblib.delayed(_get_bounds_parquet)(
+                path, file_system=file_system, pandas_fallback=pandas_fallback
+            )
+            for path in paths
+        )
+    crss = {json.dumps(x[1]) for x in bounds}
+    crss = {
+        crs
+        for crs in crss
+        if not any(str(crs).lower() == txt for txt in ["none", "null"])
+    }
+    if not crss:
+        crs = None
+    elif len(crss) == 1:
+        crs = next(iter(crss))
+    else:
+        raise ValueError(f"crs mismatch: {crss}")
+    return GeoSeries(
+        [shapely.box(*bbox[0]) if bbox[0] is not None else None for bbox in bounds],
+        index=paths,
+        crs=crs,
+    )
 
 
 def write_geopandas(
@@ -91,6 +290,7 @@ def write_geopandas(
     overwrite: bool = True,
     pandas_fallback: bool = False,
     file_system: dp.gcs.GCSFileSystem | None = None,
+    write_covering_bbox: bool = False,
     **kwargs,
 ) -> None:
     """Writes a GeoDataFrame to the speficied format.
@@ -106,6 +306,13 @@ def write_geopandas(
             not be written with geopandas and the number of rows is more than 0. If True,
             the file will be written without geo-metadata if >0 rows.
         file_system: Optional file sustem.
+        write_covering_bbox: Writes the bounding box column for each row entry with column name "bbox".
+            Writing a bbox column can be computationally expensive, but allows you to specify
+            a bbox in : func:read_parquet for filtered reading.
+            Note: this bbox column is part of the newer GeoParquet 1.1 specification and should be
+            considered as experimental. While writing the column is backwards compatible, using it
+            for filtering may not be supported by all readers.
+
         **kwargs: Additional keyword arguments passed to parquet.write_table
             (for parquet) or geopandas' to_file method (if not parquet).
     """
@@ -118,25 +325,37 @@ def write_geopandas(
     if not overwrite and exists(gcs_path):
         raise ValueError("File already exists.")
 
-    if file_system is None:
-        file_system = dp.FileClient.get_gcs_file_system()
-
     if not isinstance(df, GeoDataFrame):
         raise ValueError("DataFrame must be GeoDataFrame.")
 
+    if file_system is None:
+        file_system = dp.FileClient.get_gcs_file_system()
+
     if not len(df):
         if pandas_fallback:
-            df.geometry = df.geometry.astype(str)
             df = pd.DataFrame(df)
-        dp.write_pandas(df, gcs_path, **kwargs)
+            df.geometry = df.geometry.astype(str)
+            df.geometry = None
+        try:
+            dp.write_pandas(df, gcs_path, **kwargs)
+        except Exception as e:
+            more_txt = PANDAS_FALLBACK_INFO if not pandas_fallback else ""
+            raise e.__class__(
+                f"{e.__class__.__name__}: {e} for {df}. " + more_txt
+            ) from e
         return
 
     file_system = dp.FileClient.get_gcs_file_system()
 
     if ".parquet" in gcs_path or "prqt" in gcs_path:
         with file_system.open(gcs_path, mode="wb") as buffer:
-            table = _geopandas_to_arrow(df, index=df.index, schema_version=None)
-            parquet.write_table(table, buffer, compression="snappy", **kwargs)
+            table = _geopandas_to_arrow(
+                df,
+                index=df.index,
+                schema_version=None,
+                write_covering_bbox=write_covering_bbox,
+            )
+            pyarrow.parquet.write_table(table, buffer, compression="snappy", **kwargs)
         return
 
     layer = kwargs.pop("layer", None)

sgis/maps/examine.py

@@ -90,13 +90,13 @@ class Examine:
             **kwargs: Additional keyword arguments passed to sgis.clipmap.
 
         """
-        gdfs, column, kwargs = Map._separate_args(gdfs, column, kwargs)
-
         if mask_gdf is None:
             self.mask_gdf = gdfs[0]
         else:
             self.mask_gdf = mask_gdf
 
+        gdfs, column, kwargs = Map._separate_args(gdfs, column, kwargs)
+
         m = Explore(*gdfs, column=column, **kwargs)
 
         # m = Map(*gdfs, column=column, **kwargs)
@@ -159,14 +159,15 @@ class Examine:
             print("All rows are shown.")
             return
 
-        print(f"i == {self.i} (of {len(self.mask_gdf)})")
-        clipmap(
+        print(f"i == {self.i} (max. {len(self.mask_gdf)- 1})")
+        self.explorer = clipmap(
             self.column,
             *list(self.rasters.values()),
             **self._gdfs,
             mask=self.mask_gdf.iloc[[self.i]].buffer(self.size),
             **self.kwargs,
         )
+
         self.i += 1
 
     def sample(self, **kwargs) -> None:
@@ -182,7 +183,7 @@ class Examine:
         i = np.random.randint(0, len(self.mask_gdf))
 
         print(f"Showing index {i}")
-        clipmap(
+        self.explorer = clipmap(
             self.column,
             *list(self.rasters.values()),
             **self._gdfs,
@@ -202,7 +203,7 @@ class Examine:
             self.i = i
 
         print(f"{self.i + 1} of {len(self.mask_gdf)}")
-        clipmap(
+        self.explorer = clipmap(
             self.column,
             *list(self.rasters.values()),
             **self._gdfs,
@@ -216,7 +217,7 @@ class Examine:
             kwargs = self._fix_kwargs(kwargs)
             self.kwargs = self.kwargs | kwargs
 
-        explore(
+        self.explorer = explore(
             *list(self.rasters.values()),
             **self._gdfs,
             column=self.column,
@@ -229,7 +230,7 @@ class Examine:
             kwargs = self._fix_kwargs(kwargs)
             self.kwargs = self.kwargs | kwargs
 
-        clipmap(
+        self.explorer = clipmap(
             *list(self.rasters.values()),
             **self._gdfs,
             column=self.column,
@@ -242,7 +243,7 @@ class Examine:
             kwargs = self._fix_kwargs(kwargs)
             self.kwargs = self.kwargs | kwargs
 
-        samplemap(
+        self.explorer = samplemap(
             *list(self.rasters.values()),
             **self._gdfs,
             column=self.column,
@@ -252,7 +253,7 @@ class Examine:
     @property
     def mask(self) -> gpd.GeoDataFrame:
         """Returns a GeoDataFrame of the last shown mask geometry."""
-        return self.mask_gdf.iloc[[self.i]]
+        return self.mask_gdf.iloc[[self.i - 1]]
 
     @property
     def gdfs(self) -> dict[str, gpd.GeoDataFrame]: