join attributes columns after performing zonal stats as to avoid NULL…

… attributes in the input vector attr table (#110) Co-authored-by: Ioan Ferencik <[email protected]>
UNDP-Data · Jan 10, 2025 · 8d3d254 · 8d3d254
1 parent bccb1c9
commit 8d3d254
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 7 deletions.
diff --git a/cbsurge/initialize.py b/cbsurge/initialize.py
@@ -3,7 +3,7 @@
 import os
 import shutil
 from cbsurge.session import Session
-from cbsurge.admin import silence_httpx_az
+from cbsurge.util import silence_httpx_az
 
 logger = logging.getLogger(__name__)
 

diff --git a/cbsurge/stats/ZonalStats.py b/cbsurge/stats/ZonalStats.py
@@ -3,7 +3,7 @@
 from exactextract import exact_extract
 from cbsurge.stats.GDALRasterSource import GDALRasterSource
 from cbsurge.stats.OGRDataSource import OGRDataSource
-
+import geopandas
 
 logger = logging.getLogger(__name__)
 
@@ -76,17 +76,14 @@ def compute(self, rasters: list, operations: list = None, operation_cols: list =
                     with GDALRasterSource(raster) as raster_ds:
                         reprojected_rast = raster_ds.reproject(self.target_crs)
                         logger.debug(f"raster ({raster}) was reprojected to {self.target_crs} as {reprojected_rast}")
-                        # only first raster keep all fields
-                        include_cols = vector_fields if raster_index == 0 else None
-
 
                         gdf = exact_extract(
                             reprojected_rast,
                             reprojected_vector_path,
                             ops=operations,
-                            include_cols=include_cols,
                             include_geom=True,
-                            output="pandas"
+                            output="pandas",
+                            progress=True
                         )
                         raster_name_with_ext = os.path.basename(raster)
                         raster_file_name = os.path.splitext(raster_name_with_ext)[0]
@@ -107,6 +104,7 @@ def compute(self, rasters: list, operations: list = None, operation_cols: list =
 
                         os.remove(reprojected_rast)
                         logger.debug(f"deleted {reprojected_rast}")
+                combined_results = combined_results.merge(geopandas.read_file(reprojected_vector_path), on='geometry', how='inner')
                 os.remove(reprojected_vector_path)
                 logger.debug(f"deleted {reprojected_vector_path}")
                 logger.info(f"end computing zonal statistics for {rasters}")