feat: enhance ddbs_describe_geometry_col to improve geometry type detection and add test for handling non-spatial 'geometry' columns

Author: e-kotov

R/io_open_dataset.R

@@ -403,20 +403,26 @@ ddbs_describe_geometry_col <- function(desc, geom_col = NULL) {
   if (!is.null(geom_col) || is.null(desc) || nrow(desc) == 0) {
     return(geom_col)
   }
-
+  
   col_type <- if ("column_type" %in% names(desc)) desc$column_type else desc$data_type
-  geom_cols <- desc$column_name[grepl("^GEOMETRY(\\(|$)", col_type, ignore.case = TRUE)]
+  
+  is_geometry_type <- grepl("^GEOMETRY(\\(|$)", col_type, ignore.case = TRUE)
+  is_wkb_type <- grepl("WKB_BLOB|^BLOB$", col_type, ignore.case = TRUE)
+  is_struct_type <- grepl("^STRUCT", col_type, ignore.case = TRUE)
+  is_spatial_type <- is_geometry_type | is_wkb_type | is_struct_type
+
+  geom_cols <- desc$column_name[is_geometry_type]
   if (length(geom_cols) > 0) {
     return(geom_cols[1])
   }
 
   known <- c("geom", "geometry", "wkb_geometry")
-  found <- desc$column_name[desc$column_name %in% known]
+  found <- desc$column_name[desc$column_name %in% known & is_spatial_type]
   if (length(found) > 0) {
     return(found[1])
   }
 
-  wkb_cols <- desc$column_name[grepl("WKB_BLOB", col_type, ignore.case = TRUE)]
+  wkb_cols <- desc$column_name[is_wkb_type]
   if (length(wkb_cols) > 0) {
     return(wkb_cols[1])
   }

tests/testthat/test-ddbs_write_dataset.R

@@ -31,7 +31,6 @@ test_that("ddbs_write_dataset works for Parquet", {
 
   tmp_file <- tempfile(fileext = ".parquet")
   on.exit(unlink(tmp_file), add = TRUE)
-  on.exit(unlink(tmp_file), add = TRUE)
   expect_no_error(ddbs_write_dataset(ds, tmp_file, quiet = TRUE))
   expect_true(file.exists(tmp_file))
 

tests/testthat/test-formats_autodetect.R

@@ -119,3 +119,22 @@ test_that("ddbs_open_dataset warns when file has no CRS", {
   expect_true(is.na(sf::st_crs(attr(ds, "crs"))$epsg))
 })
 
+test_that("ddbs_open_dataset ignores non-spatial columns named 'geometry'", {
+  skip_if_not_installed("duckdb")
+  
+  conn <- tryCatch(ddbs_default_conn(), error = function(e) DBI::dbConnect(duckdb::duckdb()))
+  ddbs_install(conn, quiet = TRUE)
+  ddbs_load(conn, quiet = TRUE)
+
+  # Create a CSV with a column named "geometry" that is just text
+  tmp_csv <- tempfile(fileext = ".csv")
+  writeLines("id,geometry,value\n1,this is not a geometry,10", tmp_csv)
+  on.exit(unlink(tmp_csv), add = TRUE)
+
+  # Should open as a regular table, not a duckspatial_df
+  ds <- duckspatial::ddbs_open_dataset(tmp_csv, conn = conn)
+  
+  expect_false(inherits(ds, "duckspatial_df"))
+  expect_true(inherits(ds, "tbl_lazy"))
+})
+