Skip to contents

Perform a spatial join of two geometries

Source: R/ddbs_join.R
ddbs_join.Rd

Combines two sets of geometries based on spatial relationships, such as intersection or containment, attaching attributes from one set to the other.

Usage

ddbs_join(
  x,
  y,
  join = "intersects",
  conn = NULL,
  conn_x = NULL,
  conn_y = NULL,
  name = NULL,
  crs = NULL,
  crs_column = "crs_duckspatial",
  mode = NULL,
  overwrite = FALSE,
  quiet = FALSE
)

Arguments

x

Input spatial data. Can be:

  • A duckspatial_df object (lazy spatial data frame via dbplyr)

  • An sf object

  • A tbl_lazy from dbplyr

  • A character string naming a table/view in conn

Data is returned from this object.

y

Input spatial data. Can be:

  • A duckspatial_df object (lazy spatial data frame via dbplyr)

  • An sf object

  • A tbl_lazy from dbplyr

  • A character string naming a table/view in conn

join

A geometry predicate function. Defaults to "intersects". See the details for other options.

conn

A connection object to a DuckDB database. If NULL, the function runs on a temporary DuckDB database.

conn_x

A DBIConnection object to a DuckDB database for the input x. If NULL (default), it is resolved from conn or extracted from x.

conn_y

A DBIConnection object to a DuckDB database for the input y. If NULL (default), it is resolved from conn or extracted from y.

name

A character string of length one specifying the name of the table, or a character string of length two specifying the schema and table names. If NULL (the default), the function returns the result as an sf object

crs

Deprecated The coordinates reference system of the data. Specify if the data doesn't have a crs_column, and you know the CRS.

crs_column

Deprecated a character string of length one specifying the column storing the CRS (created automatically by ddbs_write_vector). Set to NULL if absent.

mode

Character. Controls the return type. Options:

  • "duckspatial" (default): Lazy spatial data frame backed by dbplyr/DuckDB

  • "sf": Eagerly collected sf object (uses memory)

Can be set globally via ddbs_options(mode = "...") or per-function via this argument. Per-function overrides global setting.

overwrite

Boolean. whether to overwrite the existing table if it exists. Defaults to FALSE. This argument is ignored when name is NULL.

quiet

A logical value. If TRUE, suppresses any informational messages. Defaults to FALSE.

Value

Depends on the mode argument (or global preference set by ddbs_options):

  • duckspatial (default): A duckspatial_df (lazy spatial data frame) backed by dbplyr/DuckDB.

  • sf: An eagerly collected object in R memory, that will return the same data type as the sf equivalent (e.g. sf or units vector).

When name is provided, the result is also written as a table or view in DuckDB and the function returns TRUE (invisibly).

Details

Spatial Join Predicates:

A spatial predicate is really just a function that evaluates some spatial relation between two geometries and returns true or false, e.g., “does a contain b” or “is a within distance x of b”. Here is a quick overview of the most commonly used ones, taking two geometries a and b:

  • "ST_Intersects": Whether a intersects b

  • "ST_Contains": Whether a contains b

  • "ST_ContainsProperly": Whether a contains b without b touching a's boundary

  • "ST_Within": Whether a is within b

  • "ST_Overlaps": Whether a overlaps b

  • "ST_Touches": Whether a touches b

  • "ST_Equals": Whether a is equal to b

  • "ST_Crosses": Whether a crosses b

  • "ST_Covers": Whether a covers b

  • "ST_CoveredBy": Whether a is covered by b

  • "ST_DWithin": x) Whether a is within distance x of b

Examples

if (FALSE) { # \dontrun{
# RECOMMENDED: Efficient lazy workflow using ddbs_open_dataset
library(duckspatial)

# Load data directly as lazy spatial data frames (CRS auto-detected)
countries <- ddbs_open_dataset(
  system.file("spatial/countries.geojson", package = "duckspatial")
)

# Create random points
n <- 100
points <- data.frame(
    id = 1:n,
    x = runif(n, min = -180, max = 180),
    y = runif(n, min = -90, max = 90)
) |> 
  sf::st_as_sf(coords = c("x", "y"), crs = 4326) |>
  as_duckspatial_df()

# Lazy join - computation stays in DuckDB
result <- ddbs_join(points, countries, join = "within")

# Collect to sf when needed
result_sf <- dplyr::collect(result) |> sf::st_as_sf()
plot(result_sf["CNTR_NAME"])


# Alternative: using sf objects directly (legacy compatibility)
library(sf)

countries_sf <- sf::st_read(system.file("spatial/countries.geojson", package = "duckspatial"))

output <- duckspatial::ddbs_join(
    x = points,
    y = countries_sf,
    join = "within"
)


# Alternative: using table names in a duckdb connection
conn <- duckspatial::ddbs_create_conn()

ddbs_write_table(conn, points, "points", overwrite = TRUE)
ddbs_write_table(conn, countries_sf, "countries", overwrite = TRUE)

output2 <- ddbs_join(
    conn = conn,
    x = "points",
    y = "countries",
    join = "within"
)

} # }