Read simple features or layers from file or database
Read simple features from file or database, or retrieve layer names and their geometry type(s)
Read PostGIS table directly through DBI and RPostgreSQL interface, converting Well-Know Binary geometries to sfc
st_read(dsn, layer, ...) ## S3 method for class 'character' st_read( dsn, layer, ..., query = NA, options = NULL, quiet = FALSE, geometry_column = 1L, type = 0, promote_to_multi = TRUE, stringsAsFactors = sf_stringsAsFactors(), int64_as_string = FALSE, check_ring_dir = FALSE, fid_column_name = character(0), drivers = character(0), wkt_filter = character(0) ) read_sf(..., quiet = TRUE, stringsAsFactors = FALSE, as_tibble = TRUE) ## S3 method for class 'DBIObject' st_read( dsn = NULL, layer = NULL, query = NULL, EWKB = TRUE, quiet = TRUE, as_tibble = FALSE, geometry_column = NULL, ... )
dsn |
data source name (interpretation varies by driver - for some
drivers, |
layer |
layer name (varies by driver, may be a file name without
extension); in case |
... |
parameter(s) passed on to st_as_sf |
query |
SQL query to select records; see details |
options |
character; driver dependent dataset open options, multiple options supported. For possible values, see the "Open options" section of the GDAL documentation of the corresponding driver, and https://github.com/r-spatial/sf/issues/1157 for an example. |
quiet |
logical; suppress info on name, driver, size and spatial reference, or signaling no or multiple layers |
geometry_column |
integer or character; in case of multiple geometry fields, which one to take? |
type |
integer; ISO number of desired simple feature type; see details.
If left zero, and |
promote_to_multi |
logical; in case of a mix of Point and MultiPoint, or
of LineString and MultiLineString, or of Polygon and MultiPolygon, convert
all to the Multi variety; defaults to |
stringsAsFactors |
logical; logical: should character vectors be
converted to factors? Default for |
int64_as_string |
logical; if TRUE, Int64 attributes are returned as string; if FALSE, they are returned as double and a warning is given when precision is lost (i.e., values are larger than 2^53). |
check_ring_dir |
logical; if TRUE, polygon ring directions are checked and if necessary corrected (when seen from above: exterior ring counter clockwise, holes clockwise) |
fid_column_name |
character; name of column to write feature IDs to; defaults to not doing this |
drivers |
character; limited set of driver short names to be tried (default: try all) |
wkt_filter |
character; WKT representation of a spatial filter (may be used as bounding box, selecting overlapping geometries); see examples |
as_tibble |
logical; should the returned table be of class tibble or data.frame? |
EWKB |
logical; is the WKB of type EWKB? if missing, defaults to
|
for geometry_column, see also
https://trac.osgeo.org/gdal/wiki/rfc41_multiple_geometry_fields
for values for type see
https://en.wikipedia.org/wiki/Well-known_text#Well-known_binary, but
note that not every target value may lead to successful conversion. The
typical conversion from POLYGON (3) to MULTIPOLYGON (6) should work; the
other way around (type=3), secondary rings from MULTIPOLYGONS may be dropped
without warnings. promote_to_multi is handled on a per-geometry column
basis; type may be specified for each geometry column.
Note that stray files in data source directories (such as *.dbf) may
lead to spurious errors that accompanying *.shp are missing.
In case of problems reading shapefiles from USB drives on OSX, please see https://github.com/r-spatial/sf/issues/252.
For query with a character dsn the query text is handed to
'ExecuteSQL' on the GDAL/OGR data set and will result in the creation of a
new layer (and layer is ignored). See 'OGRSQL'
https://gdal.org/user/ogr_sql_dialect.html for details. Please note that the
'FID' special field is driver-dependent, and may be either 0-based (e.g. ESRI
Shapefile), 1-based (e.g. MapInfo) or arbitrary (e.g. OSM). Other features of
OGRSQL are also likely to be driver dependent. The available layer names may
be obtained with
st_layers. Care will be required to properly escape the use of some layer names.
read_sf and write_sf are aliases for st_read and st_write, respectively, with some
modified default arguments.
read_sf and write_sf are quiet by default: they do not print information
about the data source. read_sf returns an sf-tibble rather than an sf-data.frame.
write_sf delete layers by default: it overwrites existing files without asking or warning.
if table is not given but query is, the spatial
reference system (crs) of the table queried is only available in case it
has been stored into each geometry record (e.g., by PostGIS, when using
EWKB)
The function will automatically find the 'geometry' type columns for drivers that support it. For the other drivers, it will try to cast all the character columns, which can be slow for very wide tables.
object of class sf when a layer was successfully read; in case
argument layer is missing and data source dsn does not
contain a single layer, an object of class sf_layers is returned
with the layer names, each with their geometry type(s). Note that the
number of layers may also be zero.
The use of system.file in examples make sure that examples run regardless where R is installed:
typical users will not use system.file but give the file name directly, either with full path or relative
to the current working directory (see getwd). "Shapefiles" consist of several files with the same basename
that reside in the same directory, only one of them having extension .shp.
nc = st_read(system.file("shape/nc.shp", package="sf"))
summary(nc) # note that AREA was computed using Euclidian area on lon/lat degrees
## only three fields by select clause
## only two features by where clause
nc_sql = st_read(system.file("shape/nc.shp", package="sf"),
query = "SELECT NAME, SID74, FIPS FROM \"nc\" WHERE BIR74 > 20000")
## Not run:
library(sp)
example(meuse, ask = FALSE, echo = FALSE)
try(st_write(st_as_sf(meuse), "PG:dbname=postgis", "meuse",
layer_options = "OVERWRITE=true"))
try(st_meuse <- st_read("PG:dbname=postgis", "meuse"))
if (exists("st_meuse"))
summary(st_meuse)
## End(Not run)
## Not run:
## note that we need special escaping of layer within single quotes (nc.gpkg)
## and that geom needs to be included in the select, otherwise we don't detect it
layer <- st_layers(system.file("gpkg/nc.gpkg", package = "sf"))$name[1]
nc_gpkg_sql = st_read(system.file("gpkg/nc.gpkg", package = "sf"),
query = sprintf("SELECT NAME, SID74, FIPS, geom FROM \"%s\" WHERE BIR74 > 20000", layer))
## End(Not run)
# spatial filter, as wkt:
wkt = st_as_text(st_geometry(nc[1,]))
# filter by (bbox overlaps of) first feature geometry:
read_sf(system.file("gpkg/nc.gpkg", package="sf"), wkt_filter = wkt)
# read geojson from string:
geojson_txt <- paste("{\"type\":\"MultiPoint\",\"coordinates\":",
"[[3.2,4],[3,4.6],[3.8,4.4],[3.5,3.8],[3.4,3.6],[3.9,4.5]]}")
x = read_sf(geojson_txt)
x
## Not run:
library(RPostgreSQL)
try(conn <- dbConnect(PostgreSQL(), dbname = "postgis"))
if (exists("conn") && !inherits(conn, "try-error")) {
x = st_read(conn, "meuse", query = "select * from meuse limit 3;")
x = st_read(conn, table = "public.meuse")
print(st_crs(x)) # SRID resolved by the database, not by GDAL!
dbDisconnect(conn)
}
## End(Not run)Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.