Transforming R data.frame with column of GeoJSON strings into sf object

Question

Suppose I have a data.frame object that has a column called geojson with GeoJSON strings in each row.

How can I transform this data.frame into an sf object that would allow me to perform geospatial operations on it?

Here's a quick example:

library(tidyverse)
library(dplyr)
library(sf)
library(geojsonsf)

df = data.frame(row_id = c(1,2,3),
                categories = c('Cat1','Cat1','Cat2'),
                geojson = c('{"type": "FeatureCollection","features": [{"type": "Feature","properties": {},"geometry": {"coordinates": [[0,0],[1,0]],"type": "LineString"}}]}',
                            '{"type": "FeatureCollection","features": [{"type": "Feature","properties": {},"geometry": {"coordinates": [[0,1],[1,1]],"type": "LineString"}}]}',
                            '{"type": "FeatureCollection","features": [{"type": "Feature","properties": {},"geometry": {"coordinates": [[0,2],[1,2]],"type": "LineString"}},
                                                                       {"type": "Feature","properties": {},"geometry": {"coordinates": [[0,3],[1,3]],"type": "LineString"}}]}'))

And here is my (very dirty and not fully functional) solution so far:

# Trying to create one geometry for every row in the original dataset
for (i in 1:nrow(df)){
  this_geom = df$geojson[i] %>%  # First, we access the GeoJSON text
    geojson_sf() %>%  # Then we use the use the GeoJSON text and create a whole sf dataframe
    group_by() %>% # Then we need to dissolve/unite everything into one single geometry
    summarize(geometry = st_union(geometry)) %>% 
    .$geometry[1] # Finally, I try to retrieve only the actual geometry from the result above
  df$geometry[i] = this_geom # and store it into the original dataframe in a column called `geometry`
}

# Creating an sf-aware object
gdf = df %>% 
  st_as_sf(sf_column_name = "geometry")

The first chunk of the code above (the for-loop) doesn't throw any errors, but the second bit (creating the gdf object) does. I get the following error:

Error in st_sf(x, ..., agr = agr, sf_column_name = sf_column_name) : 
  no simple features geometry column present

I think I'm messing up in the creation of the geometry column. Because when I print it out, this is what it looks like:

> print(df$geometry)
[[1]]
Geometry set for 1 feature 
Geometry type: LINESTRING
Dimension:     XY
Bounding box:  xmin: 0 ymin: 0 xmax: 1 ymax: 0
Geodetic CRS:  WGS 84
LINESTRING (0 0, 1 0)

[[2]]
Geometry set for 1 feature 
Geometry type: LINESTRING
Dimension:     XY
Bounding box:  xmin: 0 ymin: 1 xmax: 1 ymax: 1
Geodetic CRS:  WGS 84
LINESTRING (0 1, 1 1)

[[3]]
Geometry set for 1 feature 
Geometry type: LINESTRING
Dimension:     XY
Bounding box:  xmin: 0 ymin: 2 xmax: 1 ymax: 2
Geodetic CRS:  WGS 84
LINESTRING (0 2, 1 2)

It looks like it's not actually creating a proper column of data, but instead a list of items.

Is there a way that I can change this to make sf recognize the new geometries I'm creating so it'll allow me to make the gdf object?

Answer 1

First time using {geojsonsf}, so I don't have a flash of genius proposing an approach much more elegant, but I should make your current code work, I guess.

The last operation of your pipe .$geometry[1] was not executable for me, but why don't you simply use sf::st_geometry() when you're interested in extracting exactly this part of your simple feature object only?

The output looks not that bad:

# no need for {tidyverse} here
library(dplyr)
library(sf)
#> Linking to GEOS 3.11.2, GDAL 3.7.2, PROJ 9.3.0; sf_use_s2() is TRUE
library(geojsonsf)

df <- data.frame(row_id = c(1,2,3),
                categories = c('Cat1','Cat1','Cat2'),
                geojson = c('{"type": "FeatureCollection","features": [{"type": "Feature","properties": {},"geometry": {"coordinates": [[0,0],[1,0]],"type": "LineString"}}]}',
                            '{"type": "FeatureCollection","features": [{"type": "Feature","properties": {},"geometry": {"coordinates": [[0,1],[1,1]],"type": "LineString"}}]}',
                            '{"type": "FeatureCollection","features": [{"type": "Feature","properties": {},"geometry": {"coordinates": [[0,2],[1,2]],"type": "LineString"}}]}'))


for (i in 1:nrow(df)){
  
  df$geometry[i] <- df$geojson[i] |>
    geojson_sf() |> 
    group_by()  |> 
    summarize(geometry = st_union(geometry)) |> 
    st_geometry()
}

gdf <- st_as_sf(df, sf_column_name = "geometry", crs = "epsg:4326")

gdf$geometry
#> Geometry set for 3 features 
#> Geometry type: LINESTRING
#> Dimension:     XY
#> Bounding box:  xmin: 0 ymin: 0 xmax: 1 ymax: 2
#> Geodetic CRS:  WGS 84
#> LINESTRING (0 0, 1 0)
#> LINESTRING (0 1, 1 1)
#> LINESTRING (0 2, 1 2)