CODEWITHSUNDEEP
htmlrsvg
pe-perry
pe-perry

Reputation: 2621

How to get coordinates of a path from svg on a website into R

I am trying to import an online SVG map into R, to see if I can reproduce the results in R.

First, I can download the html code of the SVGs using the following code:

svg <- html("https://dce2015.thestandnews.com") %>% html_nodes("g")

and further use 

... %>% html_nodes("polygon") %>% html_attr("points")
... %>% html_nodes("polyline") %>% html_attr("points")
... %>% html_nodes("path") %>% html_attr("d")

to get the coordinates in a character strings.

I am able to deal with <polygon ...> and <polyline ...> but when I try to extract the coordinates inside <path ...>, I find there are so many curve command (e.g. s, S, c etc). Is there any method get some smoothed points lying on the curve?

Also, is there any more elegant way to import online SVG in R as polygons?

Thanks!

Upvotes: 1

Views: 839

Answers (1)

hrbrmstr
hrbrmstr

Reputation: 78832

You would be much better off trying to find an actual shapefile of the electoral districts (I didn't really try to find one but a quick google suggested that it does exist). However, this was an interesting idea.

I think it'd be ultimately cool to port svg-to-wkt to R from DOM-javascript. But, if this is a one-off solution then you can do something like:

library(rvest)
library(rgeos)

URL <- "https://dce2015.thestandnews.com/"
pg <- read_html(URL)
svg_map <- html_nodes(pg, "svg#hkmap")
writeLines(as.character(svg_map), "hk.txt")

# use http://svg-to-wkt.linfiniti.com/ with svg from hk.txt
# saving it to hk.wkt

wkt_txt <- paste0(readLines("hk.wkt"), sep="", collapse="")
svg_wkt <- readWKT(wkt_txt)

svg_wkt is a SpatialCollections object that won't work with ggplot2. But, most of the SVG got converted to polygons:

plot(svg_wkt@polyobj)

enter image description here

Note that there are some elements in svg_wkt@lineobj (which you would run into even if you did a full conversion in R IMO). Those will need to be dealt with.

You can pass in IDs to readWKT in order to have some hope of being able to more with this in R (i.e. plot the election data it seems you want to plot) but I have to go back to my original suggestion of advocating the use of a proper shapefile if you really do want to plot the election data.

Upvotes: 3

Related Questions