How to read .csv polygon data with sf?

Its close to being like a part of a WKT format string.

You have:

"5 50 0 0.5; 20 50 0 0.5; 5 40 0 0.5; 20 40 0 0.5"

and the relevant WKT string would probably be:

"POLYGON ((5 50 0 0.5, 20 50 0 0.5, 5 40 0 0.5, 20 40 0 0.5))"

To convert to that and make sf objects...

1. Replace all semicolons with commas:

gsub(";",",",S$shape)
## [1] "5 50 0 0.5, 20 50 0 0.5, 5 40 0 0.5, 20 40 0 0.5"

2. Paste on the word "POLYGON" and some parentheses:

paste0("POLYGON ((",gsub(";",",",S$shape),"))")
## [1] "POLYGON ((5 50 0 0.5, 20 50 0 0.5, 5 40 0 0.5, 20 40 0 0.5))"

3. make into a data frame and use st_as_sf:

> st_as_sf(data.frame(g=paste0("POLYGON ((",gsub(";",",",S$shape),"))")),wkt="g")
## Simple feature collection with 1 feature and 0 fields
## geometry type:  POLYGON
## dimension:      XYZM
## bbox:           xmin: 5 ymin: 40 xmax: 20 ymax: 50
## z_range:        zmin: 0 zmax: 0
## m_range:        mmin: 0.5 mmax: 0.5
## CRS:            NA
##                                g
## 1 POLYGON ZM ((5 50 0 0.5, 20...

Note this is a "POLYGON ZM" - you've got four coordinates in each part and this is treating the data as three dimensional with an "M" measurement.

and then at this point you realise that WKT needs the first point repeating in order to close the POLYGON. Oops.

So instead create it as a LINESTRING and coerce to POLYGON.

SQ = st_as_sf(
   data.frame(
    geom=paste0("LINESTRING (",gsub(";",",",S$shape),")")
   ),
   wkt="geom")

SQ = st_cast(SQ,"POLYGON")
plot(st_zm(SQ))

That's not a valid Simple Features polygon though because of the self-intersection...