You may use regular expressions. The grep identifies the interesting lines. gsub finds "x" and "y" and the corresponding values, and assembles them with a ,. strsplit splits at the comma into a list.
l <- readLines("dp.txt")
l <- setNames(do.call(rbind.data.frame,
strsplit(gsub(".+si(.)\\D*(\\d+\\.\\d+).+", "\\1, \\2",
l[grep("\\d{2,}", l)]), ",")), c("axis", "coord"))
l$coord <- as.numeric(l$coord)
l
# axis coord
# 1 x 4
# 2 y 3
# 3 x 5
# 4 y 2
# 5 x 6
# 6 y 1