I'm not sure if this is exactly what you want, but this function takes a BSgenome instance, a chromosome, and a tile width, and calculated CpG % in windows across a particular chromosome

CpG <-
    function(bsgenome, chr, tilewidth)
{
    dna <- bsgenome[[chr]]

    ## CpG on the plus and minus strand (?)
    islands <- matchPDict(DNAStringSet(c("GC", "CG")), dna)
    cvg <- coverage(islands)    # CpG island coverage

    tiles <- tileGenome(seqlengths(bsgenome)[chr], tilewidth=tilewidth,
                        cut.last.tile.in.chrom=TRUE)

    ## Average coverage in each tile
    ## Divide by 2 so each CpG counts only once
    v <- Views(cvg, ranges(tiles))
    tiles$CpG <- viewSums(v) / width(v) / 2
    tiles
}

This would seem to be a relatively effective way to quickly visualize CpG content, e.g.,

library(BSgenome.Hsapiens.UCSC.hg19)
gr <- CpG(BSgenome.Hsapiens.UCSC.hg19, "chr17", 10000)
plot(start(gr) + width(gr) / 2, gr$CpG, pch=".")

Another formulation might slide rather than tile the window across coverage, along the lines of

slidewidth = 10000
diff(cumsum(cvg), lag=slidewidth) / slidewidth / 2

I'm not sure how to visualize this with Gviz; for smaller regions one might use getSeq() to get the DNA sequence of the specific region.