#! /usr/bin/awk -f
#
# Graph a sequence of lines as prev->next relationships,
# highlighting frequencies of pairings.
#

{
    curr = $0
}

NR == 1 {
    init = curr
}

NR > 1 {
    ++nlinks[prev]
    ++nlinks_to[prev, curr]
}

{
    prev = curr
}

END {
    final = curr
    print "digraph {"
    if (init == final) {
        printf \
            "\"%s\" [penwidth=10, style=filled, fillcolor=tomato    , fontcolor=white, color=yellowgreen];\n", \
                init
    } else {
        printf \
            "\"%s\" [penwidth=0, style=filled, fillcolor=yellowgreen, fontcolor=white];\n", \
            init
        printf \
            "\"%s\" [penwidth=0, style=filled, fillcolor=tomato     , fontcolor=white];\n", \
            final
    }

    for (src_dst in nlinks_to) {
        split(src_dst, sd, SUBSEP);
        src = sd[1]
        dst = sd[2]
        m = nlinks[src]
        n = nlinks_to[src, dst]
        penwidth = num_scale(n, m, 1, 5)
        color = sprintf("/orrd9/%d", num_scale(n, m, 2, 9))
        label = sprintf("%s %.2f%%", src, (n / m) * 100)
        printf \
            "\"%s\" -> \"%s\" \
                [ label=\"%s\"\
                , fontname=monospace \
                , fontsize=8 \
                , penwidth=%d \
                , color=\"%s\" \
                , dir=both \
                , arrowtail=odot \
                ];\n", \
            src, dst, label, penwidth, color;
    }
    print "}"
}

function num_scale(src_cur, src_max, dst_min, dst_max) {
    return dst_min + ((src_cur * (dst_max - dst_min)) / src_max)
}