Simplify conversion to human-readable units

[khome.git] / home / bin / plotsi

#! /usr/bin/awk -f
#
# Plot Simply
#
# TODO: Incremental plotting (new canvas for each data line)
# TODO: Support manual set of max y

function orange(str)        {return "\033[33m"   str "\033[0m";}
function orange_bright(str) {return "\033[1;33m" str "\033[0m";}
function green(str)         {return "\033[32m"   str "\033[0m";}
function grey_dark(str)     {return "\033[30m"   str "\033[0m";}
function grey_light(str)    {return "\033[37m"   str "\033[0m";}

function abs(n) {return n >= 0 ? n : -n}
function round(n_float) {return int(n_float + 0.5)}
function log_10(n) {return log(n) / log(10)}

function width(n, w) {
    #
    # log(0) == -inf
    #
    w = 1 + round(n > 0 ? log_10(n) : (n == 0 ? 1 : log_10(-n) + 1));  # +1 for neg sign
    #printf("n: %d, w: %d\n", n, w);
    return w;
}

function bins_append_item_to_bin(bin, val,    bin_i) {
    bin_i = bin_item_count[bin]++;
    bins[bin, bin_i] = val;
}

BEGIN {
    # CLI options: w, h, p, a, b
    limits["canvas_width"]  = (w ? w : 70);
    limits["canvas_height"] = (h ? h : 20);
    char_point              = orange_bright(p ? p : ".");
    char_pad                =   grey_dark(a ? a : "");  # FIXME: non-blank pad is broken after adding the y labels
    char_blank              =   grey_dark(b ? b : "|");
    aggregation             = g ? g : "mean";
    data_field              = df ? df : 1;

    y_orig_min = 0;
    y_orig_max = 0;
}

{
    y_orig = $data_field;
    if (y_orig > y_orig_max) y_orig_max = y_orig;
    if (y_orig < y_orig_min) y_orig_min = y_orig;
    data_0[NR] = y_orig;
}

function aggregate_mean() {
}

function aggregate(array) {
    if (aggregation == "mean") {
        return aggregate_mean(array);
    } else {
        printf("Unknown aggregation: %s\n", aggregation) > "/dev/stderr";
        exit(2);
    }
}

function data_scaled_x_to_width(\
    data_in, data_out, limits,

    x_orig,
    datum,
    x_bin,
    bins,
    bin_item_counts,
    bin_sum,
    bin_item_count,
    i,
    bin_item,
    bin_mean \
) {
    # Find limit points
    x_orig_min = 0;
    x_orig_max = 0;
    for (x_orig in data_in) {
        if (x_orig > x_orig_max) x_orig_max = x_orig;
        if (x_orig < x_orig_min) x_orig_min = x_orig;
    }
    #x_orig_range = x_orig_max - x_orig_min;
    # Place items in bins
    x_bin_min = 0;
    x_bin_max = 0;
    for (x_orig in data_in) {
        datum = data_in[x_orig];
        x_bin = round(x_orig * limits["canvas_width_x"] / length(data_in));
        if (x_bin > x_bin_max) x_bin_max = x_bin;
        if (x_bin < x_bin_min) x_bin_min = x_bin;
        bin_item_count = ++bin_item_counts[x_bin];
        bins[x_bin, bin_item_count] = datum;
        #printf("x_orig: %f, x_bin: %f, bin_item_count: %f, datum: %f\n", x_orig, x_bin, bin_item_count, datum);
    }
    # Aggregate bins
    for (x_bin in bin_item_counts) {
        bin_sum = 0;
        bin_item_count = bin_item_counts[x_bin];
        for (i = 1; i <= bin_item_count; i++) {
            bin_item = bins[x_bin, i];
            bin_sum += bin_item;
            #printf("x_bin: %f, bin_item_count: %f, bin_item: %f, bin_sum: %f\n", x_bin, bin_item_count, bin_item, bin_sum);
        }
        bin_mean = bin_sum / bin_item_count;
        data_out[x_bin] = bin_mean;
        #printf("x_bin: %f, bin_mean: %f, bin_sum: %f, bin_item_count: %f\n",
               #x_bin, bin_mean, bin_sum, bin_item_count);
    }
    limits["x_min"] = x_bin_min;
    limits["x_max"] = x_bin_max;
}

function data_scaled_y_to_height(\
    data_in, data_out, limits,

    data_scaled, x, y, y_orig, y_scaled \
) {
    # Offset orig
    # TODO: Is there a better, closed-form way to get the offset?
    # TODO: Is there better way to map canvas to value ranges altogether?
    if (y_orig_min < 0) {
        offset_orig = -1 * y_orig_min;
    } else {
        offset_orig = 0;
    } 
    for (x in data_in) {
        y = data_in[x];
        #printf("x: %f, y: %f, y_min: %f, y_max: %f\n", x, y, y_min, y_max);
        data_in_offsetted[x] = y + offset_orig;
    }
    y_orig_offseted_min = y_orig_min + offset_orig;
    y_orig_offseted_max = y_orig_max + offset_orig;

    # Scale to height
    y_scaled_min = 0;
    y_scaled_max = 0;
    for (x in data_in_offsetted) {
        y_orig_offsetted = data_in_offsetted[x];
        y_scaled = \
            y_orig_offseted_max > 0 \
            ? round((y_orig_offsetted * limits["canvas_height"]) / y_orig_offseted_max) \
            : 0;
        #printf(\
            #"x: %6.2f, y_orig_offsetted: %6.2f, y_orig_max: %6.2f, y_scaled: %6.2f\n",
            #x,        y_orig_offsetted,         y_orig_max,        y_scaled);
        if (y_scaled > y_scaled_max) y_scaled_max = y_scaled;
        if (y_scaled < y_scaled_min) y_scaled_min = y_scaled;
        data_out[x] = y_scaled
    }

    # Save limits
    limits["y_min"] = y_scaled_min;
    limits["y_max"] = y_scaled_max;
    range_orig = y_orig_max - y_orig_min;
    offset_scaled = \
        range_orig > 0 \
        ? round(offset_orig * limits["canvas_height"] / range_orig) \
        : 0;
    limits["offset_scaled"] = offset_scaled;
    #printf("offset_orig: %f, offset_scaled: %f\n", offset_orig, offset_scaled);
}

function canvas_init(canvas, width, height,    row, col) {
    for (row=0; row <= height; row++) {
        for (col=0; col <= width; col++) {
            canvas[row, col] = char_pad char_blank char_pad;
        }
    }
}

function canvas_overlay_highlight_ticks_x(canvas, limits,    row, col) {
    for (col=limits["canvas_width_y"] - 1; col <= limits["canvas_width"]; col++) {
        offset = limits["offset_scaled"];
        #printf("offset: %f\n", offset);
        row = 0 + offset;
        #row = 0;
        canvas[row, col] = char_pad green("-") char_pad;
    }
}

function canvas_overlay_highlight_zero_row(canvas, limits,    row, col) {
    #print "canvas_overlay_highlight_zero_row";
    for (col=limits["canvas_width_y"] - 1; col <= limits["canvas_width"]; col++) {
        offset = limits["offset_scaled"];
        row = 0 + offset;
        #printf("col: %6.2f, row: %6.2f, offset: %f\n", col, row, offset);
        #row = 0;
        canvas[row, col] = char_pad green("-") char_pad;
    }
}

function canvas_overlay_highlight_zero_col(canvas, limits,    row, col) {
    #print "canvas_overlay_highlight_zero_col";
    for (row=0; row <= limits["canvas_height"]; row++) {
        col = limits["canvas_width_y"];  # was also -1. Why?
        # TODO: Refactor color/character configs to ease composition
        canvas[row, col] = green("|");
    }
    canvas[limits["canvas_height"], limits["canvas_width_y"]] = green("+");
    canvas[0                      , limits["canvas_width_y"]] = green("+");
}
function canvas_overlay_highlight_zero(canvas, limits) {
    #print "canvas_overlay_highlight_zero";
    canvas[0 + limits["offset_scaled"], 0 + limits["canvas_width_y"]] = green("+");
}

function canvas_overlay_data(canvas, data, limits,    x_data, x_canvas, y, yi, yj) {
    #print "canvas_overlay_data";
    for (x_data in data) {
        x_canvas = x_data + limits["canvas_width_y"] + 1;
        y = data[x_data];
        # TODO: Would be nice to scale width of all cells to the widest
        #point = y;
        #printf("canvas_width_y: %6.2f, x0: %6.2f, x1: %6.2f, x: %6.2f, y: %6.2f\n",
               #limits["canvas_width_y"], x0, x1, x, y);
        # TODO: This special case for 0 is kind of a kludge - can we do better?
        canvas[y, x_canvas] = x_data == 0 ? char_point : char_pad char_point char_pad;

        if (y > limits["offset_scaled"]) {
            for (yi = y - 1; yi >= limits["offset_scaled"]; yi--) {
                #printf("yi: %6.2f\n", yi);
                canvas[yi, x_canvas] = x_data == 0 ? orange("|") : char_pad orange("|") char_pad;
            }
        } else if (y < limits["offset_scaled"]) {
            for (yj = limits["offset_scaled"]; yj > y; yj--) {
                #printf("yj: %6.2f\n", yj);
                canvas[yj, x_canvas] = x_data == 0 ? orange("|") : char_pad orange("|") char_pad;
            }
        }
    }
}

function canvas_overlay_y_lab(canvas, limits,    y_lab_fmt, y_max_str, i) {
    y_lab_fmt = "%" limits["canvas_width_y"] - 1 "d ";
    y_max_str = sprintf(y_lab_fmt, y_orig_max);
    y_min_str = sprintf(y_lab_fmt, y_orig_min);
    #printf("y_width: %f, y_max_str: \"%s\", y_min_str: \"%s\"\n", limits["canvas_width_y"], y_max_str, y_min_str);
    for (i=1; i<=length(y_max_str); i++) {
        canvas[limits["canvas_height"], i - 1] = substr(y_max_str, i, 1);
    }
    canvas[0 + limits["offset_scaled"], 0 + limits["canvas_width_y"] - 1] = 0;
    for (i=1; i<=length(y_min_str); i++) {
        canvas[0, i - 1] = substr(y_min_str, i, 1);
    }
}

function canvas_print(canvas, limits,    row, col) {
    for (row = limits["canvas_height"]; row >= 0; row--) {
        for (col = 0; col <= limits["canvas_width"]; col++) {
            printf("%s", canvas[row, col]);
        }
        printf("\n");
    }
}

END {
    # Find maximum y number width
    y_orig_min_width = width(y_orig_min);
    y_orig_max_width = width(y_orig_max);
    if (y_orig_max_width >= y_orig_min_width) {
        y_width = y_orig_max_width;
    } else {
        y_width = y_orig_min_width;
    }
    limits["canvas_width_y"] = y_width + 1;
    limits["canvas_width_x"] = limits["canvas_width"] - limits["canvas_width_y"];

    data_scaled_x_to_width(data_0, data_1, limits);
    data_scaled_y_to_height(data_1, data_2, limits);

    canvas_init(canvas, limits["canvas_width"], limits["canvas_height"]);
    canvas_overlay_highlight_zero_row(canvas, limits);
    canvas_overlay_highlight_zero_col(canvas, limits);
    canvas_overlay_highlight_zero(canvas, limits);
    canvas_overlay_y_lab(canvas, limits);
    canvas_overlay_data(canvas, data_2, limits);
    #for (l in limits) {
        #printf("limits[%s] -> %s\n", l, limits[l]);
    #}
    canvas_print(canvas, limits);
}

# An even better way to think about scaling: ratios!!! Duh! :-D
#
# val_max / val_current = width / val_scaled
#
# val_max        width
# ----------- = ------------
# val_current    val_scaled
#
# val_max * val_scaled = val_current * width
#           val_scaled = (val_current * width) / val_max
#
#
# num_data_points    width
# ---------------- = -----
# x                  1
#
# width * x = num_data_points
#         x = num_data_points / width
#
# But that is what I already tried, and it is awkward to scale up when
# thinking thsese terms, so it is much better to first route each data
# point to an appropriate bin and then aggregate each bin:
#   1. Route: bins[scale(datum)]
#   2. Aggregate: for bin in bins: for val in bin: aggregate(val)