ggplot CLI tool

ggplot_cli.R

#!/usr/bin/env Rscript
# ggplot_cli.R — command-line ggplot2 plotting tool
#
# Usage examples:
#   cat data.csv | Rscript ggplot_cli.R -x mpg -y hp
#   Rscript ggplot_cli.R -i data.csv -o plot.png -x mpg -y hp -t line
#   Rscript ggplot_cli.R -i data.csv -x date -y revenue -t bar --yscale log10 --ylabel-fmt dollar
#   Rscript ggplot_cli.R -i data.csv -o plot.svg -x mpg -y hp -t line
#   cat data.csv | Rscript ggplot_cli.R -x weight -y price --color group -t point --xscale log10 --ylabel-fmt comma -o -
#   cat data.csv | Rscript ggplot_cli.R -x mpg -y hp -f svg -o -

suppressPackageStartupMessages({
  library(optparse)
  library(ggplot2)
  library(scales)
})

# ── CLI options ──────────────────────────────────────────────────────────────

option_list <- list(
  # I/O
  make_option(c("-i", "--input"),  default = "-",
              help = "Input CSV path. Use '-' or omit for stdin [default: stdin]"),
  make_option(c("-o", "--output"), default = "plot.png",
              help = "Output path. Use '-' for stdout [default: plot.png]"),
  make_option(c("-f", "--format"), default = NULL,
              help = "Output format: png, svg. Auto-detected from --output extension [default: png]"),


  # Aesthetics
  make_option(c("-x", "--xcol"),   help = "Column name for x-axis (required)"),
  make_option(c("-y", "--ycol"),   default = NULL,
              help = "Column name for y-axis (optional for histogram/density)"),
  make_option(c("--color"),        default = NULL,
              help = "Column name mapped to color/fill aesthetic"),
  make_option(c("--facet"),        default = NULL,
              help = "Column name for facet_wrap (e.g. 'region')"),


  # Plot type
  make_option(c("-t", "--type"),   default = "point",
              help = paste("Plot type: point, line, bar, col, histogram,",
                           "density, box, violin, area, step [default: point]")),

  # Axis scales
  make_option("--xscale",  default = "identity",
              help = "X-axis transform: identity, log10, log2, sqrt, reverse [default: identity]"),
  make_option("--yscale",  default = "identity",
              help = "Y-axis transform: identity, log10, log2, sqrt, reverse [default: identity]"),

  # Label formatters (scales package)
  make_option("--xlabel-fmt", default = NULL,
              help = paste("X-axis label format: comma, dollar, percent, scientific,",
                           "number_si, bytes, date, time, ordinal [default: none]")),
  make_option("--ylabel-fmt", default = NULL,
              help = paste("Y-axis label format: comma, dollar, percent, scientific,",
                           "number_si, bytes, date, time, ordinal [default: none]")),

  # Labels & appearance
  make_option("--title",    default = NULL, help = "Plot title"),
  make_option("--subtitle", default = NULL, help = "Plot subtitle"),
  make_option("--xlab",     default = NULL, help = "X-axis label (default: column name)"),
  make_option("--ylab",     default = NULL, help = "Y-axis label (default: column name)"),
  make_option("--theme",    default = "minimal",
              help = "ggplot2 theme: minimal, bw, classic, light, dark, void, gray [default: minimal]"),

  # Dimensions
  make_option("--width",  default = 1200, type = "integer", help = "Plot width in pixels [default: 1200]"),
  make_option("--height", default = 750,  type = "integer", help = "Plot height in pixels [default: 750]"),
  make_option("--dpi",    default = 150,  type = "integer", help = "Resolution in DPI [default: 150]"),

  # Extras
  make_option("--alpha",    default = 0.8, type = "double", help = "Point/fill alpha [default: 0.8]"),
  make_option("--size",     default = 1.5, type = "double", help = "Point/line size [default: 1.5]"),
  make_option("--bins",     default = 30,  type = "integer", help = "Bins for histogram [default: 30]"),
  make_option("--position", default = "dodge",
              help = "Bar/col position: dodge, stack, fill [default: dodge]"),
  make_option("--smooth",   default = FALSE, action = "store_true",
              help = "Add a smoothing line (geom_smooth) to scatter/line plots"),
  make_option("--se",       default = FALSE, action = "store_true",
              help = "Show confidence interval on smooth line")
)

opt <- parse_args(OptionParser(option_list = option_list))

if (is.null(opt$xcol)) {
  stop("--xcol / -x is required. Run with --help for usage.", call. = FALSE)
}

# ── Read data ────────────────────────────────────────────────────────────────

if (opt$input == "-") {
  df <- read.csv(file("stdin"), stringsAsFactors = FALSE)
} else {
  if (!file.exists(opt$input)) stop("Input file not found: ", opt$input, call. = FALSE)
  df <- read.csv(opt$input, stringsAsFactors = FALSE)
}

stopifnot(opt$xcol %in% names(df))
if (!is.null(opt$ycol)) stopifnot(opt$ycol %in% names(df))

# ── Helpers ──────────────────────────────────────────────────────────────────

get_theme <- function(name) {
  switch(name,
    minimal = theme_minimal(),
    bw      = theme_bw(),
    classic = theme_classic(),
    light   = theme_light(),
    dark    = theme_dark(),
    void    = theme_void(),
    gray    = theme_gray(),
    theme_minimal()
  )
}

get_formatter <- function(name) {
  if (is.null(name)) return(waiver())
  switch(name,
    comma      = label_comma(),
    dollar     = label_dollar(),
    percent    = label_percent(),
    scientific = label_scientific(),
    number_si  = label_number_si(),
    bytes      = label_bytes(),
    date       = label_date(),
    time       = label_time(),
    ordinal    = label_ordinal(),
    waiver()
  )
}

get_trans <- function(name) {
  switch(name,
    identity = "identity",
    log10    = "log10",
    log2     = "log2",
    sqrt     = "sqrt",
    reverse  = "reverse",
    "identity"
  )
}

# ── Build plot ───────────────────────────────────────────────────────────────

aes_map <- if (!is.null(opt$ycol)) {
  if (!is.null(opt$color)) {
    aes(x = .data[[opt$xcol]], y = .data[[opt$ycol]], color = .data[[opt$color]],
        fill = .data[[opt$color]])
  } else {
    aes(x = .data[[opt$xcol]], y = .data[[opt$ycol]])
  }
} else {
  if (!is.null(opt$color)) {
    aes(x = .data[[opt$xcol]], color = .data[[opt$color]], fill = .data[[opt$color]])
  } else {
    aes(x = .data[[opt$xcol]])
  }
}

p <- ggplot(df, aes_map)

geom_layer <- switch(opt$type,
  point     = geom_point(alpha = opt$alpha, size = opt$size),
  line      = geom_line(alpha = opt$alpha, linewidth = opt$size),
  step      = geom_step(alpha = opt$alpha, linewidth = opt$size),
  bar       = geom_bar(alpha = opt$alpha, stat = "count", position = opt$position),
  col       = geom_col(alpha = opt$alpha, position = opt$position),
  histogram = geom_histogram(alpha = opt$alpha, bins = opt$bins, position = opt$position),
  density   = geom_density(alpha = opt$alpha),
  box       = geom_boxplot(alpha = opt$alpha),
  violin    = geom_violin(alpha = opt$alpha),
  area      = geom_area(alpha = opt$alpha),
  stop("Unknown plot type: ", opt$type, call. = FALSE)
)

p <- p + geom_layer

# Optional smooth line
if (opt$smooth && opt$type %in% c("point", "line", "step")) {
  p <- p + geom_smooth(se = opt$se)
}

# ── Scales ───────────────────────────────────────────────────────────────────

if (is.numeric(df[[opt$xcol]])) {
  p <- p + scale_x_continuous(
    trans  = get_trans(opt$xscale),
    labels = get_formatter(opt$`xlabel-fmt`)
  )
}

if (!is.null(opt$ycol) && is.numeric(df[[opt$ycol]])) {
  p <- p + scale_y_continuous(
    trans  = get_trans(opt$yscale),
    labels = get_formatter(opt$`ylabel-fmt`)
  )
}

# ── Labels, theme, facets ────────────────────────────────────────────────────

p <- p + labs(
  title    = opt$title,
  subtitle = opt$subtitle,
  x        = if (!is.null(opt$xlab)) opt$xlab else opt$xcol,
  y        = if (!is.null(opt$ylab)) opt$ylab else opt$ycol
)

p <- p + get_theme(opt$theme)

if (!is.null(opt$facet)) {
  p <- p + facet_wrap(as.formula(paste("~", opt$facet)))
}

# ── Resolve output format ────────────────────────────────────────────────────

out_fmt <- if (!is.null(opt$format)) {
  match.arg(opt$format, c("png", "svg"))
} else if (opt$output != "-") {
  ext <- tolower(tools::file_ext(opt$output))
  if (ext %in% c("png", "svg")) ext else "png"
} else {
  "png"
}

# ── Save / write ─────────────────────────────────────────────────────────────

width_in  <- opt$width  / opt$dpi
height_in <- opt$height / opt$dpi
save_args <- list(plot = p, width = width_in, height = height_in, bg = "white")
if (out_fmt == "png") save_args$dpi <- opt$dpi

svg_device <- if (requireNamespace("svglite", quietly = TRUE)) svglite::svglite else grDevices::svg

if (opt$output == "-") {
  tmp <- tempfile(fileext = paste0(".", out_fmt))
  if (out_fmt == "svg") save_args$device <- svg_device
  do.call(ggsave, c(list(filename = tmp), save_args))
  raw_bytes <- readBin(tmp, "raw", file.info(tmp)$size)
  con <- pipe("cat", "wb")
  writeBin(raw_bytes, con)
  close(con)
  unlink(tmp)
} else {
  if (out_fmt == "svg") save_args$device <- svg_device
  do.call(ggsave, c(list(filename = opt$output), save_args))
  msg <- paste0("Saved: ", opt$output, "  (", opt$width, "x", opt$height, " px")
  if (out_fmt == "png") msg <- paste0(msg, " @ ", opt$dpi, " dpi")
  message(msg, ")")
}