Fixed width of legend box using ggplot, gtable and cowplot

I would like to make a plot with R that looks like the sample made with Mac's Numbers. I'm struggling with the space between the plot and the legend box. This is a sample of what I would like to achieve:

With the help of some users (see end of post for reference) I got really close already. This is my current function:

library(tidyverse)
library(cowplot)
library(gtable)
library(grid)
library(patchwork)


custom_barplot <- function(dataset, x_value, y_value, fill_value, nfill, xlab, ylab, y_limit, y_steps,legend_labels) {
# Example color set to choose from
colors=c("#CF232B","#942192","#000000","#f1eef6","#addd8e","#d0d1e6","#31a354","#a6bddb")

# user function for adjusting the size of key-polygons in legend
draw_key_polygon2 <- function(data, params, size) {
  lwd <- min(data$size, min(size) / 4)
  
  grid::rectGrob(
    width = grid::unit(0.8, "npc"),
    height = grid::unit(0.8, "npc"),
    gp = grid::gpar(
      col = data$colour,
      fill = alpha(data$fill, data$alpha),
      lty = data$linetype,
      lwd = lwd * .pt,
      linejoin = "mitre"
    ))
}
# user function for the plot itself
plot <- function(dataset, x_value, y_value, fill_value, nfill, xlab, ylab, y_limit, y_steps,legend,legend_labels) 
{ggplot(data=dataset, mapping=aes(x={{x_value}}, y={{y_value}}, fill={{fill_value}})) +
    geom_col(position=position_dodge(width=0.85),width=0.8,key_glyph="polygon2",show.legend=legend) + 
    geom_smooth(aes(color={{fill_value}}),method="lm",formula=y~x, se=FALSE,show.legend=legend, linetype="dashed") +
    labs(x=xlab,y=ylab) +
    theme(text=element_text(size=9,color="black"),
          panel.background = element_rect(fill="white"),
          panel.grid = element_line(color = "black",linetype="solid",size= 0.3),
          panel.grid.minor = element_blank(),
          panel.grid.major.x=element_blank(),
          axis.text=element_text(size=9),
          axis.line.x=element_line(color="black"),
          axis.ticks= element_blank(),
          legend.text=element_text(size=9),
          legend.position = "right",
          legend.justification = "top",
          legend.title = element_blank(),
          legend.key.size = unit(4,"mm"),
          legend.key = element_rect(fill="white"),
          plot.margin=unit(c(1,0.25,0.5,0.5),"cm")) +
    scale_y_continuous(breaks= seq(from=0, to=y_limit,by=y_steps),
                       limits=c(0,y_limit+1), 
                       expand=c(0,0)) +
    scale_x_continuous(breaks=min(data[,deparse(ensym(x_value))],na.rm=TRUE):max(data[,deparse(ensym(x_value))],na.rm=TRUE)) +
    scale_fill_manual(values = colors[1:nfill],labels={{legend_labels}})+
    scale_color_manual(values= colors[1:nfill],labels=paste("Trend ",{{legend_labels}},sep=""))+
    guides(color=guide_legend(override.aes=list(fill=NA),order=2),fill=guide_legend(override.aes = list(linetype=0),order=1))}

# taking the legend of the plot and removing the first column of the gtable within the legend
p_legend <- #cowplot::get_legend(plot(dataset, {{x_value}}, {{y_value}}, {{fill_value}}, nfill, xlab, ylab, y_limit, y_steps,legend=TRUE))
  gtable_squash_cols(cowplot::get_legend(plot(dataset, {{x_value}}, {{y_value}}, {{fill_value}},nfill, xlab, ylab, y_limit, y_steps,legend=TRUE,legend_labels)),1)

# printing the plot without legend
p_main <- plot(dataset, {{x_value}}, {{y_value}}, {{fill_value}}, nfill, xlab, ylab, y_limit, y_steps,legend=FALSE,legend_labels = NULL)

#joining it all together
Obj<- p_main+plot_spacer() + p_legend +
  plot_layout(widths=c(12.5,1.5,4))

return(Obj)

}

My problem is, that the width of the legend box seems to adjust to the size of the labels, so the distance between the plot and the legend does not remain the same.

Example Data:

set.seed(9)
data <- data.frame(Cat=c(rep("A",times=5),rep("B",times=5),rep("C", times=5)),
                   year=rep(c(2015,2016,2017,2018,2019),times=3),
                   count=c(sample(seq(60,80),replace=TRUE,size=5),sample(seq(100,140),replace=TRUE,size=5),sample(seq(20,30),replace=TRUE,size=5)))

I made four plots where only the labels are different:

plt <- custom_barplot(dataset=data %>% filter(year %in% c(2015,2016,2017)), 
     x_value=year,
     y_value=count, 
     fill_value=Cat, 
     nfill=3, 
     xlab="Year",
     ylab="Count",
     y_limit=140, 
     y_steps=20,
     legend_labels=c("A","B","C"))


plt_2 <- custom_barplot(dataset=data %>% filter(year %in% c(2015,2016,2017)), 
                        x_value=year,
                        y_value=count, 
                        fill_value=Cat, 
                        nfill=3, 
                        xlab="Year",
                        ylab="Count",
                        y_limit=140, 
                        y_steps=20,
                        legend_labels=c("Long Label A","Long Label B","Long Label C"))

plt_3 <- custom_barplot(dataset=data %>% filter(year %in% c(2015,2016,2017)), 
                        x_value=year,
                        y_value=count, 
                        fill_value=Cat, 
                        nfill=3, 
                        xlab="Year",
                        ylab="Count",
                        y_limit=140, 
                        y_steps=20,
                        legend_labels=c("Xtra Long Label A","Xtra Long Label B","Xtra Long Label C"))

plt_4 <- custom_barplot(dataset=data %>% filter(year %in% c(2015,2016,2017)), 
                        x_value=year,
                        y_value=count, 
                        fill_value=Cat, 
                        nfill=3, 
                        xlab="Year",
                        ylab="Count",
                        y_limit=140, 
                        y_steps=20,
                        legend_labels=c("Super Xtra Long Label A","Super Xtra Long Label B","Super Xtra Long Label C"))

The resulting plots look like this:

I need the space between the plot and the legend to be constant, nevermind the length of the labels in the legend. I'd rather have that label is not fully displayed (so I see I have to shorten it). The graphs are used in documents with tufte handout style and the legend should be in the same area as annotations will be.

Do you have any idea how to keep the space constant?

Reference:

• Basic method by tjebo in this post
• General adjustment of space between plot and legend by stefan in this post

I think the easiest solution is to simply apply wrapping to the text in your legend. You can do this using stringr::str_wrap() to give results like the following:

Here is a very minimal edit to your function which allows a user to control the text wrapping:

custom_barplot <- function(dataset, x_value, y_value, fill_value, nfill, xlab, ylab, y_limit, y_steps, legend_labels, wrap_labels = 20) {
  # Example color set to choose from
  colors <- c("#CF232B", "#942192", "#000000", "#f1eef6", "#addd8e", "#d0d1e6", "#31a354", "#a6bddb")

  # user function for adjusting the size of key-polygons in legend
  draw_key_polygon2 <- function(data, params, size) {
    lwd <- min(data$size, min(size) / 4)

    grid::rectGrob(
      width = grid::unit(0.8, "npc"),
      height = grid::unit(0.8, "npc"),
      gp = grid::gpar(
        col = data$colour,
        fill = alpha(data$fill, data$alpha),
        lty = data$linetype,
        lwd = lwd * .pt,
        linejoin = "mitre"
      )
    )
  }
  # user function for the plot itself
  plot <- function(dataset, x_value, y_value, fill_value, nfill, xlab, ylab, y_limit, y_steps, legend, legend_labels) {
    ggplot(data = dataset, mapping = aes(x = {{ x_value }}, y = {{ y_value }}, fill = {{ fill_value }})) +
      geom_col(position = position_dodge(width = 0.85), width = 0.8, key_glyph = "polygon2", show.legend = legend) +
      geom_smooth(aes(color = {{ fill_value }}), method = "lm", formula = y ~ x, se = FALSE, show.legend = legend, linetype = "dashed") +
      labs(x = xlab, y = ylab) +
      theme(
        text = element_text(size = 9, color = "black"),
        panel.background = element_rect(fill = "white"),
        panel.grid = element_line(color = "black", linetype = "solid", size = 0.3),
        panel.grid.minor = element_blank(),
        panel.grid.major.x = element_blank(),
        axis.text = element_text(size = 9),
        axis.line.x = element_line(color = "black"),
        axis.ticks = element_blank(),
        legend.text = element_text(size = 9),
        legend.position = "right",
        legend.justification = "top",
        legend.title = element_blank(),
        legend.key.size = unit(4, "mm"),
        legend.key = element_rect(fill = "white"),
        plot.margin = unit(c(1, 0.25, 0.5, 0.5), "cm")
      ) +
      scale_y_continuous(
        breaks = seq(from = 0, to = y_limit, by = y_steps),
        limits = c(0, y_limit + 1),
        expand = c(0, 0)
      ) +
      scale_x_continuous(breaks = min(data[, deparse(ensym(x_value))], na.rm = TRUE):max(data[, deparse(ensym(x_value))], na.rm = TRUE)) +
      scale_fill_manual(values = colors[1:nfill], labels = stringr::str_wrap({{ legend_labels }}, wrap_labels)) +
      scale_color_manual(values = colors[1:nfill], labels = stringr::str_wrap(paste("Trend ", {{ legend_labels }}, sep = ""), wrap_labels)) +
      guides(color = guide_legend(override.aes = list(fill = NA), order = 2), fill = guide_legend(override.aes = list(linetype = 0), order = 1))
  }

  # taking the legend of the plot and removing the first column of the gtable within the legend
  p_legend <- # cowplot::get_legend(plot(dataset, {{x_value}}, {{y_value}}, {{fill_value}}, nfill, xlab, ylab, y_limit, y_steps,legend=TRUE))
    gtable_squash_cols(cowplot::get_legend(plot(dataset, {{ x_value }}, {{ y_value }}, {{ fill_value }}, nfill, xlab, ylab, y_limit, y_steps, legend = TRUE, legend_labels)), 1)

  # printing the plot without legend
  p_main <- plot(dataset, {{ x_value }}, {{ y_value }}, {{ fill_value }}, nfill, xlab, ylab, y_limit, y_steps, legend = FALSE, legend_labels = NULL)

  # joining it all together
  Obj <- p_main + plot_spacer() + p_legend +
    plot_layout(widths = c(12.5, 1.5, 4))

  return(Obj)
}