top_40_bundle_source_size_changes.r

#!/usr/bin/env Rscript

app.name = "My App"
# file you gave source-map-explorer
# source-map-explorer 1.0.1/bundle.js --tsv > bundle-1.0.1.tsv
# source-map-explorer 1.5.0/bundle.js --tsv > bundle-1.5.0.tsv
asset.file = "bundle.js"
before.version = "1.0.1"
before.file = "bundle-1.0.1.tsv"
before.description = "1.0.1 (Jun 01)"
after.version = "1.5.0"
after.description = "1.5.0 (Sep 01)"
after.file = "bundle-1.5.0.tsv"

before.tsv <- read.table(file=before.file, header=TRUE, stringsAsFactors=FALSE)
after.tsv <- read.table(file=after.file, header=TRUE, stringsAsFactors=FALSE)
versions <- list(before.tsv, after.tsv)
names(versions) <- c(before.version, after.version)

library(tidyverse)

# bind before and after to one dataset, adding a new column Version
sizes <- bind_rows(versions, .id = "Version")
sizes$Version = factor(
  sizes$Version, levels=c(before.version, after.version))

# truncate the Source paths to depth
truncate_path <- function (paths, depth=2) {
  unname(sapply(paths, function (path) {
    # split path by / and take the first 2
    components <- strsplit(path, "/")[[1]][seq(depth)]
    # filter out NA if the path < depth
    components <- components[!is.na(components)]
    paste0(components, collapse="/")
  }))
}

summed_sizes <- sizes %>%
  mutate(Source = truncate_path(Source)) %>%
  # .drop=FALSE if a Source exists in one Version but
  # not the other Version it will still add a row for it with NA
  group_by(Source, Version, .drop=FALSE) %>%
  # Since we truncated the Source paths we need to sum
  # the truncated Source by Version so we have only have
  # 2 per version, we remove NA from the Sum so Size is 0
  # if a file does not exist
  summarise(Size = sum(Size, na.rm=T)) %>%
  # Group by Source and add a Change columns
  group_by(Source, Change=diff(Size), AbsChange=abs(diff(Size))) %>%
  # Arrange by absolute change for top 40
  arrange(desc(AbsChange))

# take the top 40 absolute changes by Version
# Should be 80 rows now 1 Source per Version and 2 Versions
top_40 <- summed_sizes %>%
          group_by(Version) %>%
          top_n(40, AbsChange)

# order factor this affects how ggplot orders
# the discrete x axis (currently it is alphabetical by source)
top_40$Source = factor(
  top_40$Source,
  levels=unique(top_40$Source[order(top_40$AbsChange)]))

# file size labels
library(sitools)
label_size <- function (x) {
  x[is.na(x)] <- 0
  f2si(x, unit="b")
}

abbreviate_label <- function (x) {
  abbreviate(x, method="both", minlength=50)
}

# ggplot(top_40, aes(x = Source, y = Size, color = Version)) +
#   geom_col(position="dodge") +
#   scale_x_discrete(labels = abbreviate_label) +
#   scale_y_continuous(labels = label_size) +
#   coord_flip() + theme_classic()

# collaspe out Version
# the Change and AbsChange should be the same for each Version
changes <- top_40 %>%
  group_by(Source) %>%
  summarise(Change = max(Change), AbsChange = max(AbsChange)) %>%
  # add a T/F column `Change >= 0` for maybe coloring
  mutate(Change >= 0) %>%
  arrange(desc(Change))

# calculate hi lo breaks for scale_fill_gradientn aesthetic legend
hi <- max(changes$AbsChange);
# round up to 50kb
hi <- ceiling(hi/50000) * 50000
lo <- -hi;
lims <- c(lo, hi);
breaks <- c(lo, 0, hi) # labels

# find 0 in a range of 0 to 1
mid <- 1 - hi / (hi - lo)
# values 0-1 for color points on gradient
colors <- c("orchid1", "orchid3", "purple3", "purple1")
values <- c(0,          mid,      mid+0.001, 1)

ggplot(changes, aes(x = Source, y = Change)) +
  geom_col(position="dodge", aes(fill = Change)) +
  scale_x_discrete(labels = abbreviate_label) +
  scale_y_continuous(labels = label_size) +
  scale_fill_gradientn(colors = colors,
                       limits = lims,
                       breaks = breaks,
                       values = values,
                       labels = label_size) +
  ggtitle(paste0(app.name, " Top 40 ", asset.file," Changes")) +
  ylab(paste0("Change from ", before.description, " to ", after.description)) +
  xlab(paste0("Source in ", asset.file)) +
  coord_flip() +
  theme_classic()