From bba9ced3483ac29487ba78af3b06429264e565e5 Mon Sep 17 00:00:00 2001
From: Christoph Urlacher <christoph.urlacher@protonmail.com>
Date: Fri, 24 Apr 2026 15:27:41 +0200
Subject: [PATCH] add sankey chart

---
 charts/combined_sankey.r | 80 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 80 insertions(+)
 create mode 100644 charts/combined_sankey.r

diff --git a/charts/combined_sankey.r b/charts/combined_sankey.r
new file mode 100644
index 0000000..88f6e9b
--- /dev/null
+++ b/charts/combined_sankey.r
@@ -0,0 +1,80 @@
+library(ggplot2)
+library(ggalluvial)
+
+# Usage: Rscript combined_comparion.r exp_abspath1 exp_abspath2 ...
+
+args <- commandArgs(trailingOnly = TRUE)
+argc <- length(args)
+
+if (argc != 2) {
+  print("Expecting two input files")
+  stop()
+}
+
+for (experiment in args) {
+  datafile <- paste(experiment, "/faults.csv", sep = "")
+  if (!file.exists(datafile)) {
+    print(paste("Input file", datafile, "is missing"))
+    stop()
+  }
+}
+
+resulttype_labels <- c(
+  OK_MARKER = "OK",
+  FAIL_MARKER = "FAIL",
+  DETECTED_MARKER = "DETECTED",
+  TIMEOUT = "TIMEOUT",
+  TRAP = "TRAP",
+  ACCESS_OUTERSPACE = "OUTERSPC",
+  WRITE_TEXTSEGMENT = "WRITETXT"
+)
+
+# Read data
+datafile1 <- paste(args[1], "/faults.csv", sep = "")
+data1 <- readr::read_csv(datafile1)
+data1$fault_address <- strtoi(data1$fault_address)
+data1$resulttype <- resulttype_labels[data1$resulttype]
+# tibble::glimpse(data1)
+
+datafile2 <- paste(args[2], "/faults.csv", sep = "")
+data2 <- readr::read_csv(datafile2)
+data2$fault_address <- strtoi(data2$fault_address)
+data2$resulttype <- resulttype_labels[data2$resulttype]
+# tibble::glimpse(data2)
+
+# https://corybrunson.github.io/ggalluvial/
+joined <- merge(
+  data1[, c("fault_address", "resulttype", "faults")],
+  data2[, c("fault_address", "resulttype", "faults")],
+  by = "fault_address",
+  suffixes = c("_bench1", "_bench2")
+)
+
+streams <- aggregate(
+  faults_bench2 ~ resulttype_bench1 + resulttype_bench2,
+  data = joined,
+  sum
+)
+names(streams) <- c("bench1", "bench2", "faults")
+
+plot <- ggplot(
+  data = streams,
+  aes(axis1 = bench1, axis2 = bench2, y = faults)
+) +
+  scale_x_discrete(
+    # TODO: Name the benchmarks
+    # limits = c(args[1], args[2])
+    limits = c("Bench A", "Bench B")
+  ) +
+  labs(x = "Benchmark", y = "Faults") +
+  geom_alluvium(aes(fill = bench1)) +
+  geom_stratum() +
+  geom_text(stat = "stratum", aes(label = after_stat(stratum))) +
+  theme_minimal() +
+  theme(legend.position = "none")
+
+# TODO: Name the file according to the benchmarks
+ggsave(
+  paste(args[2], "/../sankey.svg", sep = ""),
+  plot = plot,
+)