Allow to select datafile for charts/explore/compare (filtered/non-filtered)

scripts/Modules/Util.pm

@@ -281,6 +281,27 @@ sub delete_marker_info {
         "$local_archive_dir/$experiment/markers/$benchmark-$address.info" );
 }
 
+sub pick_data_file {
+    my ( $dir, $prefix ) = @_;
+
+    # \Q...\E treats ... as literal string
+    my @files = sort grep { /^\Q$prefix\E.*\.csv$/ } find_files($dir);
+
+    return "$prefix.csv" unless @files > 1;
+
+    # Make sure the unfiltered file is at the top
+    my @sorted = sort {
+            ( $a eq "$prefix.csv" ) ? -1
+          : ( $b eq "$prefix.csv" ) ? 1
+          : $a cmp $b
+    } @files;
+
+    my @selected =
+      TUI::select_from_list( "Select $prefix CSV file", 0, @sorted );
+    die "No $prefix CSV file selected" unless @selected;
+    return $selected[0];
+}
+
 sub select_experiment {
     my ($multi) = @_;
 

scripts/charts/combined_sankey.r

@@ -6,13 +6,16 @@ library(ggalluvial)
 args <- commandArgs(trailingOnly = TRUE)
 argc <- length(args)
 
-if (argc != 2) {
+if (argc < 2 || argc > 3) {
-  print("Expecting two input files")
+  print("Expecting two or three arguments: exp1 exp2 [faults_file]")
   stop()
 }
 
-for (experiment in args) {
+faults_file <- if (argc == 3) args[3] else "faults.csv"
-  datafile <- paste(experiment, "/faults.csv", sep = "")
+suffix <- gsub("^faults|\\.csv$", "", faults_file)
+
+for (experiment in args[1:2]) {
+  datafile <- file.path(experiment, faults_file)
   if (!file.exists(datafile)) {
     print(paste("Input file", datafile, "is missing"))
     stop()
@@ -30,13 +33,13 @@ resulttype_labels <- c(
 )
 
 # Read data
-datafile1 <- paste(args[1], "/faults.csv", sep = "")
+datafile1 <- file.path(args[1], faults_file)
 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 = "")
+datafile2 <- file.path(args[2], faults_file)
 data2 <- readr::read_csv(datafile2)
 data2$fault_address <- strtoi(data2$fault_address)
 data2$resulttype <- resulttype_labels[data2$resulttype]
@@ -75,6 +78,6 @@ plot <- ggplot(
 
 # TODO: Name the file according to the benchmarks
 ggsave(
-  paste(args[2], "/../sankey.svg", sep = ""),
+  paste0(args[2], "/../sankey", suffix, ".svg"),
   plot = plot,
 )

scripts/charts/single_heatmap.r

@@ -29,27 +29,84 @@ max_tile <- 0.5
 # Generate all heatmaps with crossproduct of this
 benchmarks <- c("ip", "mem", "regs")
 markers <- c(
-  "OK_MARKER", "FAIL_MARKER", "DETECTED_MARKER",
+  "OK_MARKER",
-  "ACCESS_OUTERSPACE", "WRITE_TEXTSEGMENT",
+  "FAIL_MARKER",
-  "GROUP1_MARKER", "TRAP", "TIMEOUT"
+  "DETECTED_MARKER",
+  "ACCESS_OUTERSPACE",
+  "WRITE_TEXTSEGMENT",
+  "GROUP1_MARKER",
+  "TRAP",
+  "TIMEOUT"
 )
 
 # Labels for _start/_end symbols from linker.ld
 regions <- list(
   list(label = "WAMR AOT", start = "_wamr_aot_start", end = "_wamr_aot_end"),
-  list(label = "WAMR os_mmap", start = "_wamr_mmap_start", end = "_wamr_mmap_end"),
+  list(
-  list(label = "WAMR runtime mem", start = "_wamr_runtime_pool_start", end = "_wamr_runtime_pool_end"),
+    label = "WAMR os_mmap",
-  list(label = "WAMR linear mem", start = "_wamr_linear_pool_start", end = "_wamr_linear_pool_end"),
+    start = "_wamr_mmap_start",
-  list(label = "WAMR global heap", start = "_wamr_global_heap_start", end = "_wamr_global_heap_end"),
+    end = "_wamr_mmap_end"
-  list(label = "IWASM AOT runtime", start = "_iwasm_aot_runtime_start", end = "_iwasm_aot_runtime_end"),
+  ),
-  list(label = "IWASM bh/util", start = "_iwasm_bh_start", end = "_iwasm_bh_end"),
+  list(
-  list(label = "IWASM mem_alloc", start = "_iwasm_mem_alloc_start", end = "_iwasm_mem_alloc_end"),
+    label = "WAMR runtime mem",
-  list(label = "IWASM platform", start = "_iwasm_platform_init_start", end = "_iwasm_platform_init_end"),
+    start = "_wamr_runtime_pool_start",
-  list(label = "IWASM exec_env", start = "_iwasm_exec_env_start", end = "_iwasm_exec_env_end"),
+    end = "_wamr_runtime_pool_end"
-  list(label = "IWASM interp", start = "_iwasm_interp_classic_start", end = "_iwasm_interp_classic_end"),
+  ),
-  list(label = "IWASM memory", start = "_iwasm_memory_start", end = "_iwasm_memory_end"),
+  list(
-  list(label = "IWASM native", start = "_iwasm_native_start", end = "_iwasm_native_end"),
+    label = "WAMR linear mem",
-  list(label = "IWASM runtime", start = "_iwasm_runtime_start", end = "_iwasm_runtime_end"),
+    start = "_wamr_linear_pool_start",
+    end = "_wamr_linear_pool_end"
+  ),
+  list(
+    label = "WAMR global heap",
+    start = "_wamr_global_heap_start",
+    end = "_wamr_global_heap_end"
+  ),
+  list(
+    label = "IWASM AOT runtime",
+    start = "_iwasm_aot_runtime_start",
+    end = "_iwasm_aot_runtime_end"
+  ),
+  list(
+    label = "IWASM bh/util",
+    start = "_iwasm_bh_start",
+    end = "_iwasm_bh_end"
+  ),
+  list(
+    label = "IWASM mem_alloc",
+    start = "_iwasm_mem_alloc_start",
+    end = "_iwasm_mem_alloc_end"
+  ),
+  list(
+    label = "IWASM platform",
+    start = "_iwasm_platform_init_start",
+    end = "_iwasm_platform_init_end"
+  ),
+  list(
+    label = "IWASM exec_env",
+    start = "_iwasm_exec_env_start",
+    end = "_iwasm_exec_env_end"
+  ),
+  list(
+    label = "IWASM interp",
+    start = "_iwasm_interp_classic_start",
+    end = "_iwasm_interp_classic_end"
+  ),
+  list(
+    label = "IWASM memory",
+    start = "_iwasm_memory_start",
+    end = "_iwasm_memory_end"
+  ),
+  list(
+    label = "IWASM native",
+    start = "_iwasm_native_start",
+    end = "_iwasm_native_end"
+  ),
+  list(
+    label = "IWASM runtime",
+    start = "_iwasm_runtime_start",
+    end = "_iwasm_runtime_end"
+  ),
   list(label = "TEXT", start = "_text_start", end = "_text_end"),
   list(label = "BSS", start = "_sbss", end = "_ebss")
 )
@@ -75,22 +132,27 @@ if (length(args) < 1) {
 }
 
 experiment <- args[1]
+faults_file <- if (length(args) >= 2) args[2] else "faults.csv"
+suffix <- gsub("^faults|\\.csv$", "", faults_file)
 
 # =============================================================================
 # INPUT DATA  (read once)
 # =============================================================================
 
-datafile <- file.path(experiment, "faults.csv")
+datafile <- file.path(experiment, faults_file)
 if (!file.exists(datafile)) {
   stop(paste("Input file not found:", datafile))
 }
 
-raw <- read_csv(datafile, col_types = cols(
+raw <- read_csv(
-  benchmark     = col_character(),
+  datafile,
-  resulttype    = col_character(),
+  col_types = cols(
-  faults        = col_double(),
+    benchmark = col_character(),
-  fault_address = col_character() # hex string "0x10001A"; converted below
+    resulttype = col_character(),
-))
+    faults = col_double(),
+    fault_address = col_character() # hex string "0x10001A"; converted below
+  )
+)
 
 # =============================================================================
 # ELF SYMBOLS  (parsed once)
@@ -155,10 +217,12 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
 
   # "0x10001A" -> substr strips "0x" -> strtoi parses base-16 -> integer
   aggregated <- aggregated |>
-    mutate(addr_int = strtoi(
+    mutate(
-      substr(.data$fault_address, 3L, nchar(.data$fault_address)),
+      addr_int = strtoi(
-      16L
+        substr(.data$fault_address, 3L, nchar(.data$fault_address)),
-    ))
+        16L
+      )
+    )
 
   # ===========================================================================
   # SCALE ROWS
@@ -168,17 +232,24 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
   row_width <- row_width_init
 
   # Double row_width until occupied rows <= max_rows
-  while (row_width < 65536L && n_occupied_rows(
+  while (
-    aggregated$addr_int, row_width
+    row_width < 65536L &&
-  ) > max_rows) {
+      n_occupied_rows(
+        aggregated$addr_int,
+        row_width
+      ) >
+        max_rows
+  ) {
     row_width <- row_width * 2L
   }
 
   if (row_width > row_width_init) {
     message(sprintf(
       "Note: [%s/%s] row_width auto-scaled to %d (%d occupied rows)",
-      target_resulttype, target_benchmark,
+      target_resulttype,
-      row_width, n_occupied_rows(aggregated$addr_int, row_width)
+      target_benchmark,
+      row_width,
+      n_occupied_rows(aggregated$addr_int, row_width)
     ))
   }
 
@@ -210,8 +281,8 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
   # - Adding that offset to 1...n gives the row_idx values with gap slots
   cumulative_gaps <- cumsum(has_gap_before)
   row_order <- tibble(
-    row            = rows_sorted,
+    row = rows_sorted,
-    row_idx        = seq_len(n_data_rows) + cumulative_gaps,
+    row_idx = seq_len(n_data_rows) + cumulative_gaps,
     has_gap_before = has_gap_before
   )
 
@@ -240,8 +311,8 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
 
   region_rects <- data.frame(
     label = character(0),
-    ymin  = numeric(0),
+    ymin = numeric(0),
-    ymax  = numeric(0)
+    ymax = numeric(0)
   )
 
   if (length(sym_addr) > 0) {
@@ -256,7 +327,8 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
       # Row with base address r covers bytes r ... r + row_width - 1.
       # Overlap if r < e && r + row_width > s
       overlapping <- row_order[
-        row_order$row < e & (row_order$row + row_width) > s, ,
+        row_order$row < e & (row_order$row + row_width) > s,
+        ,
         drop = FALSE
       ]
 
@@ -266,8 +338,8 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
 
       data.frame(
         label = reg$label,
-        ymin  = min(overlapping$row_idx) - 0.5,
+        ymin = min(overlapping$row_idx) - 0.5,
-        ymax  = max(overlapping$row_idx) + 0.5
+        ymax = max(overlapping$row_idx) + 0.5
       )
     })
 
@@ -310,9 +382,14 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
   # PLOT
   # ===========================================================================
 
-  plot <- ggplot(grid_complete, aes(
+  plot <- ggplot(
-    x = col, y = .data$row_idx, fill = .data$faults
+    grid_complete,
-  )) +
+    aes(
+      x = col,
+      y = .data$row_idx,
+      fill = .data$faults
+    )
+  ) +
 
     # One rectangle per (col, row_idx) tuple
     geom_tile(width = 1, height = 1, colour = NA) +
@@ -330,10 +407,10 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
 
     # Heatmap color ramp
     scale_fill_viridis_c(
-      name     = "Faults",
+      name = "Faults",
-      trans    = "log1p",
+      trans = "log1p",
       na.value = "grey85",
-      option   = "viridis"
+      option = "viridis"
     ) +
 
     # X-axis hex labels
@@ -355,10 +432,13 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
     # Title + axis labels
     labs(
       title = paste(target_resulttype, "/", target_benchmark),
-      subtitle = paste("Total:", format(
+      subtitle = paste(
-        sum(aggregated$faults, na.rm = TRUE),
+        "Total:",
-        big.mark = ","
+        format(
-      )),
+          sum(aggregated$faults, na.rm = TRUE),
+          big.mark = ","
+        )
+      ),
       x = "Byte Offset",
       y = "Base Address"
     ) +
@@ -367,7 +447,10 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
     theme_minimal() +
     theme(
       axis.text.x = element_text(
-        family = "mono", angle = 45, hjust = 1, size = 9
+        family = "mono",
+        angle = 45,
+        hjust = 1,
+        size = 9
       ),
       axis.text.y = element_text(family = "mono", size = 9),
       panel.grid = element_blank(),
@@ -399,9 +482,17 @@ make_heatmap <- function(target_resulttype, target_benchmark) {
   fig_w <- row_width * tile_size + 4.5
   fig_h <- total_slots * tile_size + 2.5
 
-  outfile <- file.path(experiment, paste0(
+  outfile <- file.path(
-    "heatmap_", target_resulttype, "_", target_benchmark, ".svg"
+    experiment,
-  ))
+    paste0(
+      "heatmap_",
+      target_resulttype,
+      "_",
+      target_benchmark,
+      suffix,
+      ".svg"
+    )
+  )
 
   ggsave(outfile, plot = plot, width = fig_w, height = fig_h, units = "in")
   message(sprintf("Saved: %s", basename(outfile)))

scripts/charts/single_result.r

@@ -1,10 +1,12 @@
 library(ggplot2)
 
-# Usage: Rscript single_result.r exp_abspath
+# Usage: Rscript single_result.r exp_abspath [resultsdata_file]
 
 args <- commandArgs(trailingOnly = TRUE)
 experiment <- args[1]
-datafile <- paste(experiment, "/resultsdata.csv", sep = "")
+resultsdata_file <- if (length(args) >= 2) args[2] else "resultsdata.csv"
+suffix <- gsub("^resultsdata|\\.csv$", "", resultsdata_file)
+datafile <- file.path(experiment, resultsdata_file)
 
 if (!file.exists(datafile)) {
   print(paste("Input file", datafile, "is missing"))
@@ -21,6 +23,6 @@ plot <- ggplot(data, aes(x = benchmark, y = faults, fill = resulttype)) +
   theme_minimal()
 
 ggsave(
-  paste(experiment, "/single_result.svg", sep = ""),
+  paste0(experiment, "/single_result", suffix, ".svg"),
   plot = plot,
 )

scripts/charts/single_scatter.r

@@ -6,7 +6,9 @@ library(ggplot2)
 
 args <- commandArgs(trailingOnly = TRUE)
 experiment <- args[1]
-datafile <- paste(experiment, "/faults.csv", sep = "")
+faults_file <- if (length(args) >= 2) args[2] else "faults.csv"
+suffix <- gsub("^faults|\\.csv$", "", faults_file)
+datafile <- file.path(experiment, faults_file)
 
 if (!file.exists(datafile)) {
   print(paste("Input file", datafile, "is missing"))
@@ -27,6 +29,6 @@ plot <- ggplot(data, aes(x = fault_address, y = faults)) +
   theme_minimal()
 
 ggsave(
-  paste(experiment, "/scatter.svg", sep = ""),
+  paste0(experiment, "/scatter", suffix, ".svg"),
   plot = plot,
 )

scripts/explore.pl

@@ -23,6 +23,9 @@ my $local_archive_dir = "$local_root/injections";
 # Select experiment to open
 my $selected_experiment = Util::select_experiment(0);
 
+my $selected_faults_csv =
+  Util::pick_data_file( "$local_archive_dir/$selected_experiment", "faults" );
+
 my $cui = TUI::init_cui();
 
 # TODO: Add a TextEditor panel beside the experiment selection for notes.
@@ -95,7 +98,7 @@ sub load_faults_csv {
 
     # Schema: benchmark, resulttype, faults, fault_address
     my $data = Text::CSV_XS::csv(
-        in      => "$local_archive_dir/$selected_experiment/faults.csv",
+        in => "$local_archive_dir/$selected_experiment/$selected_faults_csv",
         headers => 'auto'
     );
 

scripts/menu.pl

@@ -241,13 +241,18 @@ my %handlers = (
 
         my @selected_experiments = Util::select_experiment(1);
 
+       # TODO: Fails silently if not every selected experiment has this datafile
+        my $resultsdata_csv =
+          Util::pick_data_file( "$local_archive_dir/$selected_experiments[0]",
+            "resultsdata" );
+
         # Read results
         my %all_results;
         foreach my $experiment (@selected_experiments) {
 
             # Schema: benchmark, resulttype, faults
             my $data = Text::CSV_XS::csv(
-                in      => "$local_archive_dir/$experiment/resultsdata.csv",
+                in      => "$local_archive_dir/$experiment/$resultsdata_csv",
                 headers => 'auto'
             );
 
@@ -479,15 +484,38 @@ my %handlers = (
           TUI::select_from_list( "Select Plots to Generate", 1, @charts );
         die "No plot selected" unless @selected_charts;
 
+        # Need to know which chart uses which datafile
+        my @faults_charts = grep { /heatmap|scatter|sankey/ } @selected_charts;
+        my @resultsdata_charts =
+          grep { /result$|combined_comparison/ } @selected_charts;
+
+        my $faults_csv;
+        my $resultsdata_csv;
+        if (@faults_charts) {
+            $faults_csv = Util::pick_data_file(
+                "$local_archive_dir/$selected_experiments[0]", "faults" );
+        }
+        if (@resultsdata_charts) {
+            $resultsdata_csv = Util::pick_data_file(
+                "$local_archive_dir/$selected_experiments[0]",
+                "resultsdata" );
+        }
+
         my @single_charts = grep { /single/ } @selected_charts;
         foreach my $experiment (@selected_experiments) {
             foreach my $chart (@single_charts) {
                 say " - Generating plot $chart for $experiment...";
-                system(
+                my @r_args = (
                     'Rscript',
                     "$local_charts_dir/$chart.r",
                     "$local_archive_dir/$experiment"
                 );
+                push @r_args, $faults_csv
+                  if defined $faults_csv && $chart =~ /heatmap|scatter|sankey/;
+                push @r_args, $resultsdata_csv
+                  if defined $resultsdata_csv
+                  && $chart =~ /result$|combined_comparison/;
+                system(@r_args);
             }
         }
 
@@ -497,8 +525,14 @@ my %handlers = (
           map { "$local_archive_dir/$_" } @selected_experiments;
         foreach my $chart (@combined_charts) {
             say " - Generating plot $chart for ($print_experiments)...";
-            system( 'Rscript', "$local_charts_dir/$chart.r",
+            my @r_args =
-                @path_experiments );
+              ( 'Rscript', "$local_charts_dir/$chart.r", @path_experiments );
+            push @r_args, $faults_csv
+              if defined $faults_csv && $chart =~ /heatmap|scatter|sankey/;
+            push @r_args, $resultsdata_csv
+              if defined $resultsdata_csv
+              && $chart =~ /result$|combined_comparison/;
+            system(@r_args);
         }
     },