From 4c0d6c9821007c9fe99a16ae9d5cb9fb78adcb17 Mon Sep 17 00:00:00 2001
From: Christoph Urlacher <christoph.urlacher@protonmail.com>
Date: Sat, 25 Jul 2026 15:30:18 +0200
Subject: [PATCH] Add mnemonics.csv query to count mnemonics

---
 scripts/Queries/Mnemonics.pm | 50 ++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
 create mode 100644 scripts/Queries/Mnemonics.pm

diff --git a/scripts/Queries/Mnemonics.pm b/scripts/Queries/Mnemonics.pm
new file mode 100644
index 0000000..d68536c
--- /dev/null
+++ b/scripts/Queries/Mnemonics.pm
@@ -0,0 +1,50 @@
+package Mnemonics;
+
+use strict;
+use warnings;
+use diagnostics;
+
+use FindBin;
+use lib "$FindBin::Bin/../Modules";
+use Filters;
+
+use feature 'say';
+
+sub query {
+    my ( $experiment, $experiment_dir, $filter_config_name ) = @_;
+    $filter_config_name //= '';
+
+    my $filters =
+      Filters::build_filter_clause( $experiment_dir, $filter_config_name );
+
+    my $querystring = "SELECT
+CASE
+  WHEN o.disassemble IS NOT NULL AND SUBSTRING_INDEX(TRIM(o.disassemble), ' ', 1) REGEXP '^[a-zA-Z][a-zA-Z0-9 ]*\$'
+  THEN SUBSTRING_INDEX(TRIM(o.disassemble), ' ', 1)
+  ELSE NULL
+END AS mnemonic,
+COUNT(DISTINCT CONCAT(t.instr2, '_', t.data_physical_address)) AS count
+FROM variant v
+JOIN trace t ON v.id = t.variant_id
+JOIN objdump o ON o.variant_id = v.id AND o.instr_address = t.instr2_absolute
+WHERE v.variant = '$experiment'$filters
+GROUP BY mnemonic
+ORDER BY count DESC;";
+
+    say $querystring;
+
+    return $querystring;
+}
+
+sub args { return "--batch --raw"; }
+
+sub filename {
+    my ($filter_config_name) = @_;
+    $filter_config_name //= '';
+    my $suffix = length($filter_config_name) ? "_$filter_config_name" : "";
+    return "mnemonics${suffix}.csv";
+}
+
+sub postprocess { $_[0] =~ s/\t/,/g; }
+
+1;