From dc24fde7f6fa092f342dbbb7557d6fdbf6af3784 Mon Sep 17 00:00:00 2001
From: GriffinR <griffin.g.richards@gmail.com>
Date: Sun, 27 Nov 2022 17:33:09 -0500
Subject: [PATCH] Sync calcrom fixes

---
 .github/calcrom/calcrom.pl | 20 ++++++++++++--------
 1 file changed, 12 insertions(+), 8 deletions(-)

diff --git a/.github/calcrom/calcrom.pl b/.github/calcrom/calcrom.pl
index 9fbd95994..cb8c5723e 100755
--- a/.github/calcrom/calcrom.pl
+++ b/.github/calcrom/calcrom.pl
@@ -1,5 +1,11 @@
 #!/usr/bin/perl
 
+# Usage:
+#   calcrom.pl <mapfile> [--data]
+#
+#   mapfile: path to .map file output by LD
+#   data: set to output % breakdown of data
+
 use IPC::Cmd qw[ run ];
 use Getopt::Long;
 
@@ -65,12 +71,13 @@ my $base_cmd = "nm $elffname | awk '{print \$3}' | grep '^[^_].\\{4\\}' | uniq";
 
 # This looks for Unknown_, Unknown_, or sub_, followed by an address. Note that
 # it matches even if stuff precedes the unknown, like sUnknown/gUnknown.
-my $undoc_cmd = "grep '[Uu]nknown_[0-9a-fA-F]\\{5,7\\}\\|sub_[0-9a-fA-F]\\{5,7\\}'";
+my $undoc_regex = "'[Uu]nknown_[0-9a-fA-F]\\{5,7\\}\\|sub_[0-9a-fA-F]\\{5,7\\}'";
 
 # This looks for every symbol with an address at the end of it. Some things are
 # given a name based on their type / location, but still have an unknown purpose.
 # For example, FooMap_EventScript_FFFFFFF.
-my $partial_doc_cmd = "grep '_[0-28][0-9a-fA-F]\\{5,7\\}'";
+# The above may be double counted here, and will need to be filtered out.
+my $partial_doc_regex = "'_[0-28][0-9a-fA-F]\\{5,7\\}'";
 
 my $count_cmd = "wc -l";
 
@@ -87,7 +94,7 @@ my $total_syms_as_string;
 
 my $undocumented_as_string;
 (run (
-    command => "$base_cmd | $undoc_cmd | $count_cmd",
+    command => "$base_cmd | grep $undoc_regex | $count_cmd",
     buffer => \$undocumented_as_string,
     timeout => 60
 ))
@@ -95,7 +102,7 @@ my $undocumented_as_string;
 
 my $partial_documented_as_string;
 (run (
-    command => "$base_cmd | $partial_doc_cmd | $count_cmd",
+    command => "$base_cmd | grep $partial_doc_regex | grep -v $undoc_regex | $count_cmd",
     buffer => \$partial_documented_as_string,
     timeout => 60
 ))
@@ -112,7 +119,7 @@ my $undocumented = $undocumented_as_string + 0;
 $partial_documented_as_string =~ s/^\s+|\s+$//g;
 my $partial_documented = $partial_documented_as_string + 0;
 (($partial_documented != 0) or (($partial_documented == 0) and ($partial_documented_as_string eq "0")))
-	or die "ERROR: Cannot convert string to num: '$partial_documented_as_string'";
+    or die "ERROR: Cannot convert string to num: '$partial_documented_as_string'";
 
 $total_syms_as_string =~ s/^\s+|\s+$//g;
 my $total_syms = $total_syms_as_string + 0;
@@ -126,9 +133,6 @@ my $total = $src + $asm;
 my $srcPct = sprintf("%.4f", 100 * $src / $total);
 my $asmPct = sprintf("%.4f", 100 * $asm / $total);
 
-# partial_documented is double-counting the unknown_* and sub_* symbols.
-$partial_documented = $partial_documented - $undocumented;
-
 my $documented = $total_syms - ($undocumented + $partial_documented);
 my $docPct = sprintf("%.4f", 100 * $documented / $total_syms);
 my $partialPct = sprintf("%.4f", 100 * $partial_documented / $total_syms);