From 4cdeec3ccf81ab43fb20bd371b9cfb616f8f5b95 Mon Sep 17 00:00:00 2001
From: AnonymousRandomPerson <chenghanngan.us@gmail.com>
Date: Sun, 9 Jun 2024 01:48:22 -0400
Subject: [PATCH] Added documented symbol counter

---
 asm/include/main_0200C4CC.inc                 |  2 +-
 asm/main_0200C4CC.s                           |  4 +-
 asm/main_02075600.s                           |  4 +-
 .../count_documented_symbols.py               | 59 +++++++++++++++++++
 4 files changed, 64 insertions(+), 5 deletions(-)
 create mode 100644 tools/sync_pmdsky_debug/count_documented_symbols.py

diff --git a/asm/include/main_0200C4CC.inc b/asm/include/main_0200C4CC.inc
index 202f2114..8c88c94f 100644
--- a/asm/include/main_0200C4CC.inc
+++ b/asm/include/main_0200C4CC.inc
@@ -711,7 +711,7 @@
 .public _022A8994
 .public _022A92B4_JP
 .public _022A92B8_JP
-.public _022A92C4_JP
+.public _022A92C4
 .public _022A9A5C
 .public _022AA35C
 .public _022AAC64
diff --git a/asm/main_0200C4CC.s b/asm/main_0200C4CC.s
index 1b2b0888..319ce344 100644
--- a/asm/main_0200C4CC.s
+++ b/asm/main_0200C4CC.s
@@ -33112,7 +33112,7 @@ sub_02025AD8: ; 0x02025AD8
 	mov r2, #1
 	bl LoadFileFromRom
 	ldr r1, [sp]
-	ldr r0, _02025B3C ; =_022A92C4_JP
+	ldr r0, _02025B3C ; =_022A92C4
 	bl HandleSir0Translation
 	ldr r1, _02025B40 ; =_0209AC18
 	add r0, sp, #0
@@ -33184,7 +33184,7 @@ sub_02025AD8: ; 0x02025AD8
 _02025B6C: .word _0209B548_JP
 _02025B34: .word _022A7A54
 _02025B38: .word _0209AC04
-_02025B3C: .word _022A92C4_JP
+_02025B3C: .word _022A92C4
 _02025B40: .word _0209AC18
 _02025B44: .word _020AFD04
 _02025B48: .word _022A92B4_JP
diff --git a/asm/main_02075600.s b/asm/main_02075600.s
index 95c490a2..a3725cd1 100644
--- a/asm/main_02075600.s
+++ b/asm/main_02075600.s
@@ -58984,8 +58984,8 @@ _022A92B8_JP:
 	.global _022A7A54
 _022A7A54:
 	.space 0x4
-	.global _022A92C4_JP
-_022A92C4_JP:
+	.global _022A92C4
+_022A92C4:
 	.space 0x4
 #if defined(EUROPE)
 	.global _022A7A64
diff --git a/tools/sync_pmdsky_debug/count_documented_symbols.py b/tools/sync_pmdsky_debug/count_documented_symbols.py
new file mode 100644
index 00000000..5549a2ac
--- /dev/null
+++ b/tools/sync_pmdsky_debug/count_documented_symbols.py
@@ -0,0 +1,59 @@
+import re
+from typing import Dict
+from xmap_reader import read_xmap_symbols
+
+UNDOCUMENTED_SYMBOL_MATCHER = re.compile(r'_[0-9A-F]{8}$')
+
+xmap_symbols = read_xmap_symbols()
+
+def get_fraction_and_percent(documented_count, total_count):
+    return f'{documented_count}/{total_count} ({documented_count / total_count:.4%})'
+
+class SymbolCount:
+    num_documented_data_symbols: int = 0
+    total_data_symbols: int = 0
+    num_documented_symbols: int = 0
+    total_symbols: int = 0
+
+    def print_counts(self):
+        if self.total_data_symbols != self.total_symbols:
+            print(f'Total:     {get_fraction_and_percent(self.num_documented_symbols, self.total_symbols)}')
+            print(f'Functions: {get_fraction_and_percent(self.num_documented_symbols - self.num_documented_data_symbols, self.total_symbols - self.total_data_symbols)}')
+        print(f'Data:      {get_fraction_and_percent(self.num_documented_data_symbols, self.total_data_symbols)}')
+        print()
+
+symbol_counts: Dict[str, SymbolCount] = {}
+for overlay, symbols in xmap_symbols['us'].items():
+    if overlay == 'ram':
+        continue
+
+    symbol_count = SymbolCount()
+    symbol_counts[overlay] = symbol_count
+
+    symbol_count.total_symbols = len(symbols)
+
+    for symbol in symbols.values():
+        if not UNDOCUMENTED_SYMBOL_MATCHER.search(symbol.name) and 'UNKNOWN' not in symbol.name:
+            symbol_count.num_documented_symbols += 1
+            if symbol.is_data:
+                symbol_count.num_documented_data_symbols += 1
+        if symbol.is_data:
+            symbol_count.total_data_symbols += 1
+
+    if symbol_count.total_symbols > 0:
+        if overlay == 'main':
+            overlay = 'ARM9'
+        elif overlay == 'arm7':
+            overlay = 'ARM7'
+        else:
+            overlay = f'Overlay {overlay}'
+        print(overlay)
+        symbol_count.print_counts()
+
+all_symbol_count = SymbolCount()
+all_symbol_count.num_documented_data_symbols = sum([count.num_documented_data_symbols for count in symbol_counts.values()])
+all_symbol_count.num_documented_symbols = sum([count.num_documented_symbols for count in symbol_counts.values()])
+all_symbol_count.total_data_symbols = sum([count.total_data_symbols for count in symbol_counts.values()])
+all_symbol_count.total_symbols = sum([count.total_symbols for count in symbol_counts.values()])
+
+all_symbol_count.print_counts()
\ No newline at end of file