mostly finish string_util.c

7 files changed, 322 insertions, 480 deletions

diff --git a/asm/string_util.s b/asm/string_util.s
deleted file mode 100644
index 8de4972c0..000000000
--- a/asm/string_util.s
+++ /dev/null
@@ -1,368 +0,0 @@
-	.include "asm/macros.inc"
-	.include "constants/constants.inc"
-
-	.syntax unified
-
-	.text
-
-	thumb_func_start StringCopyN_Multibyte
-@ u8 *CopyString_LimitN_Multibyte(u8 *dest, u8 *src, u32 n)
-StringCopyN_Multibyte: @ 800918C
-	push {r4,r5,lr}
-	adds r4, r0, 0
-	adds r3, r1, 0
-	subs r2, 0x1
-	movs r5, 0x1
-	negs r5, r5
-	b _080091B2
-_0800919A:
-	strb r0, [r4]
-	adds r3, 0x1
-	adds r4, 0x1
-	subs r0, r3, 0x1
-	ldrb r0, [r0]
-	cmp r0, 0xF9
-	bne _080091B0
-	ldrb r0, [r3]
-	strb r0, [r4]
-	adds r3, 0x1
-	adds r4, 0x1
-_080091B0:
-	subs r2, 0x1
-_080091B2:
-	cmp r2, r5
-	beq _080091BE
-	ldrb r0, [r3]
-	adds r1, r0, 0
-	cmp r1, 0xFF
-	bne _0800919A
-_080091BE:
-	movs r0, 0xFF
-	strb r0, [r4]
-	adds r0, r4, 0
-	pop {r4,r5}
-	pop {r1}
-	bx r1
-	thumb_func_end StringCopyN_Multibyte
-
-	thumb_func_start StringLength_Multibyte
-@ u32 GetStringLength_Multibyte(u8 *s)
-StringLength_Multibyte: @ 80091CC
-	push {lr}
-	adds r2, r0, 0
-	movs r3, 0
-	b _080091DE
-_080091D4:
-	cmp r1, 0xF9
-	bne _080091DA
-	adds r2, 0x1
-_080091DA:
-	adds r2, 0x1
-	adds r3, 0x1
-_080091DE:
-	ldrb r1, [r2]
-	adds r0, r1, 0
-	cmp r0, 0xFF
-	bne _080091D4
-	adds r0, r3, 0
-	pop {r1}
-	bx r1
-	thumb_func_end StringLength_Multibyte
-
-	thumb_func_start WriteColorChangeControlCode
-@ u8 *WriteColorChangeControlCode(u8 *dest, u8 colorType, u8 color)
-WriteColorChangeControlCode: @ 80091EC
-	push {lr}
-	adds r3, r0, 0
-	lsls r2, 24
-	lsrs r2, 24
-	movs r0, 0xFC
-	strb r0, [r3]
-	adds r3, 0x1
-	cmp r1, 0x1
-	beq _0800920E
-	cmp r1, 0x1
-	bcc _08009208
-	cmp r1, 0x2
-	beq _08009214
-	b _08009218
-_08009208:
-	movs r0, 0x1
-	strb r0, [r3]
-	b _08009216
-_0800920E:
-	movs r0, 0x3
-	strb r0, [r3]
-	b _08009216
-_08009214:
-	strb r1, [r3]
-_08009216:
-	adds r3, 0x1
-_08009218:
-	strb r2, [r3]
-	adds r3, 0x1
-	movs r0, 0xFF
-	strb r0, [r3]
-	adds r0, r3, 0
-	pop {r1}
-	bx r1
-	thumb_func_end WriteColorChangeControlCode
-
-	thumb_func_start sub_8009228
-@ bool8 sub_8009228(u8 *s)
-sub_8009228: @ 8009228
-	push {lr}
-	adds r2, r0, 0
-	b _0800923E
-_0800922E:
-	adds r0, r1, 0
-	cmp r0, 0xA0
-	bhi _0800923C
-	cmp r0, 0
-	beq _0800923C
-	movs r0, 0x1
-	b _08009248
-_0800923C:
-	adds r2, 0x1
-_0800923E:
-	ldrb r1, [r2]
-	adds r0, r1, 0
-	cmp r0, 0xFF
-	bne _0800922E
-	movs r0, 0
-_08009248:
-	pop {r1}
-	bx r1
-	thumb_func_end sub_8009228
-
-	thumb_func_start sub_800924C
-@ bool8 sub_800924C(u8 *s, u32 n)
-sub_800924C: @ 800924C
-	push {r4,lr}
-	adds r3, r0, 0
-	adds r4, r1, 0
-	movs r1, 0
-	b _08009268
-_08009256:
-	adds r0, r2, 0
-	cmp r0, 0xA0
-	bhi _08009264
-	cmp r0, 0
-	beq _08009264
-	movs r0, 0x1
-	b _08009276
-_08009264:
-	adds r3, 0x1
-	adds r1, 0x1
-_08009268:
-	ldrb r2, [r3]
-	adds r0, r2, 0
-	cmp r0, 0xFF
-	beq _08009274
-	cmp r1, r4
-	blt _08009256
-_08009274:
-	movs r0, 0
-_08009276:
-	pop {r4}
-	pop {r1}
-	bx r1
-	thumb_func_end sub_800924C
-
-	thumb_func_start GetExtCtrlCodeLength
-@ u8 GetExtendedControlCodeLength(u8 code)
-GetExtCtrlCodeLength: @ 800927C
-	push {lr}
-	lsls r0, 24
-	lsrs r1, r0, 24
-	movs r0, 0
-	cmp r1, 0x18
-	bhi _0800928E
-	ldr r0, =gExtendedControlCodeLengths
-	adds r0, r1, r0
-	ldrb r0, [r0]
-_0800928E:
-	pop {r1}
-	bx r1
-	.pool
-	thumb_func_end GetExtCtrlCodeLength
-
-	thumb_func_start SkipExtCtrlCode
-@ u8 *SkipExtendedControlCode(u8 *s)
-SkipExtCtrlCode: @ 8009298
-	push {r4,lr}
-	adds r4, r0, 0
-	b _080092AC
-_0800929E:
-	adds r4, 0x1
-	ldrb r0, [r4]
-	bl GetExtCtrlCodeLength
-	lsls r0, 24
-	lsrs r0, 24
-	adds r4, r0
-_080092AC:
-	ldrb r0, [r4]
-	cmp r0, 0xFC
-	beq _0800929E
-	adds r0, r4, 0
-	pop {r4}
-	pop {r1}
-	bx r1
-	thumb_func_end SkipExtCtrlCode
-
-	thumb_func_start StringCompareWithoutExtCtrlCodes
-@ u8 CompareStringWithoutExtendedControlCodes(u8 *s1, u8 *s2)
-StringCompareWithoutExtCtrlCodes: @ 80092BC
-	push {r4-r6,lr}
-	adds r4, r0, 0
-	adds r5, r1, 0
-	movs r6, 0
-	b _080092DE
-_080092C6:
-	cmp r1, r0
-	bcs _080092D4
-	movs r6, 0x1
-	negs r6, r6
-	cmp r0, 0xFF
-	bne _080092D4
-	movs r6, 0x1
-_080092D4:
-	ldrb r0, [r4]
-	cmp r0, 0xFF
-	beq _080092FE
-	adds r4, 0x1
-	adds r5, 0x1
-_080092DE:
-	adds r0, r4, 0
-	bl SkipExtCtrlCode
-	adds r4, r0, 0
-	adds r0, r5, 0
-	bl SkipExtCtrlCode
-	adds r5, r0, 0
-	ldrb r1, [r4]
-	ldrb r0, [r5]
-	cmp r1, r0
-	bls _080092C6
-	movs r6, 0x1
-	cmp r1, 0xFF
-	bne _080092FE
-	subs r6, 0x2
-_080092FE:
-	adds r0, r6, 0
-	pop {r4-r6}
-	pop {r1}
-	bx r1
-	thumb_func_end StringCompareWithoutExtCtrlCodes
-
-	thumb_func_start ConvertInternationalString
-@ void StopMusicWhileStringIsPrinted(u8 *s, bool8 stopMusic)
-ConvertInternationalString: @ 8009308
-	push {r4,lr}
-	adds r4, r0, 0
-	lsls r1, 24
-	lsrs r1, 24
-	cmp r1, 0x1
-	bne _08009368
-	bl StripExtCtrlCodes
-	adds r0, r4, 0
-	bl StringLength
-	lsls r0, 24
-	lsrs r2, r0, 24
-	adds r1, r2, 0
-	adds r0, r1, 0x1
-	lsls r0, 24
-	lsrs r2, r0, 24
-	adds r3, r2, 0
-	adds r1, r4, r1
-	movs r0, 0xFC
-	strb r0, [r1]
-	adds r0, r3, 0x1
-	lsls r0, 24
-	lsrs r2, r0, 24
-	adds r3, r4, r3
-	movs r0, 0x16
-	strb r0, [r3]
-	adds r0, r2, 0x1
-	lsls r0, 24
-	adds r2, r4, r2
-	movs r1, 0xFF
-	strb r1, [r2]
-	movs r1, 0xFF
-	lsls r1, 24
-	adds r0, r1
-	b _0800935A
-_08009350:
-	adds r1, r2, r4
-	ldrb r0, [r1]
-	strb r0, [r1, 0x2]
-	subs r0, r2, 0x1
-	lsls r0, 24
-_0800935A:
-	lsrs r2, r0, 24
-	cmp r2, 0xFF
-	bne _08009350
-	movs r0, 0xFC
-	strb r0, [r4]
-	movs r0, 0x15
-	strb r0, [r4, 0x1]
-_08009368:
-	pop {r4}
-	pop {r0}
-	bx r0
-	thumb_func_end ConvertInternationalString
-
-	thumb_func_start StripExtCtrlCodes
-@ void StripExtendedControlCodes(u8 *s)
-StripExtCtrlCodes: @ 8009370
-	push {r4-r6,lr}
-	adds r5, r0, 0
-	movs r4, 0
-	movs r6, 0
-	ldrb r0, [r5]
-	cmp r0, 0xFF
-	beq _080093C0
-_0800937E:
-	adds r0, r5, r4
-	ldrb r0, [r0]
-	cmp r0, 0xFC
-	bne _080093A0
-	adds r0, r4, 0x1
-	lsls r0, 16
-	lsrs r4, r0, 16
-	adds r0, r5, r4
-	ldrb r0, [r0]
-	bl GetExtCtrlCodeLength
-	lsls r0, 24
-	lsrs r0, 24
-	adds r0, r4, r0
-	lsls r0, 16
-	lsrs r4, r0, 16
-	b _080093B8
-_080093A0:
-	adds r2, r6, 0
-	adds r0, r2, 0x1
-	lsls r0, 16
-	lsrs r6, r0, 16
-	adds r2, r5, r2
-	adds r1, r4, 0
-	adds r0, r1, 0x1
-	lsls r0, 16
-	lsrs r4, r0, 16
-	adds r1, r5, r1
-	ldrb r0, [r1]
-	strb r0, [r2]
-_080093B8:
-	adds r0, r5, r4
-	ldrb r0, [r0]
-	cmp r0, 0xFF
-	bne _0800937E
-_080093C0:
-	adds r1, r5, r6
-	movs r0, 0xFF
-	strb r0, [r1]
-	pop {r4-r6}
-	pop {r0}
-	bx r0
-	thumb_func_end StripExtCtrlCodes
-
-	.align 2, 0 @ Don't pad with nop.
diff --git a/data/string_util.s b/data/string_util.s
deleted file mode 100644
index 69c4c19bf..000000000
--- a/data/string_util.s
+++ /dev/null
@@ -1,70 +0,0 @@
-	.include "asm/macros.inc"
-	.include "constants/constants.inc"
-
-	.section .rodata
-
-sDigits:: @ 82EC7D4
-	.string "0123456789ABCDEF"
-
-	.align 2
-sPowersOfTen:: @ 82EC7E4
-	.4byte          1
-	.4byte         10
-	.4byte        100
-	.4byte       1000
-	.4byte      10000
-	.4byte     100000
-	.4byte    1000000
-	.4byte   10000000
-	.4byte  100000000
-	.4byte 1000000000
-
-sSetBrailleFont:: @ 82EC80C
-	.incbin "baserom.gba", 0x2ec80c, 0x4
-
-sGotoLine2:: @ 82EC810
-	.incbin "baserom.gba", 0x2ec810, 0x8
-
-	.align 2
-sExpandPlaceholderFuncs:: @ 82EC818
-	.4byte ExpandPlaceholder_UnknownStringVar
-	.4byte ExpandPlaceholder_PlayerName
-	.4byte ExpandPlaceholder_StringVar1
-	.4byte ExpandPlaceholder_StringVar2
-	.4byte ExpandPlaceholder_StringVar3
-	.4byte ExpandPlaceholder_KunChan
-	.4byte ExpandPlaceholder_RivalName
-	.4byte ExpandPlaceholder_Version
-	.4byte ExpandPlaceholder_Aqua
-	.4byte ExpandPlaceholder_Magma
-	.4byte ExpandPlaceholder_Archie
-	.4byte ExpandPlaceholder_Maxie
-	.4byte ExpandPlaceholder_Kyogre
-	.4byte ExpandPlaceholder_Groudon
-
-gExtendedControlCodeLengths:: @ 82EC850
-	.byte 1
-	.byte 2
-	.byte 2
-	.byte 2
-	.byte 4
-	.byte 2
-	.byte 2
-	.byte 1
-	.byte 2
-	.byte 1
-	.byte 1
-	.byte 3
-	.byte 2
-	.byte 2
-	.byte 2
-	.byte 1
-	.byte 3
-	.byte 2
-	.byte 2
-	.byte 2
-	.byte 2
-	.byte 1
-	.byte 1
-	.byte 1
-	.byte 1
diff --git a/include/global.h b/include/global.h
index 1f7df2ca6..b7038025b 100644
--- a/include/global.h
+++ b/include/global.h
@@ -3,6 +3,22 @@
 
 #include "gba/gba.h"
 
+// Prevent cross-jump optimization.
+#define BLOCK_CROSS_JUMP asm("");
+
+// to help in decompiling
+#define asm_comment(x) asm volatile("@ -- " x " -- ")
+
+#define ARRAY_COUNT(array) (sizeof(array) / sizeof((array)[0]))
+
+#define POKEMON_NAME_LENGTH 10
+#define OT_NAME_LENGTH 7
+
+extern u8 gStringVar1[];
+extern u8 gStringVar2[];
+extern u8 gStringVar3[];
+extern u8 gStringVar4[];
+
 enum
 {
     VERSION_SAPPHIRE = 1,
diff --git a/include/string_util.h b/include/string_util.h
index c20965380..382f52251 100644
--- a/include/string_util.h
+++ b/include/string_util.h
@@ -10,7 +10,7 @@ enum StringConvertMode
 
 u8 *StringCopy10(u8 *dest, const u8 *src);
 u8 *StringGetEnd10(u8 *str);
-u8 *StringCopy8(u8 *dest, const u8 *src);
+u8 *StringCopy7(u8 *dest, const u8 *src);
 u8 *StringCopy(u8 *dest, const u8 *src);
 u8 *StringAppend(u8 *dest, const u8 *src);
 u8 *StringCopyN(u8 *dest, const u8 *src, u8 n);
@@ -18,15 +18,25 @@ u8 *StringAppendN(u8 *dest, const u8 *src, u8 n);
 u16 StringLength(const u8 *str);
 s32 StringCompare(const u8 *str1, const u8 *str2);
 s32 StringCompareN(const u8 *str1, const u8 *str2, u32 n);
+bool8 IsStringLengthAtLeast(const u8 *str, s32 n);
 u8 *ConvertIntToDecimalStringN(u8 *dest, s32 value, enum StringConvertMode mode, u8 n);
-u8 *ConvertIntToDecimalStringN_DigitWidth6(u8 *dest, s32 value, enum StringConvertMode mode, u8 n);
+u8 *ConvertUIntToDecimalStringN(u8 *dest, u32 value, enum StringConvertMode mode, u8 n);
 u8 *ConvertIntToHexStringN(u8 *dest, s32 value, enum StringConvertMode mode, u8 n);
-u8 *ConvertIntToDecimalString(u8 *dest, s32 value);
 u8 *StringExpandPlaceholders(u8 *dest, const u8 *src);
 u8 *StringBraille(u8 *dest, const u8 *src);
 u8 *GetExpandedPlaceholder(u32 id);
 u8 *StringFill(u8 *dest, u8 c, u16 n);
 u8 *StringCopyPadded(u8 *dest, const u8 *src, u8 c, u16 n);
 u8 *StringFillWithTerminator(u8 *dest, u16 n);
+u8 *StringCopyN_Multibyte(u8 *dest, u8 *src, u32 n);
+u32 StringLength_Multibyte(u8 *str);
+u8 *WriteColorChangeControlCode(u8 *dest, u32 colorType, u8 color);
+bool32 sub_8009228(u8 *str);
+bool32 sub_800924C(u8 *str, s32 n);
+u8 GetExtCtrlCodeLength(u8 code);
+static const u8 *SkipExtCtrlCode(const u8 *s);
+s32 StringCompareWithoutExtCtrlCodes(const u8 *str1, const u8 *str2);
+void ConvertInternationalString(u8 *s, u8 language);
+void StripExtCtrlCodes(u8 *str);
 
 #endif // GUARD_STRING_UTIL_H
diff --git a/ld_script.txt b/ld_script.txt
index 83c6bc49d..6fb2fbeb8 100644
--- a/ld_script.txt
+++ b/ld_script.txt
@@ -49,7 +49,6 @@ SECTIONS {
         asm/text.o(.text);
         asm/sprite.o(.text);
         src/string_util.o(.text);
-        asm/string_util.o(.text);
         asm/link.o(.text);
         asm/rtc.o(.text);
         asm/main_menu.o(.text);
@@ -245,7 +244,7 @@ SECTIONS {
         data/window.o(.rodata);
         data/text.o(.rodata);
         data/sprite.o(.rodata);
-        data/string_util.o(.rodata);
+        src/string_util.o(.rodata);
         data/data2.o(.rodata);
         data/diploma.o(.rodata);
         data/strings.o(.rodata);
diff --git a/src/string_util.c b/src/string_util.c
index 35b48ceae..a00f261bd 100644
--- a/src/string_util.c
+++ b/src/string_util.c
@@ -2,22 +2,23 @@
 #include "string_util.h"
 #include "text.h"
 
-#define MAX_PLACEHOLDER_ID 0xD
+EWRAM_DATA u8 gUnknownStringVar[16] = {0};
 
-extern const u8 sDigits[];
-extern const s32 sPowersOfTen[];
-extern const u8 sSetBrailleFont[];
-extern const u8 sGotoLine2[];
+static const u8 sDigits[] = @"0123456789ABCDEF";
 
-typedef u8 *(*ExpandPlaceholderFunc)(void);
-
-extern const ExpandPlaceholderFunc sExpandPlaceholderFuncs[];
-
-extern u8 gUnknownStringVar[];
-extern u8 gStringVar1[];
-extern u8 gStringVar2[];
-extern u8 gStringVar3[];
-extern u8 gStringVar4[];
+static const s32 sPowersOfTen[] =
+{
+             1,
+            10,
+           100,
+          1000,
+         10000,
+        100000,
+       1000000,
+      10000000,
+     100000000,
+    1000000000,
+};
 
 extern u8 gExpandedPlaceholder_Empty[];
 extern u8 gExpandedPlaceholder_Kun[];
@@ -393,11 +394,8 @@ u8 *StringExpandPlaceholders(u8 *dest, const u8 *src)
 
 u8 *StringBraille(u8 *dest, const u8 *src)
 {
-    u8 setBrailleFont[4];
-    u8 gotoLine2[5];
-
-    memcpy(setBrailleFont, sSetBrailleFont, 4);
-    memcpy(gotoLine2, sGotoLine2, 5);
+    u8 setBrailleFont[] = { 0xFC, 0x06, 0x06, 0xFF };
+    u8 gotoLine2[5] = { 0xFE, 0xFC, 0x0E, 0x02, 0xFF };
 
     dest = StringCopy(dest, setBrailleFont);
 
@@ -421,32 +419,32 @@ u8 *StringBraille(u8 *dest, const u8 *src)
     }
 }
 
-u8 *ExpandPlaceholder_UnknownStringVar(void)
+static u8 *ExpandPlaceholder_UnknownStringVar(void)
 {
     return gUnknownStringVar;
 }
 
-u8 *ExpandPlaceholder_PlayerName(void)
+static u8 *ExpandPlaceholder_PlayerName(void)
 {
     return gSaveBlock2Ptr->playerName;
 }
 
-u8 *ExpandPlaceholder_StringVar1(void)
+static u8 *ExpandPlaceholder_StringVar1(void)
 {
     return gStringVar1;
 }
 
-u8 *ExpandPlaceholder_StringVar2(void)
+static u8 *ExpandPlaceholder_StringVar2(void)
 {
     return gStringVar2;
 }
 
-u8 *ExpandPlaceholder_StringVar3(void)
+static u8 *ExpandPlaceholder_StringVar3(void)
 {
     return gStringVar3;
 }
 
-u8 *ExpandPlaceholder_KunChan(void)
+static u8 *ExpandPlaceholder_KunChan(void)
 {
     if (gSaveBlock2Ptr->playerGender == MALE)
         return gExpandedPlaceholder_Kun;
@@ -454,7 +452,7 @@ u8 *ExpandPlaceholder_KunChan(void)
         return gExpandedPlaceholder_Chan;
 }
 
-u8 *ExpandPlaceholder_RivalName(void)
+static u8 *ExpandPlaceholder_RivalName(void)
 {
     if (gSaveBlock2Ptr->playerGender == MALE)
         return gExpandedPlaceholder_May;
@@ -462,47 +460,67 @@ u8 *ExpandPlaceholder_RivalName(void)
         return gExpandedPlaceholder_Brendan;
 }
 
-u8 *ExpandPlaceholder_Version(void)
+static u8 *ExpandPlaceholder_Version(void)
 {
     return gExpandedPlaceholder_Emerald;
 }
 
-u8 *ExpandPlaceholder_Aqua(void)
+static u8 *ExpandPlaceholder_Aqua(void)
 {
     return gExpandedPlaceholder_Aqua;
 }
 
-u8 *ExpandPlaceholder_Magma(void)
+static u8 *ExpandPlaceholder_Magma(void)
 {
     return gExpandedPlaceholder_Magma;
 }
 
-u8 *ExpandPlaceholder_Archie(void)
+static u8 *ExpandPlaceholder_Archie(void)
 {
     return gExpandedPlaceholder_Archie;
 }
 
-u8 *ExpandPlaceholder_Maxie(void)
+static u8 *ExpandPlaceholder_Maxie(void)
 {
     return gExpandedPlaceholder_Maxie;
 }
 
-u8 *ExpandPlaceholder_Kyogre(void)
+static u8 *ExpandPlaceholder_Kyogre(void)
 {
     return gExpandedPlaceholder_Kyogre;
 }
 
-u8 *ExpandPlaceholder_Groudon(void)
+static u8 *ExpandPlaceholder_Groudon(void)
 {
     return gExpandedPlaceholder_Groudon;
 }
 
 u8 *GetExpandedPlaceholder(u32 id)
 {
-    if (id > MAX_PLACEHOLDER_ID)
+    typedef u8 *(*ExpandPlaceholderFunc)(void);
+
+    static const ExpandPlaceholderFunc funcs[] =
+    {
+        ExpandPlaceholder_UnknownStringVar,
+        ExpandPlaceholder_PlayerName,
+        ExpandPlaceholder_StringVar1,
+        ExpandPlaceholder_StringVar2,
+        ExpandPlaceholder_StringVar3,
+        ExpandPlaceholder_KunChan,
+        ExpandPlaceholder_RivalName,
+        ExpandPlaceholder_Version,
+        ExpandPlaceholder_Aqua,
+        ExpandPlaceholder_Magma,
+        ExpandPlaceholder_Archie,
+        ExpandPlaceholder_Maxie,
+        ExpandPlaceholder_Kyogre,
+        ExpandPlaceholder_Groudon,
+    };
+
+    if (id >= ARRAY_COUNT(funcs))
         return gExpandedPlaceholder_Empty;
     else
-        return sExpandPlaceholderFuncs[id]();
+        return funcs[id]();
 }
 
 u8 *StringFill(u8 *dest, u8 c, u16 n)
@@ -542,3 +560,241 @@ u8 *StringFillWithTerminator(u8 *dest, u16 n)
 {
     return StringFill(dest, EOS, n);
 }
+
+__attribute__((naked))
+u8 *StringCopyN_Multibyte(u8 *dest, u8 *src, u32 n)
+{
+    asm(".syntax unified\n\
+    push {r4,r5,lr}\n\
+    adds r4, r0, 0\n\
+    adds r3, r1, 0\n\
+    subs r2, 0x1\n\
+    movs r5, 0x1\n\
+    negs r5, r5\n\
+    b _080091B2\n\
+_0800919A:\n\
+    strb r0, [r4]\n\
+    adds r3, 0x1\n\
+    adds r4, 0x1\n\
+    subs r0, r3, 0x1\n\
+    ldrb r0, [r0]\n\
+    cmp r0, 0xF9\n\
+    bne _080091B0\n\
+    ldrb r0, [r3]\n\
+    strb r0, [r4]\n\
+    adds r3, 0x1\n\
+    adds r4, 0x1\n\
+_080091B0:\n\
+    subs r2, 0x1\n\
+_080091B2:\n\
+    cmp r2, r5\n\
+    beq _080091BE\n\
+    ldrb r0, [r3]\n\
+    adds r1, r0, 0\n\
+    cmp r1, 0xFF\n\
+    bne _0800919A\n\
+_080091BE:\n\
+    movs r0, 0xFF\n\
+    strb r0, [r4]\n\
+    adds r0, r4, 0\n\
+    pop {r4,r5}\n\
+    pop {r1}\n\
+    bx r1\n\
+    .syntax divided");
+}
+
+u32 StringLength_Multibyte(u8 *str)
+{
+    u32 length = 0;
+
+    while (*str != EOS)
+    {
+        if (*str == 0xF9)
+            str++;
+        str++;
+        length++;
+    }
+
+    return length;
+}
+
+u8 *WriteColorChangeControlCode(u8 *dest, u32 colorType, u8 color)
+{
+    *dest = 0xFC;
+    dest++;
+
+    switch (colorType)
+    {
+    case 0:
+        *dest = 1;
+        dest++;
+        break;
+    case 1:
+        *dest = 3;
+        dest++;
+        break;
+    case 2:
+        *dest = 2;
+        dest++;
+        break;
+    }
+
+    *dest = color;
+    dest++;
+    *dest = EOS;
+    return dest;
+}
+
+bool32 sub_8009228(u8 *str)
+{
+    while (*str != EOS)
+    {
+        if (*str <= 0xA0)
+            if (*str != 0)
+                return TRUE;
+        str++;
+    }
+
+    return FALSE;
+}
+
+bool32 sub_800924C(u8 *str, s32 n)
+{
+    s32 i;
+
+    for (i = 0; *str != EOS && i < n; i++)
+    {
+        if (*str <= 0xA0)
+            if (*str != 0)
+                return TRUE;
+        str++;
+    }
+
+    return FALSE;
+}
+
+u8 GetExtCtrlCodeLength(u8 code)
+{
+    static const u8 lengths[] =
+    {
+        1,
+        2,
+        2,
+        2,
+        4,
+        2,
+        2,
+        1,
+        2,
+        1,
+        1,
+        3,
+        2,
+        2,
+        2,
+        1,
+        3,
+        2,
+        2,
+        2,
+        2,
+        1,
+        1,
+        1,
+        1,
+    };
+
+    u8 length = 0;
+    if (code < ARRAY_COUNT(lengths))
+        length = lengths[code];
+    return length;
+}
+
+static const u8 *SkipExtCtrlCode(const u8 *s)
+{
+    while (*s == 0xFC)
+    {
+        s++;
+        s += GetExtCtrlCodeLength(*s);
+    }
+
+    return s;
+}
+
+s32 StringCompareWithoutExtCtrlCodes(const u8 *str1, const u8 *str2)
+{
+    s32 retVal = 0;
+
+    while (1)
+    {
+        str1 = SkipExtCtrlCode(str1);
+        str2 = SkipExtCtrlCode(str2);
+
+        if (*str1 > *str2)
+            break;
+
+        if (*str1 < *str2)
+        {
+            retVal = -1;
+            if (*str2 == 0xFF)
+                retVal = 1;
+        }
+
+        if (*str1 == 0xFF)
+            return retVal;
+
+        str1++;
+        str2++;
+    }
+
+    retVal = 1;
+
+    if (*str1 == 0xFF)
+        retVal = -1;
+
+    return retVal;
+}
+
+void ConvertInternationalString(u8 *s, u8 language)
+{
+    if (language == LANGUAGE_JAPANESE)
+    {
+        u8 i;
+
+        StripExtCtrlCodes(s);
+        i = StringLength(s);
+        s[i++] = 0xFC;
+        s[i++] = 22;
+        s[i++] = 0xFF;
+
+        i--;
+
+        while (i != (u8)-1)
+        {
+            s[i + 2] = s[i];
+            i--;
+        }
+
+        s[0] = 0xFC;
+        s[1] = 21;
+    }
+}
+
+void StripExtCtrlCodes(u8 *str)
+{
+    u16 srcIndex = 0;
+    u16 destIndex = 0;
+    while (str[srcIndex] != 0xFF)
+    {
+        if (str[srcIndex] == 0xFC)
+        {
+            srcIndex++;
+            srcIndex += GetExtCtrlCodeLength(str[srcIndex]);
+        }
+        else
+        {
+            str[destIndex++] = str[srcIndex++];
+        }
+    }
+    str[destIndex] = 0xFF;
+}
diff --git a/sym_ewram.txt b/sym_ewram.txt
index fe852547a..763b7df8f 100644
--- a/sym_ewram.txt
+++ b/sym_ewram.txt
@@ -76,8 +76,7 @@ gStringVar3: @ 2021EC4
 gStringVar4: @ 2021FC4
 	.space 0x3E8
 
-gUnknownStringVar: @ 20223AC
-	.space 0x10
+	.include "src/string_util.o"
 
 gUnknown_020223BC: @ 20223BC
 	.space 0x1