Create parser that resolves the .incbin macros

2 files changed, 160 insertions, 111 deletions

diff --git a/tools/script_disassembler/incbin_parser.py b/tools/script_disassembler/incbin_parser.py
new file mode 100644
index 00000000..f20543a9
--- /dev/null
+++ b/tools/script_disassembler/incbin_parser.py
@@ -0,0 +1,47 @@
+# This python script reads the script.s file which contains all the .incbin macros
+# Then it fetches the corresponding data of the baserom, o
+TMC_FOLDER='../..'
+
+import subprocess
+from script_disassembler import disassemble_script
+
+def main():
+    # read baserom data
+    with open(f'{TMC_FOLDER}/baserom.gba', 'rb') as baserom:
+        baserom_data = bytearray(baserom.read())
+
+    # read scripts.s with incbins
+    with open(f'{TMC_FOLDER}/data/scripts.s', 'r') as scripts:
+
+        while True:
+            line = scripts.readline()
+
+            if not line:
+                # end of file
+                break
+
+            if ':: @' in line: # this might be a label
+                incbin_line = scripts.readline()
+
+                if '.incbin' in incbin_line: # found a label with incbin
+                    label = line.split('::')[0]
+                    (_, start, end) = incbin_line.split(',')
+                    start = int(start, 16)
+                    end = int(end, 16)
+                    
+                    # read data from rom
+                    data = baserom_data[start:start+end]
+
+                    print(f'DISASM {label}')
+                    disassemble_script(data)
+
+                    # print new include label
+                    print(f'.include "data/scripts/{label}.inc"')
+                else:
+                    print(line, end='')
+                    print(incbin_line, end='')
+            else:
+                print(line, end='')
+
+if __name__ == '__main__':
+    main()
\ No newline at end of file
diff --git a/tools/script_disassembler/script_disassembler.py b/tools/script_disassembler/script_disassembler.py
index 5066ce01..df2b54dc 100644
--- a/tools/script_disassembler/script_disassembler.py
+++ b/tools/script_disassembler/script_disassembler.py
@@ -26,26 +26,6 @@ def barray_to_u32_hex(barray):
     integers = struct.unpack('I'*count, barray)
     return [u32_to_hex(x) for x in integers]
 
-"""
-def GetNextScriptCommandWordAfterCommandMetadata(ctx):
-    return struct.unpack('I', ctx.data[ctx.ptr+2:ctx.ptr+6])[0]
-
-
-def sub_0807E10C(ctx):
-    meta = GetNextScriptCommandWordAfterCommandMetadata(ctx)
-    print('function', hex(meta))
-
-
-def sub_0807E5F8(ctx):
-    curPtr = start + ctx.ptr
-    print('current', hex(curPtr))
-    # gUnk_02033280.unk_00 |= GetNextScriptCommandWordAfterCommandMetadata(unk2->unk_00.raw);
-    meta = GetNextScriptCommandWordAfterCommandMetadata(ctx)
-    print('meta', meta)
-    newPtr = curPtr | meta
-    print('Start next script at: ', hex(newPtr))
-"""
-
 commands = [
     {'fun': 'nullsub_507', 'params': ''},
     {'fun': 'sub_0807E004', 'params': '', 'name': 'start executing scripts'},
@@ -65,127 +45,127 @@ commands = [
     {'fun': 'sub_0807E188', 'params': 's'},
     {'fun': 'sub_0807E1D8', 'params': 's'},
     {'fun': 'sub_0807E4AC', 'params': 's'},
-    {'fun': 'sub_0807E200'},
-    {'fun': 'sub_0807E220'},
-    {'fun': 'sub_0807E240'},
+    {'fun': 'sub_0807E200', 'params': 'w'},
+    {'fun': 'sub_0807E220', 'params': 'ss'},
+    {'fun': 'sub_0807E240', 'params': 's'},
     {'fun': 'sub_0807E260', 'params': 's'},
-    {'fun': 'sub_0807E280'},
+    {'fun': 'sub_0807E280', 'params': 'sss'},
     {'fun': 'sub_0807E2A8', 'params': 's'},
     {'fun': 'sub_0807E2E4', 'params': ''},
     {'fun': 'sub_0807E30C', 'params': ''},
-    {'fun': 'sub_0807E390'},
-    {'fun': 'sub_0807E3BC'},
-    {'fun': 'sub_0807E3E8'},
+    {'fun': 'sub_0807E390', 'params': 's'},
+    {'fun': 'sub_0807E3BC', 'params': 's'},
+    {'fun': 'sub_0807E3E8', 'params': ''},
     {'fun': 'sub_0807E40C', 'params': 's'},
-    {'fun': 'sub_0807E42C'},
-    {'fun': 'sub_0807E48C'},
-    {'fun': 'sub_0807E4CC'},
-    {'fun': 'sub_0807E4EC'},
-    {'fun': 'sub_0807E514'},
-    {'fun': 'sub_0807E538'},
-    {'fun': 'sub_0807E564'},
-    {'fun': 'sub_0807E584'},
-    {'fun': 'sub_0807E5A0'},
-    {'fun': 'sub_0807E5CC'},
+    {'fun': 'sub_0807E42C', 'params': 'ss'},
+    {'fun': 'sub_0807E48C', 'params': 's'},
+    {'fun': 'sub_0807E4CC', 'params': 'w'},
+    {'fun': 'sub_0807E4EC', 'params': 'w'},
+    {'fun': 'sub_0807E514', 'params': 's'},
+    {'fun': 'sub_0807E538', 'params':'w'},
+    {'fun': 'sub_0807E564', 'params': ''},
+    {'fun': 'sub_0807E584', 'params': ''},
+    {'fun': 'sub_0807E5A0', 'params': 's'},
+    {'fun': 'sub_0807E5CC', 'params': 's'},
     {'fun': 'sub_0807E5F8', 'params': 'w'},# 'exec': sub_0807E5F8},
-    {'fun': 'sub_0807E610'},
-    {'fun': 'sub_0807E628'},
-    {'fun': 'sub_0807E634'},
-    {'fun': 'sub_0807E644'},
+    {'fun': 'sub_0807E610', 'params': 'w'},
+    {'fun': 'sub_0807E628', 'params': 's'},
+    {'fun': 'sub_0807E634', 'params': 'ss'},
+    {'fun': 'sub_0807E644', 'params': 's'},
     {'fun': 'sub_0807E650', 'params': 's'},
-    {'fun': 'sub_0807E65C'},
-    {'fun': 'sub_0807E668'},
-    {'fun': 'sub_0807E674'},
+    {'fun': 'sub_0807E65C', 'params': 's'},
+    {'fun': 'sub_0807E668', 'params': 's'},
+    {'fun': 'sub_0807E674', 'params': 's'},
     {'fun': 'sub_0807E680', 'params': 's'},
-    {'fun': 'sub_0807E690'},
+    {'fun': 'sub_0807E690', 'params': 'w'},
     {'fun': 'sub_0807E6AC', 'params': 'w'},
     {'fun': 'sub_0807E6DC', 'params': ''},
-    {'fun': 'sub_0807E700'},
-    {'fun': 'sub_0807E72C'},
-    {'fun': 'sub_0807E75C'},
-    {'fun': 'sub_0807E778'},
-    {'fun': 'sub_0807E788'},
-    {'fun': 'sub_0807E79C'},
-    {'fun': 'nullsub_508'},
-    {'fun': 'sub_0807E7B0'},
-    {'fun': 'sub_0807E7C4'},
-    {'fun': 'sub_0807E7D8'},
-    {'fun': 'sub_0807E7EC'},
-    {'fun': 'sub_0807E800'},
-    {'fun': 'sub_0807E80C'},
-    {'fun': 'sub_0807E858'},
+    {'fun': 'sub_0807E700', 'params': ''},
+    {'fun': 'sub_0807E72C', 'params': ''},
+    {'fun': 'sub_0807E75C', 'params': ''},
+    {'fun': 'sub_0807E778', 'params': ''},
+    {'fun': 'sub_0807E788', 'params': 'w'},
+    {'fun': 'sub_0807E79C', 'params': ''},
+    {'fun': 'nullsub_508', 'params': 's'},
+    {'fun': 'sub_0807E7B0', 'params': ''},
+    {'fun': 'sub_0807E7C4', 'params': ''},
+    {'fun': 'sub_0807E7D8', 'params': ''},
+    {'fun': 'sub_0807E7EC', 'params': ''},
+    {'fun': 'sub_0807E800', 'params': 'w'},
+    {'fun': 'sub_0807E80C', 'params': 'w'},
+    {'fun': 'sub_0807E858', 'params': 's'},
     {'fun': 'sub_0807E864', 'params': ''},
     {'fun': 'sub_0807E878', 'params': ''},
-    {'fun': 'sub_0807E888'},
-    {'fun': 'sub_0807E898'},
+    {'fun': 'sub_0807E888', 'params': ''},
+    {'fun': 'sub_0807E898', 'params': 'w'},
     {'fun': 'sub_0807E8C4', 'params': 'w'},
-    {'fun': 'sub_0807E8D4'},
+    {'fun': 'sub_0807E8D4', 'params': 's'},
     {'fun': 'sub_0807E8E4_0', 'params': ''}, # duplicate
     {'fun': 'sub_0807E8E4_1', 'params': ''}, # duplicate
     {'fun': 'sub_0807E8E4_2', 'params': ''}, # duplicate
     {'fun': 'sub_0807E8E4_3', 'params': ''}, # duplicate
-    {'fun': 'sub_0807E908'},
-    {'fun': 'sub_0807E914'},
-    {'fun': 'sub_0807E924'},
+    {'fun': 'sub_0807E908', 'params': 's'},
+    {'fun': 'sub_0807E914', 'params': 'w'},
+    {'fun': 'sub_0807E924', 'params': ''},
     {'fun': 'sub_0807E930', 'params': 's'},
     {'fun': 'sub_0807E944', 'params': ''},
-    {'fun': 'sub_0807E974'},
+    {'fun': 'sub_0807E974', 'params': 's'},
     {'fun': 'sub_0807E9D4', 'params': ''},
     {'fun': 'sub_0807E9DC', 'params': ''},
-    {'fun': 'sub_0807E9E4'},
+    {'fun': 'sub_0807E9E4', 'params': 's'},
     {'fun': 'sub_0807E9F0', 'params': ''},
     {'fun': 'sub_0807EA4C', 'params': ''},
-    {'fun': 'sub_0807EA88'},
+    {'fun': 'sub_0807EA88', 'params': 's'},
     {'fun': 'sub_0807EA94', 'params': ''},
     {'fun': 'sub_0807EAB4', 'params': 's'},
-    {'fun': 'sub_0807EAC0'},
+    {'fun': 'sub_0807EAC0', 'params': 's'},
     {'fun': 'sub_0807EAD0', 'params': 'ss'},
-    {'fun': 'sub_0807EAF0'},
-    {'fun': 'sub_0807EB18'},
+    {'fun': 'sub_0807EAF0', 'params': 'w'},
+    {'fun': 'sub_0807EB18', 'params': ''},
     {'fun': 'sub_0807EB28', 'params': 's'},
-    {'fun': 'sub_0807EB38'},
+    {'fun': 'sub_0807EB38', 'params': ''},
     {'fun': 'sub_0807EB44', 'params': 's'},
-    {'fun': 'sub_0807EB4C'},
+    {'fun': 'sub_0807EB4C', 'params': 'ss'},
     {'fun': 'sub_0807EB74', 'params': ''},
     {'fun': 'sub_0807EB8C', 'params': ''},
-    {'fun': 'sub_0807EBA8'},
-    {'fun': 'sub_0807EBB0'},
+    {'fun': 'sub_0807EBA8', 'params': 's'},
+    {'fun': 'sub_0807EBB0', 'params': 's'},
     {'fun': 'sub_0807EBC0', 'params': 's'},
-    {'fun': 'sub_0807EBC8'},
-    {'fun': 'sub_0807EBD8'},
-    {'fun': 'sub_0807EBF4'},
-    {'fun': 'sub_0807EC08'},
-    {'fun': 'nullsub_509'},
+    {'fun': 'sub_0807EBC8', 'params': 'w'},
+    {'fun': 'sub_0807EBD8', 'params': 'w'},
+    {'fun': 'sub_0807EBF4', 'params': 'ss'},
+    {'fun': 'sub_0807EC08', 'params': ''},
+    {'fun': 'nullsub_509', 'params': ''},
     {'fun': 'sub_0807EC1C', 'params': 's'},
-    {'fun': 'sub_0807EC64'},
-    {'fun': 'sub_0807EC94'},
-    {'fun': 'sub_0807ECC4'},
+    {'fun': 'sub_0807EC64', 'params': 's'},
+    {'fun': 'sub_0807EC94', 'params': 's'},
+    {'fun': 'sub_0807ECC4', 'params': 's'},
     {'fun': 'sub_0807ECF4', 'params': 's'},
-    {'fun': 'sub_0807ED24'},
+    {'fun': 'sub_0807ED24', 'params': 'sss'},
     {'fun': 'sub_0807EDD4', 'params': 'ss'},
-    {'fun': 'sub_0807EE04'},
+    {'fun': 'sub_0807EE04', 'params': 'ss'},
     {'fun': 'sub_0807EE30', 'params': ''},
-    {'fun': 'sub_0807EEB4'},
-    {'fun': 'sub_0807EEF4'},
-    {'fun': 'sub_0807EF3C'},
+    {'fun': 'sub_0807EEB4', 'params': ''},
+    {'fun': 'sub_0807EEF4', 'params': ''},
+    {'fun': 'sub_0807EF3C', 'params': 'ss'},
     {'fun': 'sub_0807EF80', 'params': 's'},
-    {'fun': 'sub_0807EF90'},
-    {'fun': 'sub_0807EFA0'},
+    {'fun': 'sub_0807EF90', 'params': 's'},
+    {'fun': 'sub_0807EFA0', 'params': 's'},
     {'fun': 'sub_0807EFAC', 'params': 's'},
-    {'fun': 'sub_0807EFD4'},
+    {'fun': 'sub_0807EFD4', 'params': 'w'},
     {'fun': 'sub_0807EFE4', 'params': ''},
-    {'fun': 'sub_0807EFF4'},
-    {'fun': 'sub_0807F004'},
-    {'fun': 'sub_0807F010'},
+    {'fun': 'sub_0807EFF4', 'params': 's'},
+    {'fun': 'sub_0807F004', 'params': 's'},
+    {'fun': 'sub_0807F010', 'params': 's'},
     {'fun': 'sub_0807F034', 'params': 's'},
-    {'fun': 'sub_0807F050'},
-    {'fun': 'sub_0807F060'},
-    {'fun': 'sub_0807F078'},
-    {'fun': 'sub_0807F088'},
+    {'fun': 'sub_0807F050', 'params': 's'},
+    {'fun': 'sub_0807F060', 'params': 's'},
+    {'fun': 'sub_0807F078', 'params': 'ss'},
+    {'fun': 'sub_0807F088', 'params': 's'},
     {'fun': 'sub_0807F098', 'params': ''},
-    {'fun': 'sub_0807F0A4'},
+    {'fun': 'sub_0807F0A4', 'params': ''},
     {'fun': 'sub_0807F0B4', 'params': 's'},
-    {'fun': 'sub_0807F0C8'}
+    {'fun': 'sub_0807F0C8', 'params': 'ss'}
 ]
 
 # definitions for parameters
@@ -220,6 +200,12 @@ parameters = {
         'expr': '	.word \\a\n	.word \\b',
         'read': lambda ctx: ', '.join(barray_to_u32_hex(ctx.data[ctx.ptr+2:ctx.ptr+10]))
     },
+    'sss': {
+        'length': 3,
+        'param': 'a,b,c',
+        'expr': '	.short \\a\n	.short \\b\n	.short \\c',
+        'read': lambda ctx: ', '.join(barray_to_u16_hex(ctx.data[ctx.ptr+2:ctx.ptr+8]))
+    },
 }
 
 
@@ -229,14 +215,20 @@ def ExecuteScriptCommandSet(ctx: Context):
         return 0
 
     unk_06 = cmd >> 0xA
+    if unk_06 == 0:
+        # TODO error
+        return 0
     operationId = cmd & 0x3FF
-    #print(cmd)
-    #print('advance by', unk_06)
-    #print('command id', operationId)
+    if operationId >= len(commands):
+        print('ERR: NO CMD')
+        # TODO error
+        return 0
     command = commands[operationId]
-    #print(command)
     param_length = unk_06 - 1 
     if unk_06 > 1:
+        if ctx.ptr+2*unk_06 >= len(ctx.data):
+            # TODO raise Exception(f'Not enough data to fetch {unk_06-1} params')
+            return 0
         meta = struct.unpack(
             'H'*(unk_06-1), ctx.data[ctx.ptr+2:ctx.ptr+2*unk_06])
         #print('meta', meta)
@@ -248,7 +240,11 @@ def ExecuteScriptCommandSet(ctx: Context):
         raise Exception('Parameter configuration ' + command['params'] + ' not defined')
     params = parameters[command['params']]
     if unk_06-1 != params['length']:
-        raise Exception(f'Call {command["fun"]} with ' + str(unk_06-1) +' length, while length of ' + str(params['length'])+' defined')
+        # TODOraise Exception(f'Call {command["fun"]} with ' + str(unk_06-1) +' length, while length of ' + str(params['length'])+' defined')
+        # TEMPORARY WORKAROUND:
+        print(ctx.ptr, unk_06, cmd)
+        ctx.ptr += unk_06*2
+        return 1
 
 
     print(command['fun'] + ' ' + params['read'](ctx))
@@ -258,19 +254,24 @@ def ExecuteScriptCommandSet(ctx: Context):
     return 1
 
 
-def disassemble_script(input_data):
-    input_bytes = bytearray.fromhex(input_data)
+def disassemble_script(input_bytes):
 
     ctx = Context(0, input_bytes)
 
     while True:
+        if ctx.ptr >= len(ctx.data) - 1: # End of file (there need to be at least two bytes remaining for the next operation id)
+            break
         if not ExecuteScriptCommandSet(ctx):
             break
 
 
     # Print rest (did not manage to get there)
-
-    print('\n'.join(['.short ' + x for x in barray_to_u16_hex(ctx.data[ctx.ptr:])]))
+    if ctx.ptr < len(ctx.data):
+        if (len(ctx.data) - ctx.ptr) % 2 != 0:
+            print(ctx.data[ctx.ptr:])
+            # TODO error
+            return
+        print('\n'.join(['.short ' + x for x in barray_to_u16_hex(ctx.data[ctx.ptr:])]))
     
 
 def generate_macros():
@@ -316,6 +317,7 @@ def main():
     if input_data.strip() == 'macros':
         generate_macros()
         return
-    disassemble_script(input_data)
+    disassemble_script(bytearray.fromhex(input_data))
     
-main()
\ No newline at end of file
+if __name__ == '__main__':
+    main()
\ No newline at end of file