Split scripts at more useful places

2 files changed, 204 insertions, 79 deletions

diff --git a/tools/script_disassembler/incbin_parser.py b/tools/script_disassembler/incbin_parser.py
index e0850f13..5d184b28 100644
--- a/tools/script_disassembler/incbin_parser.py
+++ b/tools/script_disassembler/incbin_parser.py
@@ -4,49 +4,73 @@ TMC_FOLDER='../..'
 
 import subprocess
 import sys
-from script_disassembler import disassemble_script
+from script_disassembler import disassemble_script, generate_macros
 
-def main():
+ROM_OFFSET=0x08000000
+SCRIPTS_START=0x08008B5C
+SCRIPTS_END=0x08016984
+# cat data/scripts.s  | grep @ | cut -d " " -f 3 | sed  -e 's/^/0x/' | sed -e 's/\\n/, /g' > labels.txt
+
+def read_baserom():
     # 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]
-
-                    stdout = sys.stdout
-                    with open(f'{TMC_FOLDER}/data/scripts/{label}.inc','w') as out:
-                        sys.stdout = out
-                        print(f'SCRIPT_START {label}')
-                        disassemble_script(data)
-                    sys.stdout = stdout
-
-                    # print new include label
-                    print(f'.include "data/scripts/{label}.inc"')
-                else:
-                    print(line, end='')
-                    print(incbin_line, end='')
-            else:
-                print(line, end='')
+        return bytearray(baserom.read())
+
+def get_label(addr):
+    return hex(addr).upper().replace('0X', 'gUnk_0')
+
+def main():
+    baserom_data = read_baserom()
+
+    script_start = SCRIPTS_START-ROM_OFFSET
+
+    scripts = '''	.include "asm/macros.inc"
+	.include "constants/constants.inc"
+
+	.include "asm/macros/scripts.inc"
+
+	.syntax unified
+
+	.text
+    
+    '''
+
+    while script_start < SCRIPTS_END-ROM_OFFSET:
+        label = get_label(script_start+ROM_OFFSET)
+        print(f"Disassembling \033[1;34m{label}\033[0m ({script_start} / { SCRIPTS_END-ROM_OFFSET} bytes converted)...")
+        # find end of the script signified by 0xffff0000
+        script_end = baserom_data.index(b'\xff\xff\x00\x00', script_start) + 4
+
+        if script_end > SCRIPTS_END-ROM_OFFSET:
+            script_end = SCRIPTS_END-ROM_OFFSET
+
+        script_start+ROM_OFFSET
+        # read data from rom
+        data = baserom_data[script_start:script_end]
+
+
+        scripts += f'	.include "data/scripts/{label}.inc"\n'
+        stdout = sys.stdout
+        with open(f'{TMC_FOLDER}/data/scripts/{label}.inc','w') as out:
+            sys.stdout = out
+            print(f'SCRIPT_START {label}')
+            res = disassemble_script(data)
+            if res != 0:
+                # Script ended in the middle, need to create a new file
+                script_end = script_start + res
+        sys.stdout = stdout
+
+        script_start = script_end 
+    print('Writing scripts.s file...')
+    with open(f'{TMC_FOLDER}/data/scripts.s', 'w') as out:
+        out.write(scripts)
+    print('Generating asm macros...')
+    stdout = sys.stdout
+    with open(f'{TMC_FOLDER}/asm/macros/scripts.inc', 'w') as out:
+        sys.stdout = out
+        generate_macros()
+    sys.stdout = stdout
+    print('\033[1;92mDone\033[0m\n')
 
 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 330d2a7d..ae8f3272 100644
--- a/tools/script_disassembler/script_disassembler.py
+++ b/tools/script_disassembler/script_disassembler.py
@@ -5,7 +5,7 @@ import struct
 # Input 'macros' to generate the macros for the script commands
 # Input the script bytes as hex to disassemble the script
 
-# Build macros: echo "macros" | python script_disassembler.py > ~/git/tmc/github/asm/macros/script.inc 
+# Build macros: echo "macros" | python script_disassembler.py > ~/git/tmc/github/asm/macros/scripts.inc 
 
 @dataclass
 class Context:
@@ -33,25 +33,29 @@ def get_pointer(barray):
     integers = struct.unpack('I', barray)
     return 'sub_' + (struct.pack('>I', integers[0]-1).hex()).upper()
 
+def get_data_pointer(barray):
+    integers = struct.unpack('I', barray)
+    return 'gUnk_' + (struct.pack('>I', integers[0]-1).hex()).upper()
+
 commands = [
-    {'fun': 'ScriptCommandNop', 'params': ''},
+    {'fun': 'ScriptCommandNop', 'params': 'v'}, # TODO one version with length 33???
     {'fun': 'ScriptCommand_StartScript', 'params': '', 'name': 'start executing scripts'},
     {'fun': 'ScriptCommand_StopScript', 'params': '', 'name': 'stop executing scripts'},
     {'fun': 'ScriptCommand_Jump', 'params': 's', 'name': 'jump by offset'},
     {'fun': 'ScriptCommand_JumpIf', 'params': 's', 'name': 'jump if'},
     {'fun': 'ScriptCommand_JumpIfNot', 'params': 's', 'name': 'jump if not'},
-    {'fun': 'ScriptCommand_0807E078', 'params': 's'},
-    {'fun': 'ScriptCommand_JumpAbsolute', 'params': 'w','name': 'abs jump' },
-    {'fun': 'ScriptCommand_JumpAbsoluteIf', 'params': 'w', 'name': 'abs jump if'},
-    {'fun': 'ScriptCommand_JumpAbsoluteIfNot', 'params': 'w', 'name': 'abs jump if not'},
-    {'fun': 'ScriptCommand_0807E0E0', 'params': 'w'},
+    {'fun': 'ScriptCommand_0807E078', 'params': 'v'},
+    {'fun': 'ScriptCommand_JumpAbsolute', 'params': 'd','name': 'abs jump' },
+    {'fun': 'ScriptCommand_JumpAbsoluteIf', 'params': 'd', 'name': 'abs jump if'},
+    {'fun': 'ScriptCommand_JumpAbsoluteIfNot', 'params': 'd', 'name': 'abs jump if not'},
+    {'fun': 'ScriptCommand_0807E0E0', 'params': 'dd'},
     {'fun': 'ScriptCommand_Call', 'params':'p', 'name': 'Execute function via pointer'},# 'exec': ScriptCommand_Call},
-    {'fun': 'ScriptCommand_CallWithArg', 'params': 'pw'},
-    {'fun': 'ScriptCommand_LoadRoomEntityList', 'params': 'w'},
+    {'fun': 'ScriptCommand_CallWithArg', 'params': 'pv'},
+    {'fun': 'ScriptCommand_LoadRoomEntityList', 'params': 'd'},
     {'fun': 'ScriptCommand_TestBit', 'params': 'w'},
     {'fun': 'ScriptCommand_CheckInventory1', 'params': 's'},
     {'fun': 'ScriptCommand_CheckInventory2', 'params': 's'},
-    {'fun': 'ScriptCommand_HasRoomItemForSale', 'params': 's'},
+    {'fun': 'ScriptCommand_HasRoomItemForSale', 'params': ''},
     {'fun': 'ScriptCommand_CheckLocalFlag', 'params': 's'},
     {'fun': 'ScriptCommand_CheckLocalFlagByOffset', 'params': 'ss'},
     {'fun': 'ScriptCommand_CheckGlobalFlag', 'params': 's'},
@@ -68,7 +72,7 @@ commands = [
     {'fun': 'ScriptCommand_0807E48C', 'params': 's'},
     {'fun': 'ScriptCommand_0807E4CC', 'params': 'w'},
     {'fun': 'ScriptCommand_0807E4EC', 'params': 'w'},
-    {'fun': 'ScriptCommand_0807E51', 'params': 's'},
+    {'fun': 'ScriptCommand_0807E514', 'params': 'w'},
     {'fun': 'ScriptCommand_CheckPlayerFlags', 'params':'w'},
     {'fun': 'ScriptCommand_0807E564', 'params': ''},
     {'fun': 'ScriptCommand_EntityHasHeight', 'params': ''},
@@ -100,12 +104,12 @@ commands = [
     {'fun': 'ScriptCommand_DoFade7', 'params': ''},
     {'fun': 'ScriptCommand_0807E800', 'params': 'w'},
     {'fun': 'ScriptCommand_0807E80C', 'params': 'w'},
-    {'fun': 'ScriptCommand_0807E858', 'params': 's'},
+    {'fun': 'ScriptCommand_0807E858', 'params': 'v'}, # TODO why are there both 0 and 1?
     {'fun': 'ScriptCommand_0807E864', 'params': ''},
     {'fun': 'ScriptCommand_0807E878', 'params': ''},
     {'fun': 'ScriptCommand_0807E888', 'params': ''},
     {'fun': 'ScriptCommand_SetPlayerAction', 'params': 'w'},
-    {'fun': 'ScriptCommand_StartPlayerScript', 'params': 'w'},
+    {'fun': 'ScriptCommand_StartPlayerScript', 'params': 'd'},
     {'fun': 'ScriptCommand_0807E8D4', 'params': 's'},
     {'fun': 'ScriptCommand_0807E8E4_0', 'params': ''}, # duplicate
     {'fun': 'ScriptCommand_0807E8E4_1', 'params': ''}, # duplicate
@@ -126,8 +130,8 @@ commands = [
     {'fun': 'ScriptCommand_0807EA94', 'params': ''},
     {'fun': 'ScriptCommand_TextboxNoOverlapFollow', 'params': 's'},
     {'fun': 'ScriptCommand_TextboxNoOverlap', 'params': 's'},
-    {'fun': 'ScriptCommand_TextboxNoOverlapFollowPos', 'params': 'ss'},
-    {'fun': 'ScriptCommand_0807EAF0', 'params': 'w'},
+    {'fun': 'ScriptCommand_TextboxNoOverlapFollowPos', 'params': 'v'}, # TODO 1 or two
+    {'fun': 'ScriptCommand_0807EAF0', 'params': 'v'}, # TODO
     {'fun': 'ScriptCommand_TextboxNoOverlapVar', 'params': ''},
     {'fun': 'ScriptCommand_0807EB28', 'params': 's'},
     {'fun': 'ScriptCommand_0807EB38', 'params': ''},
@@ -219,12 +223,30 @@ parameters = {
         'expr': '	.word \w',
         'read': lambda ctx: get_pointer(ctx.data[ctx.ptr+2:ctx.ptr+6])
     },
-    'pw': {
+    'd': { # Data pointer
+        'length': 2,
+        'param': 'w',
+        'expr': '	.word \w',
+        'read': lambda ctx: get_data_pointer(ctx.data[ctx.ptr+2:ctx.ptr+6])
+    },
+    'pv': {
+        'length': -2,
+        'param': 'w',
+        'expr': '	.word \w',
+        'read': lambda ctx: ''
+    },
+    'dd': {
         'length': 4,
-        'param': 'a,b',
+        'param': 'a, b',
         'expr': '	.word \\a\n	.word \\b',
-        'read': lambda ctx: get_pointer(ctx.data[ctx.ptr+2:ctx.ptr+6]) + ', ' + barray_to_u32_hex(ctx.data[ctx.ptr+6:ctx.ptr+10])[0]
+        'read': lambda ctx: get_data_pointer(ctx.data[ctx.ptr+2:ctx.ptr+6]) + ', ' + get_data_pointer(ctx.data[ctx.ptr+6:ctx.ptr+10])
     },
+    'v': { # variable parameter count (TODO why?)
+        'length': -1,
+        'param': '',
+        'expr': '',
+        'read': lambda ctx: ''
+    }
 }
 
 # Remove the ScriptCommand_ prefix for the asm macros
@@ -234,31 +256,50 @@ def build_script_command(name: str):
         return '_' + name
     return name
 
+def print_rest_bytes(ctx):
+     print('\n'.join(['.byte ' + hex(x) for x in ctx.data[ctx.ptr:]]))
+
 def ExecuteScriptCommandSet(ctx: Context):
+    # print(f'@{ctx.ptr}') print offsets to debug when manually inserting labels
     cmd = struct.unpack('H', ctx.data[ctx.ptr:ctx.ptr+2])[0]
     if cmd == 0:
-        return 0
+        # this does not need to be the end of the script
+        print('.short 0x0000')
+        ctx.ptr += 2
+        return 1
+
     if cmd == 0xffff:
         ctx.ptr += 2
         print('SCRIPT_END')
-        return 0
-
-    unk_06 = cmd >> 0xA
-    if unk_06 == 0:
+        cmd = struct.unpack('H', ctx.data[ctx.ptr:ctx.ptr+2])[0]
+        if cmd == 0x0000:
+            # This is actually the end of the script
+            print('.short 0x0000')
+            ctx.ptr += 2
+            return 2
+        return 3 # There is a SCRIPT_END without 0x0000 afterwards, but still split into a new file, please
+
+    commandSize = cmd >> 0xA
+    if commandSize == 0:
+        raise Exception(f'Zero commandSize')
         # TODO error
         return 0
-    operationId = cmd & 0x3FF
-    if operationId >= len(commands):
+    commandId = cmd & 0x3FF
+    if commandId >= len(commands):
+        #print_rest_bytes(ctx)
+        print(f'.short {u16_to_hex(cmd)}')
+        ctx.ptr += 2
+        #raise Exception(f'Invalid commandId {commandId} / {len(commands)} {cmd}')
         # TODO error
-        return 0
-    command = commands[operationId]
-    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 1
+    command = commands[commandId]
+    param_length = commandSize - 1 
+    if commandSize > 1:
+        if ctx.ptr+2*commandSize > len(ctx.data):
+            raise Exception(f'Not enough data to fetch {commandSize-1} params')
             return 0
-        meta = struct.unpack(
-            'H'*(unk_06-1), ctx.data[ctx.ptr+2:ctx.ptr+2*unk_06])
+        #meta = struct.unpack(
+        #    'H'*(unk_06-1), ctx.data[ctx.ptr+2:ctx.ptr+2*unk_06])
         #print('meta', meta)
 
     # Handle parameters
@@ -266,9 +307,45 @@ def ExecuteScriptCommandSet(ctx: Context):
         raise Exception('Parameters not defined for ' + command['fun'] + '   Should be of length ' + str(param_length))
     if not command['params'] in parameters:
         raise Exception('Parameter configuration ' + command['params'] + ' not defined')
+
+    # TODO REMOVE fix pointers
+    if command['params'] == 'p':
+        command['params'] = 'w'
+    elif command['params'] == 'd':
+        command['params'] = 'w'
+    elif command['params'] == 'pv':
+        command['params'] = 'v'
+    elif command['params'] == 'dd':
+        command['params'] = 'ww'
+
     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')
+
+
+
+
+    # TODO REMOVE
+    if commandSize == 34:
+        print('@TODO FIX THIS COMMAND!')
+        commandSize = 13
+
+
+    if params['length'] == -1: # variable parameter length
+        print(build_script_command(command['fun']))
+        if commandSize > 1:
+            print('\n'.join(['.short ' + x for x in barray_to_u16_hex(ctx.data[ctx.ptr+2:ctx.ptr+commandSize*2])]))
+            print(f'@ End of {commandSize-1} parameters')
+        ctx.ptr += commandSize*2
+        return 1
+    elif params['length'] == -2: # point and var
+        print(build_script_command(command['fun']) + ' '+ get_pointer(ctx.data[ctx.ptr+2:ctx.ptr+6]))
+        if commandSize > 3:
+            print('\n'.join(['.short ' + x for x in barray_to_u16_hex(ctx.data[ctx.ptr+6:ctx.ptr+commandSize*2])]))
+            print(f'% End of {commandSize-3} parameters')
+        ctx.ptr += commandSize*2
+        return 1
+
+    if commandSize-1 != params['length']:
+        raise Exception(f'Call {command["fun"]} with ' + str(commandSize-1) +' length, while length of ' + str(params['length'])+' defined')
         return 0
         #with open('log.txt', 'a') as log:
 #            log.write(f'Call {command["fun"]} with ' + str(unk_06-1) +' length, while length of ' + str(params['length'])+' defined\n')
@@ -279,29 +356,53 @@ def ExecuteScriptCommandSet(ctx: Context):
     print(build_script_command(command['fun']) + ' ' + params['read'](ctx))
 
     # Execute script
-    ctx.ptr += unk_06*2
+    ctx.ptr += commandSize*2
     return 1
 
 
+# TODO
+# JumpAbsolute 0x08016384
+# JumpAbsoluteIf 0x08016384
+# JumpAbsoluteIfNot 0x08016384
+
 def disassemble_script(input_bytes):
 
     ctx = Context(0, input_bytes)
 
+
+    foundEnd = False
     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):
+        res = ExecuteScriptCommandSet(ctx)
+        if res == 0:
             break
+        elif res == 2:
+            foundEnd = True
+            break
+        elif res == 3:
+            # End in the middle of the script, please create a new file
+            return ctx.ptr
+
+    
 
 
     # Print rest (did not manage to get there)
     if ctx.ptr < len(ctx.data):
         if (len(ctx.data) - ctx.ptr) % 2 != 0:
-            print('\n'.join(['.byte ' + hex(x) for x in ctx.data[ctx.ptr:]]))
+            print_rest_bytes(ctx)
             # TODO error
+            raise Exception('DONT WANT EXTRA after '+ str(ctx.ptr) + ' / ' + str(len(ctx.data)))
             return
         print('\n'.join(['.short ' + x for x in barray_to_u16_hex(ctx.data[ctx.ptr:])]))
-    
+        raise Exception('DONT WANT EXTRA after '+ str(ctx.ptr) + ' / ' + str(len(ctx.data)))
+
+    if not foundEnd:
+        # Seems to happen, sadly
+        return 0
+        #print('\033[93mNo end found\033[0m')
+        #raise Exception('No end found')
+    return 0
 
 def generate_macros():
     print('@ All the macro functions for scripts')