Replace most of the address-hunting with code matches

pokedex/extract/lib/gbz80.py

@@ -186,6 +186,7 @@ gbz80_instructions = {
     0x7e: 'ld a, [hl]',
     0x7f: 'ld a, a',
 
+    # TODO understand these as both "add a, b" and "add a"
     0x80: 'add a, b',
     0x81: 'add a, c',
     0x82: 'add a, d',
@@ -305,6 +306,7 @@ gbz80_instructions = {
     0xee: 'xor #d8',
     0xef: 'rst $28',
 
+    # TODO really want better support for this
     0xf0: 'ldh a, [#a8]',
     0xf1: 'pop af',
     0xf2: 'ld a, [$ff00+c]',  # XXX table says 1 but this looks like 1 to me
@@ -599,7 +601,9 @@ def find_code(haystack, needle, **kwargs):
             if inputs is not None:
                 break
         else:
-            raise SyntaxError
+            raise SyntaxError(
+                "Can't figure out what instruction corresponds to: "
+                + instruction)
 
         instr = candidate
         pattern_chunks.append(re.escape(instr.prefix))

pokedex/extract/rby.py

@@ -1300,14 +1300,70 @@ class RBYCart:
             raise CartDetectionError("Can't find flavor text pointers")
         addresses['PokedexEntryPointers'] = idx + len(asm_DrawTileLine)
 
-        # This is a helper function for figuring out moves, followed by another
+
-        # 5-byte function, then the table of evolutions and moves.
+        match = find_code(self.data, '''
-        asm_WriteMonMoves_ShiftMoveData = bytes.fromhex('0e03 131a 220d 20fa c9')
+        ;EvolutionAfterBattle:
-        try:
+            ldh a, [#hTilesetType]
-            idx = self.data.index(asm_WriteMonMoves_ShiftMoveData)
+            push af
-        except ValueError:
+            xor a
+            ld [#wEvolutionOccurred], a
+            dec a
+            ld [#wWhichPokemon], a
+            push hl
+            push bc
+            push de
+            ld hl, #wPartyCount
+            push hl
+
+        ;Evolution_PartyMonLoop: ; loop over party mons
+            ld hl, #wWhichPokemon
+            inc [hl]
+            pop hl
+            inc hl
+            ld a, [hl]
+            cp $ff ; have we reached the end of the party?
+            jp z, #_done
+            ld [#wEvoOldSpecies], a
+            push hl
+            ld a, [#wWhichPokemon]
+            ld c, a
+            ld hl, #wCanEvolveFlags
+            ld b, #FLAG_TEST
+            call #Evolution_FlagAction
+            ld a, c
+            and a ; is the mon's bit set?
+            jp z, #Evolution_PartyMonLoop ; if not, go to the next mon
+            ld a, [#wEvoOldSpecies]
+            dec a
+            ld b, 0
+            ld hl, #EvosMovesPointerTable
+            add a, a
+            rl b
+            ld c, a
+            add hl, bc
+            ld a, [hl+]
+            ld h, [hl]
+            ld l, a
+            push hl
+            ld a, [#wcf91]
+            push af
+            xor a ; PLAYER_PARTY_DATA
+            ld [#wMonDataLocation], a
+            call #LoadMonData
+            pop af
+            ld [#wcf91], a
+            pop hl
+        ''',
+            hTilesetType=0xD7,  # FFD7
+        )
+        if not match:
             raise CartDetectionError("Can't find evolution and moveset table")
-        addresses['EvosMovesPointerTable'] = idx + len(asm_WriteMonMoves_ShiftMoveData) + 5
+        rem, inputs = match
+        # As usual, there's no bank given...  but this code has to be in the
+        # same bank as the data it loads!
+        codebank, _ = bank(rem.start())
+        addresses['EvosMovesPointerTable'] = unbank(
+            codebank, inputs['EvosMovesPointerTable'])
 
         # Several lists of names are accessed by a single function, which looks
         # through a list of pointers to find the right set of names to use.
@@ -1352,11 +1408,6 @@ class RBYCart:
             ldh a,[#H_LOADEDROMBANK]
             push af
             ld a,#BANK_MonsterNames
-            ldh [#H_LOADEDROMBANK],a
-            ld [#MBC1RomBank],a
-            ld a,[#wd11e]
-            dec a
-            ld hl,#MonsterNames
         ''',
             H_LOADEDROMBANK=0xB8,  # full address is $FFB8; ldh adds the $FF
             MBC1RomBank=0x2000,
@@ -1451,13 +1502,44 @@ class RBYCart:
         # Here's plan B: look for the function that /loads/ base stats, and
         # scrape the address out of it.  This function is a bit hairy; I've had
         # to expand some of pokered's macros and rewrite the jumps to something
-        # that the rudimentary code matcher can understand.
+        # that the rudimentary code matcher can understand.  Also, there were
-        match = find_code(self.data, '''
+        # two edits made in Yellow: bank switching is done with a function
+        # rather inlined, and Mew is no longer separate.
+        # TODO i guess it would be nice if find_code could deal with this,
+        # seeing as it IS just a regex, but i don't know how i'd cram the
+        # syntax in without a real parser™
+        rgb_bits = dict(
+            bankswitch="""
+                ldh [#H_LOADEDROMBANK], a
+                ld [#MBC1RomBank], a
+            """,
+            mewjump="""
+                cp #MEW
+                jr z,#mew
+            """,
+            mewblock="""
+                jr #done2
+                ;.mew
+                ld hl, #MewBaseStats
+                ld de, #wMonHeader
+                ld bc, #MonBaseStatsLength
+                ld a, #BANK_MewBaseStats
+                call #FarCopyData
+            """,
+        )
+        yellow_bits = dict(
+            bankswitch="""
+                call #BankswitchCommon
+            """,
+            mewjump="",
+            mewblock="",
+        )
+        for bits in (rgb_bits, yellow_bits):
+            code = '''
                 ldh a, [#H_LOADEDROMBANK]
                 push af
                 ld a, #BANK_BaseStats
-            ldh [#H_LOADEDROMBANK], a
+                {bankswitch}
-            ld [#MBC1RomBank], a
                 push bc
                 push de
                 push hl
@@ -1476,8 +1558,7 @@ class RBYCart:
                 ld b,$77 ; size of Aerodactyl fossil sprite
                 cp #FOSSIL_AERODACTYL ; Aerodactyl fossil
                 jr z,#specialID3
-            cp #MEW
+                {mewjump}
-            jr z,#mew
                 ld a, #IndexToPokedexPredef
                 call #IndexToPokedex   ; convert pokemon ID in [wd11e] to pokedex number
                 ld a,[#wd11e]
@@ -1496,14 +1577,12 @@ class RBYCart:
                 ld [hl], e ; write front sprite pointer
                 inc hl
                 ld [hl], d
-            jr #done2
+                {mewblock}
-            ;.mew
+            '''.format(**bits)
-            ld hl, #MewBaseStats
+
-            ld de, #wMonHeader
+            match = find_code(
-            ld bc, #MonBaseStatsLength
+                self.data,
-            ld a, #BANK_MewBaseStats
+                code,
-            call #FarCopyData
-        ''',
                 # These are constants; I left them in the above code for clarity
                 H_LOADEDROMBANK=0xB8,  # full address is $FFB8; ldh adds the $FF
                 MBC1RomBank=0x2000,
@@ -1512,8 +1591,14 @@ class RBYCart:
             )
             if match:
                 rem, inputs = match
-            addresses['BaseStats'] = unbank(inputs['BANK_BaseStats'], inputs['BaseStats'])
+                addresses['BaseStats'] = unbank(
-            addresses['MewBaseStats'] = unbank(inputs['BANK_MewBaseStats'], inputs['MewBaseStats'])
+                    inputs['BANK_BaseStats'], inputs['BaseStats'])
+                if 'MewBaseStats' in inputs:
+                    addresses['MewBaseStats'] = unbank(
+                        inputs['BANK_MewBaseStats'], inputs['MewBaseStats'])
+                else:
+                    addresses['MewBaseStats'] = None
+                break
         else:
             raise CartDetectionError("Can't find base stats")
 
@@ -1738,10 +1823,13 @@ class RBYCart:
     def pokemon_records(self):
         """List of pokemon_structs."""
         self.stream.seek(self.addrs['BaseStats'])
+        # Mew's data is, awkwardly, stored separately pre-Yellow
+        if self.addrs['MewBaseStats']:
             records = Array(self.NUM_POKEMON - 1, pokemon_struct).parse_stream(self.stream)
-        # Mew's data is, awkwardly, stored separately
             self.stream.seek(self.addrs['MewBaseStats'])
             records.append(pokemon_struct.parse_stream(self.stream))
+        else:
+            records = Array(self.NUM_POKEMON, pokemon_struct).parse_stream(self.stream)
 
         return records