dialog extractor for yugioh game

getting_dialog.py

@@ -0,0 +1,72 @@
+import mmap
+import struct
+from pathlib import Path
+import json
+
+def read_until_null(indx: int, data: tuple) -> int:
+    end = indx
+    while end + 1 < len(data):
+        ## Read 2 bytes at a time until null found
+        if data[end:end+2] == b"\x00\x00":
+            break
+        end += 2
+    return end
+
+# 8 bytes at a time
+cd_db = {}
+def get_props():
+    test = 0
+    for r in range(0,len(offsets_data),8):
+        test += 1
+        # offsets_data[4:8].hex()
+        card_index_name_data =  int.from_bytes( offsets_data[r:r+4], byteorder='little' )
+        if card_index_name_data == 0:
+            continue
+        #card_index_desc_data =  int.from_bytes( offsets_data[r+4:r+8], byteorder='little' )
+        #card_id = card_id[::-1]
+        #read until null found
+        # end = card_index_name_data
+        # while end + 1 < len(name_data):
+        #     ## Read 2 bytes at a time until null found
+        #     if name_data[end:end+2] == b"\x00\x00":
+        #         break
+        #     end += 2
+        # name = name_data[card_index_name_data:end].decode("utf-16le", errors="replace")
+        end_index = read_until_null(card_index_name_data, name_data)
+        name = name_data[card_index_name_data:end_index].decode("utf-16le", errors="replace")
+        print(name)
+        if name == '':
+            continue
+        #print(card_id.hex(' '))
+        # 2. Get last 14 bits using mask 0x3FFF (binary 11 1111 1111 1111)
+        #print([bin(x) for x in card_first])
+        # if test > 3:
+        #     break;
+        #print(prop_data[r:r+8])
+        if cd_db.get(test) is None:
+            cd_db[test] = {}
+        # {
+        #     f"name_{language_code}":name, 
+        #     f"desc_{language_code}": 'foo', 
+        #     f"name_addr_{language_code}": hex(card_index_name_data), 
+        #     f"desc_addr_{language_code}": hex(card_index_desc_data)
+        # }
+        cd_db[test][f"name_{language_code}"] = name 
+        #cd_db[card_id][f"desc_{language_code}"] = desc 
+        cd_db[test][f"name_addr_{language_code}"] = hex(card_index_name_data)
+        #cd_db[card_id][f"desc_addr_{language_code}"] = hex(card_index_desc_data)
+
+language_code = 'E'
+offsets_data = Path(f"YGO_2020/bin/DLG_Indx_{language_code}.bin").read_bytes() # Get all the bytes
+name_data = Path(f"YGO_2020/bin/DLG_Text_{language_code}.bin").read_bytes() # Get all the bytes
+get_props()
+
+language_code = 'J'
+offsets_data = Path(f"YGO_2020/bin/DLG_Indx_{language_code}.bin").read_bytes() # Get all the bytes
+name_data = Path(f"YGO_2020/bin/DLG_Text_{language_code}.bin").read_bytes() # Get all the bytes
+get_props()
+
+# with open("dump2.json", "w", encoding="utf-16") as outfile:
+#     json.dump(cd_db, outfile, ensure_ascii=False, indent=2)
+
+print()