Do our best to unpack mojibake'd texture manifests

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "maturin"
 
 [project]
 name = "ifstools"
-version = "2.1"
+version = "2.2"
 description = "Extractor/repacker for Konmai IFS files"
 readme = "README.md"
 authors = [

src/ifstools/handlers/generic_file.py

@@ -2,6 +2,7 @@ import os
 
 import lxml.etree as etree
 from kbinxml import KBinXML
+from kbinxml.kbinxml import KBinException
 
 from .. import utils
 from .node import Node
@@ -37,7 +38,12 @@ class GenericFile(Node):
         data = self.ifs_data.get(self.start, self.size)
 
         if convert_kbin and self.name.endswith('.xml') and KBinXML.is_binary_xml(data):
-            data = KBinXML(data).to_text().encode('utf8')
+            try:
+                data = KBinXML(data).to_text().encode('utf8')
+            except KBinException as e:
+                if 'convert_illegal_things' in e.args[0]:
+                    print("Corrupt xml found, trying to decode as utf8")
+                    data = KBinXML(data, convert_illegal_things=True).to_text().encode('utf8')
         return data
 
     def _load_from_filesystem(self, **kwargs):

src/ifstools/handlers/md5_folder.py

@@ -1,6 +1,7 @@
 from hashlib import md5
 
 from kbinxml import KBinXML
+from kbinxml.kbinxml import KBinException
 
 from .generic_folder import GenericFolder
 
@@ -20,6 +21,7 @@ class MD5Folder(GenericFolder):
 
         self.info_kbin = None
         self.info_file = None
+        self.encoding_override = None
         for filename, file in self.files.items():
             if filename.endswith('.xml'):
                 self.info_file = file
@@ -29,7 +31,15 @@ class MD5Folder(GenericFolder):
             # _super_ references to info XML breaks things - just extract what we can
             return
 
-        self.info_kbin = KBinXML(self.info_file.load(convert_kbin = False))
+        kbin_raw = self.info_file.load(convert_kbin = False)
+        try:
+            self.info_kbin = KBinXML(kbin_raw)
+        except KBinException as e:
+            if 'convert_illegal_things' in e.args[0]:
+                print("Corrupt texturelist xml found, trying to decode as utf8")
+                self.info_kbin = KBinXML(kbin_raw, convert_illegal_things=True)
+                self.encoding_override = 'utf8'
+
         self._apply_md5()
 
     def _apply_md5(self):
@@ -39,7 +49,7 @@ class MD5Folder(GenericFolder):
 
     def _apply_md5_folder(self, plain_list, folder):
         for plain in plain_list:
-            hashed = md5(plain.encode(self.info_kbin.encoding)).hexdigest()
+            hashed = md5(plain.encode(self.encoding_override or self.info_kbin.encoding)).hexdigest()
 
             if self.extension:
                 plain += self.extension