Fix encoding issue

pycdlib/pycdlib.py

@@ -718,8 +718,8 @@ def _seek_to_extent(self, extent):
         self._cdfp.seek(extent * self.logical_block_size)
 
     @lru_cache(maxsize=256)
-    def _find_iso_record(self, iso_path):
-        # type: (bytes) -> dr.DirectoryRecord
+    def _find_iso_record(self, iso_path, encoding='utf-8'):
+        # type: (bytes, str) -> dr.DirectoryRecord
         """
         An internal method to find a directory record on the ISO given an ISO
         path.  If the entry is found, it returns the directory record object
@@ -731,7 +731,7 @@ def _find_iso_record(self, iso_path):
         Returns:
          The directory record entry representing the entry on the ISO.
         """
-        return _find_dr_record_by_name(self.pvd, iso_path, 'utf-8')
+        return _find_dr_record_by_name(self.pvd, iso_path, encoding)
 
     @lru_cache(maxsize=256)
     def _find_rr_record(self, rr_path):
@@ -3487,8 +3487,8 @@ def _rm_joliet_dir(self, joliet_path):
 
         return num_bytes_to_remove
 
-    def _get_iso_entry(self, iso_path):
-        # type: (bytes) -> dr.DirectoryRecord
+    def _get_iso_entry(self, iso_path, encoding='utf-8'):
+        # type: (bytes, str) -> dr.DirectoryRecord
         """
         Internal method to get the directory record for an ISO path.
 
@@ -3500,7 +3500,7 @@ def _get_iso_entry(self, iso_path):
         if self._needs_reshuffle:
             self._reshuffle_extents()
 
-        return self._find_iso_record(iso_path)
+        return self._find_iso_record(iso_path, encoding)
 
     def _get_rr_entry(self, rr_path):
         # type: (bytes) -> dr.DirectoryRecord
@@ -5475,6 +5475,8 @@ def list_children(self, **kwargs):
             if key in ('joliet_path', 'rr_path', 'iso_path', 'udf_path'):
                 if value is not None:
                     num_paths += 1
+            elif key in ('encoding'):
+                continue
             else:
                 raise pycdlibexception.PyCdlibInvalidInput("Invalid keyword, must be one of 'iso_path', 'rr_path', 'joliet_path', or 'udf_path'")
 
@@ -5497,7 +5499,7 @@ def list_children(self, **kwargs):
                 rec = self._get_rr_entry(utils.normpath(kwargs['rr_path']))
                 use_rr = True
             else:
-                rec = self._get_iso_entry(utils.normpath(kwargs['iso_path']))
+                rec = self._get_iso_entry(utils.normpath(kwargs['iso_path']), kwargs['encoding'])
 
             for c in _yield_children(rec, use_rr):
                 yield c
@@ -5642,8 +5644,8 @@ def rm_isohybrid(self):
 
         self.isohybrid_mbr = None
 
-    def full_path_from_dirrecord(self, rec, rockridge=False):
-        # type: (Union[dr.DirectoryRecord, udfmod.UDFFileEntry], bool) -> str
+    def full_path_from_dirrecord(self, rec, user_encoding, rockridge=False):
+        # type: (Union[dr.DirectoryRecord, udfmod.UDFFileEntry], str, bool) -> str
         """
         Get the absolute path of a directory record.
 
@@ -5662,6 +5664,8 @@ def full_path_from_dirrecord(self, rec, rockridge=False):
             if self.joliet_vd is not None and id(rec.vd) == id(self.joliet_vd):
                 encoding = 'utf-16_be'
 
+            if user_encoding:
+                encoding = user_encoding
             # A root entry has no Rock Ridge entry, even on a Rock Ridge ISO.
             # Always return / here.
             if rec.is_root:
@@ -5701,6 +5705,8 @@ def full_path_from_dirrecord(self, rec, rockridge=False):
                 encoding = rec.file_ident.encoding
             else:
                 encoding = 'utf-8'
+            if user_encoding:
+                encoding = user_encoding
             udf_rec = rec  # type: Optional[udfmod.UDFFileEntry]
             while udf_rec is not None:
                 ident = udf_rec.file_identifier()
@@ -5913,13 +5919,13 @@ def walk(self, **kwargs):
         while dirs:
             dir_record = dirs.popleft()
 
-            relpath = self.full_path_from_dirrecord(dir_record,
+            relpath = self.full_path_from_dirrecord(dir_record, user_encoding,
                                                     rockridge=path_type == 'rr_path')
             dirlist = []
             filelist = []
             dirdict = {}
 
-            for child in reversed(list(self.list_children(**{path_type: relpath}))):
+            for child in reversed(list(self.list_children(**{path_type: relpath, 'encoding': kwargs['encoding']}))):
                 if child is None or child.is_dot() or child.is_dotdot():
                     continue