Merge pull request #253 from aiqwe/improve_readibility

Simple enhancement for readibility

src/datatrove/io.py

@@ -128,7 +128,10 @@ def list_files(
         Get a list of files on this directory. If `subdirectory` is given will search in `path/subdirectory`. If
         glob_pattern is given, it will only return files that match the pattern, which can be used to match a given
         extension, for example `*.myext`. Be careful with subdirectories when using glob (use ** if you want to match
-        any subpath). Args: subdirectory: str:  (Default value = "") recursive: bool:  (Default value = True)
+        any subpath).
+
+        Args: subdirectory: str:  (Default value = "")
+        recursive: bool:  (Default value = True)
         glob_pattern: str | None:  (Default value = None)
 
         Returns: a list of file paths, relative to `self.path`

src/datatrove/tools/check_dataset.py

@@ -74,7 +74,7 @@ def open_file(path):
     datafiles = input_folder.list_files(glob_pattern="*.ds")
     datafiles_index = input_folder.list_files(glob_pattern="*.ds.index")
     datafiles_loss = input_folder.list_files(glob_pattern="*.ds.loss")
-    check_loss = not not datafiles_loss
+    check_loss = bool(datafiles_loss)
     assert len(datafiles) == len(datafiles_index) and (not check_loss or len(datafiles) == len(datafiles_loss)), (
         "Mismatch between number of .ds, " ".ds.index and/or .ds.loss files"
     )