Merge pull request #623 from DagsHub/bug/handling-windows-path-yolo-conversion

 bug/handling windows path yolo conversion

Author: kbolashev

dagshub/data_engine/annotation/importer.py

@@ -1,5 +1,5 @@
 from difflib import SequenceMatcher
-from pathlib import Path, PurePosixPath
+from pathlib import Path, PurePosixPath, PurePath
 from tempfile import TemporaryDirectory
 from typing import TYPE_CHECKING, Literal, Optional, Union, Sequence, Mapping, Callable, List
 
@@ -138,7 +138,7 @@ def remap_annotations(
         """
         if remap_func is None:
             first_ann = list(annotations.keys())[0]
-            first_ann_filename = PurePosixPath(first_ann).name
+            first_ann_filename = Path(first_ann).name
             queried = self.ds["path"].endswith(first_ann_filename).select("size").all()
             dp_paths = [dp.path for dp in queried]
             remap_func = self.guess_annotation_filename_remapping(first_ann, dp_paths)
@@ -185,12 +185,14 @@ def guess_annotation_filename_remapping(
 
     @staticmethod
     def generate_path_map_func(ann_path: str, dp_path: str) -> Callable[[str], Optional[str]]:
-        ann_path_posix = PurePosixPath(ann_path)
+        # Using os-dependent path for ann_path because we're getting it from the importer,
+        # which will return os-dependent paths
+        ann_path_obj = PurePath(ann_path)
         dp_path_posix = PurePosixPath(dp_path)
 
         matcher = SequenceMatcher(
             None,
-            ann_path_posix.parts,
+            ann_path_obj.parts,
             dp_path_posix.parts,
         )
         diff = matcher.get_matching_blocks()
@@ -204,7 +206,7 @@ def generate_path_map_func(ann_path: str, dp_path: str) -> Callable[[str], Optio
 
         match = diff[0]
         # Make sure that the match goes until the end
-        if match.a + match.size != len(ann_path_posix.parts) or match.b + match.size != len(dp_path_posix.parts):
+        if match.a + match.size != len(ann_path_obj.parts) or match.b + match.size != len(dp_path_posix.parts):
             raise CannotRemapPathError(ann_path, dp_path)
         # ONE of the paths need to go until the start
         if match.a != 0 and match.b != 0:
@@ -214,6 +216,7 @@ def generate_path_map_func(ann_path: str, dp_path: str) -> Callable[[str], Optio
         if match.a == 0 and match.b == 0:
 
             def identity_func(x: str) -> str:
+                # Do a replace because we might be going from a windows path to a posix path
                 return x.replace(ann_path, dp_path)
 
             return identity_func
@@ -227,17 +230,17 @@ def identity_func(x: str) -> str:
             prefix = dp_path_posix.parts[match.a : match.b]
 
             def add_prefix(x: str) -> Optional[str]:
-                return PurePosixPath(*prefix, x).as_posix()
+                return PurePath(*prefix, x).as_posix()
 
             return add_prefix
 
         else:
             # Remove the prefix
             def remove_prefix(x: str) -> Optional[str]:
-                p = PurePosixPath(x)
+                p = PurePath(x)
                 if len(p.parts) <= match.a:
                     return None
-                return PurePosixPath(*p.parts[match.a :]).as_posix()
+                return PurePath(*p.parts[match.a :]).as_posix()
 
             return remove_prefix
 

setup.py

@@ -41,7 +41,7 @@ def get_version(rel_path: str) -> str:
     "python-dateutil",
     "boto3",
     "semver",
-    "dagshub-annotation-converter>=0.1.5",
+    "dagshub-annotation-converter>=0.1.12",
 ]
 
 extras_require = {