added wrappers for DatasetNotFound error

Author: klakhov

datumaro/components/environment.py

@@ -12,6 +12,7 @@
 
 from datumaro.components.cli_plugin import CliPlugin, plugin_types
 from datumaro.components.format_detection import RejectionReason, detect_dataset_format
+from datumaro.components.wrappers import wrap_importer
 from datumaro.util.os_util import import_foreign_module, split_path
 
 T = TypeVar("T")
@@ -226,6 +227,10 @@ def _register_plugins(self, plugins):
         self.transforms.batch_register(plugins)
         self.validators.batch_register(plugins)
 
+        for key in self.importers:
+            importer = self.importers.get(key)
+            wrap_importer(importer)
+
     def make_extractor(self, name, *args, **kwargs):
         return self.extractors.get(name)(*args, **kwargs)
 

datumaro/components/errors.py

@@ -273,9 +273,11 @@ def __str__(self):
 @define(auto_exc=False)
 class DatasetNotFoundError(DatasetImportError):
     path = field()
+    ext = field(default="")
 
     def __str__(self):
-        return f"Failed to find dataset at '{self.path}'"
+        file_ext_info = f", file '{self.ext}' was not found" if self.ext else ''
+        return f"Failed to find dataset at '{self.path}' {file_ext_info}"
 
 
 @define(auto_exc=False)

datumaro/components/extractor.py

@@ -433,13 +433,17 @@ def find_sources(cls, path) -> List[Dict]:
     def find_sources_with_params(cls, path, **extra_params) -> List[Dict]:
         return cls.find_sources(path)
 
+    @classmethod
+    def _generate_not_found_error(self, path):
+        return DatasetNotFoundError(path)
+
     def __call__(self, path, **extra_params):
         if not path or not osp.exists(path):
             raise DatasetNotFoundError(path)
 
         found_sources = self.find_sources_with_params(osp.normpath(path), **extra_params)
         if not found_sources:
-            raise DatasetNotFoundError(path)
+            raise self._generate_not_found_error(path)
 
         sources = []
         for desc in found_sources:

datumaro/components/wrappers.py

@@ -0,0 +1,41 @@
+# Copyright (C) 2024 CVAT.ai Corporation
+#
+# SPDX-License-Identifier: MIT
+
+from unittest import mock
+from typing import Optional, Callable
+
+from datumaro.components.errors import DatasetNotFoundError
+
+def wrap_find_sources_recursive(importer):
+    @classmethod
+    def updated_find_sources_recursive(
+        cls,
+        path: str,
+        ext: Optional[str],
+        extractor_name: str,
+        filename: str = "*",
+        dirname: str = "",
+        file_filter: Optional[Callable[[str], bool]] = None,
+        max_depth: int = 3,
+    ):
+        sources = super(importer, cls)._find_sources_recursive(
+            path, ext, extractor_name,
+            filename, dirname, file_filter, max_depth
+        )
+        if not sources:
+            cls._not_found_error_data = {"ext": ext}
+
+    return updated_find_sources_recursive
+
+def wrap_generate_not_found_error(importer):
+    @classmethod
+    def updated_generate_not_found_error(cls, path):
+        return DatasetNotFoundError(path, cls._not_found_error_data.get("ext"))
+
+    return updated_generate_not_found_error
+
+def wrap_importer(importer):
+    mock.patch.object(importer, '_find_sources_recursive', new=wrap_find_sources_recursive(importer)).start()
+    mock.patch.object(importer, '_generate_not_found_error', new=wrap_generate_not_found_error(importer)).start()
+    mock.patch.object(importer, '_not_found_error_data', new={"ext": ""}, create=True).start()