Enrich stack trace while importing dataset to improve user experience

CHANGELOG.md

@@ -15,6 +15,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Enhancements
 - Use autosummary for fully-automatic Python module docs generation
   (<https://github.com/openvinotoolkit/datumaro/pull/973>)
+- Enrich stack trace for better user experience when importing
+  (<https://github.com/openvinotoolkit/datumaro/pull/992>)
 
 ### Bug fixes
 - Fix Mapillary Vistas data format

datumaro/components/dataset.py

@@ -32,6 +32,7 @@
 from datumaro.components.errors import (
     CategoriesRedefinedError,
     ConflictingCategoriesError,
+    DatasetImportError,
     DatasetInfosRedefinedError,
     MediaTypeError,
     MultipleFormatsMatchError,
@@ -1249,7 +1250,11 @@ def import_from(
             if eager:
                 dataset.init_cache()
         except _ImportFail as e:
-            raise e.__cause__
+            cause = e.__cause__ if getattr(e, "__cause__", None) is not None else e
+            cause.__traceback__ = e.__traceback__
+            raise DatasetImportError(f"Failed to import dataset '{format}' at '{path}'.") from cause
+        except Exception as e:
+            raise DatasetImportError(f"Failed to import dataset '{format}' at '{path}'.") from e
 
         dataset._source_path = path
         dataset._format = format

datumaro/components/environment.py

@@ -294,7 +294,9 @@ def detect_dataset(
             else FormatDetectionConfidence.NONE
         )
 
-        return [str(format) for format in all_matched_formats if format.confidence == max_conf]
+        return sorted(
+            [str(format) for format in all_matched_formats if format.confidence == max_conf]
+        )
 
     def __reduce__(self):
         return (self.__class__, ())

datumaro/components/errors.py

@@ -272,10 +272,12 @@ def __str__(self):
 
 @define(auto_exc=False)
 class DatasetNotFoundError(DatasetImportError):
-    path = field()
+    path: str = field()
+    format: str = field()
+    template: str = field(default="Failed to find dataset '{format}' at '{path}'")
 
     def __str__(self):
-        return f"Failed to find dataset at '{self.path}'"
+        return self.template.format(path=self.path, format=self.format)
 
 
 @define(auto_exc=False)

datumaro/components/importer.py

@@ -113,11 +113,11 @@ def find_sources_with_params(cls, path: str, **extra_params) -> List[Dict]:
 
     def __call__(self, path, **extra_params):
         if not path or not osp.exists(path):
-            raise DatasetNotFoundError(path)
+            raise DatasetNotFoundError(path, self.NAME)
 
         found_sources = self.find_sources_with_params(osp.normpath(path), **extra_params)
         if not found_sources:
-            raise DatasetNotFoundError(path)
+            raise DatasetNotFoundError(path, self.NAME)
 
         sources = []
         for desc in found_sources:

datumaro/plugins/data_formats/ade20k2017.py

@@ -1,19 +1,22 @@
-# Copyright (C) 2020-2021 Intel Corporation
+# Copyright (C) 2020-2023 Intel Corporation
 #
 # SPDX-License-Identifier: MIT
 
+import errno
 import glob
 import logging as log
 import os
 import os.path as osp
 import re
+from typing import Optional
 
 import numpy as np
 
 from datumaro.components.annotation import AnnotationType, CompiledMask, LabelCategories, Mask
 from datumaro.components.dataset_base import DatasetBase, DatasetItem
+from datumaro.components.errors import InvalidAnnotationError
 from datumaro.components.format_detection import FormatDetectionContext
-from datumaro.components.importer import Importer
+from datumaro.components.importer import ImportContext, Importer
 from datumaro.components.media import Image
 from datumaro.util.image import IMAGE_EXTENSIONS, find_images, lazy_image, load_image
 from datumaro.util.meta_file_util import has_meta_file, parse_meta_file
@@ -29,16 +32,16 @@ class Ade20k2017Path:
 
 
 class Ade20k2017Base(DatasetBase):
-    def __init__(self, path):
+    def __init__(self, path: str, *, ctx: Optional[ImportContext] = None):
         if not osp.isdir(path):
-            raise FileNotFoundError("Can't read dataset directory '%s'" % path)
+            raise NotADirectoryError(errno.ENOTDIR, "Can't find dataset directory", path)
 
         # exclude dataset meta file
         subsets = [subset for subset in os.listdir(path) if osp.splitext(subset)[-1] != ".json"]
         if len(subsets) < 1:
-            raise FileNotFoundError("Can't read subsets in directory '%s'" % path)
+            raise FileNotFoundError(errno.ENOENT, "Can't find subsets in directory", path)
 
-        super().__init__(subsets=sorted(subsets))
+        super().__init__(subsets=sorted(subsets), ctx=ctx)
         self._path = path
 
         self._items = []
@@ -127,16 +130,18 @@ def _load_items(self, subset):
     def _load_item_info(self, path):
         attr_path = osp.splitext(path)[0] + "_atr.txt"
         if not osp.isfile(attr_path):
-            raise Exception("Can't find annotation file for image %s" % path)
+            raise FileNotFoundError(
+                errno.ENOENT, "Can't find annotation file for image %s" % path, attr_path
+            )
 
         item_info = []
         with open(attr_path, "r", encoding="utf-8") as f:
             for line in f:
                 columns = [s.strip() for s in line.split("#")]
                 if len(columns) != 6:
-                    raise Exception("Invalid line in %s" % attr_path)
+                    raise InvalidAnnotationError("Invalid line in %s" % attr_path)
                 if columns[5][0] != '"' or columns[5][-1] != '"':
-                    raise Exception(
+                    raise InvalidAnnotationError(
                         "Attributes column are expected \
                         in double quotes, file %s"
                         % attr_path

datumaro/plugins/data_formats/ade20k2020.py

@@ -1,12 +1,14 @@
-# Copyright (C) 2020-2022 Intel Corporation
+# Copyright (C) 2020-2023 Intel Corporation
 #
 # SPDX-License-Identifier: MIT
 
+import errno
 import glob
 import logging as log
 import os
 import os.path as osp
 import re
+from typing import Optional
 
 import numpy as np
 
@@ -19,7 +21,7 @@
 )
 from datumaro.components.dataset_base import DatasetBase, DatasetItem
 from datumaro.components.format_detection import FormatDetectionContext
-from datumaro.components.importer import Importer
+from datumaro.components.importer import ImportContext, Importer
 from datumaro.components.media import Image
 from datumaro.util import parse_json
 from datumaro.util.image import IMAGE_EXTENSIONS, find_images, lazy_image, load_image
@@ -37,16 +39,16 @@ class Ade20k2020Path:
 
 
 class Ade20k2020Base(DatasetBase):
-    def __init__(self, path):
+    def __init__(self, path: str, *, ctx: Optional[ImportContext] = None):
         if not osp.isdir(path):
-            raise FileNotFoundError("Can't read dataset directory '%s'" % path)
+            raise NotADirectoryError(errno.ENOTDIR, "Can't find dataset directory", path)
 
         # exclude dataset meta file
         subsets = [subset for subset in os.listdir(path) if osp.splitext(subset)[-1] != ".json"]
         if len(subsets) < 1:
-            raise FileNotFoundError("Can't read subsets in directory '%s'" % path)
+            raise FileNotFoundError(errno.ENOENT, "Can't find subsets in directory", path)
 
-        super().__init__(subsets=sorted(subsets))
+        super().__init__(subsets=sorted(subsets), ctx=ctx)
         self._path = path
 
         self._items = []
@@ -167,10 +169,8 @@ def _load_item_info(self, path):
         json_path = osp.splitext(path)[0] + ".json"
         item_info = []
         if not osp.isfile(json_path):
-            raise Exception(
-                "Can't find annotation file (*.json) \
-                for image %s"
-                % path
+            raise FileNotFoundError(
+                errno.ENOENT, "Can't find annotation file for image %s" % path, json_path
             )
 
         with open(json_path, "r", encoding="latin-1") as f:

datumaro/plugins/data_formats/arrow/base.py

@@ -4,11 +4,13 @@
 
 import os.path as osp
 import struct
+from typing import List, Optional
 
 import pyarrow as pa
 
 from datumaro.components.dataset_base import SubsetBase
 from datumaro.components.errors import MediaTypeError
+from datumaro.components.importer import ImportContext
 from datumaro.components.media import MediaType
 from datumaro.components.merge import get_merger
 from datumaro.plugins.data_formats.datumaro.base import DatumaroBase
@@ -19,10 +21,19 @@
 
 
 class ArrowBase(SubsetBase):
-    def __init__(self, path, ctx, subset, additional_paths=[]):
+    def __init__(
+        self,
+        path: str,
+        additional_paths: Optional[List[str]] = None,
+        *,
+        subset: Optional[str] = None,
+        ctx: Optional[ImportContext] = None,
+    ):
         super().__init__(subset=subset, ctx=ctx)
 
-        self._paths = [path] + additional_paths
+        self._paths = [path]
+        if additional_paths:
+            self._paths += additional_paths
 
         self._load()
 

datumaro/plugins/data_formats/ava/ava.py

@@ -1,19 +1,21 @@
-# Copyright (C) 2022 Intel Corporation
+# Copyright (C) 2022-2023 Intel Corporation
 #
 # SPDX-License-Identifier: MIT
 
 import csv
+import errno
 import os
 import os.path as osp
+from typing import Optional
 
 import google.protobuf.text_format as text_format
 
 from datumaro.components.annotation import AnnotationType, Bbox, LabelCategories
 from datumaro.components.dataset_base import DatasetItem, SubsetBase
-from datumaro.components.errors import DatasetImportError, MediaTypeError
+from datumaro.components.errors import MediaTypeError
 from datumaro.components.exporter import Exporter
 from datumaro.components.format_detection import FormatDetectionConfidence, FormatDetectionContext
-from datumaro.components.importer import Importer
+from datumaro.components.importer import ImportContext, Importer
 from datumaro.components.media import Image
 from datumaro.util.os_util import find_files
 
@@ -32,32 +34,41 @@ class AvaPath:
 
 
 class AvaBase(SubsetBase):
-    def __init__(self, path):
+    def __init__(
+        self,
+        path: str,
+        *,
+        subset: Optional[str] = None,
+        ctx: Optional[ImportContext] = None,
+    ):
         if not osp.isfile(path):
-            raise DatasetImportError(f"Can't find CSV file at '{path}'")
+            raise FileNotFoundError(errno.ENOENT, "Can't find CSV file", path)
         self._path = path
 
-        subset = (
-            osp.splitext(osp.basename(path))[0]
-            .replace(AvaPath.ANNOTATION_PREFIX, "")
-            .replace(AvaPath.ANNOTATION_VERSION, "")
-        )
-        super().__init__(subset=subset)
+        if not subset:
+            subset = (
+                osp.splitext(osp.basename(path))[0]
+                .replace(AvaPath.ANNOTATION_PREFIX, "")
+                .replace(AvaPath.ANNOTATION_VERSION, "")
+            )
+        super().__init__(subset=subset, ctx=ctx)
 
         if path.endswith(osp.join(AvaPath.ANNOTATION_DIR, osp.basename(path))):
             self._rootpath = path.rsplit(AvaPath.ANNOTATION_DIR, maxsplit=1)[0]
         else:
-            raise DatasetImportError(
+            raise FileNotFoundError(
+                errno.ENOENT,
                 f"Annotation path ({path}) should be under the directory which is named {AvaPath.ANNOTATION_DIR}. "
-                "If not, Datumaro fails to find the root path for this dataset."
+                "If not, Datumaro fails to find the root path for this dataset.",
             )
 
         if self._rootpath and osp.isdir(osp.join(self._rootpath, AvaPath.IMAGE_DIR)):
             self._images_dir = osp.join(self._rootpath, AvaPath.IMAGE_DIR)
         else:
-            raise DatasetImportError(
+            raise FileNotFoundError(
+                errno.ENOENT,
                 f"Root path ({self._rootpath}) should contain the directory which is named {AvaPath.IMAGE_DIR}. "
-                "If not, Datumaro fails to find the image directory path."
+                "If not, Datumaro fails to find the image directory path.",
             )
 
         self._infos = self._load_infos(osp.dirname(path))
@@ -78,9 +89,10 @@ def _load_infos(self, path):
 
     def _load_categories(self, category_path):
         if not osp.exists(category_path):
-            raise DatasetImportError(
+            raise FileNotFoundError(
+                errno.ENOENT,
                 f"Label lists cannot be found in ({category_path}). "
-                "If not, Datumaro fails to import AVA action dataset."
+                "If not, Datumaro fails to import AVA action dataset.",
             )
 
         with open(category_path, "r") as f:

datumaro/plugins/data_formats/brats.py

@@ -1,17 +1,19 @@
-# Copyright (C) 2022 Intel Corporation
+# Copyright (C) 2022-2023 Intel Corporation
 #
 # SPDX-License-Identifier: MIT
 
+import errno
 import glob
 import os.path as osp
+from typing import Optional
 
 import nibabel as nib
 import numpy as np
 
 from datumaro.components.annotation import AnnotationType, LabelCategories, Mask
 from datumaro.components.dataset_base import DatasetItem, SubsetBase
 from datumaro.components.format_detection import FormatDetectionContext
-from datumaro.components.importer import Importer
+from datumaro.components.importer import ImportContext, Importer
 from datumaro.components.media import MultiframeImage
 
 
@@ -22,17 +24,17 @@ class BratsPath:
 
 
 class BratsBase(SubsetBase):
-    def __init__(self, path):
+    def __init__(self, path: str, *, ctx: Optional[ImportContext] = None):
         if not osp.isdir(path):
-            raise FileNotFoundError("Can't read dataset directory '%s'" % path)
+            raise NotADirectoryError(errno.ENOTDIR, "Can't find dataset directory", path)
 
         self._subset_suffix = osp.basename(path)[len(BratsPath.IMAGES_DIR) :]
         subset = None
         if self._subset_suffix == "Tr":
             subset = "train"
         elif self._subset_suffix == "Ts":
             subset = "test"
-        super().__init__(subset=subset, media_type=MultiframeImage)
+        super().__init__(subset=subset, media_type=MultiframeImage, ctx=ctx)
 
         self._root_dir = osp.dirname(path)
         self._categories = self._load_categories()