Improved DatasetNotFound error message (#7923)

The recent changes enhance the dataset import functionality across various dataset formats in the CVAT application by integrating specific importers from the Datumaro library. The updates streamline the detection of datasets, improve error handling, and introduce new tests to ensure robustness against incorrect file structures during import operations.

changelog.d/20240523_091047_kirill.9992_improve_dataset_not_found.md

@@ -0,0 +1,4 @@
+### Added
+
+- Improved message of DatasetNotFoundError
+  (<https://github.com/cvat-ai/cvat/pull/7923>)

cvat-core/src/server-proxy.ts

@@ -181,7 +181,11 @@ async function chunkUpload(file: File, uploadConfig): Promise<{ uploadSentSize:
 function filterPythonTraceback(data: string): string {
     if (typeof data === 'string' && data.trim().startsWith('Traceback')) {
         const lastRow = data.split('\n').findLastIndex((str) => str.trim().length);
-        return `${data.split('\n').slice(lastRow, lastRow + 1)[0]}`;
+        let errorText = `${data.split('\n').slice(lastRow, lastRow + 1)[0]}`;
+        if (errorText.includes('CvatDatasetNotFoundError')) {
+            errorText = errorText.replace(/.*CvatDatasetNotFoundError: /, '');
+        }
+        return errorText;
     }
 
     return data;

cvat-ui/src/components/cvat-app.tsx

@@ -375,13 +375,18 @@ class CVATApplication extends React.PureComponent<CVATAppProps & RouteComponentP
         function showError(title: string, _error: Error, shouldLog?: boolean, className?: string): void {
             const error = _error?.message || _error.toString();
             const dynamicProps = typeof className === 'undefined' ? {} : { className };
+            let errorLength = error.length;
+            // Do not count the length of the link in the Markdown error message
+            if (/]\(.+\)/.test(error)) {
+                errorLength = error.replace(/]\(.+\)/, ']').length;
+            }
             notification.error({
                 ...dynamicProps,
                 message: (
                     <ReactMarkdown>{title}</ReactMarkdown>
                 ),
                 duration: null,
-                description: error.length > 300 ? 'Open the Browser Console to get details' : <ReactMarkdown>{error}</ReactMarkdown>,
+                description: errorLength > 300 ? 'Open the Browser Console to get details' : <ReactMarkdown>{error}</ReactMarkdown>,
             });
 
             if (shouldLog) {

cvat/apps/dataset_manager/bindings.py

@@ -6,6 +6,7 @@
 from __future__ import annotations
 
 import os.path as osp
+import re
 import sys
 from collections import namedtuple
 from functools import reduce
@@ -22,8 +23,12 @@
 import rq
 from attr import attrib, attrs
 from datumaro.components.media import PointCloud
+from datumaro.components.environment import Environment
+from datumaro.components.extractor import Importer
+from datumaro.components.format_detection import RejectionReason
 from django.db.models import QuerySet
 from django.utils import timezone
+from django.conf import settings
 
 from cvat.apps.dataset_manager.formats.utils import get_label_color
 from cvat.apps.dataset_manager.util import add_prefetch_fields
@@ -1664,6 +1669,33 @@ def GetCVATDataExtractor(
 class CvatImportError(Exception):
     pass
 
+@attrs
+class CvatDatasetNotFoundError(CvatImportError):
+    message: str = ""
+    reason: str = ""
+    format_name: str = ""
+    _docs_base_url = f"{settings.CVAT_DOCS_URL}/manual/advanced/formats/"
+
+    def __str__(self) -> str:
+        formatted_format_name = self._format_name_for_docs()
+        docs_message = self._docs_message(formatted_format_name)
+        display_message = self._clean_display_message()
+        return f"{docs_message}. {display_message}"
+
+    def _format_name_for_docs(self) -> str:
+        return self.format_name.replace("_", "-")
+
+    def _docs_message(self, formatted_format_name: str) -> str:
+        return f"Check [format docs]({self._docs_base_url}format-{formatted_format_name})"
+
+    def _clean_display_message(self) -> str:
+        message = re.sub(r'^.*?:', "", self.message)
+        if "dataset must contain a file matching pattern" in message:
+            message = message.replace("dataset must contain a file matching pattern", "")
+            message = message.replace("\n", "")
+            message = "Dataset must contain a file:" + message
+        return re.sub(r' +', " ", message)
+
 def mangle_image_name(name: str, subset: str, names: DefaultDict[Tuple[str, str], int]) -> str:
     name, ext = name.rsplit(osp.extsep, maxsplit=1)
 
@@ -2265,3 +2297,19 @@ def load_dataset_data(project_annotation, dataset: dm.Dataset, project_data):
             dataset_files['data_root'] = osp.commonpath(root_paths) + osp.sep
 
         project_annotation.add_task(task_fields, dataset_files, project_data)
+
+def detect_dataset(dataset_dir: str, format_name: str, importer: Importer) -> None:
+    not_found_error_instance = CvatDatasetNotFoundError()
+
+    def not_found_error(_, reason, human_message):
+        not_found_error_instance.format_name = format_name
+        not_found_error_instance.reason = reason
+        not_found_error_instance.message = human_message
+
+    detection_env = Environment()
+    detection_env.importers.items.clear()
+    detection_env.importers.register(format_name, importer)
+    detected = detection_env.detect_dataset(dataset_dir, depth=4, rejection_callback=not_found_error)
+
+    if not detected and not_found_error_instance.reason != RejectionReason.detection_unsupported:
+        raise not_found_error_instance

cvat/apps/dataset_manager/formats/camvid.py

@@ -35,6 +35,9 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     Archive(src_file.name).extractall(temp_dir)
 
+    # We do not run detect_dataset before import because the Camvid format
+    # has problem with the dataset detection in case of empty annotation file(s)
+    # Details in: https://github.com/cvat-ai/datumaro/issues/43
     dataset = Dataset.import_from(temp_dir, 'camvid', env=dm_env)
     dataset = MaskToPolygonTransformation.convert_dataset(dataset, **kwargs)
     if load_data_callback is not None:

cvat/apps/dataset_manager/formats/cityscapes.py

@@ -9,7 +9,7 @@
 from datumaro.plugins.cityscapes_format import write_label_map
 from pyunpack import Archive
 
-from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor,
+from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor, detect_dataset,
                                                 import_dm_annotations)
 from cvat.apps.dataset_manager.util import make_zip_archive
 
@@ -43,6 +43,7 @@ def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs
             for label, info in make_colormap(instance_data).items()}
         write_label_map(labelmap_file, colormap)
 
+    detect_dataset(temp_dir, format_name='cityscapes', importer= dm_env.importers.get('cityscapes'))
     dataset = Dataset.import_from(temp_dir, 'cityscapes', env=dm_env)
     dataset = MaskToPolygonTransformation.convert_dataset(dataset, **kwargs)
     if load_data_callback is not None:

cvat/apps/dataset_manager/formats/coco.py

@@ -7,8 +7,9 @@
 
 from datumaro.components.dataset import Dataset
 from datumaro.components.annotation import AnnotationType
+from datumaro.plugins.coco_format.importer import CocoImporter
 
-from cvat.apps.dataset_manager.bindings import GetCVATDataExtractor, \
+from cvat.apps.dataset_manager.bindings import GetCVATDataExtractor, detect_dataset, \
     import_dm_annotations
 from cvat.apps.dataset_manager.util import make_zip_archive
 
@@ -27,6 +28,8 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     if zipfile.is_zipfile(src_file):
         zipfile.ZipFile(src_file).extractall(temp_dir)
+        # We use coco importer because it gives better error message
+        detect_dataset(temp_dir, format_name='coco', importer=CocoImporter)
         dataset = Dataset.import_from(temp_dir, 'coco_instances', env=dm_env)
         if load_data_callback is not None:
             load_data_callback(dataset, instance_data)
@@ -55,6 +58,8 @@ def remove_extra_annotations(dataset):
 
     if zipfile.is_zipfile(src_file):
         zipfile.ZipFile(src_file).extractall(temp_dir)
+        # We use coco importer because it gives better error message
+        detect_dataset(temp_dir, format_name='coco', importer=CocoImporter)
         dataset = Dataset.import_from(temp_dir, 'coco_person_keypoints', env=dm_env)
         remove_extra_annotations(dataset)
         if load_data_callback is not None:

cvat/apps/dataset_manager/formats/cvat.py

@@ -17,10 +17,12 @@
 from datumaro.components.dataset import Dataset, DatasetItem
 from datumaro.components.extractor import (DEFAULT_SUBSET_NAME, Extractor,
                                            Importer)
+from datumaro.plugins.cvat_format.extractor import CvatImporter as _CvatImporter
+
 from datumaro.util.image import Image
 from defusedxml import ElementTree
 
-from cvat.apps.dataset_manager.bindings import (ProjectData, CommonData,
+from cvat.apps.dataset_manager.bindings import (ProjectData, CommonData, detect_dataset,
                                                 get_defaulted_subset,
                                                 import_dm_annotations,
                                                 match_dm_item)
@@ -1439,6 +1441,7 @@ def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs
         zipfile.ZipFile(src_file).extractall(temp_dir)
 
         if isinstance(instance_data, ProjectData):
+            detect_dataset(temp_dir, format_name='cvat', importer=_CvatImporter)
             dataset = Dataset.import_from(temp_dir, 'cvat', env=dm_env)
             if load_data_callback is not None:
                 load_data_callback(dataset, instance_data)

cvat/apps/dataset_manager/formats/datumaro.py

@@ -6,9 +6,10 @@
 from datumaro.components.dataset import Dataset
 from datumaro.components.extractor import ItemTransform
 from datumaro.util.image import Image
+
 from pyunpack import Archive
 
-from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor,
+from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor, detect_dataset,
     import_dm_annotations)
 from cvat.apps.dataset_manager.util import make_zip_archive
 from cvat.apps.engine.models import DimensionType
@@ -37,6 +38,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     Archive(src_file.name).extractall(temp_dir)
 
+    detect_dataset(temp_dir, format_name='datumaro', importer=dm_env.importers.get('datumaro'))
     dataset = Dataset.import_from(temp_dir, 'datumaro', env=dm_env)
 
     if load_data_callback is not None:
@@ -61,6 +63,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     Archive(src_file.name).extractall(temp_dir)
 
+    detect_dataset(temp_dir, format_name='datumaro', importer=dm_env.importers.get('datumaro'))
     dataset = Dataset.import_from(temp_dir, 'datumaro', env=dm_env)
 
     if load_data_callback is not None:

cvat/apps/dataset_manager/formats/icdar.py

@@ -88,6 +88,10 @@ def _export_recognition(dst_file, temp_dir, instance_data, save_images=False):
 @importer(name='ICDAR Recognition', ext='ZIP', version='1.0')
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     zipfile.ZipFile(src_file).extractall(temp_dir)
+
+    # We do not run detect_dataset before import because the ICDAR format
+    # has problem with the dataset detection in case of empty annotation file(s)
+    # Details in: https://github.com/cvat-ai/datumaro/issues/43
     dataset = Dataset.import_from(temp_dir, 'icdar_word_recognition', env=dm_env)
     dataset.transform(CaptionToLabel, label='icdar')
     if load_data_callback is not None:
@@ -107,6 +111,9 @@ def _export_localization(dst_file, temp_dir, instance_data, save_images=False):
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     zipfile.ZipFile(src_file).extractall(temp_dir)
 
+    # We do not run detect_dataset before import because the ICDAR format
+    # has problem with the dataset detection in case of empty annotation file(s)
+    # Details in: https://github.com/cvat-ai/datumaro/issues/43
     dataset = Dataset.import_from(temp_dir, 'icdar_text_localization', env=dm_env)
     dataset.transform(AddLabelToAnns, label='icdar')
     if load_data_callback is not None:
@@ -129,6 +136,10 @@ def _export_segmentation(dst_file, temp_dir, instance_data, save_images=False):
 @importer(name='ICDAR Segmentation', ext='ZIP', version='1.0')
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     zipfile.ZipFile(src_file).extractall(temp_dir)
+
+    # We do not run detect_dataset before import because the ICDAR format
+    # has problem with the dataset detection in case of empty annotation file(s)
+    # Details in: https://github.com/cvat-ai/datumaro/issues/43
     dataset = Dataset.import_from(temp_dir, 'icdar_text_segmentation', env=dm_env)
     dataset.transform(AddLabelToAnns, label='icdar')
     dataset = MaskToPolygonTransformation.convert_dataset(dataset, **kwargs)

cvat/apps/dataset_manager/formats/imagenet.py

@@ -30,6 +30,10 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
 @importer(name='ImageNet', ext='ZIP', version='1.0')
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     zipfile.ZipFile(src_file).extractall(temp_dir)
+
+    # We do not run detect_dataset before import because the Imagenet format
+    # has problem with the dataset detection in case of empty annotation file(s)
+    # Details in: https://github.com/cvat-ai/datumaro/issues/43
     if glob(osp.join(temp_dir, '*.txt')):
         dataset = Dataset.import_from(temp_dir, 'imagenet_txt', env=dm_env)
     else:

cvat/apps/dataset_manager/formats/kitti.py

@@ -7,9 +7,10 @@
 
 from datumaro.components.dataset import Dataset
 from datumaro.plugins.kitti_format.format import KittiPath, write_label_map
+
 from pyunpack import Archive
 
-from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor, import_dm_annotations)
+from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor, detect_dataset, import_dm_annotations)
 from cvat.apps.dataset_manager.util import make_zip_archive
 
 from .transformations import MaskToPolygonTransformation, RotatedBoxesToPolygons
@@ -41,6 +42,7 @@ def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs
     if not osp.isfile(color_map_path):
         write_label_map(color_map_path, color_map)
 
+    detect_dataset(temp_dir, format_name='kitti', importer=dm_env.importers.get('kitti'))
     dataset = Dataset.import_from(temp_dir, format='kitti', env=dm_env)
     labels_meta = instance_data.meta[instance_data.META_FIELD]['labels']
     if 'background' not in [label['name'] for _, label in labels_meta]:

cvat/apps/dataset_manager/formats/labelme.py

@@ -6,7 +6,7 @@
 from datumaro.components.dataset import Dataset
 from pyunpack import Archive
 
-from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor,
+from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor, detect_dataset,
     import_dm_annotations)
 from cvat.apps.dataset_manager.formats.transformations import MaskToPolygonTransformation
 from cvat.apps.dataset_manager.util import make_zip_archive
@@ -26,6 +26,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     Archive(src_file.name).extractall(temp_dir)
 
+    detect_dataset(temp_dir, format_name='label_me', importer=dm_env.importers.get('label_me'))
     dataset = Dataset.import_from(temp_dir, 'label_me', env=dm_env)
     dataset = MaskToPolygonTransformation.convert_dataset(dataset, **kwargs)
     if load_data_callback is not None:

cvat/apps/dataset_manager/formats/lfw.py

@@ -6,7 +6,7 @@
 from datumaro.components.dataset import Dataset
 from pyunpack import Archive
 
-from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor,
+from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor, detect_dataset,
                                                 import_dm_annotations)
 from cvat.apps.dataset_manager.util import make_zip_archive
 
@@ -17,6 +17,7 @@
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     Archive(src_file.name).extractall(temp_dir)
 
+    detect_dataset(temp_dir, format_name='lfw', importer=dm_env.importers.get('lfw'))
     dataset = Dataset.import_from(temp_dir, 'lfw')
     if load_data_callback is not None:
         load_data_callback(dataset, instance_data)

cvat/apps/dataset_manager/formats/market1501.py

@@ -10,7 +10,7 @@
 from datumaro.components.dataset import Dataset
 from datumaro.components.extractor import ItemTransform
 
-from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor,
+from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor, detect_dataset,
     import_dm_annotations)
 from cvat.apps.dataset_manager.util import make_zip_archive
 
@@ -74,6 +74,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     zipfile.ZipFile(src_file).extractall(temp_dir)
 
+    detect_dataset(temp_dir, format_name='market1501', importer=dm_env.importers.get('market1501'))
     dataset = Dataset.import_from(temp_dir, 'market1501', env=dm_env)
     dataset.transform(AttrToLabelAttr, label='market-1501')
     if load_data_callback is not None:

cvat/apps/dataset_manager/formats/mask.py

@@ -6,7 +6,7 @@
 from datumaro.components.dataset import Dataset
 from pyunpack import Archive
 
-from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor,
+from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor, detect_dataset,
     import_dm_annotations)
 from cvat.apps.dataset_manager.util import make_zip_archive
 
@@ -32,6 +32,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     Archive(src_file.name).extractall(temp_dir)
 
+    detect_dataset(temp_dir, format_name='voc', importer=dm_env.importers.get('voc'))
     dataset = Dataset.import_from(temp_dir, 'voc', env=dm_env)
     dataset = MaskToPolygonTransformation.convert_dataset(dataset, **kwargs)
     if load_data_callback is not None:

cvat/apps/dataset_manager/formats/mot.py

@@ -6,7 +6,7 @@
 import datumaro as dm
 from pyunpack import Archive
 
-from cvat.apps.dataset_manager.bindings import GetCVATDataExtractor
+from cvat.apps.dataset_manager.bindings import GetCVATDataExtractor, detect_dataset
 from cvat.apps.dataset_manager.util import make_zip_archive
 
 from .registry import dm_env, exporter, importer
@@ -105,6 +105,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     Archive(src_file.name).extractall(temp_dir)
 
+    detect_dataset(temp_dir, format_name='mot_seq', importer=dm_env.importers.get('mot_seq'))
     dataset = dm.Dataset.import_from(temp_dir, 'mot_seq', env=dm_env)
     if load_data_callback is not None:
         load_data_callback(dataset, instance_data)

cvat/apps/dataset_manager/formats/mots.py

@@ -8,7 +8,7 @@
 from datumaro.components.extractor import ItemTransform
 from pyunpack import Archive
 
-from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor,
+from cvat.apps.dataset_manager.bindings import (GetCVATDataExtractor, detect_dataset,
     find_dataset_root, match_dm_item)
 from cvat.apps.dataset_manager.util import make_zip_archive
 
@@ -110,6 +110,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
 def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
     Archive(src_file.name).extractall(temp_dir)
 
+    detect_dataset(temp_dir, format_name='mots', importer=dm_env.importers.get('mots'))
     dataset = Dataset.import_from(temp_dir, 'mots', env=dm_env)
     dataset = MaskToPolygonTransformation.convert_dataset(dataset, **kwargs)
     if load_data_callback is not None: