Add a validator to the text field

backend/api/tests/test_tasks.py

@@ -27,6 +27,7 @@ class TestIngestClassificationData(TestIngestData):
     task = DOCUMENT_CLASSIFICATION
 
     def assert_examples(self, dataset):
+        self.assertEqual(Example.objects.count(), len(dataset))
         for text, expected_labels in dataset:
             example = Example.objects.get(text=text)
             labels = set(cat.label.text for cat in example.categories.all())
@@ -151,6 +152,7 @@ class TestIngestSequenceLabelingData(TestIngestData):
     task = SEQUENCE_LABELING
 
     def assert_examples(self, dataset):
+        self.assertEqual(Example.objects.count(), len(dataset))
         for text, expected_labels in dataset:
             example = Example.objects.get(text=text)
             labels = [[span.start_offset, span.end_offset, span.label.text] for span in example.spans.all()]
@@ -193,6 +195,7 @@ class TestIngestSeq2seqData(TestIngestData):
     task = SEQ2SEQ
 
     def assert_examples(self, dataset):
+        self.assertEqual(Example.objects.count(), len(dataset))
         for text, expected_labels in dataset:
             example = Example.objects.get(text=text)
             labels = set(text_label.text for text_label in example.texts.all())

backend/api/views/upload/data.py

@@ -1,7 +1,7 @@
 import abc
 from typing import Dict
 
-from pydantic import BaseModel
+from pydantic import BaseModel, validator
 
 
 class BaseData(BaseModel, abc.ABC):
@@ -19,6 +19,13 @@ class BaseData(BaseModel, abc.ABC):
 class TextData(BaseData):
     text: str
 
+    @validator('text')
+    def text_is_not_empty(cls, value: str):
+        if value:
+            return value
+        else:
+            raise ValueError('is not empty.')
+
 
 class FileData(BaseData):
     pass

backend/api/views/upload/dataset.py

@@ -5,10 +5,10 @@ import os
 from typing import Dict, Iterator, List, Optional, Type
 
 import chardet
-import pydantic.error_wrappers
 import pyexcel
 import pyexcel.exceptions
 from chardet.universaldetector import UniversalDetector
+from pydantic import ValidationError
 from seqeval.scheme import BILOU, IOB2, IOBES, IOE2, Tokens
 
 from .data import BaseData
@@ -67,7 +67,7 @@ class Dataset:
         for filename in self.filenames:
             try:
                 yield from self.load(filename)
-            except UnicodeDecodeError as err:
+            except (UnicodeDecodeError, ValidationError) as err:
                 message = str(err)
                 raise FileParseException(filename, line_num=-1, message=message)
 
@@ -113,9 +113,15 @@ class Dataset:
         label = [label] if isinstance(label, str) else label
         try:
             label = [self.label_class.parse(o) for o in label]
-        except (pydantic.error_wrappers.ValidationError, TypeError):
+        except (ValidationError, TypeError):
             label = []
-        data = self.data_class.parse(text=text, filename=filename, meta=row)
+
+        try:
+            data = self.data_class.parse(text=text, filename=filename, meta=row)
+        except ValidationError:
+            message = 'The empty text is not allowed.'
+            raise FileParseException(filename, line_num, message)
+
         record = Record(data=data, label=label)
         return record