Add test cases for ingesting seq2seq data

backend/api/tests/data/seq2seq/example.csv

@@ -0,0 +1,5 @@
+text,label,meta
+AAA,LabelA,
+BBB,LabelB,MetaB
+CCC,LabelC
+DDD,,MetaD

backend/api/tests/data/seq2seq/example.json

@@ -0,0 +1,5 @@
+[
+  {"text": "example", "label": ["example1", "example2"]},
+  {"text": "example", "label": ["example"]},
+  {"text": "example", "label": ["example"]}
+]

backend/api/tests/data/seq2seq/example.jsonl

@@ -1,3 +1,3 @@
-{"text": "example", "labels": ["example1", "example2"], "meta": {"wikiPageID": 1}}
-{"text": "example", "labels": ["example"], "meta": {"wikiPageID": 2}}
-{"text": "example", "labels": ["example"], "meta": {"wikiPageID": 3}}
+{"text": "example", "label": ["example1", "example2"], "meta": {"wikiPageID": 1}}
+{"text": "example", "label": ["example"], "meta": {"wikiPageID": 2}}
+{"text": "example", "label": ["example"], "meta": {"wikiPageID": 3}}

backend/api/tests/test_tasks.py

@@ -2,7 +2,8 @@ import pathlib
 
 from django.test import TestCase
 
-from ..models import Category, Example, Label, Span
+from ..models import (DOCUMENT_CLASSIFICATION, SEQ2SEQ, SEQUENCE_LABELING,
+                      Category, Example, Label, Span, TextLabel)
 from ..tasks import injest_data
 from .api.utils import prepare_project
 
@@ -32,7 +33,7 @@ class TestIngestData(TestCase):
 
 
 class TestIngestClassificationData(TestIngestData):
-    task = 'DocumentClassification'
+    task = DOCUMENT_CLASSIFICATION
     annotation_class = Category
 
     def test_jsonl(self):
@@ -73,7 +74,7 @@ class TestIngestClassificationData(TestIngestData):
 
 
 class TestIngestSequenceLabelingData(TestIngestData):
-    task = 'SequenceLabeling'
+    task = SEQUENCE_LABELING
     annotation_class = Span
 
     def test_jsonl(self):
@@ -85,3 +86,23 @@ class TestIngestSequenceLabelingData(TestIngestData):
         filename = 'sequence_labeling/example.conll'
         file_format = 'CoNLL'
         self.assert_count(filename, file_format, expected_example=3, expected_label=2, expected_annotation=5)
+
+
+class TestIngestSeq2seqData(TestIngestData):
+    task = SEQ2SEQ
+    annotation_class = TextLabel
+
+    def test_jsonl(self):
+        filename = 'seq2seq/example.jsonl'
+        file_format = 'JSONL'
+        self.assert_count(filename, file_format, expected_example=3, expected_label=0, expected_annotation=4)
+
+    def test_json(self):
+        filename = 'seq2seq/example.json'
+        file_format = 'JSON'
+        self.assert_count(filename, file_format, expected_example=3, expected_label=0, expected_annotation=4)
+
+    def test_csv(self):
+        filename = 'seq2seq/example.csv'
+        file_format = 'CSV'
+        self.assert_count(filename, file_format, expected_example=4, expected_label=0, expected_annotation=3)

backend/api/views/upload/dataset.py

@@ -112,7 +112,7 @@ 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:
+        except (pydantic.error_wrappers.ValidationError, TypeError):
             label = []
         data = self.data_class.parse(text=text, filename=filename, meta=row)
         record = Record(data=data, label=label)

backend/api/views/upload/label.py

@@ -92,10 +92,10 @@ class TextLabel(Label):
 
     @classmethod
     def parse(cls, obj: Any):
-        if isinstance(obj, str):
+        if isinstance(obj, str) and obj:
             return cls(text=obj)
         else:
-            raise TypeError(f'{obj} is not str.')
+            raise TypeError(f'{obj} is not str or empty.')
 
     def replace(self, mapping: Dict[str, str]) -> 'Label':
         return self