fix: LEAP-1352: API for Bulk Annotation Configuration Validation (#6757)

Co-authored-by: hlomzik <[email protected]> Co-authored-by: MihajloHoma <[email protected]> Co-authored-by: triklozoid <[email protected]>
HumanSignal · Dec 13, 2024 · 769ced2 · 769ced2
1 parent a900be2
commit 769ced2
Show file tree

Hide file tree

Showing 2 changed files with 153 additions and 1 deletion.
diff --git a/label_studio/projects/serializers.py b/label_studio/projects/serializers.py
@@ -3,6 +3,30 @@
 import bleach
 from constants import SAFE_HTML_ATTRIBUTES, SAFE_HTML_TAGS
 from django.db.models import Q
+from label_studio_sdk.label_interface import LabelInterface
+from label_studio_sdk.label_interface.control_tags import (
+    BrushLabelsTag,
+    BrushTag,
+    ChoicesTag,
+    DateTimeTag,
+    EllipseLabelsTag,
+    EllipseTag,
+    HyperTextLabelsTag,
+    KeyPointLabelsTag,
+    KeyPointTag,
+    LabelsTag,
+    NumberTag,
+    ParagraphLabelsTag,
+    PolygonLabelsTag,
+    PolygonTag,
+    RatingTag,
+    RectangleLabelsTag,
+    RectangleTag,
+    TaxonomyTag,
+    TextAreaTag,
+    TimeSeriesLabelsTag,
+    VideoRectangleTag,
+)
 from projects.models import Project, ProjectImport, ProjectOnboarding, ProjectReimport, ProjectSummary
 from rest_flex_fields import FlexFieldsModelSerializer
 from rest_framework import serializers
@@ -66,6 +90,9 @@ class ProjectSerializer(FlexFieldsModelSerializer):
     config_has_control_tags = SerializerMethodField(
         default=None, read_only=True, help_text='Flag to detect is project ready for labeling'
     )
+    config_suitable_for_bulk_annotation = serializers.SerializerMethodField(
+        default=None, read_only=True, help_text='Flag to detect is project ready for bulk annotation'
+    )
     finished_task_number = serializers.IntegerField(default=None, read_only=True, help_text='Finished tasks')
 
     queue_total = serializers.SerializerMethodField()
@@ -82,6 +109,61 @@ def user_id(self):
     def get_config_has_control_tags(project):
         return len(project.get_parsed_config()) > 0
 
+    @staticmethod
+    def get_config_suitable_for_bulk_annotation(project):
+        li = LabelInterface(project.label_config)
+
+        # List of tags that should not be present
+        disallowed_tags = [
+            LabelsTag,
+            BrushTag,
+            BrushLabelsTag,
+            EllipseTag,
+            EllipseLabelsTag,
+            KeyPointTag,
+            KeyPointLabelsTag,
+            PolygonTag,
+            PolygonLabelsTag,
+            RectangleTag,
+            RectangleLabelsTag,
+            HyperTextLabelsTag,
+            ParagraphLabelsTag,
+            TimeSeriesLabelsTag,
+            VideoRectangleTag,
+        ]
+
+        # Return False if any disallowed tag is present
+        for tag_class in disallowed_tags:
+            if li.find_tags_by_class(tag_class):
+                return False
+
+        # Check perRegion/perItem for expanded list of tags, plus value="no" for Choices/Taxonomy
+        allowed_tags_for_checks = [ChoicesTag, TaxonomyTag, DateTimeTag, NumberTag, RatingTag, TextAreaTag]
+        for tag_class in allowed_tags_for_checks:
+            tags = li.find_tags_by_class(tag_class)
+            for tag in tags:
+                per_region = tag.attr.get('perRegion', 'false').lower() == 'true'
+                per_item = tag.attr.get('perItem', 'false').lower() == 'true'
+                if per_region or per_item:
+                    return False
+                # For ChoicesTag and TaxonomyTag, the value attribute must not be set at all
+                if tag_class in [ChoicesTag, TaxonomyTag]:
+                    if 'value' in tag.attr:
+                        return False
+
+        # For TaxonomyTag, check labeling and apiUrl
+        taxonomy_tags = li.find_tags_by_class(TaxonomyTag)
+        for tag in taxonomy_tags:
+            labeling = tag.attr.get('labeling', 'false').lower() == 'true'
+            if labeling:
+                return False
+            api_url = tag.attr.get('apiUrl', None)
+            if api_url is not None:
+                return False
+
+        # If all checks pass, return True
+        return True
+
     @staticmethod
     def get_parsed_label_config(project):
         return project.get_parsed_config()
@@ -156,6 +238,7 @@ class Meta:
             'finished_task_number',
             'queue_total',
             'queue_done',
+            'config_suitable_for_bulk_annotation',
         ]
 
     def validate_label_config(self, value):

diff --git a/label_studio/tests/config_validation.tavern.yml b/label_studio/tests/config_validation.tavern.yml
@@ -1435,4 +1435,73 @@ stages:
     method: POST
     url: '{django_live_url}/api/projects/{pk}/validate'
   response:
-    status_code: 200
+    status_code: 200
+
+---
+test_name: check_config_suitable_for_bulk_annotation
+strict: false
+marks:
+- usefixtures:
+  - django_live_url
+stages:
+
+- id: signup
+  type: ref
+
+- name: create classification project
+  request:
+    data:
+      label_config: |
+        <View>
+          <Text name="text" value="$text"/>
+          <Choices name="sentiment" toName="text" choice="single">
+            <Choice value="Positive"/>
+            <Choice value="Neutral"/>
+            <Choice value="Negative"/>
+          </Choices>
+        </View>
+    method: POST
+    url: '{django_live_url}/api/projects'
+  response:
+    status_code: 201
+    save:
+      json:
+        classification_project_id: id
+
+- name: check classification project property
+  request:
+    method: GET
+    url: '{django_live_url}/api/projects/{classification_project_id}'
+  response:
+    status_code: 200
+    json:
+      config_suitable_for_bulk_annotation: true
+
+- name: create object detection project
+  request:
+    data:
+      label_config: |
+        <View>
+          <Image name="image" value="$image"/>
+          <RectangleLabels name="label" toName="image">
+            <Label value="Car"/>
+            <Label value="Tree"/>
+            <Label value="Person"/>
+          </RectangleLabels>
+        </View>
+    method: POST
+    url: '{django_live_url}/api/projects'
+  response:
+    status_code: 201
+    save:
+      json:
+        detection_project_id: id
+
+- name: check object detection project property
+  request:
+    method: GET
+    url: '{django_live_url}/api/projects/{detection_project_id}'
+  response:
+    status_code: 200
+    json:
+      config_suitable_for_bulk_annotation: false