RDISCROWD-7779 - add duplicate field settings (#1033)

* add option to enable/disable check
* Update github-actions.yml
   * remove coverage for now. "--no-fail" option is not working.

.github/workflows/github-actions.yml

@@ -74,9 +74,3 @@ jobs:
           echo "::remove-matcher owner=python::"
           ./run_tests
 
-      - name: Run code coverage
-        run: |
-          python -m pip install coveralls
-          coveralls --service=github
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

pybossa/forms/fields/select_two.py

@@ -24,4 +24,10 @@ class Select2Field(SelectMultipleField):
     type.  The rendering logic must be handled in the client code which includes
     adding the select2 source files.
     '''
-    pass
+
+    def pre_validate(self, form):
+        if self.validate_choice and self.data:
+            values = list(c[0] for c in self.choices)
+            for d in self.data:
+                if d not in values:
+                    raise ValueError(self.gettext("'%(value)s' is not a valid choice for this field") % dict(value=d))

pybossa/forms/forms.py

@@ -148,6 +148,9 @@ class ProjectUpdateForm(ProjectForm):
     private_instance = data_access.data_access_levels.get("valid_access_levels", []) == ["L1", "L2", "L3", "L4"]
     if not private_instance:
         allow_taskrun_edit = BooleanField(lazy_gettext('Allow Editing of Task Submissions'))
+    duplicate_task_check_duplicate_fields = Select2Field(lazy_gettext('Task.info field names to be included in duplication check'), choices=[], validate_choice=False)
+    duplicate_task_check_completed_tasks = BooleanField(lazy_gettext('Include completed tasks in task duplication check'))
+    enable_duplicate_task_check = BooleanField(lazy_gettext('Enable task duplication check'))
     zip_download = BooleanField(lazy_gettext('Allow ZIP data download'))
     category_id = SelectField(lazy_gettext('Category'), coerce=int)
     hidden = BooleanField(lazy_gettext('Hide?'))
@@ -164,6 +167,9 @@ class ProjectUpdateForm(ProjectForm):
                         [pb_validator.Webhook()])
     sync_enabled = BooleanField(lazy_gettext('Enable Project Syncing'))
 
+    def set_duplicate_task_check_duplicate_fields_options(self, options):
+        self.duplicate_task_check_duplicate_fields.choices = options
+
 class AnnotationForm(Form):
     dataset_description = TextAreaField(lazy_gettext('Dataset Description'))
     provider = create_nullable_select(

pybossa/view/projects.py

@@ -810,7 +810,6 @@ def delete(short_name):
 @blueprint.route('/<short_name>/update', methods=['GET', 'POST'])
 @login_required
 def update(short_name):
-
     sync_enabled = current_app.config.get('SYNC_ENABLED')
     project, owner, ps = project_by_shortname(short_name)
 
@@ -847,6 +846,13 @@ def handle_valid_form(form):
             'input_data': form.input_data_class.data,
             'output_data': form.output_data_class.data
         }
+        if form.enable_duplicate_task_check.data:
+            new_project.info['duplicate_task_check'] = {
+                'duplicate_fields': form.duplicate_task_check_duplicate_fields.data,
+                'completed_tasks': form.duplicate_task_check_completed_tasks.data
+            }
+        else:
+            new_project.info.pop('duplicate_task_check', None)
         if "allow_taskrun_edit" in form:
             new_project.info["allow_taskrun_edit"] = form.allow_taskrun_edit.data
 
@@ -880,6 +886,10 @@ def handle_valid_form(form):
         project.input_data_class = project.info.get('data_classification', {}).get('input_data')
         project.output_data_class = project.info.get('data_classification', {}).get('output_data')
         project.allow_taskrun_edit = project.info.get("allow_taskrun_edit", False)
+        duplicate_task_check_duplicate_fields = project.info.get("duplicate_task_check", {}).get("duplicate_fields", [])
+        project.duplicate_task_check_duplicate_fields = duplicate_task_check_duplicate_fields
+        project.duplicate_task_check_completed_tasks = project.info.get("duplicate_task_check", {}).get("completed_tasks", False)
+        project.enable_duplicate_task_check = True if project.info.get("duplicate_task_check") else False
         ensure_amp_config_applied_to_project(project, project.info.get('annotation_config', {}))
         form = dynamic_project_form(ProjectUpdateForm, None, data_access_levels, obj=project,
                                     products=prodsubprods, data_classes=data_classes)
@@ -891,6 +901,7 @@ def handle_valid_form(form):
         if project.category_id is None:
             project.category_id = categories[0].id
         form.populate_obj(project)
+        form.set_duplicate_task_check_duplicate_fields_options([(field, field) for field in duplicate_task_check_duplicate_fields])
 
 
     if request.method == 'POST':

test/test_view/test_project_settings.py

@@ -2,14 +2,19 @@
 import json
 from unittest.mock import patch
 
-from test import with_context
+from test import db, with_context
 from test.factories import ProjectFactory
 from test.helper import web
 from pybossa import data_access
+from pybossa.repositories import ProjectRepository
 
 
 class TestProjectSettings(web.Helper):
 
+    def setUp(self):
+        super(TestProjectSettings, self).setUp()
+        self.project_repo = ProjectRepository(db)
+
     @staticmethod
     def patched_levels(**kwargs):
         patch_data_access_levels = dict(
@@ -22,10 +27,10 @@ def patched_levels(**kwargs):
         )
         patch_data_access_levels.update(kwargs)
         return patch_data_access_levels
-    
+
     @with_context
     def test_project_update_amp_store(self):
-        with patch.dict(data_access.data_access_levels, self.patched_levels()):        
+        with patch.dict(data_access.data_access_levels, self.patched_levels()):
             project = ProjectFactory.create(
                 published=True,
                 info={
@@ -39,3 +44,19 @@ def test_project_update_amp_store(self):
             res = self.app_get_json(url)
             data = json.loads(res.data)
             assert data['form']['amp_store'], 'opt-in amp store should be checked'
+
+    @with_context
+    def test_project_get_request_update_duplicate_tasks_setting(self):
+        project = ProjectFactory.create(
+            published=True,
+            info={
+                'duplicate_task_check': {
+                    'duplicate_fields': ["company_name", "company_website"],
+                    'completed_tasks': True
+                }
+            })
+        url = f'/project/{project.short_name}/update?api_key={project.owner.api_key}'
+        res = self.app_get_json(url)
+        data = json.loads(res.data)
+        assert data['form']['duplicate_task_check_completed_tasks'] == True, data['form']
+        assert data['form']['duplicate_task_check_duplicate_fields'] == ['company_name', 'company_website'], data['form']