workflows: add uk curation step

inspirehep/modules/workflows/tasks/actions.py

@@ -1131,6 +1131,15 @@ def check_if_france_in_raw_affiliations(obj, eng):
             return True
 
 
+def check_if_core_and_uk_in_fulltext(obj, eng):
+    fulltext = get_fulltext(obj)
+    if not fulltext or not is_core(obj, eng):
+        return
+    regex = re.compile(
+        r"\b(UK|United\s+Kingdom|England|Scotland|Northern\s+Ireland)\b", re.UNICODE | re.IGNORECASE)
+    return regex.search(fulltext)
+
+
 def load_record_from_hep(obj, wf):
     control_number = obj.data['control_number']
     pid_type = get_pid_type_from_schema(obj.data['$schema'])

inspirehep/modules/workflows/workflows/article.py

@@ -76,7 +76,8 @@
     create_core_selection_wf,
     check_if_france_in_fulltext,
     check_if_france_in_raw_affiliations,
-    link_institutions_with_affiliations
+    link_institutions_with_affiliations,
+    check_if_core_and_uk_in_fulltext
 )
 
 from inspirehep.modules.workflows.tasks.classifier import (
@@ -261,15 +262,26 @@
         is_marked('is-update'),
         IF_ELSE(
             is_arxiv_paper,
-            IF(
-                check_if_france_in_fulltext,
-                create_ticket(
-                    template='literaturesuggest/tickets/curation_core.html',
-                    queue='HAL_curation',
-                    context_factory=curation_ticket_context,
-                    ticket_id_key='curation_ticket_id',
+            [
+                IF(
+                    check_if_france_in_fulltext,
+                    create_ticket(
+                        template='literaturesuggest/tickets/curation_core.html',
+                        queue='HAL_curation',
+                        context_factory=curation_ticket_context,
+                        ticket_id_key='curation_ticket_id',
+                    ),
                 ),
-            ),
+                IF(
+                    check_if_core_and_uk_in_fulltext,
+                    create_ticket(
+                        template='literaturesuggest/tickets/curation_core.html',
+                        queue='UK_curation',
+                        context_factory=curation_ticket_context,
+                        ticket_id_key='curation_ticket_id',
+                    ),
+                )
+            ],
             IF(
                 check_if_france_in_raw_affiliations,
                 create_ticket(

tests/integration/workflows/test_article_workflow.py

@@ -152,6 +152,10 @@ def test_create_ticket_when_source_is_not_publishing(
     "inspirehep.modules.workflows.tasks.actions.check_if_france_in_fulltext",
     return_value=False,
 )
+@mock.patch(
+    "inspirehep.modules.workflows.tasks.actions.check_if_core_and_uk_in_fulltext",
+    return_value=False,
+)
 @mock.patch("inspirehep.modules.workflows.tasks.submission.send_robotupload")
 def test_set_fermilab_collection_from_report_number(
     mocked_api_request_magpie,

tests/unit/workflows/test_workflows_actions.py

@@ -38,7 +38,7 @@
 
 from inspirehep.modules.workflows.tasks.actions import jlab_ticket_needed, load_from_source_data, \
     extract_authors_from_pdf, is_suitable_for_pdf_authors_extraction, is_fermilab_report, add_collection, \
-    check_if_france_in_fulltext, check_if_france_in_raw_affiliations
+    check_if_france_in_fulltext, check_if_france_in_raw_affiliations, check_if_core_and_uk_in_fulltext
 
 
 def test_match_approval_gets_match_recid():
@@ -579,3 +579,81 @@ def test_check_if_france_in_fulltext_when_france_in_text_body(mocked_get_documen
                 france_in_fulltext = check_if_france_in_fulltext(obj, eng)
 
     assert france_in_fulltext
+
+
+@patch("inspirehep.modules.workflows.tasks.actions.get_document_in_workflow")
+def test_check_if_uk_in_fulltext_not_core(mocked_get_document, app):
+    fake_grobid_response = "<country key=\"UK\">England</country>"
+    obj = MagicMock()
+    obj.data = {
+        'core': False
+    }
+    obj.extra_data = {}
+    eng = None
+    new_config = {"GROBID_URL": "http://grobid_url.local"}
+    with patch.dict(current_app.config, new_config):
+        with requests_mock.Mocker() as requests_mocker:
+            requests_mocker.register_uri(
+                'POST', 'http://grobid_url.local/api/processFulltextDocument',
+                text=fake_grobid_response,
+                headers={'content-type': 'application/xml'},
+                status_code=200,
+            )
+            with tempfile.NamedTemporaryFile() as tmp_file:
+                mocked_get_document.return_value.__enter__.return_value = tmp_file.name
+                uk_in_fulltext_and_core = check_if_core_and_uk_in_fulltext(
+                    obj, eng)
+
+    assert not uk_in_fulltext_and_core
+
+
+@patch("inspirehep.modules.workflows.tasks.actions.get_document_in_workflow")
+def test_check_if_uk_in_fulltext_core(mocked_get_document, app):
+    fake_grobid_response = "<country key=\"UK\">England</country>"
+    obj = MagicMock()
+    obj.data = {
+        'core': True
+    }
+    obj.extra_data = {}
+    eng = None
+    new_config = {"GROBID_URL": "http://grobid_url.local"}
+    with patch.dict(current_app.config, new_config):
+        with requests_mock.Mocker() as requests_mocker:
+            requests_mocker.register_uri(
+                'POST', 'http://grobid_url.local/api/processFulltextDocument',
+                text=fake_grobid_response,
+                headers={'content-type': 'application/xml'},
+                status_code=200,
+            )
+            with tempfile.NamedTemporaryFile() as tmp_file:
+                mocked_get_document.return_value.__enter__.return_value = tmp_file.name
+                uk_in_fulltext_and_core = check_if_core_and_uk_in_fulltext(
+                    obj, eng)
+
+    assert uk_in_fulltext_and_core
+
+
+@patch("inspirehep.modules.workflows.tasks.actions.get_document_in_workflow")
+def test_check_if_uk_in_fulltext_core_case_insensitive(mocked_get_document, app):
+    fake_grobid_response = "<country>unitEd KiNgdOm</country>"
+    obj = MagicMock()
+    obj.data = {
+        'core': True
+    }
+    obj.extra_data = {}
+    eng = None
+    new_config = {"GROBID_URL": "http://grobid_url.local"}
+    with patch.dict(current_app.config, new_config):
+        with requests_mock.Mocker() as requests_mocker:
+            requests_mocker.register_uri(
+                'POST', 'http://grobid_url.local/api/processFulltextDocument',
+                text=fake_grobid_response,
+                headers={'content-type': 'application/xml'},
+                status_code=200,
+            )
+            with tempfile.NamedTemporaryFile() as tmp_file:
+                mocked_get_document.return_value.__enter__.return_value = tmp_file.name
+                uk_in_fulltext_and_core = check_if_core_and_uk_in_fulltext(
+                    obj, eng)
+
+    assert uk_in_fulltext_and_core