updated processing of exceptions to be a bit more flexible and added …

…test cases
4dn-dcic · Aug 23, 2023 · 89dd17d · 89dd17d
1 parent cfd4ba7
commit 89dd17d
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 25 deletions.
diff --git a/tests/test_import_data.py b/tests/test_import_data.py
@@ -280,21 +280,32 @@ def test_combine_set_expsets_with_existing():
 
 
 def test_error_report(connection_mock):
-    # There are 3 errors, 2 of them are legit, one needs to be checked afains the all aliases list, and excluded
-    err_dict = {"title": "Unprocessable Entity",
-                "status": "error",
-                "errors": [
-                    {"name": "protocol_documents",
-                     "description": "'dcic:insituhicagar' not found", "location": "body"},
-                    {"name": "age",
-                     "description": "'at' is not of type 'number'", "location": "body"},
-                    {"name": "sex",
-                     "description": "'green' is not one of ['male', 'female', 'unknown', 'mixed']", "location": "body"}],
-                "code": 422,
-                "@type": ["ValidationFailure", "Error"],
-                "description": "Failed validation"}
-    rep = imp.error_report(err_dict, "Vendor", ['dcic:insituhicagar'], connection_mock)
+    # There are x errors, x of them are legit,  need to be checked against the all aliases list, and excluded
+    err_dict = {
+                    "title": "Unprocessable Entity",
+                    "status": "error",
+                    "errors": [
+                        {"name": "Schema: ", "location": "body",
+                         "description": "Unable to resolve link: siyuan-wang-lab:region_1MB_TAD_1"},
+                        {"name": "Schema: ", "location": "body",
+                         "description": "Unable to resolve link: siyuan-wang-lab:region_5MB_TAD_2"},
+                        {"location": "body", "name": "Schema: genome_location.1",
+                         "description": "'siyuan-wang-lab:region_5MB_TAD_2' not found"},
+                        {"name": "protocol_documents",
+                         "description": "'dcic:insituhicagar' not found", "location": "body"},
+                        {"name": "age",
+                         "description": "'at' is not of type 'number'", "location": "body"},
+                        {"name": "sex",
+                         "description": "'green' is not one of ['male', 'female', 'unknown', 'mixed']", "location": "body"}
+                    ],
+                    "code": 422,
+                    "@type": ["ValidationFailure", "Error"],
+                    "description": "Failed validation"
+                }
+    rep = imp.error_report(err_dict, "Vendor", ['dcic:insituhicagar', 'siyuan-wang-lab:region_1MB_TAD_1'], connection_mock)
     message = '''
+ERROR vendor                  Field 'Schema: ': Unable to resolve link: siyuan-wang-lab:region_5MB_TAD_2
+ERROR vendor                  Field 'Schema: genome_location.1': 'siyuan-wang-lab:region_5MB_TAD_2' not found
 ERROR vendor                  Field 'age': 'at' is not of type 'number'
 ERROR vendor                  Field 'sex': 'green' is not one of ['male', 'female', 'unknown', 'mixed']
 '''

diff --git a/wranglertools/import_data.py b/wranglertools/import_data.py
@@ -779,23 +779,29 @@ def error_report(error_dic, sheet, all_aliases, connection, error_id=''):
                     'title': 'Unprocessable Entity', 'description': 'Failed validation'}
     report = []
     if all(item in error_dic.items() for item in error_header.items()):
-        for err in error_dic['errors']:
-            error_description = err['description']
-            # if no field specified in the error, schema wide error
-            if not err['name']:
+        # deal with Validation errors
+        for err in error_dic.get('errors'):
+            error_description = err.get('description')
+            # this may no longer ever happen?
+            if 'name' not in err or not err.get('name'):
                 report.append("{sheet:<30}{des}"
                               .format(des=error_description, sheet="ERROR " + sheet.lower()))
             else:
-                # field errors
-                not_found = None
+                # deal with errors about linked objects not in db - checking for those with
+                # aliases present in the workbook that should be ignored
                 utrl_txt = 'Unable to resolve link:'
-                if utrl_txt in error_description:
+                nf_txt = 'not found'
+                not_found = None
+                alias_bit = None
+                if error_id:
+                    alias_bit = error_id
+                elif utrl_txt in error_description:
                     alias_bit = error_description.replace(utrl_txt, '')
+                elif error_description.endswith(nf_txt):
+                    alias_bit = error_description.replace(nf_txt, '').replace("'", '')
+                if alias_bit:
                     not_found = alias_bit.strip()
-                elif error_description[-9:] == 'not found':
-                    # if error is about object connections, check all aliases
-                    # ignore ones about existing aliases
-                    not_found = error_description[1:-11]
+                # ignore ones about existing aliases
                 if not_found and not_found in all_aliases:
                     continue
                 error_field = err['name']