Merge pull request #188 from uc-cdis/chore/update-vlmd-validation

Chore/update vlmd validation

.secrets.baseline

@@ -1,9 +1,9 @@
 {
   "exclude": {
-    "files": "^.secrets.baseline$",
+    "files": null,
     "lines": null
   },
-  "generated_at": "2024-06-06T20:34:12Z",
+  "generated_at": "2024-06-17T22:02:51Z",
   "plugins_used": [
     {
       "name": "AWSKeyDetector"
@@ -468,17 +468,17 @@
     ],
     "vlmd-submission-tools/poetry.lock": [
       {
-        "hashed_secret": "5b240644452ed40dfe194673b7db6b641971c720",
+        "hashed_secret": "e1df343623dcc5d44e3a7da0e09ee4b0c980b52f",
         "is_verified": false,
-        "line_number": 1221,
+        "line_number": 1519,
         "type": "Hex High Entropy String"
       }
     ],
     "vlmd-submission-tools/tests/test_common_utils.py": [
       {
         "hashed_secret": "8318df9ecda039deac9868adf1944a29a95c7114",
         "is_verified": false,
-        "line_number": 100,
+        "line_number": 102,
         "type": "Secret Keyword"
       }
     ],
@@ -494,7 +494,7 @@
       {
         "hashed_secret": "8318df9ecda039deac9868adf1944a29a95c7114",
         "is_verified": false,
-        "line_number": 129,
+        "line_number": 163,
         "type": "Secret Keyword"
       }
     ]

vlmd-submission-tools/pyproject.toml

@@ -28,6 +28,8 @@ frictionless = "^5.12.1"
 parameterized = "^0.9.0"
 pytest = "^7.4.2"
 pytest-cov = "^4.1.0"
+healdata-utils = "^0.5.1"
+requests-mock = "^1.12.1"
 
 [tool.poetry.dev-dependencies]
 

vlmd-submission-tools/tests/templates/template_submission_invalid.csv

@@ -0,0 +1,8 @@
+name,description,type
+participant_id,Unique identifier for participant,character
+race,Self-reported race,integer
+,,integer
+hispanic,"Are you of Hispanic, Latino, or Spanish origin?",boolean
+sex_at_birth,The self-reported sex of the participant/subject at birth,
+SU4,During the past 30 days how many days did you use heroin (alone or mixed with other drugs)? ] [Write 0 days if no use],integer
+pulse_rate,Heart rate measured at systemic artery,number

vlmd-submission-tools/tests/templates/template_submission_invalid.json

@@ -0,0 +1,16 @@
+{
+    "title": "Example VLMD",
+    "description": "This is an example",
+    "fields": [
+        {
+            "name": "participant_id",
+            "description": "Unique identifier for participant",
+            "type": "character"
+        },
+        {
+            "name": "race",
+            "description": "Self-reported race",
+            "type": "integer"
+        }
+    ]
+}

vlmd-submission-tools/tests/templates/template_submission_invalid.tsv

@@ -0,0 +1,7 @@
+name	title	description	type
+participant_id	Participant Id	Unique identifier for participant	character
+race	Race	Self-reported race	integer
+age	Age	What is your age? (age at enrollment)	integer
+hispanic	"Hispanic	 Latino	 or Spanish Origin"	"Are you of Hispanic	 Latino	 or Spanish origin?"	boolean
+sex_at_birth	Sex at Birth	The self-reported sex of the participant/subject at birth	string
+SU4	Heroin Days Used	During the past 30 days how many days did you use heroin (alone or mixed with other drugs)? ] [Write 0 days if no use]	integer

vlmd-submission-tools/tests/templates/template_submission_small.csv

@@ -0,0 +1,3 @@
+module,name,title,description,type,format,
+,participant_id,Participant Id,Unique identifier for participant,string,,
+,race,Race,Self-reported race,integer,,

vlmd-submission-tools/tests/templates/template_submission_small.json

@@ -0,0 +1,18 @@
+{
+    "title": "Example VLMD",
+    "description": "This is an example",
+    "fields": [
+        {
+            "name": "participant_id",
+            "title": "Participant Id",
+            "description": "Unique identifier for participant",
+            "type": "string"
+        },
+        {
+            "name": "race",
+            "title": "Race",
+            "description": "Self-reported race",
+            "type": "integer"
+        }
+    ]
+}

vlmd-submission-tools/tests/templates/template_submission_small.tsv

@@ -0,0 +1,3 @@
+name	title	description	type
+participant_id	Participant Id	Unique identifier for participant	string
+race	Race	Self-reported race	integer

vlmd-submission-tools/tests/test_common_utils.py

@@ -43,14 +43,16 @@ def test_check_mds_study_id(self, mocked_post):
         hostname = "mycommons.planx-pla.net"
         study_id = "my_study_id"
         expected_data_dictionaries = {
-            "my first dictionary": "guid1",
-            "my second dictionary": "guid2"
+            "data_dictionaries" : {
+                "my first dictionary": "guid1",
+                "my second dictionary": "guid2"
+            }
         }
         mock_mds_response = MagicMock(requests.Response)
         mock_mds_response.status_code = 200
         mock_mds_response.json.return_value = {
             "_guid_type": "discovery_metadata",
-            "data_dictionaries": expected_data_dictionaries
+            "variable_level_metadata": expected_data_dictionaries
         }
         mocked_post.return_value = mock_mds_response