Merge pull request #12 from mgcam/improve_tag_validation

Added validation for multiple tags.
wtsi-npg · Apr 27, 2023 · 8ae6d78 · 8ae6d78
2 parents 607d5a2 + ffb5440
commit 8ae6d78
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 9 deletions.
diff --git a/npg_id_generation/pac_bio.py b/npg_id_generation/pac_bio.py
@@ -78,7 +78,7 @@ def well_label_conforms_to_pattern(cls, v):
 
     @validator("tags")
     def tags_have_correct_characters(cls, v):
-        if (v is not None) and (not re.match("^[ACGT,]+$", v)):
+        if (v is not None) and (not re.match("^[ACGT]+(,[ACGT]+)*$", v)):
             raise ValueError(
                 "Tags should be a comma separated list of uppercase DNA sequences"
             )

diff --git a/tests/test_hashing.py b/tests/test_hashing.py
@@ -108,21 +108,25 @@ def test_well_label_conforms_to_pattern():
 
 
 def test_tags_have_correct_characters():
-    bad_tags = ["ABCD", "ACGT.AGTC", " ACGT", "ACGT ", "acgt"]
+    bad_tags = [
+        "ABCD",
+        "ACGT.AGTC",
+        " ACGT",
+        "ACGT ",
+        "acgt",
+        ",",
+        " ACCTG ",
+        "ACCTG,",
+        ",ACCTG",
+        "ACCTG,,GGTAC",
+    ]
     for tag in bad_tags:
         with pytest.raises(ValidationError) as excinfo:
             PacBioEntity(run_name="MARATHON", well_label="A1", tags=tag)
         assert (
             "Tags should be a comma separated list of uppercase DNA sequences"
             in str(excinfo.value)
         )
-    with pytest.raises(ValidationError) as excinfo:
-        PacBioEntity.parse_raw(
-            '{"run_name":"MARATHON", "well_label":"A1", "tags":"ABCD"}'
-        )
-    assert "Tags should be a comma separated list of uppercase DNA sequences" in str(
-        excinfo.value
-    )
 
 
 def test_expected_hashes():