bugfix RegexProcessor

h5rdmtoolbox/conventions/generate_utils.py

@@ -10,12 +10,20 @@ def get_regex_name():
 
 
 class RegexProcessor:
+    """Process regex validator.
+
+    Parameters
+    ----------
+    standard_attribute: Dict
+        Standard attribute dictionary, e.g. {'validator': 'regex(r"^[a-zA-Z0-9_]*$")'}
+    """
     def __init__(self, standard_attribute: Dict):
         regex_validator = get_regex_name()
         validator = standard_attribute['validator']
         self.name = regex_validator
-        match = re.search(r'regex\((.*?)\)', validator)
-        re_pattern = match.group(1)
+        re_pattern = validator.split('regex(', 1)[1].rsplit(')', 1)[0]
+        # match = re.search(r'regex\((.*?)\)', validator)
+        # re_pattern = match.group(1)
 
         self.standard_attribute = standard_attribute.copy()
         if re_pattern.startswith("r'") and re_pattern.endswith("'"):

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = h5rdmtoolbox
-version = 0.11.1
+version = 0.11.2
 author = Matthias Probst
 author_email = matthias.probst@kit.edu
 description = Supporting a FAIR Research Data lifecycle using Python and HDF5.

tests/conventions/test_toolbox_validators.py

@@ -36,6 +36,15 @@ class MyStr(BaseModel):
                 self.assertListEqual(['1', '2'], h5.keywords)
         cv.delete()
 
+    def test_validate_regex(self):
+        from h5rdmtoolbox.conventions.generate_utils import RegexProcessor
+
+        rp = RegexProcessor({'validator': 'regex(r"^[a-zA-Z0-9_]*$")'})
+        self.assertEqual('r"^[a-zA-Z0-9_]*$"', rp.re_pattern)
+
+        rp = RegexProcessor({'validator': '$regex(^[a-zA-Z].*(?<!\s)$)'})
+        self.assertEqual('^[a-zA-Z].*(?<!\s)$', rp.re_pattern)
+
     def test_validate_orcid(self):
         class Validator(BaseModel):
             orcid: toolbox_validators.orcid