Raise error for regex operator

Author: jeromekelleher

tests/test_filter.py

@@ -11,6 +11,10 @@
 
 
 class TestFilterExpressionParser:
+    @pytest.fixture()
+    def parser(self):
+        return filter_mod.make_bcftools_filter_parser(map_vcf_identifiers=False)
+
     @pytest.mark.parametrize(
         "expression",
         [
@@ -20,11 +24,20 @@ class TestFilterExpressionParser:
             '"stri + 2',
         ],
     )
-    def test_invalid_expressions(self, expression):
-        parser = filter_mod.make_bcftools_filter_parser(map_vcf_identifiers=False)
+    def test_invalid_expressions(self, parser, expression):
         with pytest.raises(pp.ParseException):
             parser.parse_string(expression, parse_all=True)
 
+    @pytest.mark.parametrize(
+        ("expression", "exception_class"),
+        [
+            ('INFO/HAYSTACK ~ "needle"', filter_mod.UnsupportedRegexError),
+        ],
+    )
+    def test_unsupported_syntax(self, parser, expression, exception_class):
+        with pytest.raises(exception_class):
+            parser.parse_string(expression, parse_all=True)
+
 
 class TestFilterExpressionSample:
     @pytest.mark.parametrize(

vcztools/filter.py

@@ -13,6 +13,20 @@
 pp.ParserElement.enablePackrat()
 
 
+class UnsupportedFilteringFeatureError(ValueError):
+    def __init__(self):
+        super().__init__(
+            f"Unsupported filtering feature: {self.feature}. Please see "
+            f"https://github.com/sgkit-dev/vcztools/issues/{self.issue} "
+            "for details and let us know if this is important to you."
+        )
+
+
+class UnsupportedRegexError(UnsupportedFilteringFeatureError):
+    issue = "174"
+    feature = "Regular expressions"
+
+
 # The parser and evaluation model here are based on the eval_arith example
 # in the pyparsing docs:
 # https://github.com/pyparsing/pyparsing/blob/master/examples/eval_arith.py
@@ -27,18 +41,18 @@ class EvaluationNode:
     def __init__(self, tokens):
         self.tokens = tokens[0]
 
-
-class Constant(EvaluationNode):
-    def eval(self, data):
-        return self.tokens
-
     def __repr__(self):
         return repr(self.tokens)
 
     def referenced_fields(self):
         return frozenset()
 
 
+class Constant(EvaluationNode):
+    def eval(self, data):
+        return self.tokens
+
+
 class Identifier(EvaluationNode):
     def __init__(self, mapper, tokens):
         self.field_name = mapper(tokens[0])
@@ -55,6 +69,11 @@ def referenced_fields(self):
         return frozenset([self.field_name])
 
 
+class RegexOperator(EvaluationNode):
+    def __init__(self, tokens):
+        raise UnsupportedRegexError()
+
+
 # NOTE we should perhaps add a Operator superclass of UnaryMinus,
 # BinaryOperator and ComparisonOperator to reduce duplication
 # when doing things like referenced_fields. We should probably
@@ -175,6 +194,9 @@ def make_bcftools_filter_parser(all_fields=None, map_vcf_identifiers=True):
             (pp.Keyword("&&"), 2, pp.OpAssoc.LEFT, BinaryOperator),
             (pp.Keyword("|"), 2, pp.OpAssoc.LEFT, BinaryOperator),
             (pp.Keyword("||"), 2, pp.OpAssoc.LEFT, BinaryOperator),
+            # NOTE Putting the Regex operator at the end for now as
+            # I haven't figured out what the actual precedence is.
+            (pp.one_of("~ !~"), 2, pp.OpAssoc.LEFT, RegexOperator),
         ],
     )
     return filter_expression