Raise an error for all string values on input

Author: jeromekelleher

tests/test_bcftools_validation.py

@@ -46,7 +46,9 @@ def run_vcztools(args: str, expect_error=False) -> tuple[str, str]:
         ("view --no-version", "sample.vcf.gz"),
         ("view --no-version", "chr22.vcf.gz"),
         ("view --no-version", "msprime_diploid.vcf.gz"),
-        ("view --no-version -i 'ID == \"rs6054257\"'", "sample.vcf.gz"),
+        # Filters based on strings disabled:
+        # https://github.com/sgkit-dev/vcztools/issues/189
+        # ("view --no-version -i 'ID == \"rs6054257\"'", "sample.vcf.gz"),
         ("view --no-version -i 'INFO/DP > 10'", "sample.vcf.gz"),
         # Filters based on FMT values are currently disabled.
         # https://github.com/sgkit-dev/vcztools/issues/180

tests/test_filter.py

@@ -31,7 +31,12 @@ def test_invalid_expressions(self, parser, expression):
     @pytest.mark.parametrize(
         ("expression", "exception_class"),
         [
-            ('INFO/HAYSTACK ~ "needle"', filter_mod.UnsupportedRegexError),
+            # NOTE: using an integer here so that we don't trigger the
+            # generic string issue. Can fix this later when we've gotten
+            # some partial string handling implemented
+            ("INFO/HAYSTACK ~ 0", filter_mod.UnsupportedRegexError),
+            ('CHROM="1"', filter_mod.UnsupportedStringsError),
+            ('FILTER="PASS"', filter_mod.UnsupportedStringsError),
             ("INFO/X[0] == 1", filter_mod.UnsupportedArraySubscriptError),
             ("INFO/AF[0] > 0.3", filter_mod.UnsupportedArraySubscriptError),
             ("FORMAT/AD[0:0] > 30", filter_mod.UnsupportedArraySubscriptError),
@@ -135,7 +140,9 @@ def test_referenced_fields(self, expr, expected):
             ("a + 1 + 2", "(variant_a)+(1)+(2)"),
             ("a + (1 + 2)", "(variant_a)+((1)+(2))"),
             ("POS<10", "(variant_position)<(10)"),
-            ('CHROM=="chr1"', "(variant_contig)==('chr1')"),
+            # Filters based on strings disabled:
+            # https://github.com/sgkit-dev/vcztools/issues/189
+            # ('CHROM=="chr1"', "(variant_contig)==('chr1')"),
         ],
     )
     def test_repr(self, expr, expected):
@@ -148,8 +155,6 @@ class TestBcftoolsParser:
         "expr",
         [
             "2",
-            '"x"',
-            '"INFO/STRING"',
             "2 + 2",
             "(2 + 3) / 2",
             "2 / (2 + 3)",
@@ -167,8 +172,12 @@ class TestBcftoolsParser:
             "1 + 2 == 1 + 2 + 3",
             "(1 + 2) == (1 + 2 + 3)",
             "(1 == 1) != (2 == 2)",
-            '("x" == "x")',
             "-1 == 1 + 2 - 4",
+            # Filters based on strings disabled:
+            # https://github.com/sgkit-dev/vcztools/issues/189
+            # '("x" == "x")',
+            # '"x"',
+            # '"INFO/STRING"',
         ],
     )
     def test_python_arithmetic_expressions(self, expr):
@@ -177,6 +186,24 @@ def test_python_arithmetic_expressions(self, expr):
         result = parsed[0].eval({})
         assert result == eval(expr)
 
+    @pytest.mark.parametrize(
+        ("expr", "data"),
+        [
+            ('("x" == "x")', {}),
+            ('"x"', {}),
+            ('"INFO/STRING"', {}),
+            ('a == "string"', {"a": "string"}),
+        ],
+    )
+    def test_python_string_expressions_data(self, expr, data):
+        # Filters based on strings disabled:
+        # https://github.com/sgkit-dev/vcztools/issues/189
+        parser = filter_mod.make_bcftools_filter_parser()
+        with pytest.raises(filter_mod.UnsupportedStringsError):
+            parser.parse_string(expr, parse_all=True)
+            # result = parsed[0].eval({})
+            # assert result == eval(expr)
+
     @pytest.mark.parametrize(
         ("expr", "data"),
         [
@@ -189,7 +216,6 @@ def test_python_arithmetic_expressions(self, expr):
             ("a == a", {"a": 1}),
             ("-a == -a", {"a": 1}),
             ("-a == b", {"a": 1, "b": -1}),
-            ('a == "string"', {"a": "string"}),
         ],
     )
     def test_python_arithmetic_expressions_data(self, expr, data):

vcztools/filter.py

@@ -37,6 +37,11 @@ class UnsupportedArraySubscriptError(UnsupportedFilteringFeatureError):
     feature = "Array subscripts"
 
 
+class UnsupportedStringsError(UnsupportedFilteringFeatureError):
+    issue = "189"
+    feature = "String values currently not supported"
+
+
 # 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
@@ -63,6 +68,15 @@ def eval(self, data):
         return self.tokens
 
 
+class Number(Constant):
+    pass
+
+
+class String(Constant):
+    def __init__(self, tokens):
+        raise UnsupportedStringsError()
+
+
 class Identifier(EvaluationNode):
     def __init__(self, mapper, tokens):
         self.field_name = mapper(tokens[0])
@@ -189,9 +203,11 @@ def make_bcftools_filter_parser(all_fields=None, map_vcf_identifiers=True):
     if all_fields is None:
         all_fields = set()
 
-    constant = (pp.common.number | pp.QuotedString('"')).set_parse_action(Constant)
-    identifier = pp.common.identifier()
+    number = pp.common.number.set_parse_action(Number)
+    string = pp.QuotedString('"').set_parse_action(String)
+    constant = number | string
 
+    identifier = pp.common.identifier()
     vcf_prefixes = pp.Literal("INFO/") | pp.Literal("FORMAT/") | pp.Literal("FMT/")
     vcf_identifier = pp.Combine(vcf_prefixes + identifier) | identifier