reformat validator errors

Author: nystrom

sdks/python/src/lqp/proto_validator.py

@@ -427,14 +427,14 @@ def get_original_name(self, rid: logic_pb2.RelationId) -> str:
         key = relation_id_key(rid)
         return self.original_names.get(key, relation_id_hex(rid))
 
-    def _location_str(self, node: Message) -> str:
-        """Return ' at file:line:col' for the given node, or '' if unavailable."""
+    def _location_prefix(self, node: Message) -> str:
+        """Return 'file:line:col: ' for the given node, or '' if unavailable."""
         span = self._provenance.get(id(node))
         if span is not None:
             loc = span.start
             if self._filename:
-                return f" at {self._filename}:{loc.line}:{loc.column}"
-            return f" at {loc.line}:{loc.column}"
+                return f"{self._filename}:{loc.line}:{loc.column}: "
+            return f"{loc.line}:{loc.column}: "
         return ""
 
     def _resolve_visitor(self, type_name: str):
@@ -491,7 +491,7 @@ def _mark_var_used(self, var_name: str, node: Message):
                 used.add(var_name)
                 return
         raise ValidationError(
-            f"Undeclared variable used{self._location_str(node)}: '{var_name}'"
+            f"ERROR: {self._location_prefix(node)}Undeclared variable used: '{var_name}'"
         )
 
     def visit_Abstraction(self, node: logic_pb2.Abstraction, *args: Any):
@@ -509,7 +509,7 @@ def visit_Abstraction(self, node: logic_pb2.Abstraction, *args: Any):
                 continue
             b = binding_map[var_name]
             raise ValidationError(
-                f"Unused variable declared{self._location_str(b)}: '{var_name}'"
+                f"ERROR: {self._location_prefix(b)}Unused variable declared: '{var_name}'"
             )
 
     def visit_Var(self, node: logic_pb2.Var, *args: Any):
@@ -532,7 +532,7 @@ def visit_Abstraction(self, node: logic_pb2.Abstraction, *args: Any):
             name = binding.var.name
             if name in in_scope_names:
                 raise ValidationError(
-                    f"Shadowed variable{self._location_str(binding)}: '{name}'"
+                    f"ERROR: {self._location_prefix(binding)}Shadowed variable: '{name}'"
                 )
         new_scope = in_scope_names | {b.var.name for b in node.vars}
         self.visit(node.value, new_scope)
@@ -556,12 +556,12 @@ def _check_relation_id(self, rid: logic_pb2.RelationId, node: Message):
             if self.curr_fragment != seen_frag:
                 original_name = self.get_original_name(rid)
                 raise ValidationError(
-                    f"Duplicate declaration across fragments{self._location_str(node)}: '{original_name}'"
+                    f"ERROR: {self._location_prefix(node)}Duplicate declaration across fragments: '{original_name}'"
                 )
             elif self.curr_epoch == seen_epoch:
                 original_name = self.get_original_name(rid)
                 raise ValidationError(
-                    f"Duplicate declaration within fragment in epoch{self._location_str(node)}: '{original_name}'"
+                    f"ERROR: {self._location_prefix(node)}Duplicate declaration within fragment in epoch: '{original_name}'"
                 )
         self.seen_ids[key] = (self.curr_epoch, self.curr_fragment)
 
@@ -579,7 +579,7 @@ def visit_Algorithm(self, node: logic_pb2.Algorithm, *args: Any):
             if key in self.seen_ids:
                 original_name = self.get_original_name(rid)
                 raise ValidationError(
-                    f"Duplicate declaration{self._location_str(rid)}: '{original_name}'"
+                    f"ERROR: {self._location_prefix(rid)}Duplicate declaration: '{original_name}'"
                 )
             else:
                 self.seen_ids[key] = (self.curr_epoch, self.curr_fragment)
@@ -600,7 +600,7 @@ def visit_Define(self, node: transactions_pb2.Define, *args: Any):
         if frag_id in self.seen_ids:
             id_str = frag_id.decode("utf-8")
             raise ValidationError(
-                f"Duplicate fragment within an epoch{self._location_str(node)}: '{id_str}'"
+                f"ERROR: {self._location_prefix(node)}Duplicate fragment within an epoch: '{id_str}'"
             )
         else:
             self.seen_ids.add(frag_id)
@@ -724,7 +724,8 @@ def visit_Atom(self, node: logic_pb2.Atom, *args: Any):
         if atom_arity != relation_arity:
             original_name = self.get_original_name(node.name)
             raise ValidationError(
-                f"Incorrect arity for '{original_name}' atom{self._location_str(node)}: "
+                f"ERROR: {self._location_prefix(node)}"
+                f"Incorrect arity for '{original_name}' atom: "
                 f"expected {relation_arity} term{'' if relation_arity == 1 else 's'}, got {atom_arity}"
             )
 
@@ -742,7 +743,8 @@ def visit_Atom(self, node: logic_pb2.Atom, *args: Any):
                 original_name = self.get_original_name(node.name)
                 pretty_term = proto_term_str(term)
                 raise ValidationError(
-                    f"Incorrect type for '{original_name}' atom at index {i} ('{pretty_term}'){self._location_str(node)}: "
+                    f"ERROR: {self._location_prefix(node)}"
+                    f"Incorrect type for '{original_name}' atom at index {i} ('{pretty_term}'): "
                     f"expected {expected_type} term, got {term_type}"
                 )
 
@@ -758,7 +760,7 @@ def visit_Loop(self, node: logic_pb2.Loop, *args: Any):
                 brk = getattr(instr_wrapper, "break")
                 original_name = self.get_original_name(brk.name)
                 raise ValidationError(
-                    f"Break rule found outside of body{self._location_str(brk)}: '{original_name}'"
+                    f"ERROR: {self._location_prefix(brk)}Break rule found outside of body: '{original_name}'"
                 )
 
 
@@ -811,7 +813,7 @@ def visit_Loop(self, node: logic_pb2.Loop, *args: Any):
                     if key in self.globals and key not in self.init:
                         original_name = self.get_original_name(actual.name)  # type: ignore[union-attr]
                         raise ValidationError(
-                            f"Global rule found in body{self._location_str(actual)}: '{original_name}'"
+                            f"ERROR: {self._location_prefix(actual)}Global rule found in body: '{original_name}'"
                         )
 
 
@@ -825,7 +827,7 @@ def _check_atom_body(self, node: Message, instr_type_name: str):
         which = formula.WhichOneof("formula_type")
         if which != "atom":
             raise ValidationError(
-                f"{instr_type_name}{self._location_str(node)} must have an Atom as its value"
+                f"ERROR: {self._location_prefix(node)}{instr_type_name} must have an Atom as its value"
             )
 
     def visit_Upsert(self, node: logic_pb2.Upsert, *args: Any):
@@ -851,24 +853,24 @@ def __init__(self, txn: transactions_pb2.Transaction, **kwargs: Any):
         self.visit(txn)
 
     def visit_ExportCSVConfig(self, node: transactions_pb2.ExportCSVConfig, *args: Any):
-        loc = self._location_str(node)
+        loc = self._location_prefix(node)
         if node.HasField("syntax_delim") and len(node.syntax_delim) != 1:
             raise ValidationError(
-                f"CSV delimiter should be a single character{loc}, got '{node.syntax_delim}'"
+                f"ERROR: {loc}CSV delimiter should be a single character, got '{node.syntax_delim}'"
             )
         if node.HasField("syntax_quotechar") and len(node.syntax_quotechar) != 1:
             raise ValidationError(
-                f"CSV quotechar should be a single character{loc}, got '{node.syntax_quotechar}'"
+                f"ERROR: {loc}CSV quotechar should be a single character, got '{node.syntax_quotechar}'"
             )
         if node.HasField("syntax_escapechar") and len(node.syntax_escapechar) != 1:
             raise ValidationError(
-                f"CSV escapechar should be a single character{loc}, got '{node.syntax_escapechar}'"
+                f"ERROR: {loc}CSV escapechar should be a single character, got '{node.syntax_escapechar}'"
             )
 
         valid_compressions = {"", "gzip"}
         if node.HasField("compression") and node.compression not in valid_compressions:
             raise ValidationError(
-                f"CSV compression should be one of {valid_compressions}{loc}, got '{node.compression}'"
+                f"ERROR: {loc}CSV compression should be one of {valid_compressions}, got '{node.compression}'"
             )
 
         column_0_key_types: list[str] | None = None
@@ -881,7 +883,7 @@ def visit_ExportCSVConfig(self, node: transactions_pb2.ExportCSVConfig, *args: A
             column_types = self.relation_types[key]
             if len(column_types) < 1:
                 raise ValidationError(
-                    f"Data column relation must have at least one column{loc}, "
+                    f"ERROR: {loc}Data column relation must have at least one column, "
                     f"got zero columns in '{self.get_original_name(column.column_data)}'"
                 )
             key_types = column_types[:-1]
@@ -891,7 +893,7 @@ def visit_ExportCSVConfig(self, node: transactions_pb2.ExportCSVConfig, *args: A
             else:
                 if column_0_key_types != key_types:
                     raise ValidationError(
-                        f"All data columns in ExportCSVConfig{loc} must have the same key types. "
+                        f"ERROR: {loc}All data columns in ExportCSVConfig must have the same key types. "
                         f"Got '{column_0_name}' with key types {[str(t) for t in column_0_key_types]} "
                         f"and '{self.get_original_name(column.column_data)}' with key types {[str(t) for t in key_types]}."
                     )
@@ -909,12 +911,12 @@ def visit_FunctionalDependency(
         for var in node.keys:
             if var.name not in guard_var_names:
                 raise ValidationError(
-                    f"Key variable '{var.name}' not declared in guard{self._location_str(var)}"
+                    f"ERROR: {self._location_prefix(var)}Key variable '{var.name}' not declared in guard"
                 )
         for var in node.values:
             if var.name not in guard_var_names:
                 raise ValidationError(
-                    f"Value variable '{var.name}' not declared in guard{self._location_str(var)}"
+                    f"ERROR: {self._location_prefix(var)}Value variable '{var.name}' not declared in guard"
                 )
 
 

sdks/python/tests/test_proto_validator.py

@@ -66,7 +66,7 @@ def test_proto_validator_error_has_location(validator_file):
         pytest.skip(f"No expected error comment found in {validator_file}")
         return
     # Skip files whose expected error doesn't include a location.
-    if not re.search(r"at\s+\S+:\d+:\d+", expected_error):
+    if not re.search(r"\S+:\d+:\d+:", expected_error):
         pytest.skip(f"Expected error has no location: {validator_file}")
         return
     with open(file_path) as f:
@@ -75,6 +75,6 @@ def test_proto_validator_error_has_location(validator_file):
     with pytest.raises(ValidationError) as exc_info:
         validate_proto(txn_proto, provenance=provenance, filename=validator_file)
     error_message = str(exc_info.value)
-    assert re.search(r"at\s+\S+:\d+:\d+", error_message), (
+    assert re.search(r"\S+:\d+:\d+:", error_message), (
         f"Error message missing location: '{error_message}'"
     )

sdks/python/tests/utils.py

@@ -34,7 +34,7 @@ def extract_expected_error(file_path) -> str | None:
     """Extract the expected error from a ;; ERROR: comment in the file."""
     with open(file_path) as f:
         content = f.read()
-    error_match = re.search(r";;\s*ERROR:\s*(.+)(?:\n|\r\n?)", content)
+    error_match = re.search(r";;\s*(.+?)(?:\n|\r\n?|\Z)", content)
     if error_match:
         return error_match.group(1).strip()
     return None

sdks/python/tests/validator/fail_atom_arity_1.lqp

@@ -1,4 +1,4 @@
-;; ERROR: Incorrect arity for 'foo' atom at fail_atom_arity_1.lqp:8:25: expected 0 terms, got 1
+;; ERROR: fail_atom_arity_1.lqp:8:25: Incorrect arity for 'foo' atom: expected 0 terms, got 1
 (transaction
   (epoch
     (writes

sdks/python/tests/validator/fail_atom_arity_2.lqp

@@ -1,4 +1,4 @@
-;; ERROR: Incorrect arity for 'foo' atom at fail_atom_arity_2.lqp:8:25: expected 2 terms, got 1
+;; ERROR: fail_atom_arity_2.lqp:8:25: Incorrect arity for 'foo' atom: expected 2 terms, got 1
 (transaction
   (epoch
     (writes

sdks/python/tests/validator/fail_atom_arity_3.lqp

@@ -1,4 +1,4 @@
-;; ERROR: Incorrect arity for 'foo2' atom at fail_atom_arity_3.lqp:15:26: expected 0 terms, got 1
+;; ERROR: fail_atom_arity_3.lqp:15:26: Incorrect arity for 'foo2' atom: expected 0 terms, got 1
 (transaction
   (epoch
     (writes

sdks/python/tests/validator/fail_atom_type_1.lqp

@@ -1,4 +1,4 @@
-;; ERROR: Incorrect type for 'foo' atom at index 1 ('2') at fail_atom_type_1.lqp:8:25: expected STRING term, got INT
+;; ERROR: fail_atom_type_1.lqp:8:25: Incorrect type for 'foo' atom at index 1 ('2'): expected STRING term, got INT
 (transaction
   (epoch
     (writes

sdks/python/tests/validator/fail_atom_type_2.lqp

@@ -1,4 +1,4 @@
-;; ERROR: Incorrect type for 'foo' atom at index 0 ('z') at fail_atom_type_2.lqp:8:31: expected FLOAT term, got INT
+;; ERROR: fail_atom_type_2.lqp:8:31: Incorrect type for 'foo' atom at index 0 ('z'): expected FLOAT term, got INT
 (transaction
   (epoch
     (writes

sdks/python/tests/validator/fail_atom_type_3.lqp

@@ -1,4 +1,4 @@
-;; ERROR: Incorrect type for 'foo' atom at index 0 ('1') at fail_atom_type_3.lqp:8:25: expected STRING term, got INT
+;; ERROR: fail_atom_type_3.lqp:8:25: Incorrect type for 'foo' atom at index 0 ('1'): expected STRING term, got INT
 (transaction
   (epoch
     (writes

sdks/python/tests/validator/fail_atom_type_4.lqp

@@ -1,4 +1,4 @@
-;; ERROR: Incorrect type for 'foo' atom at index 1 ('1') at fail_atom_type_4.lqp:15:26: expected STRING term, got INT
+;; ERROR: fail_atom_type_4.lqp:15:26: Incorrect type for 'foo' atom at index 1 ('1'): expected STRING term, got INT
 (transaction
   (epoch
     (writes