feat(linter): add format_type_mismatch rule to detect invalid format usage

src/extension/alterschema/CMakeLists.txt

@@ -85,6 +85,7 @@ sourcemeta_library(NAMESPACE sourcemeta PROJECT core NAME alterschema
     linter/enum_to_const.h
     linter/equal_numeric_bounds_to_const.h
     linter/forbid_empty_enum.h
+    linter/format_type_mismatch.h
     linter/incoherent_min_max_contains.h
     linter/invalid_external_ref.h
     linter/items_array_default.h

src/extension/alterschema/alterschema.cc

@@ -114,6 +114,7 @@ inline auto APPLIES_TO_POINTERS(std::vector<Pointer> &&keywords)
 #include "linter/enum_to_const.h"
 #include "linter/equal_numeric_bounds_to_const.h"
 #include "linter/forbid_empty_enum.h"
+#include "linter/format_type_mismatch.h"
 #include "linter/incoherent_min_max_contains.h"
 #include "linter/invalid_external_ref.h"
 #include "linter/items_array_default.h"
@@ -236,6 +237,7 @@ auto add(SchemaTransformer &bundle, const AlterSchemaMode mode) -> void {
     bundle.add<UnsatisfiableMinProperties>();
     bundle.add<EnumToConst>();
     bundle.add<ForbidEmptyEnum>();
+    bundle.add<FormatTypeMismatch>();
     bundle.add<TopLevelTitle>();
     bundle.add<TopLevelDescription>();
     bundle.add<TopLevelExamples>();

src/extension/alterschema/linter/format_type_mismatch.h

@@ -0,0 +1,34 @@
+class FormatTypeMismatch final : public SchemaTransformRule {
+public:
+  using mutates = std::false_type;
+  using reframe_after_transform = std::false_type;
+  FormatTypeMismatch()
+      : SchemaTransformRule{
+            "format_type_mismatch",
+            "The `format` keyword validates string instances but `type` "
+            "is not `string`"} {};
+
+  [[nodiscard]] auto
+  condition(const sourcemeta::core::JSON &schema,
+            const sourcemeta::core::JSON &,
+            const sourcemeta::core::Vocabularies &vocabularies,
+            const sourcemeta::core::SchemaFrame &,
+            const sourcemeta::core::SchemaFrame::Location &,
+            const sourcemeta::core::SchemaWalker &,
+            const sourcemeta::core::SchemaResolver &) const
+      -> sourcemeta::core::SchemaTransformRule::Result override {
+    ONLY_CONTINUE_IF(
+        vocabularies.contains_any(
+            {Vocabularies::Known::JSON_Schema_2020_12_Validation,
+             Vocabularies::Known::JSON_Schema_2019_09_Validation,
+             Vocabularies::Known::JSON_Schema_Draft_7,
+             Vocabularies::Known::JSON_Schema_Draft_6,
+             Vocabularies::Known::JSON_Schema_Draft_4,
+             Vocabularies::Known::JSON_Schema_Draft_3}) &&
+        schema.is_object() && schema.defines("type") &&
+        schema.at("type").is_string() &&
+        schema.at("type").to_string() != "string" &&
+        schema.defines("format") && schema.at("format").is_string());
+    return APPLIES_TO_KEYWORDS("format", "type");
+  }
+};

test/alterschema/alterschema_lint_2019_09_test.cc

@@ -5092,3 +5092,199 @@ TEST(AlterSchema_lint_2019_09,
 
   EXPECT_EQ(document, expected);
 }
+
+TEST(AlterSchema_lint_2019_09, format_type_mismatch_1) {
+  const sourcemeta::core::JSON document = sourcemeta::core::parse_json(R"JSON({
+    "$schema": "https://json-schema.org/draft/2019-09/schema",
+    "title": "Test",
+    "description": "A test schema",
+    "examples": [ 1 ],
+    "type": "integer",
+    "format": "email"
+  })JSON");
+
+  LINT_WITHOUT_FIX(document, result, traces);
+
+  EXPECT_FALSE(result.first);
+  EXPECT_EQ(traces.size(), 2);
+  EXPECT_LINT_TRACE(traces, 0, "", "non_applicable_type_specific_keywords",
+                    "Avoid keywords that don't apply to the type or "
+                    "types that the current subschema expects",
+                    true);
+  EXPECT_LINT_TRACE(traces, 1, "", "format_type_mismatch",
+                    "The `format` keyword validates string instances but "
+                    "`type` is not `string`",
+                    false);
+}
+
+TEST(AlterSchema_lint_2019_09, format_type_mismatch_2) {
+  const sourcemeta::core::JSON document = sourcemeta::core::parse_json(R"JSON({
+    "$schema": "https://json-schema.org/draft/2019-09/schema",
+    "title": "Test",
+    "description": "A test schema",
+    "examples": [ "foo@bar.com" ],
+    "type": "string",
+    "format": "email"
+  })JSON");
+
+  LINT_WITHOUT_FIX(document, result, traces);
+
+  EXPECT_TRUE(result.first);
+  EXPECT_EQ(traces.size(), 0);
+}
+
+TEST(AlterSchema_lint_2019_09, format_type_mismatch_3) {
+  const sourcemeta::core::JSON document = sourcemeta::core::parse_json(R"JSON({
+    "$schema": "https://json-schema.org/draft/2019-09/schema",
+    "title": "Test",
+    "description": "A test schema",
+    "examples": [ "foo@bar.com" ],
+    "format": "email"
+  })JSON");
+
+  LINT_WITHOUT_FIX(document, result, traces);
+
+  EXPECT_TRUE(result.first);
+  EXPECT_EQ(traces.size(), 0);
+}
+
+TEST(AlterSchema_lint_2019_09, format_type_mismatch_4) {
+  const sourcemeta::core::JSON document = sourcemeta::core::parse_json(R"JSON({
+    "$schema": "https://json-schema.org/draft/2019-09/schema",
+    "title": "Test",
+    "description": "A test schema",
+    "examples": [ {} ],
+    "properties": {
+      "foo": {
+        "type": "integer",
+        "format": "email"
+      }
+    }
+  })JSON");
+
+  LINT_WITHOUT_FIX(document, result, traces);
+
+  EXPECT_FALSE(result.first);
+  EXPECT_EQ(traces.size(), 2);
+  EXPECT_LINT_TRACE(traces, 0, "/properties/foo",
+                    "non_applicable_type_specific_keywords",
+                    "Avoid keywords that don't apply to the type or "
+                    "types that the current subschema expects",
+                    true);
+  EXPECT_LINT_TRACE(traces, 1, "/properties/foo", "format_type_mismatch",
+                    "The `format` keyword validates string instances but "
+                    "`type` is not `string`",
+                    false);
+}
+
+TEST(AlterSchema_lint_2019_09, format_type_mismatch_5) {
+  const sourcemeta::core::JSON document = sourcemeta::core::parse_json(R"JSON({
+    "$schema": "https://json-schema.org/draft/2019-09/schema",
+    "title": "Test",
+    "description": "A test schema",
+    "examples": [ 1 ],
+    "type": "integer"
+  })JSON");
+
+  LINT_WITHOUT_FIX(document, result, traces);
+
+  EXPECT_TRUE(result.first);
+  EXPECT_EQ(traces.size(), 0);
+}
+
+TEST(AlterSchema_lint_2019_09, format_type_mismatch_6) {
+  const sourcemeta::core::JSON document = sourcemeta::core::parse_json(R"JSON({
+    "$schema": "https://json-schema.org/draft/2019-09/schema",
+    "title": "Test",
+    "description": "A test schema",
+    "examples": [ [] ],
+    "type": "array",
+    "items": {
+      "type": "number",
+      "format": "date-time"
+    }
+  })JSON");
+
+  LINT_WITHOUT_FIX(document, result, traces);
+
+  EXPECT_FALSE(result.first);
+  EXPECT_EQ(traces.size(), 2);
+  EXPECT_LINT_TRACE(traces, 0, "/items",
+                    "non_applicable_type_specific_keywords",
+                    "Avoid keywords that don't apply to the type or "
+                    "types that the current subschema expects",
+                    true);
+  EXPECT_LINT_TRACE(traces, 1, "/items", "format_type_mismatch",
+                    "The `format` keyword validates string instances but "
+                    "`type` is not `string`",
+                    false);
+}
+
+TEST(AlterSchema_lint_2019_09, format_type_mismatch_7) {
+  const sourcemeta::core::JSON document = sourcemeta::core::parse_json(R"JSON({
+    "$schema": "https://json-schema.org/draft/2019-09/schema",
+    "title": "Test",
+    "description": "A test schema",
+    "examples": [ {} ],
+    "$ref": "#/$defs/foo",
+    "$defs": {
+      "foo": {
+        "type": "integer",
+        "format": "uri"
+      }
+    }
+  })JSON");
+
+  LINT_WITHOUT_FIX(document, result, traces);
+
+  EXPECT_FALSE(result.first);
+  EXPECT_EQ(traces.size(), 1);
+  EXPECT_LINT_TRACE(traces, 0, "/$defs/foo", "format_type_mismatch",
+                    "The `format` keyword validates string instances but "
+                    "`type` is not `string`",
+                    false);
+}
+
+TEST(AlterSchema_lint_2019_09, format_type_mismatch_8) {
+  const sourcemeta::core::JSON document = sourcemeta::core::parse_json(R"JSON({
+    "$schema": "https://json-schema.org/draft/2019-09/schema",
+    "title": "Test",
+    "description": "A test schema",
+    "examples": [ {} ],
+    "$ref": "#/$defs/A/properties/bar",
+    "$defs": {
+      "A": {
+        "type": "integer",
+        "format": "uri",
+        "properties": {
+          "bar": { "type": "string" }
+        }
+      }
+    }
+  })JSON");
+
+  LINT_WITHOUT_FIX(document, result, traces);
+
+  EXPECT_FALSE(result.first);
+  EXPECT_EQ(traces.size(), 1);
+  EXPECT_LINT_TRACE(traces, 0, "/$defs/A", "format_type_mismatch",
+                    "The `format` keyword validates string instances but "
+                    "`type` is not `string`",
+                    false);
+}
+
+TEST(AlterSchema_lint_2019_09, format_type_mismatch_9) {
+  const sourcemeta::core::JSON document = sourcemeta::core::parse_json(R"JSON({
+    "$schema": "https://json-schema.org/draft/2019-09/schema",
+    "title": "Test",
+    "description": "A test schema",
+    "examples": [ 1 ],
+    "type": [ "integer", "string" ],
+    "format": "email"
+  })JSON");
+
+  LINT_WITHOUT_FIX(document, result, traces);
+
+  EXPECT_TRUE(result.first);
+  EXPECT_EQ(traces.size(), 0);
+}