Consider jsonschema.json extensions when loading direct schema paths (#561)

jviotti · web-flow · commit 9e4df075a3f3 · 2025-12-01T15:42:54.000-04:00
Signed-off-by: Juan Cruz Viotti &lt;jv@jviotti.com&gt;
diff --git a/DEPENDENCIES b/DEPENDENCIES
@@ -1,5 +1,5 @@
 vendorpull https://github.com/sourcemeta/vendorpull 1dcbac42809cf87cb5b045106b863e17ad84ba02
-core https://github.com/sourcemeta/core 626949ede75fb77ac4a919a24af0ade6f776751d
+core https://github.com/sourcemeta/core 81cf10ad7243e799991e5c692cd3d4a6a6e5cc9f
 jsonbinpack https://github.com/sourcemeta/jsonbinpack abd40e41050d14d74af1fddb5c397de5cca3b13c
 blaze https://github.com/sourcemeta/blaze 53d6ba77c26e613b2803f044c2852d4441bfb0a1
 hydra https://github.com/sourcemeta/hydra af9f2c54709d620872ead0c3f8f683c15a0fa702
diff --git a/src/command_bundle.cc b/src/command_bundle.cc
@@ -23,7 +23,8 @@ auto sourcemeta::jsonschema::bundle(const sourcemeta::core::Options &options)
 
   const std::filesystem::path schema_path{options.positional().front()};
   const auto configuration_path{find_configuration(schema_path)};
-  const auto &configuration{read_configuration(options, configuration_path)};
+  const auto &configuration{
+      read_configuration(options, configuration_path, schema_path)};
   const auto dialect{default_dialect(options, configuration)};
   auto schema{sourcemeta::core::read_yaml_or_json(schema_path)};
 
diff --git a/src/command_compile.cc b/src/command_compile.cc
@@ -22,7 +22,8 @@ auto sourcemeta::jsonschema::compile(const sourcemeta::core::Options &options)
 
   const auto &schema_path{options.positional().at(0)};
   const auto configuration_path{find_configuration(schema_path)};
-  const auto &configuration{read_configuration(options, configuration_path)};
+  const auto &configuration{
+      read_configuration(options, configuration_path, schema_path)};
   const auto dialect{default_dialect(options, configuration)};
 
   const auto schema{sourcemeta::core::read_yaml_or_json(schema_path)};
diff --git a/src/command_fmt.cc b/src/command_fmt.cc
@@ -36,7 +36,7 @@ auto sourcemeta::jsonschema::fmt(const sourcemeta::core::Options &options)
     try {
       const auto configuration_path{find_configuration(entry.first)};
       const auto &configuration{
-          read_configuration(options, configuration_path)};
+          read_configuration(options, configuration_path, entry.first)};
       const auto dialect{default_dialect(options, configuration)};
       const auto &custom_resolver{
           resolver(options, options.contains("http"), dialect, configuration)};
diff --git a/src/command_inspect.cc b/src/command_inspect.cc
@@ -157,7 +157,8 @@ auto sourcemeta::jsonschema::inspect(const sourcemeta::core::Options &options)
       sourcemeta::core::read_yaml_or_json(schema_path, std::ref(positions))};
 
   const auto configuration_path{find_configuration(schema_path)};
-  const auto &configuration{read_configuration(options, configuration_path)};
+  const auto &configuration{
+      read_configuration(options, configuration_path, schema_path)};
   const auto dialect{default_dialect(options, configuration)};
 
   sourcemeta::core::SchemaFrame frame{
diff --git a/src/command_lint.cc b/src/command_lint.cc
@@ -196,7 +196,7 @@ auto sourcemeta::jsonschema::lint(const sourcemeta::core::Options &options)
     for (const auto &entry : for_each_json(options)) {
       const auto configuration_path{find_configuration(entry.first)};
       const auto &configuration{
-          read_configuration(options, configuration_path)};
+          read_configuration(options, configuration_path, entry.first)};
       const auto dialect{default_dialect(options, configuration)};
 
       const auto &custom_resolver{
@@ -255,7 +255,7 @@ auto sourcemeta::jsonschema::lint(const sourcemeta::core::Options &options)
     for (const auto &entry : for_each_json(options)) {
       const auto configuration_path{find_configuration(entry.first)};
       const auto &configuration{
-          read_configuration(options, configuration_path)};
+          read_configuration(options, configuration_path, entry.first)};
       const auto dialect{default_dialect(options, configuration)};
       const auto &custom_resolver{
           resolver(options, options.contains("http"), dialect, configuration)};
diff --git a/src/command_metaschema.cc b/src/command_metaschema.cc
@@ -36,7 +36,8 @@ auto sourcemeta::jsonschema::metaschema(
     }
 
     const auto configuration_path{find_configuration(entry.first)};
-    const auto &configuration{read_configuration(options, configuration_path)};
+    const auto &configuration{
+        read_configuration(options, configuration_path, entry.first)};
     const auto default_dialect_option{default_dialect(options, configuration)};
 
     const auto &custom_resolver{resolver(options, options.contains("http"),
diff --git a/src/command_validate.cc b/src/command_validate.cc
@@ -149,7 +149,8 @@ auto sourcemeta::jsonschema::validate(const sourcemeta::core::Options &options)
   }
 
   const auto configuration_path{find_configuration(schema_path)};
-  const auto &configuration{read_configuration(options, configuration_path)};
+  const auto &configuration{
+      read_configuration(options, configuration_path, schema_path)};
   const auto dialect{default_dialect(options, configuration)};
 
   const auto schema{sourcemeta::core::read_yaml_or_json(schema_path)};
diff --git a/src/configuration.h b/src/configuration.h
@@ -21,7 +21,8 @@ inline auto find_configuration(const std::filesystem::path &path)
 
 inline auto read_configuration(
     const sourcemeta::core::Options &options,
-    const std::optional<std::filesystem::path> &configuration_path)
+    const std::optional<std::filesystem::path> &configuration_path,
+    const std::optional<std::filesystem::path> &schema_path = std::nullopt)
     -> const std::optional<sourcemeta::core::SchemaConfig> & {
   using CacheKey = std::optional<std::filesystem::path>;
   static std::map<CacheKey, std::optional<sourcemeta::core::SchemaConfig>>
@@ -48,6 +49,17 @@ inline auto read_configuration(
       throw FileError<sourcemeta::core::SchemaConfigParseError>(
           configuration_path.value(), error);
     }
+
+    assert(result.has_value());
+    if (schema_path.has_value() &&
+        !result.value().applies_to(schema_path.value())) {
+      LOG_VERBOSE(options)
+          << "Ignoring configuration file given extensions mismatch: "
+          << sourcemeta::core::weakly_canonical(configuration_path.value())
+                 .string()
+          << "\n";
+      result = std::nullopt;
+    }
   }
 
   auto [inserted_iterator, inserted] =
diff --git a/src/input.h b/src/input.h
@@ -156,6 +156,7 @@ inline auto for_each_json(const std::vector<std::string_view> &arguments,
     const auto configuration_path{find_configuration(current_path)};
     const auto &configuration{read_configuration(options, configuration_path)};
     const auto extensions{parse_extensions(options, configuration)};
+
     handle_json_entry(configuration.has_value()
                           ? configuration.value().absolute_path
                           : current_path,
diff --git a/test/CMakeLists.txt b/test/CMakeLists.txt
@@ -167,6 +167,7 @@ add_jsonschema_test_unix(validate/fail_benchmark_directory)
 add_jsonschema_test_unix(validate/pass_benchmark_loop)
 add_jsonschema_test_unix(validate/pass_benchmark_loop_jsonl)
 add_jsonschema_test_unix(validate/fail_benchmark_zero)
+add_jsonschema_test_unix(validate/fail_default_dialect_config_extension_mismatch)
 add_jsonschema_test_unix(validate/pass_resolve_remap)
 add_jsonschema_test_unix(validate/pass_resolve_remap_relative)
 
@@ -178,6 +179,7 @@ add_jsonschema_test_unix(metaschema/fail_single)
 add_jsonschema_test_unix(metaschema/fail_yaml)
 add_jsonschema_test_unix(metaschema/fail_non_schema)
 add_jsonschema_test_unix(metaschema/fail_no_dialect)
+add_jsonschema_test_unix(metaschema/fail_default_dialect_config_extension_mismatch)
 add_jsonschema_test_unix(metaschema/pass_cwd)
 add_jsonschema_test_unix(metaschema/pass_single)
 add_jsonschema_test_unix(metaschema/pass_2020_12)
@@ -267,6 +269,7 @@ add_jsonschema_test_unix(bundle/fail_resolve_duplicate)
 add_jsonschema_test_unix(bundle/fail_resolve_invalid_json)
 add_jsonschema_test_unix(bundle/fail_schema_invalid_json)
 add_jsonschema_test_unix(bundle/fail_unknown_metaschema)
+add_jsonschema_test_unix(bundle/fail_default_dialect_config_extension_mismatch)
 add_jsonschema_test_unix(bundle/pass_bigint)
 add_jsonschema_test_unix(bundle/pass_resolve_default_dialect_config)
 add_jsonschema_test_unix(bundle/pass_resolve_default_dialect_config_relative)
@@ -285,6 +288,7 @@ add_jsonschema_test_unix(inspect/pass_json_output)
 add_jsonschema_test_unix(inspect/fail_no_schema)
 add_jsonschema_test_unix(inspect/fail_schema_invalid_json)
 add_jsonschema_test_unix(inspect/fail_unknown_metaschema)
+add_jsonschema_test_unix(inspect/fail_default_dialect_config_extension_mismatch)
 add_jsonschema_test_unix(inspect/fail_relative_file_metaschema_ref)
 
 # Compile
@@ -300,6 +304,7 @@ add_jsonschema_test_unix(compile/pass_yaml)
 add_jsonschema_test_unix(compile/fail_no_schema)
 add_jsonschema_test_unix(compile/fail_schema_invalid_json)
 add_jsonschema_test_unix(compile/fail_unknown_metaschema)
+add_jsonschema_test_unix(compile/fail_default_dialect_config_extension_mismatch)
 add_jsonschema_test_unix(compile/pass_resolve_remap)
 add_jsonschema_test_unix(compile/pass_resolve_remap_relative)
 
diff --git a/test/bundle/fail_default_dialect_config_extension_mismatch.sh b/test/bundle/fail_default_dialect_config_extension_mismatch.sh
@@ -0,0 +1,48 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << 'EOF' > "$TMP/document.json"
+{ "foo": 1 }
+EOF
+
+cat << 'EOF' > "$TMP/jsonschema.json"
+{
+  "extension": [".schema.json"],
+  "defaultDialect": "https://json-schema.org/draft/2020-12/schema"
+}
+EOF
+
+"$1" bundle "$TMP/document.json" --verbose 2>"$TMP/stderr.txt" && CODE="$?" || CODE="$?"
+test "$CODE" = "1" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+Using configuration file: $(realpath "$TMP")/jsonschema.json
+Ignoring configuration file given extensions mismatch: $(realpath "$TMP")/jsonschema.json
+error: Could not determine the base dialect of the schema
+  at file path $(realpath "$TMP")/document.json
+
+Are you sure the input is a valid JSON Schema and its base dialect is known?
+If the input does not declare the \`\$schema\` keyword, you might want to
+explicitly declare a default dialect using \`--default-dialect/-d\`
+EOF
+
+diff "$TMP/stderr.txt" "$TMP/expected.txt"
+
+# JSON error
+"$1" bundle "$TMP/document.json" --json >"$TMP/stdout.txt" && CODE="$?" || CODE="$?"
+test "$CODE" = "1" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+{
+  "error": "Could not determine the base dialect of the schema",
+  "filePath": "$(realpath "$TMP")/document.json"
+}
+EOF
+
+diff "$TMP/stdout.txt" "$TMP/expected.txt"
diff --git a/test/compile/fail_default_dialect_config_extension_mismatch.sh b/test/compile/fail_default_dialect_config_extension_mismatch.sh
@@ -0,0 +1,48 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << 'EOF' > "$TMP/document.json"
+{ "foo": 1 }
+EOF
+
+cat << 'EOF' > "$TMP/jsonschema.json"
+{
+  "extension": [".schema.json"],
+  "defaultDialect": "https://json-schema.org/draft/2020-12/schema"
+}
+EOF
+
+"$1" compile "$TMP/document.json" --verbose 2>"$TMP/stderr.txt" && CODE="$?" || CODE="$?"
+test "$CODE" = "1" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+Using configuration file: $(realpath "$TMP")/jsonschema.json
+Ignoring configuration file given extensions mismatch: $(realpath "$TMP")/jsonschema.json
+error: Could not determine the base dialect of the schema
+  at file path $(realpath "$TMP")/document.json
+
+Are you sure the input is a valid JSON Schema and its base dialect is known?
+If the input does not declare the \`\$schema\` keyword, you might want to
+explicitly declare a default dialect using \`--default-dialect/-d\`
+EOF
+
+diff "$TMP/stderr.txt" "$TMP/expected.txt"
+
+# JSON error
+"$1" compile "$TMP/document.json" --json >"$TMP/stdout.txt" && CODE="$?" || CODE="$?"
+test "$CODE" = "1" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+{
+  "error": "Could not determine the base dialect of the schema",
+  "filePath": "$(realpath "$TMP")/document.json"
+}
+EOF
+
+diff "$TMP/stdout.txt" "$TMP/expected.txt"
diff --git a/test/inspect/fail_default_dialect_config_extension_mismatch.sh b/test/inspect/fail_default_dialect_config_extension_mismatch.sh
@@ -0,0 +1,48 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << 'EOF' > "$TMP/document.json"
+{ "foo": 1 }
+EOF
+
+cat << 'EOF' > "$TMP/jsonschema.json"
+{
+  "extension": [".schema.json"],
+  "defaultDialect": "https://json-schema.org/draft/2020-12/schema"
+}
+EOF
+
+"$1" inspect "$TMP/document.json" --verbose 2>"$TMP/stderr.txt" && CODE="$?" || CODE="$?"
+test "$CODE" = "1" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+Using configuration file: $(realpath "$TMP")/jsonschema.json
+Ignoring configuration file given extensions mismatch: $(realpath "$TMP")/jsonschema.json
+error: Could not determine the base dialect of the schema
+  at file path $(realpath "$TMP")/document.json
+
+Are you sure the input is a valid JSON Schema and its base dialect is known?
+If the input does not declare the \`\$schema\` keyword, you might want to
+explicitly declare a default dialect using \`--default-dialect/-d\`
+EOF
+
+diff "$TMP/stderr.txt" "$TMP/expected.txt"
+
+# JSON error
+"$1" inspect "$TMP/document.json" --json >"$TMP/stdout.txt" && CODE="$?" || CODE="$?"
+test "$CODE" = "1" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+{
+  "error": "Could not determine the base dialect of the schema",
+  "filePath": "$(realpath "$TMP")/document.json"
+}
+EOF
+
+diff "$TMP/stdout.txt" "$TMP/expected.txt"
diff --git a/test/metaschema/fail_default_dialect_config_extension_mismatch.sh b/test/metaschema/fail_default_dialect_config_extension_mismatch.sh
@@ -0,0 +1,48 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << 'EOF' > "$TMP/document.json"
+{ "foo": 1 }
+EOF
+
+cat << 'EOF' > "$TMP/jsonschema.json"
+{
+  "extension": [".schema.json"],
+  "defaultDialect": "https://json-schema.org/draft/2020-12/schema"
+}
+EOF
+
+"$1" metaschema "$TMP/document.json" --verbose 2>"$TMP/stderr.txt" && CODE="$?" || CODE="$?"
+test "$CODE" = "1" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+Using configuration file: $(realpath "$TMP")/jsonschema.json
+Ignoring configuration file given extensions mismatch: $(realpath "$TMP")/jsonschema.json
+error: Could not determine the base dialect of the schema
+  at file path $(realpath "$TMP/document.json")
+
+Are you sure the input is a valid JSON Schema and its base dialect is known?
+If the input does not declare the \`\$schema\` keyword, you might want to
+explicitly declare a default dialect using \`--default-dialect/-d\`
+EOF
+
+diff "$TMP/stderr.txt" "$TMP/expected.txt"
+
+# JSON error
+"$1" metaschema "$TMP/document.json" --json >"$TMP/stdout.txt" && CODE="$?" || CODE="$?"
+test "$CODE" = "1" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+{
+  "error": "Could not determine the base dialect of the schema",
+  "filePath": "$(realpath "$TMP/document.json")"
+}
+EOF
+
+diff "$TMP/stdout.txt" "$TMP/expected.txt"
diff --git a/test/validate/fail_default_dialect_config_extension_mismatch.sh b/test/validate/fail_default_dialect_config_extension_mismatch.sh
diff --git a/vendor/core/src/extension/schemaconfig/CMakeLists.txt b/vendor/core/src/extension/schemaconfig/CMakeLists.txt
diff --git a/vendor/core/src/extension/schemaconfig/include/sourcemeta/core/schemaconfig.h b/vendor/core/src/extension/schemaconfig/include/sourcemeta/core/schemaconfig.h
diff --git a/vendor/core/src/extension/schemaconfig/schemaconfig.cc b/vendor/core/src/extension/schemaconfig/schemaconfig.cc

Original file line number	Diff line number	Diff line change
`@@ -36,7 +36,8 @@ auto sourcemeta::jsonschema::metaschema(`
`36`	`36`	`}`
`37`	`37`
`38`	`38`	`const auto configuration_path{find_configuration(entry.first)};`
`39`		`- const auto &configuration{read_configuration(options, configuration_path)};`
	`39`	`+ const auto &configuration{`
	`40`	`+ read_configuration(options, configuration_path, entry.first)};`
`40`	`41`	`const auto default_dialect_option{default_dialect(options, configuration)};`
`41`	`42`
`42`	`43`	`const auto &custom_resolver{resolver(options, options.contains("http"),`
Original file line number	Diff line number	Diff line change
`@@ -149,7 +149,8 @@ auto sourcemeta::jsonschema::validate(const sourcemeta::core::Options &options)`
`149`	`149`	`}`
`150`	`150`
`151`	`151`	`const auto configuration_path{find_configuration(schema_path)};`
`152`		`- const auto &configuration{read_configuration(options, configuration_path)};`
	`152`	`+ const auto &configuration{`
	`153`	`+ read_configuration(options, configuration_path, schema_path)};`
`153`	`154`	`const auto dialect{default_dialect(options, configuration)};`
`154`	`155`
`155`	`156`	`const auto schema{sourcemeta::core::read_yaml_or_json(schema_path)};`