Generalise relativize() as reference_visit() (#1504)

Signed-off-by: Juan Cruz Viotti <jv@jviotti.com>

src/core/jsonschema/CMakeLists.txt

@@ -8,7 +8,7 @@ sourcemeta_library(NAMESPACE sourcemeta PROJECT core NAME jsonschema
     keywords.h transform.h
   SOURCES jsonschema.cc default_walker.cc frame.cc
     resolver.cc walker.cc bundle.cc
-    unevaluated.cc relativize.cc unidentify.cc
+    unevaluated.cc unidentify.cc
     transform_rule.cc transformer.cc
     "${CMAKE_CURRENT_BINARY_DIR}/official_resolver.cc")
 

src/core/jsonschema/include/sourcemeta/core/jsonschema.h

@@ -15,9 +15,10 @@
 #include <sourcemeta/core/jsonschema_unevaluated.h>
 #include <sourcemeta/core/jsonschema_walker.h>
 
-#include <map>      // std::map
-#include <optional> // std::optional
-#include <string>   // std::string
+#include <functional> // std::function
+#include <map>        // std::map
+#include <optional>   // std::optional
+#include <string>     // std::string
 
 /// @defgroup jsonschema JSON Schema
 /// @brief A set of JSON Schema utilities across draft versions.
@@ -323,8 +324,8 @@ auto schema_format_compare(const JSON::String &left, const JSON::String &right)
 
 /// @ingroup jsonschema
 ///
-/// Try to turn every possible absolute reference in a schema into a relative
-/// one. For example:
+/// Remove every identifer from a schema, rephrasing references (if any) as
+/// needed. For example:
 ///
 /// ```cpp
 /// #include <sourcemeta/core/json.h>
@@ -335,32 +336,42 @@ auto schema_format_compare(const JSON::String &left, const JSON::String &right)
 ///   sourcemeta::core::parse_json(R"JSON({
 ///   "$id": "https://www.example.com/schema",
 ///   "$schema": "https://json-schema.org/draft/2020-12/schema",
-///   "$ref": "https://www.example.com/another",
+///   "$ref": "another",
 /// })JSON");
 ///
-/// sourcemeta::core::relativize(schema,
+/// sourcemeta::core::unidentify(schema,
 ///   sourcemeta::core::schema_official_walker,
 ///   sourcemeta::core::official_resolver);
 ///
 /// const sourcemeta::core::JSON expected =
 ///   sourcemeta::core::parse_json(R"JSON({
-///   "$id": "https://www.example.com/schema",
 ///   "$schema": "https://json-schema.org/draft/2020-12/schema",
-///   "$ref": "another",
+///   "$ref": "https://www.example.com/another",
 /// })JSON");
 ///
 /// assert(schema == expected);
 /// ```
 SOURCEMETA_CORE_JSONSCHEMA_EXPORT
-auto relativize(
+auto unidentify(
     JSON &schema, const SchemaWalker &walker, const SchemaResolver &resolver,
-    const std::optional<std::string> &default_dialect = std::nullopt,
-    const std::optional<std::string> &default_id = std::nullopt) -> void;
+    const std::optional<std::string> &default_dialect = std::nullopt) -> void;
 
 /// @ingroup jsonschema
 ///
-/// Remove every identifer from a schema, rephrasing references (if any) as
-/// needed. For example:
+/// Visit every reference in a schema. The arguments are as follows:
+///
+/// - The current subschema
+/// - The base URI of the current subschema
+/// - The reference vocabulary
+/// - The reference keyword name
+/// - The reference reference destination
+using SchemaVisitorReference = std::function<void(
+    JSON &, const URI &, const JSON::String &, const JSON::String &, URI &)>;
+
+/// @ingroup jsonschema
+///
+/// A reference visitor to try to turn every possible absolute reference in a
+/// schema into a relative one. For example:
 ///
 /// ```cpp
 /// #include <sourcemeta/core/json.h>
@@ -371,25 +382,70 @@ auto relativize(
 ///   sourcemeta::core::parse_json(R"JSON({
 ///   "$id": "https://www.example.com/schema",
 ///   "$schema": "https://json-schema.org/draft/2020-12/schema",
-///   "$ref": "another",
+///   "$ref": "https://www.example.com/another",
 /// })JSON");
 ///
-/// sourcemeta::core::unidentify(schema,
+/// sourcemeta::core::reference_visit(schema,
 ///   sourcemeta::core::schema_official_walker,
-///   sourcemeta::core::official_resolver);
+///   sourcemeta::core::official_resolver,
+///   sourcemeta::core::reference_visitor_relativize);
 ///
 /// const sourcemeta::core::JSON expected =
 ///   sourcemeta::core::parse_json(R"JSON({
+///   "$id": "https://www.example.com/schema",
 ///   "$schema": "https://json-schema.org/draft/2020-12/schema",
-///   "$ref": "https://www.example.com/another",
+///   "$ref": "another",
 /// })JSON");
 ///
 /// assert(schema == expected);
 /// ```
 SOURCEMETA_CORE_JSONSCHEMA_EXPORT
-auto unidentify(
+auto reference_visitor_relativize(JSON &subschema, const URI &base,
+                                  const JSON::String &vocabulary,
+                                  const JSON::String &keyword, URI &value)
+    -> void;
+
+/// @ingroup jsonschema
+///
+/// A utility function to loop over every reference in a schema, allowing
+/// modifications to their subschemas if desired. Note that the consumer is
+/// responsible for not making the schema invalid. For example:
+///
+/// ```cpp
+/// #include <sourcemeta/core/json.h>
+/// #include <sourcemeta/core/jsonschema.h>
+///
+/// sourcemeta::core::JSON schema =
+///   sourcemeta::core::parse_json(R"JSON({
+///   "$id": "https://www.example.com/schema",
+///   "$schema": "https://json-schema.org/draft/2020-12/schema",
+///   "$ref": "https://www.example.com/another",
+/// })JSON");
+///
+/// static auto visitor(JSON &subschema,
+///                     const URI &base,
+///                     const JSON::String &vocabulary,
+///                     const JSON::String &keyword,
+///                     URI &value) -> void {
+///   sourcemeta::core::prettify(subschema, std::cerr);
+///   std::cerr << "\n";
+///   std::cerr << base.recompose() << "\n";
+///   std::cerr << vocabulary << "\n";
+///   std::cerr << keyword << "\n";
+///   std::cerr << value.recompose() << "\n";
+/// }
+///
+/// sourcemeta::core::reference_visit(schema,
+///   sourcemeta::core::schema_official_walker,
+///   sourcemeta::core::official_resolver,
+///   visitor);
+/// ```
+SOURCEMETA_CORE_JSONSCHEMA_EXPORT
+auto reference_visit(
     JSON &schema, const SchemaWalker &walker, const SchemaResolver &resolver,
-    const std::optional<std::string> &default_dialect = std::nullopt) -> void;
+    const SchemaVisitorReference &callback,
+    const std::optional<std::string> &default_dialect = std::nullopt,
+    const std::optional<std::string> &default_id = std::nullopt) -> void;
 
 } // namespace sourcemeta::core
 

src/core/jsonschema/jsonschema.cc

@@ -547,3 +547,68 @@ auto sourcemeta::core::schema_format_compare(
     return left < right;
   }
 }
+
+auto sourcemeta::core::reference_visit(
+    sourcemeta::core::JSON &schema,
+    const sourcemeta::core::SchemaWalker &walker,
+    const sourcemeta::core::SchemaResolver &resolver,
+    const sourcemeta::core::SchemaVisitorReference &callback,
+    const std::optional<std::string> &default_dialect,
+    const std::optional<std::string> &default_id) -> void {
+  sourcemeta::core::SchemaFrame frame;
+  frame.analyse(schema, walker, resolver, default_dialect, default_id);
+  for (const auto &entry : frame.locations()) {
+    if (entry.second.type !=
+            sourcemeta::core::SchemaFrame::LocationType::Resource &&
+        entry.second.type !=
+            sourcemeta::core::SchemaFrame::LocationType::Subschema) {
+      continue;
+    }
+
+    auto &subschema{sourcemeta::core::get(schema, entry.second.pointer)};
+    assert(sourcemeta::core::is_schema(subschema));
+    if (!subschema.is_object()) {
+      continue;
+    }
+
+    const sourcemeta::core::URI base{entry.second.base};
+    // Assume the base is canonicalized already
+    assert(
+        sourcemeta::core::URI{entry.second.base}.canonicalize().recompose() ==
+        base.recompose());
+    for (const auto &property : subschema.as_object()) {
+      const auto walker_result{
+          walker(property.first, frame.vocabularies(entry.second, resolver))};
+      if (walker_result.type !=
+              sourcemeta::core::SchemaKeywordType::Reference ||
+          !property.second.is_string()) {
+        continue;
+      }
+
+      assert(property.second.is_string());
+      assert(walker_result.vocabulary.has_value());
+      sourcemeta::core::URI reference{property.second.to_string()};
+      callback(subschema, base, walker_result.vocabulary.value(),
+               property.first, reference);
+    }
+  }
+}
+
+auto sourcemeta::core::reference_visitor_relativize(
+    sourcemeta::core::JSON &subschema, const sourcemeta::core::URI &base,
+    const sourcemeta::core::JSON::String &vocabulary,
+    const sourcemeta::core::JSON::String &keyword, URI &reference) -> void {
+  // In 2019-09, `$recursiveRef` can only be `#`, so there
+  // is nothing else we can possibly do
+  if (vocabulary == "https://json-schema.org/draft/2019-09/vocab/core" &&
+      keyword == "$recursiveRef") {
+    return;
+  }
+
+  reference.relative_to(base);
+  reference.canonicalize();
+
+  if (reference.is_relative()) {
+    subschema.assign(keyword, sourcemeta::core::JSON{reference.recompose()});
+  }
+}

src/core/jsonschema/resolver.cc

@@ -150,9 +150,10 @@ auto SchemaFlatFileResolver::operator()(std::string_view identifier) const
                                  *this, result->second.default_dialect);
     // Because we allow re-identification, we can get into issues unless we
     // always try to relativize references
-    sourcemeta::core::relativize(schema, schema_official_walker, *this,
-                                 result->second.default_dialect,
-                                 result->second.original_identifier);
+    sourcemeta::core::reference_visit(
+        schema, schema_official_walker, *this,
+        sourcemeta::core::reference_visitor_relativize,
+        result->second.default_dialect, result->second.original_identifier);
     sourcemeta::core::reidentify(schema, result->first, *this,
                                  result->second.default_dialect);