From 52fb83fc4c23b4035521c4edb274668c942ec782 Mon Sep 17 00:00:00 2001
From: David Harsha <davishmcclurg@gmail.com>
Date: Thu, 23 Nov 2023 09:26:20 -0800
Subject: [PATCH] Fix keyword refs

This fixes two problems with JSON pointer refs and known keywords:

- If the ref resolved to a `Keyword` object with a schema value (eg,
  `#/items`), `unknown_schema!` was called, which was only defined for
  `UnknownKeyword`. I changed the name to `parsed_schema` and defined it
  for all keywords, though it still feels brittle since it requires
  keywords to pre-parse their values into schema objects.

- If the ref resolved through a `Keyword` object with a schema value
  (eg, '#/items/not`), `fetch` was called on the keyword's parsed
  schema, which wasn't defined. I settled on a `fetch` method for
  `Keyword` and `Schema` to find child objects by key/index. That way a
  keyword with a schema value will delegate `fetch` to its parsed schema
  object. `UnknownKeyword` also becomes less of a special case.

I also added a `Schema` type check after pointer resolution to help
people identify misdirected pointers.

Closes: https://github.com/davishmcclurg/json_schemer/issues/159
---
 lib/json_schemer/draft202012/vocab/core.rb |  6 +-
 lib/json_schemer/keyword.rb                |  8 +++
 lib/json_schemer/schema.rb                 | 15 ++--
 test/ref_test.rb                           | 80 ++++++++++++++++++++++
 4 files changed, 98 insertions(+), 11 deletions(-)

diff --git a/lib/json_schemer/draft202012/vocab/core.rb b/lib/json_schemer/draft202012/vocab/core.rb
index 66fea2c..c5d1182 100644
--- a/lib/json_schemer/draft202012/vocab/core.rb
+++ b/lib/json_schemer/draft202012/vocab/core.rb
@@ -136,7 +136,7 @@ def parse
             end
           end
 
-          def fetch_unknown!(token)
+          def fetch(token)
             if value.is_a?(Hash)
               parsed[token] ||= JSONSchemer::Schema::UNKNOWN_KEYWORD_CLASS.new(value.fetch(token), self, token, schema)
             elsif value.is_a?(Array)
@@ -146,8 +146,8 @@ def fetch_unknown!(token)
             end
           end
 
-          def unknown_schema!
-            @unknown_schema ||= subschema(value)
+          def parsed_schema
+            @parsed_schema ||= subschema(value)
           end
 
           def validate(instance, instance_location, keyword_location, _context)
diff --git a/lib/json_schemer/keyword.rb b/lib/json_schemer/keyword.rb
index 091d7d8..f61ea99 100644
--- a/lib/json_schemer/keyword.rb
+++ b/lib/json_schemer/keyword.rb
@@ -30,6 +30,14 @@ def error_key
       keyword
     end
 
+    def fetch(key)
+      parsed.fetch(parsed.is_a?(Array) ? key.to_i : key)
+    end
+
+    def parsed_schema
+      parsed.is_a?(Schema) ? parsed : nil
+    end
+
   private
 
     def parse
diff --git a/lib/json_schemer/schema.rb b/lib/json_schemer/schema.rb
index d77bfa7..baf1ec2 100644
--- a/lib/json_schemer/schema.rb
+++ b/lib/json_schemer/schema.rb
@@ -203,18 +203,13 @@ def resolve_ref(uri)
       end
 
       schema = Hana::Pointer.parse(pointer).reduce(schema) do |obj, token|
-        if obj.is_a?(UNKNOWN_KEYWORD_CLASS)
-          obj.fetch_unknown!(token)
-        elsif obj.parsed.is_a?(Array)
-          obj.parsed.fetch(token.to_i)
-        else
-          obj.parsed.fetch(token)
-        end
+        obj.fetch(token)
       rescue IndexError
         raise InvalidRefPointer, pointer
       end
 
-      schema = schema.unknown_schema! unless schema.is_a?(Schema)
+      schema = schema.parsed_schema if schema.is_a?(Keyword)
+      raise InvalidRefPointer, pointer unless schema.is_a?(Schema)
 
       schema
     end
@@ -299,6 +294,10 @@ def error_key
       '^'
     end
 
+    def fetch(key)
+      parsed.fetch(key)
+    end
+
     def fetch_format(format, *args, &block)
       if meta_schema == self
         formats.fetch(format, *args, &block)
diff --git a/test/ref_test.rb b/test/ref_test.rb
index 1cf8ede..ac8905c 100644
--- a/test/ref_test.rb
+++ b/test/ref_test.rb
@@ -395,4 +395,84 @@ def test_exclusive_ref_supports_definitions_with_id_and_json_pointer
     assert(schema.valid?(1))
     refute(schema.valid?('1'))
   end
+
+  def test_keyword_ref
+    schemer = JSONSchemer.schema({
+      '$ref' => '#/items',
+      'items' => {
+        'type' => 'integer'
+      }
+    })
+    assert(schemer.valid?(1))
+    refute(schemer.valid?('1'))
+  end
+
+  def test_nested_keyword_ref
+    schemer = JSONSchemer.schema({
+      '$ref' => '#/items/not',
+      'items' => {
+        'not' => {
+          'type' => 'integer'
+        }
+      }
+    })
+    assert(schemer.valid?(1))
+    refute(schemer.valid?('1'))
+
+    schemer = JSONSchemer.schema({
+      '$ref' => '#/properties/a',
+      'properties' => {
+        'a' => {
+          'type' => 'integer'
+        }
+      }
+    })
+    assert(schemer.valid?(1))
+    refute(schemer.valid?('1'))
+  end
+
+  def test_complex_nested_keyword_ref
+    schemer = JSONSchemer.schema({
+      '$ref' => '#/definitions/a/allOf/0/items/properties/b/unknown-array/0/unknown-hash/c',
+      'definitions' => {
+        'a' => {
+          'allOf' => [
+            {
+              'items' => {
+                'properties' => {
+                  'b' => {
+                    'unknown-array' => [
+                      {
+                        'unknown-hash' => {
+                          'c' => {
+                            'type' => 'integer'
+                          }
+                        }
+                      }
+                    ]
+                  }
+                }
+              }
+            }
+          ]
+        }
+      }
+    })
+    assert(schemer.valid?(1))
+    refute(schemer.valid?('1'))
+  end
+
+  def test_non_schema_ref_pointer
+    schemer = JSONSchemer.schema({
+      '$ref' => '#/allOf',
+      'allOf' => [true]
+    })
+    assert_raises(JSONSchemer::InvalidRefPointer) { schemer.valid?(1) }
+
+    schemer = JSONSchemer.schema({
+      '$ref' => '#/type/0',
+      'type' => ['integer', 'string']
+    })
+    assert_raises(JSONSchemer::InvalidRefPointer) { schemer.valid?(1) }
+  end
 end