json-validator/include/jvalidate/detail/vocabulary.h

#pragma once

#include <string_view>
#include <unordered_map>
#include <unordered_set>

#include <jvalidate/enum.h>
#include <jvalidate/forward.h>

namespace jvalidate::detail {
template <Adapter A> struct ParserContext;
template <Adapter A> class Vocabulary {
public:
  friend class ConstraintFactory<A>;
  using pConstraint = std::unique_ptr<constraint::Constraint>;
  using MakeConstraint = std::function<pConstraint(ParserContext<A> const &)>;

private:
  schema::Version version_;
  std::unordered_map<std::string_view, MakeConstraint> make_;
  std::unordered_set<std::string_view> permitted_;
  std::unordered_set<std::string> vocabularies_;

  // TODO(samjaffe): Migrate this back to constraintsfactory
  std::unordered_set<std::string_view> keywords_{"$defs",
                                                 "additionalItems",
                                                 "additionalProperties",
                                                 "allOf",
                                                 "anyOf",
                                                 "definitions",
                                                 "dependencies",
                                                 "dependentSchemas",
                                                 "else",
                                                 "extends",
                                                 "if",
                                                 "items",
                                                 "not",
                                                 "oneOf",
                                                 "patternProperties",
                                                 "prefixItems",
                                                 "properties",
                                                 "then",
                                                 "unevaluatedItems",
                                                 "unevaluatedProperties"};
  std::unordered_set<std::string_view> property_keywords_{
      "$defs",     "definitions", "dependencies", "dependentSchemas", "patternProperties",
      "properties"};
  std::unordered_set<std::string_view> post_constraints_{"unevaluatedItems",
                                                         "unevaluatedProperties"};

public:
  Vocabulary() = default;
  Vocabulary(schema::Version version, std::unordered_map<std::string_view, MakeConstraint> make)
      : version_(version), make_(std::move(make)) {
    for (auto const & [keyword, _] : make_) {
      permitted_.emplace(keyword);
    }
  }

  void restrict(std::unordered_set<std::string> const & permitted_keywords,
                std::unordered_set<std::string> const & vocabularies) & {
    permitted_.clear();
    vocabularies_ = vocabularies;
    for (auto const & [keyword, _] : make_) {
      if (permitted_keywords.contains(std::string(keyword))) {
        permitted_.insert(keyword);
      }
    }
  }

  schema::Version version() const { return version_; }

  bool is_format_assertion() const {
    // In Draft07 and prior - format assertions were considered enabled by
    // default. This is - of course - problematic because very few
    // implementations actually had full support for format constraints.
    if (version_ < schema::Version::Draft2019_09) {
      return true;
    }

    // Some implementations wouldn't even bother with format constraints, and
    // others would provide implementations that either missed a number of edge
    // cases or were flat-out wrong on certail matters.
    // Therefore - starting in Draft 2019-09, the format keyword is an
    // annotation by default, instead of an assertion.
    if (version_ == schema::Version::Draft2019_09) {
      return permitted_.contains("/vocab/format");
    }

    // Draft 2020-12 makes this even more explicit - having separate vocabulary
    // documents for "format as assertion" and "format as annotation". Allowing
    // validators to add format constraints that are only used for annotating
    // results.
    return vocabularies_.contains("/vocab/format-assertion");
  }

  /**
   * @brief Is the given "key"word actually a keyword? As in, would
   * I expect to resolve a constraint out of it.
   */
  bool is_keyword(std::string_view word) const {
    return permitted_.contains(word) && make_.contains(word) && keywords_.contains(word);
  }

  /**
   * @brief Does the given "key"word represent a property object - that is to
   * say, an object containing some number of schemas mapped by arbitrary keys
   */
  bool is_property_keyword(std::string_view word) const {
    return is_keyword(word) && property_keywords_.contains(word);
  }

  bool is_constraint(std::string_view word) const {
    return permitted_.contains(word) && make_.contains(word) && make_.at(word);
  }

  auto constraint(std::string_view word, ParserContext<A> const & context) const {
    return std::make_pair(is_constraint(word) ? make_.at(word)(context) : nullptr,
                          post_constraints_.contains(word));
  }
};
}