json-validator/include/jvalidate/validation_visitor.h

#pragma once

#include <unordered_map>

#include <jvalidate/constraint/array_constraint.h>
#include <jvalidate/constraint/general_constraint.h>
#include <jvalidate/constraint/number_constraint.h>
#include <jvalidate/constraint/object_constraint.h>
#include <jvalidate/constraint/string_constraint.h>
#include <jvalidate/constraint/visitor.h>
#include <jvalidate/detail/expect.h>
#include <jvalidate/forward.h>
#include <jvalidate/schema.h>
#include <jvalidate/status.h>
#include <jvalidate/validation_config.h>
#include <jvalidate/validation_result.h>

#define NOOP_UNLESS_TYPE(etype)                                                                    \
  RETURN_UNLESS(document_.type() == adapter::Type::etype, Status::Noop)

#define BREAK_EARLY_IF_NO_RESULT_TREE()                                                            \
  do {                                                                                             \
    if (rval == Status::Reject and not result_) {                                                  \
      break;                                                                                       \
    }                                                                                              \
  } while (false)

namespace jvalidate {
template <Adapter A, RegexEngine RE>
class ValidationVisitor : public constraint::ConstraintVisitor {
private:
  A document_;
  detail::Pointer where_;

  schema::Node const & schema_;

  ValidationResult * result_;
  ValidationResult * local_result_;

  ValidationConfig const & cfg_;
  std::unordered_map<std::string, RE> & regex_cache_;

public:
  ValidationVisitor(A const & json, schema::Node const & schema, ValidationConfig const & cfg,
                    std::unordered_map<std::string, RE> & regex_cache, ValidationResult * result)
      : ValidationVisitor(json, schema, cfg, regex_cache, {}, result) {}

  Status visit(constraint::TypeConstraint const & cons) const {
    adapter::Type const type = document_.type();
    for (adapter::Type const accept : cons.types) {
      if (accept & type) {
        return Status::Accept;
      }
    }
    return Status::Reject;
  }

  Status visit(constraint::ExtensionConstraint const & cons) const {
    return cons.validate(document_, where_, result_);
  }

  Status visit(constraint::EnumConstraint const & cons) const {
    auto is_equal = [this](auto const & frozen) {
      return document_.equals(frozen, cfg_.strict_equality);
    };
    for (auto const & option : cons.enumeration) {
      if (option->apply(is_equal)) {
        return Status::Accept;
      }
    }
    return Status::Reject;
  }

  Status visit(constraint::AllOfConstraint const & cons) const {
    Status rval = Status::Accept;

    for (schema::Node const * subschema : cons.children) {
      rval &= validate_subschema(subschema);
      BREAK_EARLY_IF_NO_RESULT_TREE();
    }

    return rval;
  }

  Status visit(constraint::AnyOfConstraint const & cons) const {
    for (schema::Node const * subschema : cons.children) {
      if (validate_subschema(subschema)) {
        return Status::Accept;
      }
    }

    return Status::Reject;
  }

  Status visit(constraint::OneOfConstraint const & cons) const {
    size_t matches = 0;
    for (schema::Node const * subschema : cons.children) {
      if (validate_subschema(subschema)) {
        ++matches;
      }
    }

    return matches == 1 ? Status::Accept : Status::Reject;
  }

  Status visit(constraint::NotConstraint const & cons) const {
    return validate_subschema(cons.child) == Status::Reject;
  }

  Status visit(constraint::ConditionalConstraint const & cons) const {
    if (validate_subschema(cons.if_constraint)) {
      return validate_subschema(cons.then_constraint);
    }
    return validate_subschema(cons.else_constraint);
  }

  Status visit(constraint::MaximumConstraint const & cons) const {
    switch (document_.type()) {
    case adapter::Type::Integer:
      return cons(document_.as_integer());
    case adapter::Type::Number:
      return cons(document_.as_number());
    default:
      return Status::Noop;
    }
  }

  Status visit(constraint::MinimumConstraint const & cons) const {
    switch (document_.type()) {
    case adapter::Type::Integer:
      return cons(document_.as_integer());
    case adapter::Type::Number:
      return cons(document_.as_number());
    default:
      return Status::Noop;
    }
  }

  Status visit(constraint::MultipleOfConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Integer);
    return cons(document_.as_integer());
  }

  Status visit(constraint::MaxLengthConstraint const & cons) const {
    NOOP_UNLESS_TYPE(String);
    return cons(document_.as_string());
  }

  Status visit(constraint::MinLengthConstraint const & cons) const {
    NOOP_UNLESS_TYPE(String);
    return cons(document_.as_string());
  }

  Status visit(constraint::PatternConstraint const & cons) const {
    NOOP_UNLESS_TYPE(String);

    RE const & regex = regex_cache_.try_emplace(cons.regex, cons.regex).first->second;
    return regex.search(document_.as_string());
  }

  Status visit(constraint::AdditionalItemsConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Array);

    auto array = document_.as_array();

    Status rval = Status::Accept;
    for (size_t i = cons.applies_after_nth; i < array.size(); ++i) {
      rval &= validate_subschema_on(cons.subschema, array[i], i);
      BREAK_EARLY_IF_NO_RESULT_TREE();
    }

    return rval;
  }

  Status visit(constraint::ContainsConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Array);

    auto array = document_.as_array();
    size_t const minimum = cons.minimum.value_or(1);
    size_t const maximum = cons.maximum.value_or(array.size());
    size_t matches = 0;
    for (size_t i = 0; i < array.size(); ++i) {
      if (validate_subschema_on(cons.subschema, array[i], i)) {
        ++matches;
      }
    }

    if (matches < minimum) {
      return Status::Reject;
    }
    if (matches > maximum) {
      return Status::Reject;
    }
    return Status::Accept;
  }

  Status visit(constraint::MaxItemsConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Array);
    return cons(document_.as_array());
  }

  Status visit(constraint::MinItemsConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Array);
    return cons(document_.as_array());
  }

  Status visit(constraint::TupleConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Array);

    Status rval = Status::Accept;

    auto array = document_.as_array();
    size_t const n = std::min(cons.items.size(), array.size());
    for (size_t i = 0; i < n; ++i) {
      rval &= validate_subschema_on(cons.items[i], array[i], i);
      BREAK_EARLY_IF_NO_RESULT_TREE();
    }

    return rval;
  }

  Status visit(constraint::UniqueItemsConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Array);

    if constexpr (std::totally_ordered<A>) {
      std::set<A> cache;
      for (A const & elem : document_.as_array()) {
        if (not cache.insert(elem).second) {
          return Status::Reject;
        }
      }
    } else {
      auto array = document_.as_array();
      for (size_t i = 0; i < array.size(); ++i) {
        for (size_t j = i + 1; j < array.size(); ++j) {
          if (array[i].equals(array[j], true)) {
            return Status::Reject;
          }
        }
      }
    }

    return Status::Accept;
  }

  Status visit(constraint::AdditionalPropertiesConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Object);

    auto matches_any_pattern = [this, &cons](std::string const & key) {
      for (auto & pattern : cons.patterns) {
        RE const & regex = regex_cache_.try_emplace(pattern, pattern).first->second;
        if (regex.search(key)) {
          return true;
        }
      }
      return false;
    };

    Status rval = Status::Accept;
    for (auto const & [key, elem] : document_.as_object()) {
      if (not cons.properties.contains(key) && not matches_any_pattern(key)) {
        rval &= validate_subschema_on(cons.subschema, elem, key);
      }
      BREAK_EARLY_IF_NO_RESULT_TREE();
    }

    return rval;
  }

  Status visit(constraint::DependenciesConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Object);

    auto object = document_.as_object();
    Status rval = Status::Accept;
    for (auto const & [key, subschema] : cons.subschemas) {
      if (not object.contains(key)) {
        continue;
      }

      rval &= validate_subschema(subschema);
      BREAK_EARLY_IF_NO_RESULT_TREE();
    }

    for (auto [key, required] : cons.required) {
      if (not object.contains(key)) {
        continue;
      }

      for (auto const & [key, _] : object) {
        required.erase(key);
      }

      rval &= required.empty();
      BREAK_EARLY_IF_NO_RESULT_TREE();
    }

    return rval;
  }

  Status visit(constraint::MaxPropertiesConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Object);
    return cons(document_.as_object());
  }

  Status visit(constraint::MinPropertiesConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Object);
    return cons(document_.as_object());
  }

  Status visit(constraint::PatternPropertiesConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Object);

    Status rval = Status::Accept;
    for (auto const & [pattern, subschema] : cons.properties) {
      RE const & regex = regex_cache_.try_emplace(pattern, pattern).first->second;
      for (auto const & [key, elem] : document_.as_object()) {
        if (regex.search(key)) {
          rval &= validate_subschema_on(subschema, elem, key);
        }
        BREAK_EARLY_IF_NO_RESULT_TREE();
      }
    }

    return rval;
  }

  Status visit(constraint::PropertiesConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Object);

    Status rval = Status::Accept;
    auto object = document_.as_object();

    if constexpr (MutableAdapter<A>) {
      for (auto const & [key, subschema] : cons.properties) {
        auto const * default_value = subschema->default_value();
        if (default_value && not object.contains(key)) {
          object.assign(key, *default_value);
        }
      }
    }

    for (auto const & [key, elem] : object) {
      if (auto it = cons.properties.find(key); it != cons.properties.end()) {
        rval &= validate_subschema_on(it->second, elem, key);
      }
      BREAK_EARLY_IF_NO_RESULT_TREE();
    }

    return rval;
  }

  Status visit(constraint::PropertyNamesConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Object);

    Status rval = Status::Accept;
    for (auto const & [key, _] : document_.as_object()) {
      // TODO(samjaffe): Should we prefer a std::string adapter like valijson?
      typename A::value_type key_json{key};
      rval &= validate_subschema_on(cons.key_schema, A(key_json), "$$key");
    }
    return rval;
  }

  Status visit(constraint::RequiredConstraint const & cons) const {
    NOOP_UNLESS_TYPE(Object);

    auto required = cons.properties;
    for (auto const & [key, _] : document_.as_object()) {
      required.erase(key);
    }

    return required.empty();
  }

  Status visit(constraint::UnevaluatedItemsConstraint const & cons) const {
    EXPECT_M(local_result_, "Invalid State - no result object for post-constraint");
    NOOP_UNLESS_TYPE(Array);

    Status rval = Status::Accept;
    auto array = document_.as_array();
    for (size_t i = 0; i < array.size(); ++i) {
      if (not local_result_->has_visited(i)) {
        rval &= validate_subschema_on(cons.subschema, array[i], i);
      }
      BREAK_EARLY_IF_NO_RESULT_TREE();
    }
  }

  Status visit(constraint::UnevaluatedPropertiesConstraint const & cons) const {
    EXPECT_M(local_result_, "Invalid State - no result object for post-constraint");
    NOOP_UNLESS_TYPE(Object);

    Status rval = Status::Accept;
    for (auto const & [key, elem] : document_.as_object()) {
      if (not local_result_->has_visited(key)) {
        rval &= validate_subschema_on(cons.subschema, elem, key);
      }
      BREAK_EARLY_IF_NO_RESULT_TREE();
    }
  }

  Status validate() {
    if (schema_.rejects_all()) {
      return Status::Reject;
    }

    ValidationResult local_result;
    if (schema_.requires_result_context() && not local_result_) {
      // Ensure that we store results even if there aren't any...
      local_result_ = &local_result;
    }

    Status rval = Status::Noop;
    if (auto ref = schema_.reference_schema()) {
      rval = validate_subschema(*ref);
    }

    for (auto const & [key, p_constraint] : schema_.constraints()) {
      BREAK_EARLY_IF_NO_RESULT_TREE();
      if (result_) {
        result_->constraint(key);
      }
      rval &= p_constraint->accept(*this);
    }

    for (auto const & [key, p_constraint] : schema_.post_constraints()) {
      BREAK_EARLY_IF_NO_RESULT_TREE();
      if (result_) {
        result_->constraint(key);
      }
      rval &= p_constraint->accept(*this);
    }

    return rval;
  }

private:
  template <typename... Args> void add_error(Args &&... args) const {
    if (not result_) {
      return;
    }
    std::stringstream ss;
    ss << (std::forward<Args>(args) << ...);
    result_->message(ss.str());
  }

  ValidationVisitor(A const & json, schema::Node const & schema, ValidationConfig const & cfg,
                    std::unordered_map<std::string, RE> & regex_cache,
                    detail::Pointer const & where, ValidationResult * result,
                    ValidationResult * local_result = nullptr)
      : document_(json), where_(where), schema_(schema), cfg_(cfg), regex_cache_(regex_cache),
        result_(result), local_result_(local_result ?: result_) {}

  Status validate_subschema(schema::Node const * subschema) const {
    return ValidationVisitor(document_, *subschema, cfg_, regex_cache_, where_, result_,
                             local_result_)
        .validate();
  }

  template <typename K>
  Status validate_subschema_on(schema::Node const * subschema, A const & document,
                               K const & key) const {
    ValidationResult next;
    ValidationResult * pnext = result_ ? &next : nullptr;

    auto status =
        ValidationVisitor(document, *subschema, cfg_, regex_cache_, where_ / key, pnext).validate();
    if (status != Status::Noop and local_result_) {
      local_result_->visit(key);
    }
    if (status == Status::Reject and result_) {
      result_->error(key, std::move(next));
    }
    return status;
  }
};
}