validation_visitor.h 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531
  1. #pragma once
  2. #include <tuple>
  3. #include <unordered_map>
  4. #include <jvalidate/constraint/array_constraint.h>
  5. #include <jvalidate/constraint/general_constraint.h>
  6. #include <jvalidate/constraint/number_constraint.h>
  7. #include <jvalidate/constraint/object_constraint.h>
  8. #include <jvalidate/constraint/string_constraint.h>
  9. #include <jvalidate/constraint/visitor.h>
  10. #include <jvalidate/detail/expect.h>
  11. #include <jvalidate/detail/iostream.h>
  12. #include <jvalidate/detail/pointer.h>
  13. #include <jvalidate/forward.h>
  14. #include <jvalidate/schema.h>
  15. #include <jvalidate/status.h>
  16. #include <jvalidate/validation_config.h>
  17. #include <jvalidate/validation_result.h>
  18. #define VISITED(type) std::get<std::unordered_set<type>>(*visited_)
  19. #define NOOP_UNLESS_TYPE(etype) RETURN_UNLESS(adapter::Type::etype & document_.type(), Status::Noop)
  20. #define BREAK_EARLY_IF_NO_RESULT_TREE() \
  21. do { \
  22. if (rval == Status::Reject and not result_) { \
  23. break; \
  24. } \
  25. } while (false)
  26. namespace jvalidate {
  27. template <Adapter A, RegexEngine RE>
  28. class ValidationVisitor : public constraint::ConstraintVisitor {
  29. private:
  30. using VisitedAnnotation = std::tuple<std::unordered_set<size_t>, std::unordered_set<std::string>>;
  31. private:
  32. A document_;
  33. detail::Pointer where_;
  34. detail::Pointer schema_path_;
  35. schema::Node const * schema_;
  36. ValidationResult * result_;
  37. ValidationConfig const & cfg_;
  38. std::unordered_map<std::string, RE> & regex_cache_;
  39. mutable VisitedAnnotation * visited_ = nullptr;
  40. public:
  41. ValidationVisitor(A const & json, schema::Node const & schema, ValidationConfig const & cfg,
  42. std::unordered_map<std::string, RE> & regex_cache, ValidationResult * result)
  43. : document_(json), schema_(&schema), result_(result), cfg_(cfg), regex_cache_(regex_cache) {}
  44. Status visit(constraint::TypeConstraint const & cons) const {
  45. adapter::Type const type = document_.type();
  46. for (adapter::Type const accept : cons.types) {
  47. if (accept & type) {
  48. return Status::Accept;
  49. }
  50. }
  51. add_error("type ", type, " is not one of {", cons.types, '}');
  52. return Status::Reject;
  53. }
  54. Status visit(constraint::ExtensionConstraint const & cons) const {
  55. return cons.validate(document_, where_, result_);
  56. }
  57. Status visit(constraint::EnumConstraint const & cons) const {
  58. auto is_equal = [this](auto const & frozen) {
  59. return document_.equals(frozen, cfg_.strict_equality);
  60. };
  61. for (auto const & option : cons.enumeration) {
  62. if (option->apply(is_equal)) {
  63. return Status::Accept;
  64. }
  65. }
  66. add_error("equals none of the values");
  67. return Status::Reject;
  68. }
  69. Status visit(constraint::AllOfConstraint const & cons) const {
  70. Status rval = Status::Accept;
  71. size_t i = 0;
  72. for (schema::Node const * subschema : cons.children) {
  73. rval &= validate_subschema(subschema, i);
  74. ++i;
  75. BREAK_EARLY_IF_NO_RESULT_TREE();
  76. }
  77. return rval;
  78. }
  79. Status visit(constraint::AnyOfConstraint const & cons) const {
  80. size_t i = 0;
  81. Status rval = Status::Reject;
  82. for (schema::Node const * subschema : cons.children) {
  83. if (validate_subschema(subschema, i)) {
  84. // TODO(samjaffe): Perhaps we should store some UnevaluatedTracking object for short-circuit
  85. rval = Status::Accept;
  86. }
  87. ++i;
  88. }
  89. return rval;
  90. }
  91. Status visit(constraint::OneOfConstraint const & cons) const {
  92. size_t matches = 0;
  93. size_t i = 0;
  94. for (schema::Node const * subschema : cons.children) {
  95. if (validate_subschema(subschema, i)) {
  96. ++matches;
  97. }
  98. ++i;
  99. }
  100. return matches == 1 ? Status::Accept : Status::Reject;
  101. }
  102. Status visit(constraint::NotConstraint const & cons) const {
  103. VisitedAnnotation * suppress = nullptr;
  104. std::swap(suppress, visited_);
  105. auto rval = validate_subschema(cons.child, "") == Status::Reject;
  106. std::swap(suppress, visited_);
  107. return rval;
  108. }
  109. Status visit(constraint::ConditionalConstraint const & cons) const {
  110. if (validate_subschema(cons.if_constraint, "")) {
  111. return validate_subschema(cons.then_constraint, "then");
  112. }
  113. return validate_subschema(cons.else_constraint, "else");
  114. }
  115. Status visit(constraint::MaximumConstraint const & cons) const {
  116. switch (document_.type()) {
  117. case adapter::Type::Integer:
  118. return cons(document_.as_integer());
  119. case adapter::Type::Number:
  120. return cons(document_.as_number());
  121. default:
  122. return Status::Noop;
  123. }
  124. }
  125. Status visit(constraint::MinimumConstraint const & cons) const {
  126. switch (document_.type()) {
  127. case adapter::Type::Integer:
  128. return cons(document_.as_integer());
  129. case adapter::Type::Number:
  130. return cons(document_.as_number());
  131. default:
  132. return Status::Noop;
  133. }
  134. }
  135. Status visit(constraint::MultipleOfConstraint const & cons) const {
  136. NOOP_UNLESS_TYPE(Number);
  137. return cons(document_.as_number());
  138. }
  139. Status visit(constraint::MaxLengthConstraint const & cons) const {
  140. NOOP_UNLESS_TYPE(String);
  141. return cons(document_.as_string());
  142. }
  143. Status visit(constraint::MinLengthConstraint const & cons) const {
  144. NOOP_UNLESS_TYPE(String);
  145. return cons(document_.as_string());
  146. }
  147. Status visit(constraint::PatternConstraint const & cons) const {
  148. NOOP_UNLESS_TYPE(String);
  149. RE const & regex = regex_cache_.try_emplace(cons.regex, cons.regex).first->second;
  150. return regex.search(document_.as_string());
  151. }
  152. Status visit(constraint::AdditionalItemsConstraint const & cons) const {
  153. NOOP_UNLESS_TYPE(Array);
  154. auto array = document_.as_array();
  155. Status rval = Status::Accept;
  156. for (size_t i = cons.applies_after_nth; i < array.size(); ++i) {
  157. rval &= validate_subschema_on(cons.subschema, array[i], i);
  158. BREAK_EARLY_IF_NO_RESULT_TREE();
  159. }
  160. return rval;
  161. }
  162. Status visit(constraint::ContainsConstraint const & cons) const {
  163. NOOP_UNLESS_TYPE(Array);
  164. auto array = document_.as_array();
  165. size_t const minimum = cons.minimum.value_or(1);
  166. size_t const maximum = cons.maximum.value_or(array.size());
  167. size_t matches = 0;
  168. for (size_t i = 0; i < array.size(); ++i) {
  169. if (validate_subschema_on(cons.subschema, array[i], i)) {
  170. ++matches;
  171. }
  172. }
  173. if (matches < minimum) {
  174. return Status::Reject;
  175. }
  176. if (matches > maximum) {
  177. return Status::Reject;
  178. }
  179. return Status::Accept;
  180. }
  181. Status visit(constraint::MaxItemsConstraint const & cons) const {
  182. NOOP_UNLESS_TYPE(Array);
  183. return cons(document_.as_array());
  184. }
  185. Status visit(constraint::MinItemsConstraint const & cons) const {
  186. NOOP_UNLESS_TYPE(Array);
  187. return cons(document_.as_array());
  188. }
  189. Status visit(constraint::TupleConstraint const & cons) const {
  190. NOOP_UNLESS_TYPE(Array);
  191. Status rval = Status::Accept;
  192. auto array = document_.as_array();
  193. size_t const n = std::min(cons.items.size(), array.size());
  194. for (size_t i = 0; i < n; ++i) {
  195. rval &= validate_subschema_on(cons.items[i], array[i], i);
  196. BREAK_EARLY_IF_NO_RESULT_TREE();
  197. }
  198. return rval;
  199. }
  200. Status visit(constraint::UniqueItemsConstraint const & cons) const {
  201. NOOP_UNLESS_TYPE(Array);
  202. if constexpr (std::totally_ordered<A>) {
  203. std::set<A> cache;
  204. for (A const & elem : document_.as_array()) {
  205. if (not cache.insert(elem).second) {
  206. return Status::Reject;
  207. }
  208. }
  209. } else {
  210. auto array = document_.as_array();
  211. for (size_t i = 0; i < array.size(); ++i) {
  212. for (size_t j = i + 1; j < array.size(); ++j) {
  213. if (array[i].equals(array[j], true)) {
  214. return Status::Reject;
  215. }
  216. }
  217. }
  218. }
  219. return Status::Accept;
  220. }
  221. Status visit(constraint::AdditionalPropertiesConstraint const & cons) const {
  222. NOOP_UNLESS_TYPE(Object);
  223. auto matches_any_pattern = [this, &cons](std::string const & key) {
  224. for (auto & pattern : cons.patterns) {
  225. RE const & regex = regex_cache_.try_emplace(pattern, pattern).first->second;
  226. if (regex.search(key)) {
  227. return true;
  228. }
  229. }
  230. return false;
  231. };
  232. Status rval = Status::Accept;
  233. for (auto const & [key, elem] : document_.as_object()) {
  234. if (not cons.properties.contains(key) && not matches_any_pattern(key)) {
  235. rval &= validate_subschema_on(cons.subschema, elem, key);
  236. }
  237. BREAK_EARLY_IF_NO_RESULT_TREE();
  238. }
  239. return rval;
  240. }
  241. Status visit(constraint::DependenciesConstraint const & cons) const {
  242. NOOP_UNLESS_TYPE(Object);
  243. auto object = document_.as_object();
  244. Status rval = Status::Accept;
  245. for (auto const & [key, subschema] : cons.subschemas) {
  246. if (not object.contains(key)) {
  247. continue;
  248. }
  249. rval &= validate_subschema(subschema, key);
  250. BREAK_EARLY_IF_NO_RESULT_TREE();
  251. }
  252. for (auto [key, required] : cons.required) {
  253. if (not object.contains(key)) {
  254. continue;
  255. }
  256. for (auto const & [key, _] : object) {
  257. required.erase(key);
  258. }
  259. rval &= required.empty();
  260. BREAK_EARLY_IF_NO_RESULT_TREE();
  261. }
  262. return rval;
  263. }
  264. Status visit(constraint::MaxPropertiesConstraint const & cons) const {
  265. NOOP_UNLESS_TYPE(Object);
  266. return cons(document_.as_object());
  267. }
  268. Status visit(constraint::MinPropertiesConstraint const & cons) const {
  269. NOOP_UNLESS_TYPE(Object);
  270. return cons(document_.as_object());
  271. }
  272. Status visit(constraint::PatternPropertiesConstraint const & cons) const {
  273. NOOP_UNLESS_TYPE(Object);
  274. Status rval = Status::Accept;
  275. for (auto const & [pattern, subschema] : cons.properties) {
  276. RE const & regex = regex_cache_.try_emplace(pattern, pattern).first->second;
  277. for (auto const & [key, elem] : document_.as_object()) {
  278. if (regex.search(key)) {
  279. rval &= validate_subschema_on(subschema, elem, key);
  280. }
  281. BREAK_EARLY_IF_NO_RESULT_TREE();
  282. }
  283. }
  284. return rval;
  285. }
  286. Status visit(constraint::PropertiesConstraint const & cons) const {
  287. NOOP_UNLESS_TYPE(Object);
  288. Status rval = Status::Accept;
  289. auto object = document_.as_object();
  290. if constexpr (MutableAdapter<A>) {
  291. for (auto const & [key, subschema] : cons.properties) {
  292. auto const * default_value = subschema->default_value();
  293. if (default_value && not object.contains(key)) {
  294. object.assign(key, *default_value);
  295. }
  296. }
  297. }
  298. for (auto const & [key, elem] : object) {
  299. if (auto it = cons.properties.find(key); it != cons.properties.end()) {
  300. rval &= validate_subschema_on(it->second, elem, key);
  301. }
  302. BREAK_EARLY_IF_NO_RESULT_TREE();
  303. }
  304. return rval;
  305. }
  306. Status visit(constraint::PropertyNamesConstraint const & cons) const {
  307. NOOP_UNLESS_TYPE(Object);
  308. Status rval = Status::Accept;
  309. for (auto const & [key, _] : document_.as_object()) {
  310. // TODO(samjaffe): Should we prefer a std::string adapter like valijson?
  311. typename A::value_type key_json{key};
  312. rval &= validate_subschema_on(cons.key_schema, A(key_json), std::string("$$key"));
  313. }
  314. return rval;
  315. }
  316. Status visit(constraint::RequiredConstraint const & cons) const {
  317. NOOP_UNLESS_TYPE(Object);
  318. auto required = cons.properties;
  319. for (auto const & [key, _] : document_.as_object()) {
  320. required.erase(key);
  321. }
  322. if (required.empty()) {
  323. return Status::Accept;
  324. }
  325. add_error("missing required properties ", required);
  326. return Status::Reject;
  327. }
  328. Status visit(constraint::UnevaluatedItemsConstraint const & cons) const {
  329. NOOP_UNLESS_TYPE(Array);
  330. if (not visited_) {
  331. return Status::Reject;
  332. }
  333. Status rval = Status::Accept;
  334. auto array = document_.as_array();
  335. for (size_t i = 0; i < array.size(); ++i) {
  336. if (not VISITED(size_t).contains(i)) {
  337. rval &= validate_subschema_on(cons.subschema, array[i], i);
  338. }
  339. BREAK_EARLY_IF_NO_RESULT_TREE();
  340. }
  341. return rval;
  342. }
  343. Status visit(constraint::UnevaluatedPropertiesConstraint const & cons) const {
  344. NOOP_UNLESS_TYPE(Object);
  345. if (not visited_) {
  346. return Status::Reject;
  347. }
  348. Status rval = Status::Accept;
  349. for (auto const & [key, elem] : document_.as_object()) {
  350. if (not VISITED(std::string).contains(key)) {
  351. rval &= validate_subschema_on(cons.subschema, elem, key);
  352. }
  353. BREAK_EARLY_IF_NO_RESULT_TREE();
  354. }
  355. return rval;
  356. }
  357. Status validate() {
  358. if (auto const & reject = schema_->rejects_all()) {
  359. add_error(*reject);
  360. return Status::Reject;
  361. }
  362. if (schema_->accepts_all()) {
  363. // An accept-all schema is not No-Op for the purpose of unevaluated*
  364. return Status::Accept;
  365. }
  366. VisitedAnnotation annotate;
  367. if (schema_->requires_result_context() and not visited_) {
  368. visited_ = &annotate;
  369. }
  370. Status rval = Status::Noop;
  371. if (auto ref = schema_->reference_schema()) {
  372. rval = validate_subschema(*ref, "$ref");
  373. }
  374. detail::Pointer const current_schema = schema_path_;
  375. for (auto const & [key, p_constraint] : schema_->constraints()) {
  376. BREAK_EARLY_IF_NO_RESULT_TREE();
  377. schema_path_ = current_schema / key;
  378. rval &= p_constraint->accept(*this);
  379. }
  380. for (auto const & [key, p_constraint] : schema_->post_constraints()) {
  381. BREAK_EARLY_IF_NO_RESULT_TREE();
  382. schema_path_ = current_schema / key;
  383. rval &= p_constraint->accept(*this);
  384. }
  385. return rval;
  386. }
  387. private:
  388. template <typename... Args> void add_error(Args &&... args) const {
  389. if (not result_) {
  390. return;
  391. }
  392. std::stringstream ss;
  393. using ::jvalidate::operator<<;
  394. [[maybe_unused]] int _[] = {(ss << std::forward<Args>(args), 0)...};
  395. result_->add_error(where_, schema_path_, ss.str());
  396. }
  397. template <typename C> static void merge_visited(C & to, C const & from) {
  398. to.insert(from.begin(), from.end());
  399. }
  400. template <typename K>
  401. Status validate_subschema(schema::Node const * subschema, K const & key) const {
  402. EXPECT(subschema != schema_); // TODO(samjaffe) - Figure out what's causing this infinite loop
  403. VisitedAnnotation annotate;
  404. ValidationVisitor next = *this;
  405. next.schema_path_ /= key;
  406. std::tie(next.schema_, next.visited_) =
  407. std::forward_as_tuple(subschema, visited_ ? &annotate : nullptr);
  408. Status rval = next.validate();
  409. if (rval == Status::Accept and visited_) {
  410. merge_visited(std::get<0>(*visited_), std::get<0>(annotate));
  411. merge_visited(std::get<1>(*visited_), std::get<1>(annotate));
  412. }
  413. return rval;
  414. }
  415. template <typename K>
  416. Status validate_subschema_on(schema::Node const * subschema, A const & document,
  417. K const & key) const {
  418. ValidationResult result;
  419. ValidationVisitor next = *this;
  420. next.where_ /= key;
  421. std::tie(next.document_, next.schema_, next.result_, next.visited_) =
  422. std::forward_as_tuple(document, subschema, result_ ? &result : nullptr, nullptr);
  423. auto status = next.validate();
  424. if (status == Status::Accept and visited_) {
  425. VISITED(K).insert(key);
  426. }
  427. if (status == Status::Reject and result_) {
  428. result_->add_error(std::move(result));
  429. }
  430. return status;
  431. }
  432. };
  433. }