validation_visitor.h 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508
  1. #pragma once
  2. #include <unordered_map>
  3. #include <jvalidate/constraint/array_constraint.h>
  4. #include <jvalidate/constraint/general_constraint.h>
  5. #include <jvalidate/constraint/number_constraint.h>
  6. #include <jvalidate/constraint/object_constraint.h>
  7. #include <jvalidate/constraint/string_constraint.h>
  8. #include <jvalidate/constraint/visitor.h>
  9. #include <jvalidate/detail/expect.h>
  10. #include <jvalidate/detail/iostream.h>
  11. #include <jvalidate/detail/pointer.h>
  12. #include <jvalidate/forward.h>
  13. #include <jvalidate/schema.h>
  14. #include <jvalidate/status.h>
  15. #include <jvalidate/validation_config.h>
  16. #include <jvalidate/validation_result.h>
  17. #define VISITED(type) std::get<std::unordered_set<type>>(visited_)
  18. #define NOOP_UNLESS_TYPE(etype) RETURN_UNLESS(adapter::Type::etype & document_.type(), Status::Noop)
  19. #define BREAK_EARLY_IF_NO_RESULT_TREE() \
  20. do { \
  21. if (rval == Status::Reject and not result_) { \
  22. break; \
  23. } \
  24. } while (false)
  25. namespace jvalidate {
  26. template <Adapter A, RegexEngine RE>
  27. class ValidationVisitor : public constraint::ConstraintVisitor {
  28. private:
  29. A document_;
  30. detail::Pointer where_;
  31. detail::Pointer schema_path_;
  32. schema::Node const & schema_;
  33. ValidationResult * result_;
  34. ValidationConfig const & cfg_;
  35. std::unordered_map<std::string, RE> & regex_cache_;
  36. mutable std::tuple<std::unordered_set<size_t>, std::unordered_set<std::string>> visited_;
  37. public:
  38. ValidationVisitor(A const & json, schema::Node const & schema, ValidationConfig const & cfg,
  39. std::unordered_map<std::string, RE> & regex_cache, ValidationResult * result)
  40. : ValidationVisitor(json, schema, cfg, regex_cache, {}, {}, result) {}
  41. Status visit(constraint::TypeConstraint const & cons) const {
  42. adapter::Type const type = document_.type();
  43. for (adapter::Type const accept : cons.types) {
  44. if (accept & type) {
  45. return Status::Accept;
  46. }
  47. }
  48. add_error("type ", type, " is not one of {", cons.types, '}');
  49. return Status::Reject;
  50. }
  51. Status visit(constraint::ExtensionConstraint const & cons) const {
  52. return cons.validate(document_, where_, result_);
  53. }
  54. Status visit(constraint::EnumConstraint const & cons) const {
  55. auto is_equal = [this](auto const & frozen) {
  56. return document_.equals(frozen, cfg_.strict_equality);
  57. };
  58. for (auto const & option : cons.enumeration) {
  59. if (option->apply(is_equal)) {
  60. return Status::Accept;
  61. }
  62. }
  63. add_error("equals none of the values");
  64. return Status::Reject;
  65. }
  66. Status visit(constraint::AllOfConstraint const & cons) const {
  67. Status rval = Status::Accept;
  68. size_t i = 0;
  69. for (schema::Node const * subschema : cons.children) {
  70. rval &= validate_subschema(subschema, i);
  71. ++i;
  72. BREAK_EARLY_IF_NO_RESULT_TREE();
  73. }
  74. return rval;
  75. }
  76. Status visit(constraint::AnyOfConstraint const & cons) const {
  77. size_t i = 0;
  78. Status rval = Status::Reject;
  79. for (schema::Node const * subschema : cons.children) {
  80. if (validate_subschema(subschema, i)) {
  81. // TODO(samjaffe): Perhaps we should store some UnevaluatedTracking object for short-circuit
  82. rval = Status::Accept;
  83. }
  84. ++i;
  85. }
  86. return rval;
  87. }
  88. Status visit(constraint::OneOfConstraint const & cons) const {
  89. size_t matches = 0;
  90. size_t i = 0;
  91. for (schema::Node const * subschema : cons.children) {
  92. if (validate_subschema(subschema, i)) {
  93. ++matches;
  94. }
  95. ++i;
  96. }
  97. return matches == 1 ? Status::Accept : Status::Reject;
  98. }
  99. Status visit(constraint::NotConstraint const & cons) const {
  100. return validate_subschema(cons.child, "") == Status::Reject;
  101. }
  102. Status visit(constraint::ConditionalConstraint const & cons) const {
  103. if (validate_subschema(cons.if_constraint, "")) {
  104. return validate_subschema(cons.then_constraint, "then");
  105. }
  106. return validate_subschema(cons.else_constraint, "else");
  107. }
  108. Status visit(constraint::MaximumConstraint const & cons) const {
  109. switch (document_.type()) {
  110. case adapter::Type::Integer:
  111. return cons(document_.as_integer());
  112. case adapter::Type::Number:
  113. return cons(document_.as_number());
  114. default:
  115. return Status::Noop;
  116. }
  117. }
  118. Status visit(constraint::MinimumConstraint const & cons) const {
  119. switch (document_.type()) {
  120. case adapter::Type::Integer:
  121. return cons(document_.as_integer());
  122. case adapter::Type::Number:
  123. return cons(document_.as_number());
  124. default:
  125. return Status::Noop;
  126. }
  127. }
  128. Status visit(constraint::MultipleOfConstraint const & cons) const {
  129. NOOP_UNLESS_TYPE(Number);
  130. return cons(document_.as_number());
  131. }
  132. Status visit(constraint::MaxLengthConstraint const & cons) const {
  133. NOOP_UNLESS_TYPE(String);
  134. return cons(document_.as_string());
  135. }
  136. Status visit(constraint::MinLengthConstraint const & cons) const {
  137. NOOP_UNLESS_TYPE(String);
  138. return cons(document_.as_string());
  139. }
  140. Status visit(constraint::PatternConstraint const & cons) const {
  141. NOOP_UNLESS_TYPE(String);
  142. RE const & regex = regex_cache_.try_emplace(cons.regex, cons.regex).first->second;
  143. return regex.search(document_.as_string());
  144. }
  145. Status visit(constraint::AdditionalItemsConstraint const & cons) const {
  146. NOOP_UNLESS_TYPE(Array);
  147. auto array = document_.as_array();
  148. Status rval = Status::Accept;
  149. for (size_t i = cons.applies_after_nth; i < array.size(); ++i) {
  150. rval &= validate_subschema_on(cons.subschema, array[i], i);
  151. BREAK_EARLY_IF_NO_RESULT_TREE();
  152. }
  153. return rval;
  154. }
  155. Status visit(constraint::ContainsConstraint const & cons) const {
  156. NOOP_UNLESS_TYPE(Array);
  157. auto array = document_.as_array();
  158. size_t const minimum = cons.minimum.value_or(1);
  159. size_t const maximum = cons.maximum.value_or(array.size());
  160. size_t matches = 0;
  161. for (size_t i = 0; i < array.size(); ++i) {
  162. if (validate_subschema_on(cons.subschema, array[i], i)) {
  163. ++matches;
  164. }
  165. }
  166. if (matches < minimum) {
  167. return Status::Reject;
  168. }
  169. if (matches > maximum) {
  170. return Status::Reject;
  171. }
  172. return Status::Accept;
  173. }
  174. Status visit(constraint::MaxItemsConstraint const & cons) const {
  175. NOOP_UNLESS_TYPE(Array);
  176. return cons(document_.as_array());
  177. }
  178. Status visit(constraint::MinItemsConstraint const & cons) const {
  179. NOOP_UNLESS_TYPE(Array);
  180. return cons(document_.as_array());
  181. }
  182. Status visit(constraint::TupleConstraint const & cons) const {
  183. NOOP_UNLESS_TYPE(Array);
  184. Status rval = Status::Accept;
  185. auto array = document_.as_array();
  186. size_t const n = std::min(cons.items.size(), array.size());
  187. for (size_t i = 0; i < n; ++i) {
  188. rval &= validate_subschema_on(cons.items[i], array[i], i);
  189. BREAK_EARLY_IF_NO_RESULT_TREE();
  190. }
  191. return rval;
  192. }
  193. Status visit(constraint::UniqueItemsConstraint const & cons) const {
  194. NOOP_UNLESS_TYPE(Array);
  195. if constexpr (std::totally_ordered<A>) {
  196. std::set<A> cache;
  197. for (A const & elem : document_.as_array()) {
  198. if (not cache.insert(elem).second) {
  199. return Status::Reject;
  200. }
  201. }
  202. } else {
  203. auto array = document_.as_array();
  204. for (size_t i = 0; i < array.size(); ++i) {
  205. for (size_t j = i + 1; j < array.size(); ++j) {
  206. if (array[i].equals(array[j], true)) {
  207. return Status::Reject;
  208. }
  209. }
  210. }
  211. }
  212. return Status::Accept;
  213. }
  214. Status visit(constraint::AdditionalPropertiesConstraint const & cons) const {
  215. NOOP_UNLESS_TYPE(Object);
  216. auto matches_any_pattern = [this, &cons](std::string const & key) {
  217. for (auto & pattern : cons.patterns) {
  218. RE const & regex = regex_cache_.try_emplace(pattern, pattern).first->second;
  219. if (regex.search(key)) {
  220. return true;
  221. }
  222. }
  223. return false;
  224. };
  225. Status rval = Status::Accept;
  226. for (auto const & [key, elem] : document_.as_object()) {
  227. if (not cons.properties.contains(key) && not matches_any_pattern(key)) {
  228. rval &= validate_subschema_on(cons.subschema, elem, key);
  229. }
  230. BREAK_EARLY_IF_NO_RESULT_TREE();
  231. }
  232. return rval;
  233. }
  234. Status visit(constraint::DependenciesConstraint const & cons) const {
  235. NOOP_UNLESS_TYPE(Object);
  236. auto object = document_.as_object();
  237. Status rval = Status::Accept;
  238. for (auto const & [key, subschema] : cons.subschemas) {
  239. if (not object.contains(key)) {
  240. continue;
  241. }
  242. rval &= validate_subschema(subschema, key);
  243. BREAK_EARLY_IF_NO_RESULT_TREE();
  244. }
  245. for (auto [key, required] : cons.required) {
  246. if (not object.contains(key)) {
  247. continue;
  248. }
  249. for (auto const & [key, _] : object) {
  250. required.erase(key);
  251. }
  252. rval &= required.empty();
  253. BREAK_EARLY_IF_NO_RESULT_TREE();
  254. }
  255. return rval;
  256. }
  257. Status visit(constraint::MaxPropertiesConstraint const & cons) const {
  258. NOOP_UNLESS_TYPE(Object);
  259. return cons(document_.as_object());
  260. }
  261. Status visit(constraint::MinPropertiesConstraint const & cons) const {
  262. NOOP_UNLESS_TYPE(Object);
  263. return cons(document_.as_object());
  264. }
  265. Status visit(constraint::PatternPropertiesConstraint const & cons) const {
  266. NOOP_UNLESS_TYPE(Object);
  267. Status rval = Status::Accept;
  268. for (auto const & [pattern, subschema] : cons.properties) {
  269. RE const & regex = regex_cache_.try_emplace(pattern, pattern).first->second;
  270. for (auto const & [key, elem] : document_.as_object()) {
  271. if (regex.search(key)) {
  272. rval &= validate_subschema_on(subschema, elem, key);
  273. }
  274. BREAK_EARLY_IF_NO_RESULT_TREE();
  275. }
  276. }
  277. return rval;
  278. }
  279. Status visit(constraint::PropertiesConstraint const & cons) const {
  280. NOOP_UNLESS_TYPE(Object);
  281. Status rval = Status::Accept;
  282. auto object = document_.as_object();
  283. if constexpr (MutableAdapter<A>) {
  284. for (auto const & [key, subschema] : cons.properties) {
  285. auto const * default_value = subschema->default_value();
  286. if (default_value && not object.contains(key)) {
  287. object.assign(key, *default_value);
  288. }
  289. }
  290. }
  291. for (auto const & [key, elem] : object) {
  292. if (auto it = cons.properties.find(key); it != cons.properties.end()) {
  293. rval &= validate_subschema_on(it->second, elem, key);
  294. }
  295. BREAK_EARLY_IF_NO_RESULT_TREE();
  296. }
  297. return rval;
  298. }
  299. Status visit(constraint::PropertyNamesConstraint const & cons) const {
  300. NOOP_UNLESS_TYPE(Object);
  301. Status rval = Status::Accept;
  302. for (auto const & [key, _] : document_.as_object()) {
  303. // TODO(samjaffe): Should we prefer a std::string adapter like valijson?
  304. typename A::value_type key_json{key};
  305. rval &= validate_subschema_on(cons.key_schema, A(key_json), std::string("$$key"));
  306. }
  307. return rval;
  308. }
  309. Status visit(constraint::RequiredConstraint const & cons) const {
  310. NOOP_UNLESS_TYPE(Object);
  311. auto required = cons.properties;
  312. for (auto const & [key, _] : document_.as_object()) {
  313. required.erase(key);
  314. }
  315. if (required.empty()) {
  316. return Status::Accept;
  317. }
  318. add_error("missing required properties ", required);
  319. return Status::Reject;
  320. }
  321. Status visit(constraint::UnevaluatedItemsConstraint const & cons) const {
  322. NOOP_UNLESS_TYPE(Array);
  323. Status rval = Status::Accept;
  324. auto array = document_.as_array();
  325. for (size_t i = 0; i < array.size(); ++i) {
  326. if (not VISITED(size_t).contains(i)) {
  327. rval &= validate_subschema_on(cons.subschema, array[i], i);
  328. }
  329. BREAK_EARLY_IF_NO_RESULT_TREE();
  330. }
  331. return rval;
  332. }
  333. Status visit(constraint::UnevaluatedPropertiesConstraint const & cons) const {
  334. NOOP_UNLESS_TYPE(Object);
  335. Status rval = Status::Accept;
  336. for (auto const & [key, elem] : document_.as_object()) {
  337. if (not VISITED(std::string).contains(key)) {
  338. rval &= validate_subschema_on(cons.subschema, elem, key);
  339. }
  340. BREAK_EARLY_IF_NO_RESULT_TREE();
  341. }
  342. return rval;
  343. }
  344. Status validate() {
  345. if (auto const & reject = schema_.rejects_all()) {
  346. add_error(*reject);
  347. return Status::Reject;
  348. }
  349. if (schema_.accepts_all()) {
  350. // An accept-all schema is not No-Op for the purpose of unevaluated*
  351. return Status::Accept;
  352. }
  353. Status rval = Status::Noop;
  354. if (auto ref = schema_.reference_schema()) {
  355. rval = validate_subschema(*ref, "$ref");
  356. }
  357. detail::Pointer const current_schema = schema_path_;
  358. for (auto const & [key, p_constraint] : schema_.constraints()) {
  359. BREAK_EARLY_IF_NO_RESULT_TREE();
  360. schema_path_ = current_schema / key;
  361. rval &= p_constraint->accept(*this);
  362. }
  363. for (auto const & [key, p_constraint] : schema_.post_constraints()) {
  364. BREAK_EARLY_IF_NO_RESULT_TREE();
  365. schema_path_ = current_schema / key;
  366. rval &= p_constraint->accept(*this);
  367. }
  368. return rval;
  369. }
  370. private:
  371. template <typename... Args> void add_error(Args &&... args) const {
  372. if (not result_) {
  373. return;
  374. }
  375. std::stringstream ss;
  376. using ::jvalidate::operator<<;
  377. [[maybe_unused]] int _[] = {(ss << std::forward<Args>(args), 0)...};
  378. result_->add_error(where_, schema_path_, ss.str());
  379. }
  380. template <typename C> static void merge_visited(C & to, C const & from) {
  381. to.insert(from.begin(), from.end());
  382. }
  383. ValidationVisitor(A const & json, schema::Node const & schema, ValidationConfig const & cfg,
  384. std::unordered_map<std::string, RE> & regex_cache,
  385. detail::Pointer const & where, detail::Pointer const & schema_path,
  386. ValidationResult * result)
  387. : document_(json), where_(where), schema_path_(schema_path), schema_(schema), cfg_(cfg),
  388. regex_cache_(regex_cache), result_(result) {}
  389. template <typename K>
  390. Status validate_subschema(schema::Node const * subschema, K const & key) const {
  391. EXPECT(subschema != &schema_); // TODO(samjaffe) - Figure out what's causing this infinite loop
  392. ValidationVisitor next(document_, *subschema, cfg_, regex_cache_, where_, schema_path_ / key,
  393. result_);
  394. Status rval = next.validate();
  395. merge_visited(std::get<0>(visited_), std::get<0>(next.visited_));
  396. merge_visited(std::get<1>(visited_), std::get<1>(next.visited_));
  397. return rval;
  398. }
  399. template <typename K>
  400. Status validate_subschema_on(schema::Node const * subschema, A const & document,
  401. K const & key) const {
  402. ValidationResult next;
  403. ValidationResult * pnext = result_ ? &next : nullptr;
  404. auto status = ValidationVisitor(document, *subschema, cfg_, regex_cache_, where_ / key,
  405. schema_path_, pnext)
  406. .validate();
  407. if (status != Status::Noop) {
  408. VISITED(K).insert(key);
  409. }
  410. if (status == Status::Reject and result_) {
  411. result_->add_error(std::move(next));
  412. }
  413. return status;
  414. }
  415. };
  416. }