Implements issue 1491 (check for unique keys)

doc/basics.md

@@ -193,6 +193,17 @@ support for users who avoid exceptions. See [the simdjson error handling documen
   > double y = doc["y"]; // The cursor is now after the 2 (at })
   > double x = doc["x"]; // Success: [] loops back around to find "x"
   > ```
+  >
+  > By default, the simdjson does not check that keys are unique. If you are concerned
+  > about repeated keys, you may use the `find_unique_field` method instead.
+    > ```c++
+  > ondemand::parser parser;
+  > auto json = R"(  { "x": 1, "y": 2, "y": 3 }  )"_padded;
+  > auto doc = parser.iterate(json);
+  > double x = doc.find_unique_field["x"]; // return 2
+  > double y = doc.find_unique_field["y"]; // will throw with NON_UNIQUE_FIELD error
+  > ```
+  > Users should be aware that `find_unique_field` could be significantly less performant.
 * **Array Iteration:** To iterate through an array, use `for (auto value : array) { ... }`. This will
   step through each value in the JSON array.
 

include/simdjson/error.h

@@ -37,6 +37,7 @@ enum error_code {
   PARSER_IN_USE,            ///< parser is already in use.
   OUT_OF_ORDER_ITERATION,   ///< tried to iterate an array or object out of order
   INSUFFICIENT_PADDING,     ///< The JSON doesn't have enough padding for simdjson to safely parse it.
+  NON_UNIQUE_FIELD,         ///< repeated key in JSON object
   NUM_ERROR_CODES
 };
 

include/simdjson/generic/ondemand/document-inl.h

@@ -98,6 +98,12 @@ simdjson_really_inline simdjson_result<value> document::find_field(std::string_v
 simdjson_really_inline simdjson_result<value> document::find_field(const char *key) & noexcept {
   return resume_value().find_field(key);
 }
+simdjson_really_inline simdjson_result<value> document::find_unique_field(std::string_view key) & noexcept {
+  return resume_value().find_unique_field(key);
+}
+simdjson_really_inline simdjson_result<value> document::find_unique_field(const char *key) & noexcept {
+  return resume_value().find_unique_field(key);
+}
 simdjson_really_inline simdjson_result<value> document::find_field_unordered(std::string_view key) & noexcept {
   return resume_value().find_field_unordered(key);
 }
@@ -150,6 +156,14 @@ simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array_
 simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::array_iterator> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::document>::end() & noexcept {
   return {};
 }
+simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::document>::find_unique_field(std::string_view key) & noexcept {
+  if (error()) { return error(); }
+  return first.find_unique_field(key);
+}
+simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::document>::find_unique_field(const char *key) & noexcept {
+  if (error()) { return error(); }
+  return first.find_unique_field(key);
+}
 simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::document>::find_field_unordered(std::string_view key) & noexcept {
   if (error()) { return error(); }
   return first.find_field_unordered(key);

include/simdjson/generic/ondemand/document.h

@@ -236,6 +236,16 @@ class document {
   /** @overload simdjson_really_inline simdjson_result<value> find_field(std::string_view key) & noexcept; */
   simdjson_really_inline simdjson_result<value> find_field(const char *key) & noexcept;
 
+  /**
+   * Look up a field by name on an object. This function is similar to find_field_unordered but
+   * returns a NON_UNIQUE_FIELD if the key appears more than once.
+   *
+   * @param key The key to look up.
+   * @returns The value of the field, or NO_SUCH_FIELD if the field is not in the object, or NON_UNIQUE_FIELD if the key appears more than once.
+   */
+  simdjson_really_inline simdjson_result<value> find_unique_field(std::string_view key) & noexcept;
+  /** @overload simdjson_really_inline simdjson_result<value> find_unique_field(std::string_view key) & noexcept; */
+  simdjson_really_inline simdjson_result<value> find_unique_field(const char *key) & noexcept;
   /**
    * Look up a field by name on an object, without regard to key order.
    *
@@ -369,6 +379,8 @@ struct simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::document> : public SIM
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_field(const char *key) & noexcept;
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> operator[](std::string_view key) & noexcept;
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> operator[](const char *key) & noexcept;
+  simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_unique_field(std::string_view key) & noexcept;
+  simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_unique_field(const char *key) & noexcept;
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_field_unordered(std::string_view key) & noexcept;
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_field_unordered(const char *key) & noexcept;
 

include/simdjson/generic/ondemand/object-inl.h

@@ -32,7 +32,18 @@ simdjson_really_inline simdjson_result<value> object::find_field(const std::stri
   if (!has_value) { return NO_SUCH_FIELD; }
   return value(iter.child());
 }
-
+simdjson_really_inline simdjson_result<value> object::find_unique_field(const std::string_view key) & noexcept {
+  bool has_value;
+  SIMDJSON_TRY( iter.find_unique_field_raw(key).get(has_value) );
+  if (!has_value) { return NO_SUCH_FIELD; }
+  return value(iter.child());
+}
+simdjson_really_inline simdjson_result<value> object::find_unique_field(const std::string_view key) && noexcept {
+  bool has_value;
+  SIMDJSON_TRY( iter.find_unique_field_raw(key).get(has_value) );
+  if (!has_value) { return NO_SUCH_FIELD; }
+  return value(iter.child());
+}
 simdjson_really_inline simdjson_result<object> object::start(value_iterator &iter) noexcept {
   // We don't need to know if the object is empty to start iteration, but we do want to know if there
   // is an error--thus `simdjson_unused`.
@@ -111,5 +122,13 @@ simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value>
   if (error()) { return error(); }
   return std::forward<SIMDJSON_IMPLEMENTATION::ondemand::object>(first).find_field(key);
 }
+simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::object>::find_unique_field(std::string_view key) & noexcept {
+  if (error()) { return error(); }
+  return first.find_unique_field(key);
+}
+simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::object>::find_unique_field(std::string_view key) && noexcept {
+  if (error()) { return error(); }
+  return std::forward<SIMDJSON_IMPLEMENTATION::ondemand::object>(first).find_unique_field(key);
+}
 
 } // namespace simdjson

include/simdjson/generic/ondemand/object.h

@@ -44,6 +44,16 @@ class object {
   simdjson_really_inline simdjson_result<value> find_field(std::string_view key) & noexcept;
   /** @overload simdjson_really_inline simdjson_result<value> find_field(std::string_view key) & noexcept; */
   simdjson_really_inline simdjson_result<value> find_field(std::string_view key) && noexcept;
+  /**
+   * Look up a field by name on an object. This function is similar to find_field_unordered but
+   * returns a NON_UNIQUE_FIELD if the key appears more than once.
+   *
+   * @param key The key to look up.
+   * @returns The value of the field, or NO_SUCH_FIELD if the field is not in the object, or NON_UNIQUE_FIELD if the key appears more than once.
+   */
+  simdjson_really_inline simdjson_result<value> find_unique_field(std::string_view key) & noexcept;
+  /** @overload simdjson_really_inline simdjson_result<value> find_unique_field(std::string_view key) & noexcept; */
+  simdjson_really_inline simdjson_result<value> find_unique_field(std::string_view key) && noexcept;
 
   /**
    * Look up a field by name on an object, without regard to key order.
@@ -82,8 +92,6 @@ class object {
   static simdjson_really_inline object resume(const value_iterator &iter) noexcept;
   simdjson_really_inline object(const value_iterator &iter) noexcept;
 
-  simdjson_warn_unused simdjson_really_inline error_code find_field_raw(const std::string_view key) noexcept;
-
   value_iterator iter{};
 
   friend class value;
@@ -108,6 +116,8 @@ struct simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::object> : public SIMDJ
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::object_iterator> end() noexcept;
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_field(std::string_view key) & noexcept;
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_field(std::string_view key) && noexcept;
+  simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_unique_field(std::string_view key) & noexcept;
+  simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_unique_field(std::string_view key) && noexcept;
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_field_unordered(std::string_view key) & noexcept;
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_field_unordered(std::string_view key) && noexcept;
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> operator[](std::string_view key) & noexcept;

include/simdjson/generic/ondemand/value-inl.h

@@ -102,7 +102,12 @@ simdjson_really_inline simdjson_result<value> value::find_field(std::string_view
 simdjson_really_inline simdjson_result<value> value::find_field(const char *key) noexcept {
   return start_or_resume_object().find_field(key);
 }
-
+simdjson_really_inline simdjson_result<value> value::find_unique_field(std::string_view key) noexcept {
+  return start_or_resume_object().find_unique_field(key);
+}
+simdjson_really_inline simdjson_result<value> value::find_unique_field(const char *key) noexcept {
+  return start_or_resume_object().find_unique_field(key);
+}
 simdjson_really_inline simdjson_result<value> value::find_field_unordered(std::string_view key) noexcept {
   return start_or_resume_object().find_field_unordered(key);
 }
@@ -164,6 +169,15 @@ simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value>
   return first.find_field(key);
 }
 
+simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value>::find_unique_field(std::string_view key) noexcept {
+  if (error()) { return error(); }
+  return first.find_unique_field(key);
+}
+simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value>::find_unique_field(const char *key) noexcept {
+  if (error()) { return error(); }
+  return first.find_unique_field(key);
+}
+
 simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value>::find_field_unordered(std::string_view key) noexcept {
   if (error()) { return error(); }
   return first.find_field_unordered(key);

include/simdjson/generic/ondemand/value.h

@@ -234,6 +234,16 @@ class value {
   simdjson_really_inline simdjson_result<value> find_field(std::string_view key) noexcept;
   /** @overload simdjson_really_inline simdjson_result<value> find_field(std::string_view key) noexcept; */
   simdjson_really_inline simdjson_result<value> find_field(const char *key) noexcept;
+  /**
+   * Look up a field by name on an object. This function is similar to find_field_unordered but
+   * returns a NON_UNIQUE_FIELD if the key appears more than once.
+   *
+   * @param key The key to look up.
+   * @returns The value of the field, or NO_SUCH_FIELD if the field is not in the object, or NON_UNIQUE_FIELD if the key appears more than once.
+   */
+  simdjson_really_inline simdjson_result<value> find_unique_field(std::string_view key) noexcept;
+  /** @overload simdjson_really_inline simdjson_result<value> find_field(std::string_view key) noexcept; */
+  simdjson_really_inline simdjson_result<value> find_unique_field(const char *key) noexcept;
 
   /**
    * Look up a field by name on an object, without regard to key order.
@@ -410,6 +420,16 @@ struct simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> : public SIMDJS
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_field(std::string_view key) noexcept;
   /** @overload simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_field(std::string_view key) noexcept; */
   simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_field(const char *key) noexcept;
+  /**
+   * Look up a field by name on an object. This function is similar to find_field_unordered but
+   * returns a NON_UNIQUE_FIELD if the key appears more than once.
+   *
+   * @param key The key to look up.
+   * @returns The value of the field, or NO_SUCH_FIELD if the field is not in the object, or NON_UNIQUE_FIELD if the key appears more than once.
+   */
+  simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_unique_field(std::string_view key) noexcept;
+  /** @overload simdjson_really_inline simdjson_result<value> find_field(std::string_view key) noexcept; */
+  simdjson_really_inline simdjson_result<SIMDJSON_IMPLEMENTATION::ondemand::value> find_unique_field(const char *key) noexcept;
 
   /**
    * Look up a field by name on an object, without regard to key order.