diff --git a/Makefile b/Makefile index 086b07f1..24c77f51 100644 --- a/Makefile +++ b/Makefile @@ -442,7 +442,7 @@ fuzzing-stop: # call cppcheck on the main header file cppcheck: - cppcheck --enable=warning --inconclusive --force --std=c++11 $(SRCS) --error-exitcode=1 + cppcheck --enable=warning --inline-suppr --inconclusive --force --std=c++11 $(SRCS) --error-exitcode=1 # compile and check with Clang Static Analyzer clang_analyze: diff --git a/include/nlohmann/detail/input/parser.hpp b/include/nlohmann/detail/input/parser.hpp index a3072de4..6c03181e 100644 --- a/include/nlohmann/detail/input/parser.hpp +++ b/include/nlohmann/detail/input/parser.hpp @@ -262,14 +262,13 @@ class parser m_lexer.get_token_string(), out_of_range::create(406, "number overflow parsing '" + m_lexer.get_token_string() + "'")); } - else + + if (JSON_UNLIKELY(not sax->number_float(res, m_lexer.get_string()))) { - if (JSON_UNLIKELY(not sax->number_float(res, m_lexer.get_string()))) - { - return false; - } - break; + return false; } + + break; } case token_type::literal_false: @@ -355,103 +354,95 @@ class parser // empty stack: we reached the end of the hierarchy: done return true; } - else + + if (states.back()) // array { - if (states.back()) // array + // comma -> next value + if (get_token() == token_type::value_separator) { - // comma -> next value - if (get_token() == token_type::value_separator) + // parse a new value + get_token(); + continue; + } + + // closing ] + if (JSON_LIKELY(last_token == token_type::end_array)) + { + if (JSON_UNLIKELY(not sax->end_array())) { - // parse a new value - get_token(); - continue; + return false; } - // closing ] - if (JSON_LIKELY(last_token == token_type::end_array)) - { - if (JSON_UNLIKELY(not sax->end_array())) - { - return false; - } + // We are done with this array. Before we can parse a + // new value, we need to evaluate the new state first. + // By setting skip_to_state_evaluation to false, we + // are effectively jumping to the beginning of this if. + assert(not states.empty()); + states.pop_back(); + skip_to_state_evaluation = true; + continue; + } - // We are done with this array. Before we can parse a - // new value, we need to evaluate the new state first. - // By setting skip_to_state_evaluation to false, we - // are effectively jumping to the beginning of this if. - assert(not states.empty()); - states.pop_back(); - skip_to_state_evaluation = true; - continue; - } - else + return sax->parse_error(m_lexer.get_position(), + m_lexer.get_token_string(), + parse_error::create(101, m_lexer.get_position(), + exception_message(token_type::end_array, "array"))); + } + else // object + { + // comma -> next value + if (get_token() == token_type::value_separator) + { + // parse key + if (JSON_UNLIKELY(get_token() != token_type::value_string)) { return sax->parse_error(m_lexer.get_position(), m_lexer.get_token_string(), parse_error::create(101, m_lexer.get_position(), - exception_message(token_type::end_array, "array"))); + exception_message(token_type::value_string, "object key"))); } - } - else // object - { - // comma -> next value - if (get_token() == token_type::value_separator) + + if (JSON_UNLIKELY(not sax->key(m_lexer.get_string()))) { - // parse key - if (JSON_UNLIKELY(get_token() != token_type::value_string)) - { - return sax->parse_error(m_lexer.get_position(), - m_lexer.get_token_string(), - parse_error::create(101, m_lexer.get_position(), - exception_message(token_type::value_string, "object key"))); - } - else - { - if (JSON_UNLIKELY(not sax->key(m_lexer.get_string()))) - { - return false; - } - } - - // parse separator (:) - if (JSON_UNLIKELY(get_token() != token_type::name_separator)) - { - return sax->parse_error(m_lexer.get_position(), - m_lexer.get_token_string(), - parse_error::create(101, m_lexer.get_position(), - exception_message(token_type::name_separator, "object separator"))); - } - - // parse values - get_token(); - continue; + return false; } - // closing } - if (JSON_LIKELY(last_token == token_type::end_object)) - { - if (JSON_UNLIKELY(not sax->end_object())) - { - return false; - } - - // We are done with this object. Before we can parse a - // new value, we need to evaluate the new state first. - // By setting skip_to_state_evaluation to false, we - // are effectively jumping to the beginning of this if. - assert(not states.empty()); - states.pop_back(); - skip_to_state_evaluation = true; - continue; - } - else + // parse separator (:) + if (JSON_UNLIKELY(get_token() != token_type::name_separator)) { return sax->parse_error(m_lexer.get_position(), m_lexer.get_token_string(), parse_error::create(101, m_lexer.get_position(), - exception_message(token_type::end_object, "object"))); + exception_message(token_type::name_separator, "object separator"))); } + + // parse values + get_token(); + continue; } + + // closing } + if (JSON_LIKELY(last_token == token_type::end_object)) + { + if (JSON_UNLIKELY(not sax->end_object())) + { + return false; + } + + // We are done with this object. Before we can parse a + // new value, we need to evaluate the new state first. + // By setting skip_to_state_evaluation to false, we + // are effectively jumping to the beginning of this if. + assert(not states.empty()); + states.pop_back(); + skip_to_state_evaluation = true; + continue; + } + + return sax->parse_error(m_lexer.get_position(), + m_lexer.get_token_string(), + parse_error::create(101, m_lexer.get_position(), + exception_message(token_type::end_object, "object"))); } } } diff --git a/include/nlohmann/json.hpp b/include/nlohmann/json.hpp index cbe30fe2..6a5c1075 100644 --- a/include/nlohmann/json.hpp +++ b/include/nlohmann/json.hpp @@ -4777,7 +4777,8 @@ class basic_json // add element to array (move semantics) m_value.array->push_back(std::move(val)); - // invalidate object + // invalidate object: mark it null so we do not call the destructor + // cppcheck-suppress accessMoved val.m_type = value_t::null; } diff --git a/single_include/nlohmann/json.hpp b/single_include/nlohmann/json.hpp index 92293fb0..7509df49 100644 --- a/single_include/nlohmann/json.hpp +++ b/single_include/nlohmann/json.hpp @@ -7276,14 +7276,13 @@ class parser m_lexer.get_token_string(), out_of_range::create(406, "number overflow parsing '" + m_lexer.get_token_string() + "'")); } - else + + if (JSON_UNLIKELY(not sax->number_float(res, m_lexer.get_string()))) { - if (JSON_UNLIKELY(not sax->number_float(res, m_lexer.get_string()))) - { - return false; - } - break; + return false; } + + break; } case token_type::literal_false: @@ -7369,103 +7368,95 @@ class parser // empty stack: we reached the end of the hierarchy: done return true; } - else + + if (states.back()) // array { - if (states.back()) // array + // comma -> next value + if (get_token() == token_type::value_separator) { - // comma -> next value - if (get_token() == token_type::value_separator) + // parse a new value + get_token(); + continue; + } + + // closing ] + if (JSON_LIKELY(last_token == token_type::end_array)) + { + if (JSON_UNLIKELY(not sax->end_array())) { - // parse a new value - get_token(); - continue; + return false; } - // closing ] - if (JSON_LIKELY(last_token == token_type::end_array)) - { - if (JSON_UNLIKELY(not sax->end_array())) - { - return false; - } + // We are done with this array. Before we can parse a + // new value, we need to evaluate the new state first. + // By setting skip_to_state_evaluation to false, we + // are effectively jumping to the beginning of this if. + assert(not states.empty()); + states.pop_back(); + skip_to_state_evaluation = true; + continue; + } - // We are done with this array. Before we can parse a - // new value, we need to evaluate the new state first. - // By setting skip_to_state_evaluation to false, we - // are effectively jumping to the beginning of this if. - assert(not states.empty()); - states.pop_back(); - skip_to_state_evaluation = true; - continue; - } - else + return sax->parse_error(m_lexer.get_position(), + m_lexer.get_token_string(), + parse_error::create(101, m_lexer.get_position(), + exception_message(token_type::end_array, "array"))); + } + else // object + { + // comma -> next value + if (get_token() == token_type::value_separator) + { + // parse key + if (JSON_UNLIKELY(get_token() != token_type::value_string)) { return sax->parse_error(m_lexer.get_position(), m_lexer.get_token_string(), parse_error::create(101, m_lexer.get_position(), - exception_message(token_type::end_array, "array"))); + exception_message(token_type::value_string, "object key"))); } - } - else // object - { - // comma -> next value - if (get_token() == token_type::value_separator) + + if (JSON_UNLIKELY(not sax->key(m_lexer.get_string()))) { - // parse key - if (JSON_UNLIKELY(get_token() != token_type::value_string)) - { - return sax->parse_error(m_lexer.get_position(), - m_lexer.get_token_string(), - parse_error::create(101, m_lexer.get_position(), - exception_message(token_type::value_string, "object key"))); - } - else - { - if (JSON_UNLIKELY(not sax->key(m_lexer.get_string()))) - { - return false; - } - } - - // parse separator (:) - if (JSON_UNLIKELY(get_token() != token_type::name_separator)) - { - return sax->parse_error(m_lexer.get_position(), - m_lexer.get_token_string(), - parse_error::create(101, m_lexer.get_position(), - exception_message(token_type::name_separator, "object separator"))); - } - - // parse values - get_token(); - continue; + return false; } - // closing } - if (JSON_LIKELY(last_token == token_type::end_object)) - { - if (JSON_UNLIKELY(not sax->end_object())) - { - return false; - } - - // We are done with this object. Before we can parse a - // new value, we need to evaluate the new state first. - // By setting skip_to_state_evaluation to false, we - // are effectively jumping to the beginning of this if. - assert(not states.empty()); - states.pop_back(); - skip_to_state_evaluation = true; - continue; - } - else + // parse separator (:) + if (JSON_UNLIKELY(get_token() != token_type::name_separator)) { return sax->parse_error(m_lexer.get_position(), m_lexer.get_token_string(), parse_error::create(101, m_lexer.get_position(), - exception_message(token_type::end_object, "object"))); + exception_message(token_type::name_separator, "object separator"))); } + + // parse values + get_token(); + continue; } + + // closing } + if (JSON_LIKELY(last_token == token_type::end_object)) + { + if (JSON_UNLIKELY(not sax->end_object())) + { + return false; + } + + // We are done with this object. Before we can parse a + // new value, we need to evaluate the new state first. + // By setting skip_to_state_evaluation to false, we + // are effectively jumping to the beginning of this if. + assert(not states.empty()); + states.pop_back(); + skip_to_state_evaluation = true; + continue; + } + + return sax->parse_error(m_lexer.get_position(), + m_lexer.get_token_string(), + parse_error::create(101, m_lexer.get_position(), + exception_message(token_type::end_object, "object"))); } } } @@ -17420,7 +17411,8 @@ class basic_json // add element to array (move semantics) m_value.array->push_back(std::move(val)); - // invalidate object + // invalidate object: mark it null so we do not call the destructor + // cppcheck-suppress accessMoved val.m_type = value_t::null; }