Niels Lohmann
330fbead59
* Document clang-tidy exclusions and drop unneeded ones Re-enable every excluded check in a run of clang-tidy 22 over all tests and remove the seven that never fire (cert-err58-cpp, clang-analyzer-webkit.NoUncountedMemberChecker, cppcoreguidelines-virtual-class-destructor, google-readability-function-size, hicpp-function-size, hicpp-no-assembler, readability-function-size). Document the reason for every remaining exclusion at the top of the file. Exclude in advance the checks that only exist in clang-tidy 23 and are noise here (bugprone-signed-bitwise, *-explicit-constructor, readability-trailing-comma, modernize-use-string-view), plus modernize-use-integer-sign-comparison, which needs C++20 std::cmp_less and flags deliberate casts. Signed-off-by: Niels Lohmann <mail@nlohmann.me> * Fix clang-tidy findings in documentation examples and replace redundant aliases Apply clang-tidy fixes to the documentation examples: - performance-avoid-endl: std::endl -> '\n' - readability-qualified-auto: auto -> auto* / const auto* - performance-for-range-copy: copy -> const reference - google-readability-casting: C-style cast -> static_cast Replace redundant aliases (readability-redundant-qualified-alias) such as "using json = nlohmann::json;" with "using nlohmann::json;" in the documentation, README, examples, tests, and benchmarks. Aliases at class scope and aliases of other types (with_base_class_t etc.) are kept. No changes to the library or its public API. All changed examples were compiled and run; their output matches the stored .output files (except max_size and meta, which depend on the platform and compiler). Signed-off-by: Niels Lohmann <mail@nlohmann.me> * Format doc examples with the pinned astyle Signed-off-by: Niels Lohmann <mail@nlohmann.me> --------- Signed-off-by: Niels Lohmann <mail@nlohmann.me>
45 lines
1.4 KiB
C++
45 lines
1.4 KiB
C++
#include <iostream>
|
|
#include <nlohmann/json_view.hpp>
|
|
|
|
using nlohmann::json_document;
|
|
|
|
int main()
|
|
{
|
|
// two user records from a large API response; only the fields that are
|
|
// actually read are ever touched, and no nlohmann::json tree is built
|
|
// for the batch
|
|
json_document batch = json_document::parse(R"(
|
|
[
|
|
{"name": "Alice", "email": "alice@example.com", "tags": ["admin", "ops"]},
|
|
{"name": "Bob", "tags": []}
|
|
]
|
|
)");
|
|
|
|
const auto users = batch.root();
|
|
for (std::size_t i = 0; i < users.size(); ++i)
|
|
{
|
|
const auto user = users[i];
|
|
std::cout << user["name"].materialize().dump();
|
|
|
|
// operator[] on a missing object key gives a discarded view -- test
|
|
// it with is_discarded(). The const overload of json::operator[]
|
|
// would instead be undefined behavior (guarded by an assertion) for
|
|
// a missing key
|
|
const auto email = user["email"];
|
|
if (!email.is_discarded())
|
|
{
|
|
std::cout << " <" << email.materialize().dump() << ">";
|
|
}
|
|
|
|
// the same holds for an array index past the end: a discarded view,
|
|
// not undefined behavior
|
|
const auto first_tag = user["tags"][0];
|
|
if (!first_tag.is_discarded())
|
|
{
|
|
std::cout << " #" << first_tag.materialize().dump();
|
|
}
|
|
|
|
std::cout << '\n';
|
|
}
|
|
}
|