Recover from Parsing Errors and Continue JSON Processing With Ondemand Parser #2347
Description
Request: Recover from Parsing Errors and Continue JSON Processing
Description
Hi~, I am using simdjson::ondemand::parser to parse JSON data, but sometimes my JSON objects may contain malformed elements (e.g., trailing commas). While I understand that such structures are invalid according to strict JSON specifications, I would like to have an option to skip or recover from errors rather than having the entire parsing process fail.
Some JSON libraries, such as yyjson, provide support for trailing commas or other minor syntax relaxations. My main concern is not just about trailing commas but about having a way to recover from an error and continue parsing the rest of the JSON object.
Example Code
simdjson::ondemand::parser parser;
simdjson::ondemand::document doc;
std::string data = R"({
"errorObj": {
"key" : "value", // Trailing comma, invalid JSON
},
"extData": {
"key2" : "some data in here"
}
})";
auto error = parser.iterate(data).get(doc);
auto root = doc.get_object().value();
// If I try to access `errorObj` first, my code crashes.
// However, if I avoid accessing `errorObj`, the code runs fine.
// try {
// auto errorObj = root["errorObj"].take_value();
// auto size = errorObj.count_fields().take_value();
// std::cout << "error object size = " << size << std::endl;
// } catch (const std::exception & e) {
// std::cout << "error: " << e.what() << std::endl;
// // Is there a way to reset the error and continue parsing the rest of the JSON?
// }
try {
auto extData = root["extData"].take_value();
auto size = extData.count_fields();
std::cout << "extData object size = " << size;
} catch (const std::exception & e) {
std::cout << "error: " << e.what() << std::endl;
}I hope that the ondemand parser could incorporate some level of error tolerance and recovery capabilities to support more complex and customized scenarios for users.
Thank you for considering this feature request. Your work on simdjson is greatly appreciated!