Filter löschen
Filter löschen

jsondecode not suitable to parse uint64 array values

10 Ansichten (letzte 30 Tage)
GonzaloMadrono
GonzaloMadrono am 23 Jan. 2024
Beantwortet: Sachin Lodhi am 12 Jun. 2024
Why jsondecode function is not able to correctly parse uint64 values? Is MathWorks going to solve this issue?
I was reading a json file with timestamps in nanoseconds:
json_data = fileread('my_json_file.json');
data = jsondecode(json_data);
When accessing to the value in matlab, it returned 1704972586391907328. My original value was 1704972586391907240.
So I asked myself, how is this possible. It turns out that what "jsondecode" function is doing is:
str2double(raw_values_str)
As Numerics of class double only have 15 decimal digits of precision (52 bits), it is not precisely representable using a double.
To be able to retrieve the uint64 value I did a manual parsing of json file:
function values = extract_uint64_array(json_data, field_name)
% Construct the field search string
field_search_str = sprintf('"%s": [', field_name);
% Extract the specified field manually
start_index = strfind(json_data, field_search_str) + length(field_search_str);
end_index = strfind(json_data(start_index:end), ']') - 2 + start_index;
raw_values_str = json_data(start_index:end_index);
% Convert the field values to uint64 array using sscanf and a loop
raw_values_cell = strsplit(raw_values_str, ',');
values = uint64(zeros(1, numel(raw_values_cell)));
for i = 1:numel(raw_values_cell)
values(i) = sscanf(raw_values_cell{i}, '%lu');
end
end
I hope it helps some of you. Thank you for reading!

Antworten (1)

Sachin Lodhi
Sachin Lodhi am 12 Jun. 2024
Hello GonzaloMadrono,
I understand that you are trying to parse uint64 values using "jsondecode" function. I too was facing this issue and with the following workaround I was able to accurately parse uint64 values from JSON data: 
function values = extract_uint64_array(json_data, field_name)
% Construct the field search string
field_search_str = sprintf('"%s": [', field_name);
% Extract the specified field manually
start_index = strfind(json_data, field_search_str) + length(field_search_str);
end_index = strfind(json_data(start_index:end), ']') - 2 + start_index;
raw_values_str = json_data(start_index:end_index);
% Convert the field values to uint64 array using sscanf and a loop
raw_values_cell = strsplit(raw_values_str, ',');
values = uint64(zeros(1, numel(raw_values_cell)));
for i = 1:numel(raw_values_cell)
values(i) = sscanf(raw_values_cell{i}, '%lu');
end
end
This function manually extracts the array of uint64 values from the raw JSON data, ensuring accurate representation of the large integers.
I hope this helps!

Produkte


Version

R2023b

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by