Take string input and reformat as cell array

I would like to convert a string into a cell array. Here is a small example of string:

[{"ts": "2018-07-03T06:30:07", "v": 1.0}, {"ts": "2018-07-03T06:30:12", "v": 0.0}, {"ts": "2018-07-03T08:41:06", "v": 1.0}]

I would like to convert this into the following format:

[ datetime ] [number]

The ts and v refer to timestamp and value, and are not important to keep.

I believe using the textscan is an efficient way to do it, however I cannot figure out how to use it properly, any help is appreciated!

[Stephen23]

@Connor Brackley: please upload a sample of your data by clicking the paperclip button.

[Connor Brackley]

Here is a sample, thank you

Answer 1

[Stephen23]

>> str = fileread('SampleData.txt');
>> str = regexprep(str,{'",','[":{}\[\]]+'},{' ',''});
>> fmt = '%*s%{yyyy-MM-dd''T''HHmmss}D%*s%f';
>> out = textscan(str,fmt,'EndOfLine',',');

Checking the output:

>> out{1}
ans = 
   2018-07-01T101102
   2018-07-01T101107
   2018-07-01T101109
   2018-07-01T101114
   2018-07-01T101120
   2018-07-01T101125
etc.

[Stephen23]

Do you have any control over the file format?

If it was possible to change the character between each set of {...} (e.g. to ';' or similar) then it would be possible to use textscan directly on the file itself, which would likely be much faster.

[Connor Brackley]

No, this format is direclty from an API of a service being used to store the data. For my purposes this is fast enough, but thank you for trying to improve it.

Answer 2

[Ameer Hamza]

Try this

str = fileread('test.txt');

dts = regexp(str, '"ts": "([0-9:T-]*)"', 'tokens');
vs = regexp(str, '"v": ([0-9.]*)', 'tokens');

dts = cellfun(@(x) datetime(x, 'InputFormat', 'yyyy-MM-dd''T''HH:mm:ss'), [dts{:}]);
vs = cellfun(@str2double, [vs{:}]);

T = table(dts.', vs.', 'VariableNames', {'DateTime', 'V'});
disp(T)

Result

          DateTime          V
    ____________________    _
    03-Jul-2018 06:30:07    1
    03-Jul-2018 06:30:12    0
    03-Jul-2018 08:41:06    1

[Connor Brackley]

Thank you! This worked, however @Stephen Cobeldick's answer performed the task much faster. I had 400,000+ datapoints to process so speed was important :)