How to identify and homogenize date format of instances in a string? - string

Question

I can't find a way to identify the date formats of a string in MATLAB and put all of them in the same format. I have the following cell array:
list = {'01-Sep-1882'; ...
'01-Aug-1895'; ...
'04/01/1912'; ...
'Tue, 05/28/46'; ...
'Tue, 03/10/53'; ...
'06/20/58'; ...
'Thu, 09/20/73'; ...
'Fri, 08/15/75'; ...
'Sun, 12/01/1996'};
If I do datenum(list) there's an error message because all the rows don't have the same date format. Can you think of a way to circumvent this?

Answer 1

You can do this by successively applying datetime to convert each format and isnat to identify those dates that didn't convert properly. In addition, you can specify days of the week in the date format string and what pivot year to use for years with only the last two numbers. Starting with the sample data and expected date formats in your question, here's the code to do it:
% Input:
list = {'01-Sep-1882'; ...
'01-Aug-1895'; ...
'04/01/1912'; ...
'Tue, 05/28/46'; ...
'Tue, 03/10/53'; ...
'06/20/58'; ...
'Thu, 09/20/73'; ...
'Fri, 08/15/75'; ...
'Sun, 12/01/1996'};
% Conversion code:
dt = datetime(list, 'Format', 'dd-MMM-yyyy');
index = isnat(dt);
dt(index) = datetime(list(index), 'Format', 'MM/dd/yy', 'PivotYear', 1900);
index = isnat(dt);
dt(index) = datetime(list(index), 'Format', 'eee, MM/dd/yy', 'PivotYear', 1900)
% Output:
dt =
9×1 datetime array
01-Sep-1882
01-Aug-1895
01-Apr-1912
28-May-1946
10-Mar-1953
20-Jun-1958
20-Sep-1973
15-Aug-1975
01-Dec-1996
Now you can convert these to numeric values with datenum:
dnum = datenum(dt);

Answer 2

Part of the issue here is that MATLAB's datenum cannot understand this format: Tue, 05/28/46. So let's clean the original list so that it can understand.
% original list
list = {'01-Sep-1882','01-Aug-1895','04/01/1912','Tue, 05/28/46','Tue, 03/10/53','06/20/58','Thu, 09/20/73','Fri, 08/15/75','Sun, 12/01/1996'}
% split each cell in the list
list_split = cellfun(#(x) strsplit(x,' '), list, 'UniformOutput', false);
% now detect where there is unusual format, this will give logical array
abnormal_idx = cellfun(#(x) length(x) == 2, list_split,'UniformOutput', true)
% make copy
clean_list = list;
% now at abnormal indices retain only the part that MATLAB understands
clean_list(abnormal_idx) = cellfun(#(x) x{2}, list_split(abnormal_idx), 'UniformOutput', false);
% now run datenum on clean list
date_num = cellfun(#datenum, clean_list, 'UniformOutput', true);