Readall stops because of non-existing data, Can I skip these?
39 views (last 30 days)
Show older comments
Hi y'all,
When I run my code I get an error:
Error using matlab.io.datastore.TabularDatastore/readall (line 196)
Unable to parse a "Numeric" field when reading row 39898, field 7.
Actual Text: "N,07121.27550,W,2,28,0.9,0.79,M,,,5,0131*01"
Expected: A number or literal "NaN", "Inf". (possibly signed, case insensitive
Is there a way to have the readall function skip these ,,, non-existing values in my data?
Code:
function [] = each_day_table(folder_referenced, output_folder_location)
dimention = size(folder_referenced) ;
for i = 1:dimention(2)
datastore_result = datastore(folder_referenced(i)) ;
original_data = readall(datastore_result) ;
new_table = new_table_useful_data(original_data) ;
[~, name, ext] = fileparts(folder_referenced(i)) ;
output_filename = fullfile(output_folder_location, "new_" + name + ext) ;
writetable(new_table, output_filename)
end
folder_with_files = dir(input_folder_location) ; % define directory where files are
filenames = fullfile({folder_with_files.folder}, {folder_with_files.name}) ; % access file path and name information
csvFiles = endsWith(filenames, '.csv') ; % use logi → determining which values end in .csv (fullfile provides axtra info we don't want)
filenames = filenames(csvFiles) ; % create cell array with all file names
each_day_table(filenames, output_folder_location) ; % applies all the functions to the files
I've attached a ference file.
3 Comments
dpb
on 19 Sep 2025 at 22:03
Edited: dpb
on 20 Sep 2025 at 15:11
Similar, but not the same as, Walter. There there were additional page header lines in the file that were specific text that could be identified and those particular values then set as
% ...,'TreatAsMissing',[TreatAsMissing={'Time','Board0_Ai0'}, ...
One would have thought the default of "" and NaN for default 'MissingValue' should have worked but apprently not.
There's the 'MissingRule' in an import options object that I guess one could see if
% ...,'MissingRule','omitrow', ...
would be accepted, but I'd guess it unlikely and afaict there's no facility to use an import object here.
Well, let's just see...
ds=datastore('20250214RAW.log.csv','MissingRule','omitrow')
So that isn't allowed and
ds=datastore('20250214RAW.log.csv');
data=readall(ds,'MissingRule','omitrow');
fails as well as it says the 'UseParallel' is the only recognized named parameter.
I'm sure there must be a way, but it certainly isn't clear to me how to beat it into submission. A regular call to readtable with the import options object would work, but there's no way to use one here that I can see.
Accepted Answer
dpb
on 19 Sep 2025 at 23:30
Edited: dpb
on 21 Sep 2025 at 15:55
A workaround until somebody can come up with the clean answer
function fixupmissingfields(infile,outfile)
% substitute NA into missing comma-delimited fields in input file
% and write to output file
fidi=fopen(infile,'r');
fido=fopen(outfile,'w');
while ~feof(fidi)
l=fgetl(fidi); % read line excluding terminator
w=split(l,','); % get fields
w(strlength(w)==0)={'NA'}; % insert the missing indicator in empty fields
l=char(join(w,',')); % put back together
fprintf(fido,'%s\n',l); % output to new file with newline
end
fclose(fidi);
fclose(fido);
end
Alternatively, you could mimic the 'omitrow' of readtable with
function omitmissingrows(infile,outfile)
% skip records with missing comma-delimited fields in input file
% and write to output file
fidi=fopen(infile,'r');
fido=fopen(outfile,'w');
while ~feof(fidi)
l=fgets(fidi); % read line including terminator
if contains(l,',,'); % at least one missing field
continue % skip this record from output
end
fprintf(fido,'%s',l); % output to new file
end
fclose(fidi)
fclose(fido)
end
Illustrate on the attached file adding NA...
infile='20250214RAW.log.csv';
outfile=strrep(infile,'RAW','CLEANED');
fixupmissingfields(infile,outfile)
type(outfile);
1 Comment
dpb
on 20 Sep 2025 at 17:53
Edited: dpb
on 20 Sep 2025 at 21:41
ADDENDUM
The above assume file may be too large to fit in memory; if it can be read all at once, then the first can be vectorized as
function fixupmissingfields2(infile,outfile)
% substitute NA into missing comma-delimited fields in input file
% and write to output file - in memory version
w=split(readlines(infile),',');
w(strlength(w)==0)={'NA'};
writematrix(w,outfile,'Delimiter',',','QuoteStrings',0);
end
infile='20250214RAW.log.csv';
outfile=strrep(infile,'RAW','CLEANED');
fixupmissingfields2(infile,outfile);
type(outfile)
More Answers (0)
See Also
Categories
Find more on Data Import and Analysis in Help Center and File Exchange
Products
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!