Help reading and cleaning data from text file
2 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Hello everyone,
I am currently having difficulties trying to read a text file in Matlab. A small example is shown below.
My aim is to;
-Open the file
-Ignore the header (first n lines)
-Replace the Bad entries with NaN
-Replace any corrupted entries with NaN (as entry (3,1))
----
I tried the following code with which I had some success if I removed the header first but I am struggling to get it all to work together.
buffer=fileread('C:\Users\Jack\Desktop\ContourGT\13.07.2013_header_removed.txt);
buffer = strrep(buffer, 'Bad', 'NaN') ;
surface=sscanf(buffer, '%f',[width length]);
---
Wyko ASCII Data File Format 0 0 1
X Size 640
Y Size 3015
Block Name Type Length Value
Wavelength 7 4 200.746124
Bad Bad Bad Bad Bad Bad
Bad -15216.897461 Bad Bad Bad Bad
-15s83.3t93555 Bad Bad Bad Bad Bad
Bad Bad -15169.881836 Bad Bad Bad
Bad Bad Bad Bad Bad Bad
Bad Bad Bad Bad Bad Bad
---
Any ideas? Any thoughts would be gratefully appreciated.
Thanks in advance
Jack
0 Kommentare
Akzeptierte Antwort
dpb
am 30 Aug. 2013
>> fmt=repmat('%s',1,6);
>> c=textscan(fid,fmt,'delimiter',' ','headerlines',5,'collectoutput',true);
>> fid=fclose(fid);
>> strrep(c{:},'Bad','NaN')
ans =
'NaN' 'NaN' 'NaN' 'NaN' 'NaN' 'NaN'
'NaN' '-15216.897461' 'NaN' 'NaN' 'NaN' 'NaN'
'-15s83.3t93555' 'NaN' 'NaN' 'NaN' 'NaN' 'NaN'
'NaN' 'NaN' '-15169.881836' 'NaN' 'NaN' 'NaN'
'NaN' 'NaN' 'NaN' 'NaN' 'NaN' 'NaN'
'NaN' 'NaN' 'NaN' 'NaN' 'NaN' 'NaN'
>>
Now either use regexp or cellfun w/ a pattern on the cell content to replace any string that isn't all numeric w/ 'NaN' as well. Then you can use str2num()
The only assumption is that the corruption is limited as your sample so that there are a consistent number of fields per record.
0 Kommentare
Weitere Antworten (0)
Siehe auch
Kategorien
Mehr zu String Parsing finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!