how to get distict values from many text files (please help)

i have many text file like i shared below and each of them has same type. i want to get some values from them and put those values into the calculation. For example i need the value of Area, , xbar,ybar,IXXw and Iyyw . how can i get those values from all files and find mean value of the

Work Part FA00AAC41319/002 : 08/29/13 11:07 Information Units kg - mm

Perimeter = 765.13878188660 Area = 11875.0

First Moments MY = -109375.0 MX = -1062500.0

Center of Mass Xbar = -9.21052631578950 Ybar = -89.4736842105260

Moments of Inertia (Work) Ixxw = 110677083.333330 Iyyw = 25716145.8333330

Moments of Inertia (Centroidal) Ixx = 15611293.8596490 Iyy = 24708744.5175440

Moment of Inertia (Polar) = 40320038.3771930

Product of Inertia (Work) Pxyw = 20507812.50

Product of Inertia (Centroidal) Pxy = 10721628.2894740

Radii of Gyration (Work) Rgxw = 96.5410557151540 Rgyw = 46.5356871168630

Radii of Gyration (Centroidal) Rgx = 36.2578994481410 Rgy = 45.6150893940230

Radii of Gyration (Polar) = 58.2698176830530

Principal Axes Xp(X) = -0.55201395809680 Xp(Y) = 0.83383486978316 Yp(X) = 0.83383486978316 Yp(Y) = 0.55201395809680

Principal Moments of Inertia Ixxp = 31806658.8454180 Iyyp = 8513379.53177490

Circle Size Center XC = 50.0 YC = -100.0 Diameter = 269.258240356730

[per isakson]

Is there a varying number of "=" per line or is it a problem with line breaks?

What is the typical size of the file? Is performance an issue?

Answer 1

[Cedric]

The following approach is based on regular expressions:

 files  = {'data_01.txt', 'data_02.txt'} ;
 nFiles = numel(files) ;

 area   = zeros(nFiles, 1) ;                  % Prealloc.
 xbar   = zeros(nFiles, 1) ;
 ybar   = zeros(nFiles, 1) ;
 ixxw   = zeros(nFiles, 1) ;
 iyyw   = zeros(nFiles, 1) ;

 for fId = 1 : nFiles
    content   = fileread(files{fId}) ;
    area(fId) = str2double(regexp(content, '(?<=Area \= )[\-\d\.]+', ...
                                  'match', 'once')) ;
    xbar(fId) = str2double(regexp(content, '(?<=Xbar \= )[\-\d\.]+', ...
                                  'match', 'once')) ;
    ybar(fId) = str2double(regexp(content, '(?<=Ybar\= )[\-\d\.]+', ...
                                  'match', 'once')) ;
    ixxw(fId) = str2double(regexp(content, '(?<=Ixxw \= )[\-\d\.]+', ...
                                  'match', 'once')) ;
    iyyw(fId) = str2double(regexp(content, '(?<=Iyyz \= )[\-\d\.]+', ...
                                  'match', 'once')) ;
 end

EDIT : here is a little more elaborate version

 files  = {'data_01.txt', 'data_02.txt'} ;
 items  = {'Area', 'Xbar', 'Ybar', 'Ixxw', 'Iyyw'} ;

 nFiles = numel(files) ;
 nItems = numel(items) ;
 data   = struct('fId', {}) ;

 for fId = 1 : nFiles
    content = fileread(files{fId}) ;
    data(fId).fId = fId ;
    for iId = 1 : nItems
        data(fId).(items{iId}) = str2double(regexp(content, ...
            sprintf('(?<=%s \\= )[\\-\\d\\.]+', items{iId}), ...
            'match', 'once')) ;
    end
 end

which generates a struct array, e.g., after running this code using two fake files built for the example:

 >> data
 data = 
 1x2 struct array with fields:
    fId
    Area
    Xbar
    Ybar
    Ixxw
    Iyyw

 >> data(1)
 ans = 
     fId: 1
    Area: 11875
    Xbar: -9.2105
    Ybar: -89.4737
    Ixxw: 1.1068e+08
    Iyyw: 2.5716e+07

 >> data(2)
 ans = 
     fId: 2
    Area: 11876
    Xbar: -9.4105
    Ybar: -80.4737
    Ixxw: 2.1068e+08
    Iyyw: 1.5716e+07

 >> mean([data.Xbar])
 ans =
   -9.3105

Answer 2

[Yusuf]

for fId = 1 : nFiles content = fileread(files{fId}) ; data(fId).fId = fId ; for iId = 1 : nItems data(fId).(items{iId}) = str2double(regexp(content, ... sprintf('(?<=%s \\= )[\\-\\d\\.]+', items{iId}), ... 'match', 'once')) ; i just dont understand this part can you explain little bit more?

[Cedric]

Do you understand the simplest solution involving e.g.

 area(fId) = str2double(regexp(content, '(?<=Area \= )[\-\d\.]+', ...
                                  'match', 'once')) ;

If not, I can explain that to you (just ask). If you do understand it, then the only difference in the second solution is that instead of having as many arrays as items, we build a struct array whose fields are item names, and we build the regexp pattern using corresponding item names. So what you see in

 data(fId).(items{iId}) = str2double(regexp(content, ...
            sprintf('(?<=%s \\= )[\\-\\d\\.]+', items{iId}), ...
            'match', 'once')) ;

is dynamic field names and a generation of the pattern using SPRINTF.

About dynamic field names, if s is a struct

 >> s = struct() ;

you can access fields as follows

 >> s.age = 65 ;
 >> s.name = 'Jon' ;

As you can see, field names are hardcoded. Now if a field name must be defined (dynamically) as a string stored in some variable, we can use the following type of expression

 >> fName = 'city' ;
 >> s.(fName) = 'New York' ;
 >> s
 s = 
     age: 65
    name: 'Jon'
    city: 'New York'

Here you can see that the field name 'city' was stored in a char variable fName (it is its content) and could be used to index s. This is what we do in

 data(fId).(items{iId}) = ...

where we use items{iId} as a field name. We also want to use items{iId} for building the regexp pattern and we do it with SPRINTF:

 sprintf('(?<=%s \\= )[\\-\\d\\.]+', items{iId})

which is almost the same as in the simple example with the difference that, as \ is an escape character for SPRINTF, we have the escape it so it becomes a \ character in the output of SPRINTF.

Hope it helps; let me know if you still don't understand.

[Yusuf]

thank you for detailed explanation but i am beginner on matlab and i used your code but i got this error;

Error using fileread (line 27) Could not open file data_01.txt. No such file or directory.

so just to make sure that the name of text files you used here are data_01 and data_02, arent they?

and do i need to change any variable here? for example "n" depends on the number of my files in those codes? nFiles = numel(files) for fId = 1 : nFiles

[Cedric]

The only thing that you should do is to define the cell array files with your own file names instead of

 files  = {'data_01.txt', 'data_02.txt'} ;

I chose these file names for the example actually. Other than that, there is nothing to do except to understand how to use the data container in which numbers are stored. The first solution is the simplest on that matter, as it involves basic numeric arrays.

If you had hundreds of files in a folder and didn't want to list them all by hand for defining the cell array files, you could use a call to DIR to retrieve a directory listing. For example, if you had this MATLAB script in a folder C:\Yusuf and all data files in C:\Yusuf\data with the extension .txt, you could replace the line

 files = {'data_01.txt', 'data_02.txt'} ;

by

 files = dir(fullfile('data', '*.txt')) ;

and then the line

 content = fileread(files{fId}) ;

by

 content = fileread(fullfile('data', files(fId).name)) ;

This way, running the script would provide you with a result based on all .txt files in C:\Yusuf\data.