Extracting multiple datasets from a text file

I am trying to extract data from a text file in the following format. 1st line (whatever characters it contains) needs to be the name a structure. each data set that I am trying to store in a given text file is preceded by a line description, part of that line needs to be the name of the sub matrix within that structure (see comments below). The order of datasets across multiple text files could vary... The number of data set is large (always mx3 though), I truncated it for a demo purpose here. I put comments as labels below for clarification...

The output would be one structure containing 5 matrices with a set of points stored within.

SAW_KNEE_Name % all of this needs to be the name of the structure
::PrepStep_TTTgeneric_fe_ Ant-Chmf_nar_IA_CR_fem %if this dataset containts Ant-Chmf, then create a sub matric within structure with name 'Ant-Chmf' 
-51.402984619140625 -46.14201354980469 -157.0766143798828
-3.12049627304077148 -0.78153657913208008 1.52892947196960449
::PrepStep_TTTgeneric_fe_ Ant_nar_IA_CR_fem %if this dataset containts Ant, then create a sub matric within structure with name 'Ant-Chmf' 
-51.402984619140625 -46.14201354980469 -157.0766143798828
-3.12049627304077148 -0.78153657913208008 1.52892947196960449
::PrepStep_TTTgeneric_fe_ Pst_nar_IA_CR_fem %if this dataset containts Pst, then create a sub matric within structure with name 'Pst'
-51.402984619140625 -46.14201354980469 -157.0766143798828
-3.12049627304077148 -0.78153657913208008 1.52892947196960449 
::PrepStep_TTTgeneric_fe_ Pst-Chmf_nar_IA_CR_fem %if this dataset containts Pst-Chmf, then create a sub matric within structure with name 'Pst-Chmf'
-51.402984619140625 -46.14201354980469 -157.0766143798828
-3.12049627304077148 -0.78153657913208008 1.52892947196960449
::PrepStep_TTTgeneric_fe_ Dst_nar_IA_CR_fem %if this dataset containts Dst, then create a sub matric within structure with name 'Dst'
-51.402984619140625 -46.14201354980469 -157.0766143798828
-3.12049627304077148 -0.78153657913208008 1.52892947196960449

[dpb]

"1st line ... needs to be the name a structure."

SAW_KNEE_Name % this needs to be the name of the structure

"This way be dragons!"

If you go this route, you're falling into the trap of there being no generic way to subsequently address the structure except via eval

Rethink your "needs" and consider instead a top-level data structure with named fields that can be dynamically addressed.

[per isakson]

"containts Ant, then create a sub matric within structure with name 'Ant-Chmf'" &nbsp does this contain more than one typing mistake?

[Stephen23]

Read this before following this approach:

https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval

Answer 1

[per isakson]

"the following format" &nbsp Your description is incomplete. Thus, I need to make a number of assumptions:

1. the entire file fits in memory together with the resulting structure
2. "::" appears only as the first and second character of the "line description".
3. The "line description" consists of two text strings separated by one space. The characters left of the first underscore of the second text string controls the name of the structure field. ("-" is not allowed in Matlab names.)
4. The comments, which are displayed in the example, don't exist in the data file.

Here is one way to read the file. Try

>> sas = cssm( 'SAW_KNEE_Name.txt' )
sas = 
       Name: 'SAW_KNEE_Name'
    AntChmf: [-51.4030 -46.1420 -157.0766]
        Ant: [3x3 double]
        Pst: [2x3 double]
    PstChmf: [2x3 double]
        Dst: [4x3 double]

where

function    sas = cssm( filespec )    
%%SAW_KNEE_Name
    cac = {
        'Ant-Chmf', 'AntChmf' 
        'Ant'     , 'Ant' 
        'Pst'     , 'Pst'
        'Pst-Chmf', 'PstChmf'
        'Dst'     , 'Dst'   
        };
    field_names = containers.Map(  cac(:,1), cac(:,2) );
    str = fileread( filespec );
    sas.Name = regexp( str, '^\S+', 'match', 'once' );
    buffer   = regexp( str, '(?<=::).+?(?=(::)|$)', 'match' );
    for ca = buffer
        key = regexp( ca{:}, '(?<=^\S+ )[^_]+', 'match', 'once' );
        val = field_names( key );
        num = textscan( ca{:}, '%f%f%f', 'Headerlines',1, 'CollectOutput',true );
        sas.(val) = num{:};
    end
end

and SAW_KNEE_Name.txt contains

SAW_KNEE_Name 
::PrepStep_TTTgeneric_fe_ Ant-Chmf_nar_IA_CR_fem 
-51.402984619140625 -46.14201354980469 -157.0766143798828
::PrepStep_TTTgeneric_fe_ Ant_nar_IA_CR_fem 
-51.402984619140625 -46.14201354980469 -157.0766143798828
-3.12049627304077148 -0.78153657913208008 1.52892947196960449
-3.12049627304077148 -0.78153657913208008 1.52892947196960449
::PrepStep_TTTgeneric_fe_ Pst_nar_IA_CR_fem 
-51.402984619140625 -46.14201354980469 -157.0766143798828
-3.12049627304077148 -0.78153657913208008 1.52892947196960449 
::PrepStep_TTTgeneric_fe_ Pst-Chmf_nar_IA_CR_fem 
-51.402984619140625 -46.14201354980469 -157.0766143798828
-3.12049627304077148 -0.78153657913208008 1.52892947196960449
::PrepStep_TTTgeneric_fe_ Dst_nar_IA_CR_fem 
-51.402984619140625 -46.14201354980469 -157.0766143798828
-3.12049627304077148 -0.78153657913208008 1.52892947196960449
-3.12049627304077148 -0.78153657913208008 1.52892947196960449
-3.12049627304077148 -0.78153657913208008 1.52892947196960449

&nbsp

and finally

>> eval( [sas.Name, ' = sas;'] )
>> whos(sas.Name)
  Name               Size            Bytes  Class     Attributes
  SAW_KNEE_Name      1x1              1370  struct

which dpb wisely advised against. It really is asking for trouble to create variables this way.