Extracting number after a string in header of CSV file

Hello, I have a CSV file that contains information in the headers and then numerical data.

I have no problem loading the numerical tabulated data into a uitable via readtable, but I want to pick out several strings within the header too.

I particularly want to extract the numbers in these lines

1: Memory Length,1000000,

2: Horizontal Scale,2.000E-04,

3: Sampling Period,2.000E-09,

Im nearly there but just can't quite get it:

fullpath=fullfile(folder,file)
l=readlines(fullpath);
Key = 'Memory Length';
line_idx=find(startsWith(l,Key))
Str=l(line_idx)
class(Str)
Index = strfind(Str, Key)
Str(Index(1) + length(Key):end)

line_idx =

2

Str =

"Memory Length,1000000,"

ans =

'string'

Index =

1

ans =

1×0 empty string array

Here is the typical file (also attached)

Format,1.0B,
Memory Length,1000000,
IntpDistance,0,
Trigger Address,-1,
Trigger Level,2.200E+00,
Source,CH1,
Vertical Units,V,
Horizontal Units,S,
Horizontal Scale,2.000E-04,
Sampling Period,2.000E-09,
Horizontal Old Scale,2.000E-04,
Horizontal Old Position,1.000E-03,
Firmware,V1.27.001,
Mode,Detail,
Waveform Data,
6.891787e-11,5.00e+00,
2.068918e-09,5.00e+00,
4.068918e-09,4.76e+00,
6.068918e-09,4.24e+00,
8.068918e-09,4.24e+00,

[Jason]

Oooh, just seen using readlines gives a string array. So can use this:

pat = digitsPattern
newStr = extract(Str,pat)

But it falls over with the E here "Horizontal Scale,2.000E-04"

l=readlines(fullpath);                

Key = 'Memory Length';
line_idx=find(startsWith(l,Key))
Str=l(line_idx)
pat = digitsPattern
MemoryLength = str2double(extract(Str,pat))
Mem_Million=MemoryLength/(10^6)



Key='Horizontal Scale';
line_idx=find(startsWith(l,Key))
Str=l(line_idx)
pat = digitsPattern
H_Scale = str2double(extract(Str,pat))

Str =

"Horizontal Scale,2.000E-04,"

pat =

pattern

Matching:

digitsPattern

H_Scale =

2

0

4

Answer 1

[Star Strider]

One approach —

C1 = readcell('DummyData.txt')

C1 = 43x2 cell array
    {'Format'                   }    {'1.0B'      }
    {'Memory Length'            }    {[   1000000]}
    {'IntpDistance'             }    {[         0]}
    {'Trigger Address'          }    {[        -1]}
    {'Trigger Level'            }    {[    2.2000]}
    {'Source'                   }    {'CH1'       }
    {'Vertical Units'           }    {'V'         }
    {'Vertical Units Div'       }    {[         0]}
    {'Vertical Units Extend Div'}    {[        16]}
    {'Label'                    }    {[<missing> ]}
    {'Probe Type'               }    {[         0]}
    {'Probe Ratio'              }    {[        10]}
    {'Vertical Scale'           }    {[         1]}
    {'Vertical Position'        }    {[   -1.7600]}
    {'Horizontal Units'         }    {'S'         }
    {'Horizontal Scale'         }    {[2.0000e-04]}

str = {'Memory Length','Horizontal Scale','Sampling Period'}';
idx = cellfun(@(x)strcmp(C1(:,1), x), str, 'Unif',0);
for k = 1:numel(idx)
    val(k,:) = C1{idx{k},2};
end

Results = table(str, val)

Results = 3x2 table
            str              val  
    ____________________    ______

    {'Memory Length'   }     1e+06
    {'Horizontal Scale'}    0.0002
    {'Sampling Period' }     2e-09

.

[Jason]

Wow. I found the below works with exponential form, but its not always exponential. so I prefer your solution. Thanks again

H_Scale=sscanf(Str,'Horizontal Scale,%e')  

[Star Strider]

As always, my pleasure!

[Jason]

Is there a reason why the "table" isn't showing up in the command window?

      fullpath=fullfile(folder,file)                   
      opts = detectImportOptions(fullpath);
      opts.DataLines=[1,100];    % limit the read to just the headers as 1M lines
      C1 = readcell(fullpath,opts);
      str = {'Memory Length','Horizontal Scale','Sampling Period'}'
      idx = cellfun(@(x)strcmp(C1(:,1), x), str, 'Unif',0);
      for k = 1:numel(idx)
         val(k,:) = C1{idx{k},2}
      end
      Results = table(str, val)

fullpath =

'F:\IMAGES\ATS_IMAGES\2025\ScopeReadings\DS0030.CSV'

str =

3×1 cell array

{'Memory Length' }

{'Horizontal Scale'}

{'Sampling Period' }

val =

1000000

val =

1.0e+06 *

1.0000

0.0000

val =

1.0e+06 *

1.0000

0.0000

0.0000

[Star Strider]

It should.

It does when I run it on my computer offline (R2024b):

Results =

3×2 table

str val

____________________ ______

{'Memory Length' } 1e+06

{'Horizontal Scale'} 0.0002

{'Sampling Period' } 2e-09

I copied that from my Command Window, and pasted it here. This should also work in R2023b, although I can’t specifically test it.

If all else fails, perhaps writetable and type would work.

.

Answer 2

[Cris LaPierre]

There are lots of ways to do this. Here is one modified from code I shared here: https://www.mathworks.com/matlabcentral/answers/2172834-2d-matrix-resize-or-interpolation#comment_3323358

data = readData('DummyData.txt')

data = struct with fields:
                    Format: '1.0B'
              MemoryLength: 1000000
              IntpDistance: 0
            TriggerAddress: -1
              TriggerLevel: 2.2000
                    Source: 'CH1'
             VerticalUnits: 'V'
          VerticalUnitsDiv: 0
    VerticalUnitsExtendDiv: 16
                     Label: ' '
                 ProbeType: 0
                ProbeRatio: 10
             VerticalScale: 1
          VerticalPosition: -1.7600
           HorizontalUnits: 'S'
           HorizontalScale: 2.0000e-04
        HorizontalPosition: 1.0000e-03
            HorizontalMode: 'Main'
                SincETMode: 'Real Time'
            SamplingPeriod: 2.0000e-09
        HorizontalOldScale: 2.0000e-04
     HorizontalOldPosition: 1.0000e-03
                  Firmware: 'V1.27.001'
                      Mode: 'Detail'
                  Waveform: [18x2 double]

plot(data.Waveform(:,1),data.Waveform(:,2))
xlabel("Time - "+data.HorizontalUnits)
ylabel("Amplitude - "+data.VerticalUnits)

function data = readData(fname)
r=0;
data = struct;
fid = fopen(fname);
while ~feof(fid)
    ln = fgetl(fid);
    [param,val] = strtok(ln,",");
    switch param
        case 'Waveform Data'
            fseek(fid,0,'eof');
            fgetl(fid);
        otherwise
            val = strtok(val,',');
            pat = asManyOfPattern(characterListPattern("0123456789-+.E"));
            nums = extract(val,pat);
            Param = matlab.lang.makeValidName(param);
            if length(val)==length(nums{1})
                data.(Param) = str2double(val);
            else
                data.(Param) = val;
            end
    end
    r = r+1;
end
fclose(fid);
val = readmatrix(fname,'FileType','text','NumHeaderLines',r);
data.Waveform = val;
end

[Jason]

Thankyou, this is also very nice and will be useful!

[Jason]

Could you expand on this line please

Param = matlab.lang.makeValidName(param);

[Cris LaPierre]

You can read about what this is doing here: https://www.mathworks.com/help/matlab/ref/matlab.lang.makevalidname.html

As an example, 'Vertical Units' cannot be used as a fieldname since it is not a valid variable name. This command converts it into a valid variable name so that the parameter name can be used as the field name in the returned structure.

matlab.lang.makeValidName('Vertical Units')

ans = 'VerticalUnits'