Regexp expression to handle changing format

%dummy data

% t,00000000CIB0000004001,0.47,L,000 00:00:00.00,343 19:54:20.684 8,22.501

% t,00000000CIB0000004001,0.47,L,000 00 00:00:00.00,21 343 19:54:20.684 8,22.501

S=fileread(filename);

myexpression = ['(?<tvar>w*,'...

'(?<tmCodeRdr>\w*),'...

'(?<tmCodLvl>\w*\.*\w*),'...

'(?<HNL>\w*),'...

'(?<codeTm>\w*\s*\d*\:*\d*\:*\d*\.*\d*,'... % <== This line handles the first line of dummy data

'(?<caprTm>\w*\s*\d*\:*\d*\:*\d*\.*\d*\s*\d*,'... % <== This line handles the first line of dummy data

'(?<logAt>\w*\.*\w*']

parts = regexp(filtered,myexpression,'names')

The third and second to last variables (codeTm, caprTm) change formats within the data. How can I modify or add logic to accept 2 to 3 spaced values within the variable "codeTm" and 3 to 4 spaced values within variable "caprTm"???

2 spaced valued variable (000 00:00:00.00)

3 spaced valued variable (000 00 00:00:00.00) or (343 19:54:20.684 8)

4 spaced valued variable (21 343 19:54:20.684 8)

Thank you for the help. My apologies for making my expresion so complicated. Still learning the in's and out's for expression formats for regexp to read data.

[Stephen23]

It is not clear why you are using regular expressions for importing this data: READTABLE et al have options for handling missing field data. Having you considered using the inbuilt data importing functions?

[jimmy zubiate]

In the process of learning Matlab. Persued regexp function to create a structure array where I could maneuver through the values to perform analysis needed.

What I'm thinking I should pursue is prep file to remove unwanted white space, headers and other non-useful data and import as a comma space delimited file. Then I can count items inside each variable, marked by spaces and then off to the next step.

Other option is pursue fgetl function and implement logic to read useful data gracefully. I'm attaching dummy test data for your viewing. Thanks.

Answer 1

[Stephen23]

You can easily make a group optional or occur a specific number of times using any suitable quantifier, for example:

(..)? % zero or one time
(..)* % zero or more times
(..){2,4} % two to four times
etc.

https://www.mathworks.com/help/matlab/matlab_prog/regular-expressions.html#f0-43073

https://www.mathworks.com/help/matlab/matlab_prog/regular-expressions.html#f0-42884

However, rather than trying to match specific groups of characters I would use a simpler approach of matching sets of characters. I had to fix several other bugs in your regular expression to get this working, mostly missing backslashes and parentheses.

str = fileread('test.txt')

str = 
    't,00000000CIB0000004001,0.47,L,000 00:00:00.00,343 19:54:20.684 8,22.501
     t,00000000CIB0000004001,0.47,L,000 00 00:00:00.00,21 343 19:54:20.684 8,22.501'

rgx = ['^\s*(?<tvar>\w*),'...
    '(?<tmCodeRdr>\w*),'...
    '(?<tmCodLvl>\d*\.?\d*),'...
    '(?<HNL>\w*),'...
    '(?<codeTm>[ :\w\.]*),'...
    '(?<caprTm>[ :\w\.]*),'...
    '(?<logAt>\d*\.?\d*)'];
parts = regexp(str,rgx,'names','lineanchors')

parts = 1×2 struct array with fields:
    tvar
    tmCodeRdr
    tmCodLvl
    HNL
    codeTm
    caprTm
    logAt

parts.codeTm

ans = '000 00:00:00.00'

ans = '000 00 00:00:00.00'

But personally I would not try and reinvent the wheel for such a data file, READTABLE is much simpler:

tbl = readtable('test.txt','delimiter',',');
tbl.Properties.VariableNames = {'tvar','tmCodeRdr','tmCodLv','HNL','codeTm','caprTm','logAt'}

tbl = 2×7 table
    tvar             tmCodeRdr            tmCodLv     HNL             codeTm                     caprTm              logAt 
    _____    _________________________    _______    _____    ______________________    _________________________    ______

    {'t'}    {'00000000CIB0000004001'}     0.47      {'L'}    {'000 00:00:00.00'   }    {'343 19:54:20.684 8'   }    22.501
    {'t'}    {'00000000CIB0000004001'}     0.47      {'L'}    {'000 00 00:00:00.00'}    {'21 343 19:54:20.684 8'}    22.501

[jimmy zubiate]

That should work. Let me try to implement on my side and see what I get. Thanks Stephen!