How to use regexp to read numbers preceded by text out of a data file

Hello, i tried to get numbers out of a data file with regexp but i just cant get it to work.

The file is structured like this:

Time = 10001

smoothSolver:  Solving for Ux, Initial residual = 1.660195e-07, Final residual = 3.2635415e-09, No Iterations 2
smoothSolver:  Solving for Uy, Initial residual = 0.0010739333, Final residual = 1.1651965e-07, No Iterations 1
smoothSolver:  Solving for Uz, Initial residual = 1.1728104e-06, Final residual = 2.3091287e-08, No Iterations 2
GAMG:  Solving for p, Initial residual = 1.0923774e-05, Final residual = 7.2745974e-07, No Iterations 3
time step continuity errors : sum local = 0.34757443, global = 0.0029670501, cumulative = 0.0029670501
ExecutionTime = 0.92 s  ClockTime = 3 s

i want to get out the values behind "Solving for Ux, Initial residual =", "Solving for Uy, Initial residual =", "Solving for Uz, Initial residual =", "Solving for p, Initial residual" and "ClockTime ="

i tried testing this with the following code:

S = fileread(path);

residuals = regexp(S,"(?<=ClockTime\s=\s)\d*");

This should find any number of consecutive digits following the string "ClockTime =" but instead it gives me the numbers 595 and 1193 which the file does not even contain as consecutive digits.

I read this already but it is not really helping me.

Thanks for any help :)

Answer 1

[Walter Roberson]

residuals = regexp(S,"(?<=ClockTime\s=\s)\d+", 'match');

The numbers you are getting back are the default outputs, which is the position relative to the start of the string.

Note: for your work you might want to use named tokens.

residuals = regexp(S,"(?<=residual\s =)(?<XR>)\S+).*(?<=residual\s =)(?<YR>)\S+).*(?<=residual\s =)(?<ZR>)\S+).*(?<=residual\s =)(?<PR>)\S+)(?<=ClockTime\s=\s)(?<CT>\d+)", 'names');

if all went well, if I did not make mistakes in the pattern, this would return a struct array with fields XR YR ZR PR CT each of which is a character vector, with XR corresponding to 'Solving for x', YR for the next line, ZR for the third line, PR for the GAMG line, and CT for the ClockTIme.

You would then

Nresiduals = structfun(@double, residuals, 'uniform', 0);
Ux = [Nresiduals.XR]; Uy = [Nresiduals.YR]; Uz = [NResiduals.ZR]; p = [NResiduals.PR];
clocks = [NResiduals.CT];

[Jean Volkmar]

Thanks for your help ! Unfortunately the regexp does not work like this. How exactly is "residual\s" enogh information to indentify which value it is, i.e Ux,Uy..... ? Doesn't the expression need to know the exaxt preceding text to find the right value ? Also how is " \S+).* " defining where to stop reading the number ?

The clockTime expression works tough.

[Walter Roberson]

The search is done from the beginning. The first residual encountered is saved in XR, the second in YR, and so on. Ah, I guess I should modify to

residuals = regexp(S,"(?<=residual\s =)(?<XR>)\S+).*?(?<=residual\s =)(?<YR>)\S+).*?(?<=residual\s =)(?<ZR>)\S+).*?(?<=residual\s =)(?<PR>)\S+).*?(?<=ClockTime\s=\s)(?<CT>\d+)", 'names');

to move only to the nearest "residuals" afterwards instead of going to end of file. Also I was missing a .* before clock time.

\S+ is "all non-space". Ummm, that's going to grab the commas too... okay, modify that to end at comma...

residuals = regexp(S,"(?<=residual\s =)(?<XR>)[^,]+).*?(?<=residual\s =)(?<YR>)[^,]+).*?(?<=residual\s =)(?<ZR>)[^,]+).*?(?<=residual\s =)(?<PR>)[^,]+).*?(?<=ClockTime\s=\s)(?<CT>\S+)", 'names');

Non-comma should be able to handle reading 1.660195e-07 (exponential notation) and 0.0010739333 (not exponential) and even possible plain integers (I can tell that a %g format is being used and so if the time just happened to be integer it would not emit any decimal place at all)

The last match I changed from \d+ to \S+ in case at some point there is a non-integer time in the file. It does not need to check for comma as only the residuals have comma.

[Jean Volkmar]

Alright thanks, but it still only returns a 0x0 struct. From my understanding it should be "\S+[^,]+)" so it saves all the spaces after the equal sign up to the commma to XR . Also i think there must be a "/s" at the end of (?<=residual\s =), like this: (?<=residual\s =\s) but after trying this it still does not return any values.

[Walter Roberson]

I will experiment later after I have had some sleep.

[Jean Volkmar]

Okay so i also had a go at experimenting, in the end i did not get it to work as nicely but better than nothing.

Here is what i did:

path(1) = "";

residuals = cell(length(path),5); %Preallocate

for j = 1:1:length(path)
    
S = fileread(path(j)); %read file

residuals{j,1} = regexp(S,"(?<=Ux,\sInitial\sresidual\s=\s)[^,]+","match"); %Find a number of consecutive charakters after a matching string "Ux, Initial residual = " 1 or more times consecutively, and exclude the comma. 
residuals{j,2} = regexp(S,"(?<=Uy,\sInitial\sresidual\s=\s)[^,]+","match"); %Find a number of consecutive charakters after a matching string "Uy, Initial residual = " 1 or more times consecutively, and exclude the comma. 
residuals{j,3} = regexp(S,"(?<=Uz,\sInitial\sresidual\s=\s)[^,]+","match"); %Find a number of consecutive charakters after a matching string "Uz, Initial residual = " 1 or more times consecutively, and exclude the comma. 
residuals{j,4} = regexp(S,"(?<=p,\sInitial\sresidual\s=\s)[^,]+","match"); %Find a number of consecutive charakters after a matching string "p, Initial residual = " 1 or more times consecutively, and exclude the comma. 
residuals{j,5} = regexp(S,"(?<=ClockTime\s=\s)\d+","match"); %Find a number of consecutive digits after a matching string "ClockTime = " 1 or more times consecutively, and exclude the comma. 

end

After that i saved the numbers into an array like this:

ux = zeros(length(path),length(residuals{1})); %Preallocate

for k = 1:1:length(path)
    for j = 1:1:length(residuals{1})
        ux(k,j) = str2double(residuals{k,1}(1,j)); %Convert string inside the cell to double
    end
end