How do i find one string with another?

Hi Guys,

I am trying to sequentially look for a string in a document and was wondering how I would go about doing that.

Essentially I have a large file called A.csv with a bunch of columns [date], [Open], [Low], [Close], [Volume], [Adj.Close], [Ret]

I want to write a script that will find a date 4/5/2000 and will pull the corresponding return for that date.

This is the trick: the day is a variable. All of the dates have a different month and year so it looks something like 1/3/2000, 2/4/2000, 3/1/2000. How do I find a match using the year and month. For example, I want to pull 1/*/2006, 7/*/2007, but I don't know what the * is (it could be 1, 2, 3, 4, 5, etc...)

The first row for example (skipping the header), looks like: 1/3/2000,78.75,78.94,58.13,66.19,1642300,62.37,0.569183903

Thank you for all of your help guys!

-Larry G.

[Laurentiu Galan]

I want to be able to pull the return (0.569183903) and place it in a vector.

Answer 1

[Andrew Newell]

The trick is to use regular expressions. The first line below searches for any string that has one or more integers between '1/' and '/2000'. One line at a time is examined and the number extracted if there is a match.

match_str = '1/[0-9]+/2000';
match_vector = zeros(32000,1); % Use whatever size you're sure is large enough
fid=fopen('yourfile.m');
count=0;
tline = fgetl(fid);
while ischar(tline)
  if regexp(tline,match_str)
      A = textscan(tline,'%*s %*f %*f %*f %*f %*d %*f %f','delimiter',',');
      count = count+1;
      match_vector(count) = A{1};
  end
end
fclose(fid);

[Laurentiu Galan]

This one looks like it will help me the most. I get an error in the line match_vector = [match_vector; A{1}];

[Laurentiu Galan]

Btw, is there a way to speed this up if I will be doing 32,000 matches across something like 8,000 different files?

Thanks!

[Andrew Newell]

What sort of error? To speed it up (and perhaps correct the error), I'll modify it to preallocate match_vector.

[Andrew Newell]

Ah - I found the error: I should be updating match_vector inside the IF-END block.

[Laurentiu Galan]

attached update below

Answer 2

[Walter Roberson]

regexp(STRING, '^(?<=1/\d+/2006/.*,)[^,]+$', 'match', 'dotexceptnewline', 'lineanchors')

Answer 3

[Laurentiu Galan]

Thanks Andrew!! This is great. Now comes the really tough part. I want to loop the whole process.

Basically I want to open a file 'A' which contains a series of Ticker symbols (A.csv, B.csv, (about 8000 of them) and month and date information).

Then I want to open each individual file in a directory which are named based on the ticker symbols in file 'A'. Finally, I want to pull returns from each file using the month and date information also located in file A.

I don't expect you to be able to help me with this task as it is really extensive, but I was wondering if you wouldn't mind some sharing some insight as to how I can improve the whole process?

Your answer was more than sufficient as it has helped organize some of my thinking process. Thanks a bunch!! Any additional insight is greatly appreciated.

I attached my existing code to try and better explain what I mean:

%Code to Get Matrix

%fid=fopen('C:\Users\Laurentiu Galan\Desktop\pca1.csv');
C = textscan(fid, '%s %s %s %*s %*s %*s %*s', 'delimiter', ',', ...
  'HeaderLines', '1');
fclose(fid);

%Strcat Identifier
    tickername=C{1}
    year=C{2};
    month=C{3};

%Get Size of Loop for filepath
D=size(tickername);
numval=D(1,1);

%Create Loop for filepaths
for i=1:numval;
  filepath(i,1) = strcat('C:\Users\Laurentiu Galan\Desktop\', tickername(i,1), '.csv');
end;

%Create Matching value
for i=1:numval;
    ssearch(i,1) = strcat(month(i,1), '/[0-9]+/', year(i,1));
end;

%Open file where search will take place (path name will be looped)
match_str = '1/[0-9]+/2000'; %(This will also be looped based ssearch
match_vector = zeros(32000,1); 
fid=fopen('-------'); %<- Loop for each file goes here
count=0;
tline = fgetl(fid);
while ischar(tline)
  if regexp(tline,match_str)
      A = textscan(tline,'%*s %*f %*f %*f %*f %*d %*f %f','delimiter',',');
      count = count+1;
      match_vector(count) = A{1};
  end
end
fclose(fid);

%Then I want to output a file with all the returns and ticker symbols on
%the desktop

[Andrew Newell]

I don't see anything obvious. For any code I suggest the following sequence: (1) test it thoroughly to make sure it works; (2) run it with the MATLAB Profiler and see where the code is spending most of its time; and (3) look for ways to speed up that part of the code.