howto process non-numeric text file and convert into struct most efficiently? - matlab

Question

I have a bunch of .txt files that have the following format:
|file | time | color | tags |
|1 | 1:10 | red | ok, correct|
|2 | 2:20 | blue | bad |
|3 | 1:20 | yellow | sometag |
The first row specifies the column names.
The subsequent rows are 'database' entries.
I want to read in this file, and put all the information into a Matlab structure. I'm wondering what the most efficient way of doing this is.
textread with 'delimiter', '\n' and process each line individually?
textread with 'delimiter', '|' and having to determine which entries belong together?
fread line by line?
I love the convenience of textread with 'delimiter', '\n' but then it's quite a pain to get out the individual entries for each column (with a for loop). Alternatively I can split each row up using regexp:
regexp(file{1}, '\|', 'split')
But this will only split up each row, and won't take care of the whitespaces (for which I would need another regexp call to in order to get rid of).
So what's the most straight forward (and maybe even most efficient) way of doing this?
EDIT1 I want to build a struct like db = struct('file', [], 'time', [], 'color', [], 'tags') which is easy to create once I've read-in the first line.

Answer 1

Let's use textscan:
fid = fopen('asdf.txt');
header = textscan(fid, '%s', 5, 'delimiter', '|');
data=textscan(fid, '|%d %s %s %s','delimiter','|');
fclose(fid);
which gives:
data =
[3x1 double] {3x1 cell} {3x1 cell} {3x1 cell}
From here it's easy to go to the struct you wanted. The strings also have some extra spaces, which need trimming:
data{2} = strtrim(data{2});