Reading nth value in a line starting with a word

I am trying to read a text file that's designed more for human than for computer reading, i.e. it has large sections of copyright information, full sentences, etc. But there is one line I want to read that has this structure:

Ivac, m/s 3 4

The "3" and "4" are placeholders for real numerical values. I want to extract the value in Column 4 from this line starting with "Ivac," but don't know how. I have tried to look up how to do it, but I can't really understand any of the information about how to do this with regular expressions, and the other methods fail on the first line of my text file, which is just blank (right before a lot of copyright info).

Any assistance in resolving this problem would be greatly appreciated.

Answer 1

[Walter Roberson]

S = fileread('FilenameGoesHere.txt');

parts = regexp(S, 'lvac,\s+m/s\s+(?<col3>\S+)\s+(?<col4>\S+)', 'names', 'once');
col3 = str2double(parts.col3);
col4 = str2double(parts.col4);

This code does not assume integers, but it does assume that there are no stray characters immediately adjacent. For example,

 lvac, m/s   3°  17.2µm

would fail.

[Alex Dworzanczyk]

Thanks! So this gives me two issues. First, this is part of a larger thing where I'm trying to generate about 5,000 output files to process like this and pick the 'best' one. Unfortunately, not all the output files will even have that line, so I'll have to include a conditional If before using your approach (I think I can handle that).

Also, can you explain some of the regular expression language you use? I suppose the \s establishes that there's a space after each column, while, within the parentheses, the ? means "indefinitely large number of white spaces before" and then the \S+ means "indefinitely large number of white spaces after"? To what do 'names' and 'once' refer?

[Walter Roberson]

\s does not mean space, exactly: it means any kind of "whitespace", which includes space, tab, newline, formfeed. \s+ means one or more whitespace characters are to be matched.

\S means anything that is not whitespace, so \S+ means one or more non-whitespace characters. Basically \S+ matches a column of space (or tab) delimited data.

'once' as an option means that MATLAB should stop searching once the pattern has been found once. It also has some effect on how output is returned when the user has asked for matched text to be returned as character vectors: with 'once' it returns one character vector for each input character vector, whereas without 'once', it returns a cell array of character vectors for each input character vector, even if only one result is found.

regexp() uses the syntax (?SOMETHING) for a number of advanced features. The ? does not act as a quantifier for (?SOMETHING) syntax.

In particular, (?<NAME>PATTERN) means that what is matched by the given PATTERN is to be stored in a struct field named NAME . So (?<col3>\S+) means to match an entire non-blank column, and to store what was found in a struct field named col3 . Thus, the effect of this regexp is to request that a struct be returned that has a field named col3 that matches the column after "lvac, m/s" and a field named col4 that matches the column after that.

'names' as an option tells it to return the structure mentioned before. When 'names' is requested and 'once' is not present in the options, then the struct returned can be a non-scalar struct; with 'once' it will either be an empty struct or a scalar struct.

You can tell whether there was any match by checking isempty(parts)

S = fileread('FilenameGoesHere.txt');

parts = regexp(S, 'lvac,\s+m/s\s+(?<col3>\S+)\s+(?<col4>\S+)', 'names', 'once');
if isempty(parts)
    %the line is not present, do something appropriate
else
  col3 = str2double(parts.col3);
  col4 = str2double(parts.col4);
end