Re^2: Regular Expressions Matching with Perl

Is it guaranteed that the compression method field will never nest a space, like "LZW cmp" or something?

When building a regexp against sample data (as opposed to "against a specification") my approach tends to be exactly the opposite of Fletch's - make the regexp constrain as much as possible, so that I can warn if I ever see new data that violates my expectations:

  $line =~ m{^
    \s* \d+              # size?
    \s+ \w+              # compression method
    \s+ \d+              # compressed size?
    \s+ \d+ %            # compression ratio
    \s+ \d+ - \d+ - \d+  # date
    \s+ \d+ : \d+        # time
    \s+ [0-9a-f]{8}      # checksum?
    \s+ (.*)             # filename
  $}xi or warn "Couldn't match input line '$line'";
  $filename = $1;
[download]

It is worth checking whether it is possible to store a filename with some odd characters to see what happens, such as a newline, backslash etc. Similarly it is worth looking for boundary conditions on other fields - if the size is more than 8 digits does it still retain at least one following space?

Hugo

Comment on Re^2: Regular Expressions Matching with Perl Download Code

Replies are listed 'Best First'.
Re^3: Regular Expressions Matching with Perl by ikegami (Patriarch) on Apr 21, 2005 at 14:18 UTC
Is it guaranteed that the compression method field will never nest a space, like "LZW cmp" or something? yes, I think it's always one word (and probably specifically for easy parsing, judging by the odd names). When building a regexp against sample data my approach tends to be exactly the opposite of Fletch's I call the two approaches "Extraction" (`/:.{15}(.)/`) and "Validation" (your's). Which I use is determined by the situation. Sometimes, there's a happy middle that's a mixture of both (Fletch's `/[[:hexdigit:]]{8}\s+(.)$/`).	[reply] [d/l] [select]

Replies are listed 'Best First'.

Re^3: Regular Expressions Matching with Perl
by ikegami (Patriarch) on Apr 21, 2005 at 14:18 UTC

Is it guaranteed that the compression method field will never nest a space, like "LZW cmp" or something?

yes, I think it's always one word (and probably specifically for easy parsing, judging by the odd names).

When building a regexp against sample data my approach tends to be exactly the opposite of Fletch's

I call the two approaches "Extraction" (/:.{15}(.*)/) and "Validation" (your's). Which I use is determined by the situation. Sometimes, there's a happy middle that's a mixture of both (Fletch's /[[:hexdigit:]]{8}\s+(.*)$/).

[reply]
[d/l]
[select]