Correct Regex to match directories

Informative post for the archives.

When you are trying to match paths using Regex, it is tempting to use .*? to match a single directory, and while this works in most cases, in certain circumstances it will match multiple directories. It is better to use the match [^/]* as it will match ONLY a single directory.

If I am trying to match the category and type directories under templatedata and I have created an archive directory with a data directory underneath it, the following code and results illustrate the incorrect match.

$path = "/templatedata/cat/type/Archive/data/done.dcr";

$path = ~ m|/templatedata/(.*?)/(.*?)/data/|;

print "match: $1 | $2\n";

---------- Captured Output ----------
match: cat | type/Archive

If instead you use the [^/]* pattern you get the following.

$path = "/templatedata/cat/type/Archive/data/done.dcr";

$path =~ m|/templatedata/([^/]*)/([^/]*)/data/|;

print "match: $1 | $2\n";
---------- Captured Output ----------
match: |

while this does not match anything, it is correct as it should not match.

-------------------------------
E-Mail : farnsaw@stonedoor.com

Find more posts tagged with

Comments

Adam Stoller

$path = "/templatedata/cat/type/Archive/data/done.dcr";
$path = ~ m|/templatedata/(.*?)/(.*?)/data/|;
print "match: $1 | $2\n";

match: cat | type/Archive

$path =~ m|/templatedata/([^/]*)/([^/]*)/data/|;
print "match: $1 | $2\n";

match: | The first case works as desired because the pattern is read as:
Match against /templatedata/ followed by a non-greedy match of anything followed by / (cat) followed by a non-greedy match of anything followed by /data/ (type/Archive).

The second case doesn't work as desired because the pattern is read as:
Match against /templatedata/ followed by zero-or-more-non-slash characters (tentatively "cat") followed by a / followed by zero-or-more-non-slash charcters followed by /data/. This last part is what causes the entire match to fail - and therefore nothing is stored in either registry ($1 or $2).

In order to work, the second regex should have been: m|/templatedata/([^/]+)/([^/]+)/([^/]+)/data/|

Does that help make sesnse of what you're seeing?

--fish
Senior Consultant, Quotient Inc.
http://www.quotient-inc.com

Migrateduser

You are correct in your analysis, however, what I was trying to point out is that if you use the "non-greedy any character" match and you have a data directory buried some place under another directory, you can get an invalid "successful" match where the second match will only return valid "successful" matches or undef which you should then check for and handle as an error.

A more appropriate match to get the category and data type for templating would be just to match like so:

$path =~ m|/templatedata/([^/]+)/([^/]+)/|;

This will always return the category and data type correctly if there is one and return null otherwise. It does not validate that you are using the path of a DCR, however, it would be conceivable that you are looking for the category / type for a DCT or a TPL. This also forces there to be at least one character between the directory delimiter, a slash in this case, for both category and data type.

-------------------------------
E-Mail : farnsaw@stonedoor.com

jbonifaci

I believe what Adam was pointing out was that you were getting exactly what you were asking for in both of your regex's. They are both simple regex's and if one has RTFM'd about regex they would know what each is doing.

~Jeff

nipper

I believe what Adam was pointing out was that you were getting exactly what you were asking for in both of your regex's.

D**N computer, do what I mean, not what I say.