regular expresssion to match something and not match something

Question

I want an nginx location directive that would match all urls with "xyz" and not match ".php"

I tried this

location ~* /(.*)xyz/(.*)(?!\.php)(.*)$ {}

but for example it always seem to match both /xyz/1 and /xyz/1.php but it should have matched only /xyz/1

Answer 1

I don't think you can do this without a separate regex checking to make sure that ".php" is absent. The problem is that constructs like .* will swallow up enough characters to make sure that your negative lookahead never applies.

Answer 2

Well, as Michael said, though .*? (note the '?') will make the search non-greedy and it could work.

...Or, you could try (rather hackish):

[^.][^p][^h][^p]

Might works, might not... It definately is sort of a hack.

Answer 3

You should be more specific about the begin and end of the location. If .php should not appear at the end of the location, put $ in the look-ahead assertion:

location ~* /(.*)xyz/(.*)(?!.*\.php$)(.*)$ {}

And if xyz should be a proper path segment and the first path segment (so it must start with /xyz/), use this:

location ~* ^/xyz/(?!.*\.php$)(.*)$ {}

Answer 4

This one is simple, yet it can be tough to see. The second (.*) in the regex is "greedy", hence, it will capture everything including the ".php" so, to the parser's "eyes" there is no ".php" after it (it's already passed) and the result is a match. :(

Just make the second .* lazy by changing it to .*? and it will solve your problem :

location ~* /(.*)xyz/(.*?)(?!\.php)(.*)$ {}

I suggest you read this article, it explains it in a much better way than I. :D

Answer 5

Thanks all, we finally came up with this

location ~* ^/xyz([^.]*)/([^.]*)$ {
            rewrite  ^/xyz([^.]*)/([^.]*)$ /xyz$1/index.php/$2 last;
}

and it works.