基于 PCRE 的用于重定向的正则表达式

Question

I have some links as follow.

http://www.example.com/excel.aspx
http://www.example.com:80/word.aspx
http://www.example.com:80/powerpoint.aspx
http://www.example.com:80/pdf.aspx
http://www.example.com:80/barcode.aspx
http://www.example.com/ocr.aspx
http://www.example.com/email.aspx
http://www.example.com/project.aspx

& I want to rewrite the URLs as following in a single regex pattern or single regex line.

https://www.example.com/excel
https://www.example.com/word
https://www.example.com/powerpoint
https://www.example.com/pdf
https://www.example.com/barcode
https://www.example.com/ocr
https://www.example.com/outlook
https://www.example.com/project

If you check & compare, I have replaced email with outlook in substitution URL.

I can match patterns as follow .*/(\\w+).aspx and substitute as https://www.example.com/ $1 but I cannot find a way to replace email with outlook . Demo https://regex101.com/r/71huIx/24

Something similar to http://www.rexegg.com/regex-trick-conditional-replacement.html but without a pool or dictionary. However, if the pool or dictionary can be part of regex itself then it will OK. Reason being, I have txt file with rules, where line1 is regex & line2 is a substitution.

Any ideas how can I do that?

Answer 1

Here's an Apache httpd config that should work:

RewriteEngine On
RewriteRule   ^/outlook$        https://example.com/email.aspx  [P,L]
RewriteRule   ^/([^./])/?$      https://example.com/$1.aspx     [P,L]

This uses a reverse proxy to serve pages like email.aspx and foobar.aspx when given "outlook" and "foobar" without revealing their extensions to users.

Though it looks like you're trying to go backwards for some reason. In that case, you'd want:

RewriteEngine On
RewriteRule   ^/email\.aspx$    https://example.com/outlook     [L]
RewriteRule   ^/([^.]+)\.aspx$  https://example.com/$1          [L]

Regex conditional expressions cannot perform your task. The structure of a regex conditional is:

(?(condition)yes-pattern|no-pattern)

• condition is either
  • a zero-width assertion (typically a lookahead)
  • a parenthesized integer, denoting whether that numbered capture has content
  • a capture group name in angle brackets or single quotes
  • (R) which is true when evaluated inside a recursion or eval
• yes-pattern is a regex to match given the condition is true
• no-pattern is an optional regex to match given the condition is false

Therefore, you can't shove new text into the match using a conditional.