Java Regex的低位字母下划线，左下划线，中间等于下划线，右下划线

Question

im reading a file line for line and want only the lines like:

gui_test=Hallo
gui_test_hello=Hello

So on the left side are only small letter words separated by _ . In the middle is always the = and on the right side can be any.

I made this regex:

^(([a-z]+_[a-z]+)+=.*)$

The Problem is, that it is working for

gui_pfc_button_ok=Ok!

But not working for:

gui_pfcuser_opportunitydetails_label_title=label Title

I don't know where my Problem is.

Answer 1

Brief

The problem you're facing is that regex only captures the last match into the group (at least in most engines; .net is an exception here). You're using ([az]+_[az]+)+ and, while this is valid, you're only matching the last occurrence that matches this pattern, thus, you're only getting l_title . Since the match is greedy you're getting l_ because the previous match took as many characters as it could. Breaking this down you're actually matching in the following way:

• gui_pfcuse
• r_opportunitydetail
• s_labe
• l_title

Your current regex will also fail if you only have 1 letter between underscores such as something_here_a_test . See your regex in use here

---

Code

See regex in use here

^([a-z]+(?:_[a-z]+)*)=(.*)$

You can also use ^((?:[az]+_)*[az]+)=(.*)$ but it's less efficient (uses more steps than the regex above).

---

Results

Input

gui_test=Hallo gui_test_hello=Hello
gui_pfcuser_opportunitydetails_label_title=label Title
gui_pfc_button_ok=Ok!
something_here_a_test=More words

Output

1. Match: gui_test=Hallo gui_test_hello=Hello
   • Group 1: gui_test
   • Group 2: Hallo gui_test_hello=Hello
2. Match: gui_pfcuser_opportunitydetails_label_title=label Title
   • Group 1: gui_pfcuser_opportunitydetails_label_title
   • Group 2: label Title
3. Match: gui_pfc_button_ok=Ok!
   • Group 1: gui_pfc_button_ok
   • Group 2: Ok!
4. Match: something_here_a_test=More words
   • Group 1: something_here_a_test
   • Group 2: More words

---

Explanation

• ^ Assert position at the start of the line
• ([az]+(?:_[az]+)*) Capture the following into capture group 1
  • [az]+ Match any lowercase ASCII letter one or more times
  • (?:_[az]+)* Match the following any number of times. If you require at least one match you can change * to + such that you end up with (?:_[az]+)+
    • _ Match this literally
    • [az]+ Match any lowercase ASCII letter one or more times
• = Match this literally
• (.*) Capture any character (except newline characters) into capture group 2
• $ Assert position at the end of the line

Answer 2

^(([a-z]+_)+[a-z]+\=.*) $

Tested on https://regex101.com/ where the various parts are:

^          Asserts position at start of a line
([a-z]+_)+ One or more sequences of the form <one_or_more_lower_case>_
[a-z]+     One or more lower case letter
\= escaped equals sign
.*         zero or more characters
$          end of line

Answer 3

Just move your underscore into the character class brackets, and you're home. You can remove the + for the inner group as well if you want...

^(([a-z]+[_a-z]+)=.*)$

If you want, you can try it out at regex101.com .

If you don't want to match lines with two consecutive underscores, you can group on the underscore part, in pseudocode xxx(_xxx)* , which would then become

^([a-z]+(_[a-z]+)*+=.*)$

Also testable at regex101.com .