[Solved] Regular Expression matched value replacement

[Phillip]

I want to re-arrange
=IF(ISNUMBER(E5); IF(NOW()>AgeT+E5;"#";"");"")
to be
=IF(ISNUMBER(E5); IF(NOW()>E5+AgeT;"#";"");"")

However Find & Replace - using Regular Expressions in More Options
Search for: >AgeT\+(.*);
Replace with: >$1+AgeT;
Results in: =IF(ISNUMBER(E5); IF(NOW()>E5;"#";"")+AgeT;"")

It would seem that the "(.*);" matches all characters after "AgeT+" until the last ";" instead of the first ";". Why?
NB.Search for: >AgeT\+([^;].*); does the same !

[Jan_J]

Regular expressions behave greedy, i.e. they try to find the widest acceptable phrase
You can go two ways:
(1) as you tried >AgeT\+([^;].*); but there you did not express what you wanted. Correct to >AgeT\+([^;]*);
or
(2) convert pattern to nongreedy using ? specifier: >AgeT\+(.*?);
// corrected the former misspell >AgeT\+(.*);?

Both should be equivalent.

[Phillip]

Thanks Jan-J - my error in adding the extra "."
When I tried: >AgeT\+([^;]*);
Works OK - Result: =IF(ISNUMBER(E5); IF(NOW()>E5+AgeT;"#";"");"")

NB Also tried: >AgeT\+(.*);?
Didn't work - Result: =IF(ISNUMBER(E5); IF(NOW()>E5;"#";"");"")+AgeT;
Perhaps because you are on Unix and I am on Windows?

[Jan_J]

my error in adding the extra "."

Indeed. [^;].* means `any characters in a sequence of arbitrary length, preceded by single non-semicolon`.
[^;]* means `any sequence of non-semicolon characters`.
[^;]+ is the same, except that the phrase must have nonzero length.

Also tried: >AgeT\+(.*);?

Well, my fault in hurry. Should be >AgeT\+(.*?);

U**X, Windows: no matter. The regex library is the same in OpenOffice.

[Phillip]

Many thanks. Tried: >AgeT\+(.*?); worked ok. I could not find an explanation for the "?", will look again.
I use regex a lot with SED on windows - one difference is Windows requires a " instead of ' as in sed -i -r "s/^[0-9]+.*-/,-/g" file.txt
Apologies for "U**X" just saw the reference to "Unix tools".

[robleyd]

I could not find an explanation for the "?"

From the HELP:

?   Finds zero or one of the characters in front of the "?". For example, "Texts?"
    finds "Text" and "Texts" and "x(ab|c)?y" finds "xy", "xaby", or "xcy".

[Jan_J]

More precisely, or more subtle: ? also changes context of pattern matching from greedy (find the widest case) to conservative (find the narrowest one).
It is not explicitly stated in OO/LO help. However, there's a link from LO help (https://help.libreoffice.org/latest/en- ... 00001.html?)

For a full list of supported metacharacters and syntax, see ICU Regular Expressions documentation

to ICU Regex service (https://unicode-org.github.io/icu/userg ... characters), and there you read:

? Match zero or one times. Prefer one.
*? Match 0 or more times. Match as few times as possible.
+? Match 1 or more times. Match as few times as possible.

That's it.