Replace characters only when a certain number exist

Alan Kilborn

@Jim-Miller

This will find the lines with exactly 5 | characters:

find: ^(?:([^|\r\n]*)\|){5}(?1)\R
search mode: regular expression

But I’m not sure how to replace arbitrary ones of them! :-)

More info in THIS thread about this kind of thing.

Terry R

@Jim-Miller said in Replace characters only when a certain number exist:

I want to look for lines that have 5 pipes | and if 5 exist, replace number 1 and 3 with a comma.

I believe I have created a regex (regular exression) which will replace the numbers 1 and 3 if there are 5 | pipes in a line. However some caveats.

If more than 5 pipes it will still replace any 1 and/or 3 on that line.
It needs to run multiple times as ONLY replaces 1 number on each line for each run. If your lines will ONLY have 1 of the numbers then possibly only 1 run required.
As it uses the \K combination it MUST be run with the “Replace All” button.

Given that, here is what I have:
Find What:(?-s)^(?=.*(?:\|.*){5}).*\K(1|3)
Replace With:,

As stated before it’s a regex so search mode must be “regular expression”. Keep clicking on the “Replace All” button until the bottom of the window states “Replace All: 0 occurances were replaced in entire file”.

See if that helps at all.

Terry

PS edit, I assumed the actual number 1 and 3 were to be replaced, however I think I see @Alan-Kilborn assumes that the 1st and 3rd instance of the pipe character are to be replaced. Which is it?

guy038

Hello, @jim-miller, @Alan-kilborn, @terry-r and All,

I assume, like @alan-kilborn, that you want change the first and third | char, of any line, with a comma ,, ONLY IF current line contains, at least, five | characters

I propose this regex version :

SEARCH ^([^|\r\n]*)\K\|((?1)\|(?1))\|(?=(?2)\|)

REPLACE ,\2,

You may use as well, for the search regex, the free-spacing mode, below, with some explanations , in comments. Just select all that block and open the Replace dialog ( Ctrl + H )

(?x)               # FREE-SPACING mode
^                  # BEGINNING of line
(                  # START Group 1
[^|\r\n]*          #   ANY range, possibly NULL, of chars, DIFFERENT from PIPE | and EOL chars, till ...
)                  # END   Group 1
\K                 # RESET of current MATCH, so far
\|                 # ... a LITERAL PIPE char
(                  # START Group 2
(?1)  \|  (?1)     #   "Group 1 REGEX", followed by a LITERAL PIPE char and followed by an other "Group 1 REGEX", till ...
)                  # END   Group 2
\|                 # ... a LITERAL PIPE char
(?= (?2)  \| )     # ONLY IF "Group 2 REGEX", followed by a LITERAL PIPE char ( The FIFTH PIPE char ) is found, RIGHT AFTER

Important :

Whatever the search syntax used, You must click, exclusively, on the Replace All button ( Do not use the Replace button )
Select the Regular expression search mode, of course
Tick, preferably, the Wrap around option

Best Regards,

guy038

Jim Miller

@Terry-R First and third instance of pipe

Jim Miller

@guy038 said in Replace characters only when a certain number exist:

Am unsure if this will have worked but it causes my NP++ to crash. I updated to the latest version as I knew that I was a few versions behind but this gave the same result.

Alan Kilborn

I don’t see a Notepad++ crash with @guy038 's regex:

(?x)^([^|\r\n]*)\K\|((?1)\|(?1))\|(?=(?2)\|)

In fact, I don’t know of any regex you could specify that has been known to cause a N++ crash before??

But anyway, RegexBuddy complains about the regex, saying for the second (?1) that: Recursive calls need to be optional or the called group needs an alternative without recursion for the group to be able to match anything

…so, I think there is something funky going on with this.

guy038

Hi, @jim-miller, @Alan-kilborn, @terry-r and All,

To @alan-kilborn :

I don’t really understand the RegexBuddy comment ?!

Below, I indicated :

In the second line, where the two groups 1 and 2 are defined
In the third line, the regex
In the fourth line, where the subroutine calls are used

Definitions:      Gr_1           Gr_2
                _________      __________
Regex:        ^([^|\r\n]*)\K\|((?1)\|(?1))\|(?=(?2)\|)
                               ¯¯¯¯  ¯¯¯¯      ¯¯¯¯
Utilizations:                  Gr_1  Gr_1      Gr_2

And, we cannot see any subroutine call (?#), located within its own group # (...). So this clearly shows that this regex does not use any recursive syntax !

In order to get a recursive syntax, you need, for instance, this kind of regex construction : .......(......(?1)..)... , with no parenthesis in any “dot” zones, with the (?1) located inside the parentheses zone defining the Group1 !

To all :

Moreover, I tried with a small set of lines with pipe chars that I duplicated many times in order to get a 12,397,000 bytes file, containing about 506,000 lines.

Applying the regex S/R, with the free-spacing mode, against that file was successful, after 303,600 replacements in 57 s, due to my old Win XP SP3 machine ;-) And, of course, N++ did not crash at all !

Best Regards

guy038

P.S. :

In the short regex form, without the free-spacing mode, I simply omitted the useless modifier (?x), as there is no space char !

Alan Kilborn

@guy038 said in Replace characters only when a certain number exist:

I don’t really understand the RegexBuddy comment

Ha. Me either!

guy038

Hi, @alan-kilborn and All,

To be more accurate :

As a matter of fact, the RegexBuddy’s statement “Recursive calls need to be optional OR the called group needs an alternative without recursion for the group to be able to match anything” is totally exact

But I still do not see how this could apply to my regular expression, given the observations mentioned in my previous post. !

BR

guy038

Jim Miller

@guy038
I tried to run it on another machine and it executed successfully. I suppose I should have tried the first ;-)
Anyway, thank you very much for your help