Remove duplicate lines not possible?

Cletos

@Alan-Kilborn said in Remove duplicate lines not possible?:

It is a tough problem to solve with Notepad++ alone

There once was such option to remove spread duplicates, if I remember it right.

sometimes the techniques to do it work, sometimes they don’t. It is data-dependent

Yes, I understand. But why replace an option doing both (spread and following lines) with one doing only one of them.

So it is not possible with Notepad at the moment.

Many thanks!

Alan Kilborn

@Cletos said in Remove duplicate lines not possible?:

There once was such option to remove spread duplicates, if I remember it right.

No, only a way to do it via regular expressions discussed here on the Community – that’s probably what you remember.

So it is not possible with Notepad++ at the moment.

Well, you can try it with the regular expression technique; search the Community site and you’ll rediscover the links with instructions.

Cletos

Alright, thank you very much!

guy038

Hi, @cletos, @alan-kilborn and All,

Alan, as you know, I’ve certainly answered this question, many times ! But, I’m a bit lazy and, instead of finding the different links, for the OP, I prefer to “re-invent the wheel” ;-))

So @cletos, here is the magic regular expression S/R, which deletes all duplicates lines, without changing the order of the lines

SEARCH (?-s)^(.+\R)(?=(?s:.*)^\1)
REPLACE Leave EMPTY
Tick the Match case option, if you prefer a case detection
Tick the Wrap around option, preferably
Select the Regular expression search mode
Click on the Replace All button ( or use the “step by step” Replace button to verify how the regex works ! )

Remark :

Let’s suppose that your initial text is :

aaa
bbb
ccc
ddd
bbb
bbb
eee
fff
bbb
ggg
bbb
hhh
iii

Then this regex S/R will delete :

The bbb line between lines aaa and ccc
The bbb line between lines ddd and bbb
The bbb line between lines bbb and eee
The bbb line between lines fff and ggg

And keeps, only the line bbb, located between lines ggg and hhh

So, to sum up, this regex S/R keep all the last duplicate lines found, in the input text !

So your final text becomes :

aaa
ccc
ddd
eee
fff
ggg
bbb
hhh
iii

I cannot get an other layout, with a correct regex S/R ! ( For instance, keeping the line bbb between lines aaa and ccc and deleting all subsequent bbb lines ) Sorry for this limitation !

IMPORTANT :

The last line of your list must always be followed with a line-break
Be aware that the behaviour of this regex S/R is rather weird ! It works nice with small or middle-size text to process. But :
- If your file has a big size, over 10 Mb about, even not concerned with duplicates lines, OR
- If 2 duplicate lines are separated with, let’s say, more than 10,000 lines

It may happen that this S/R is completely wrong, with an extra occurrence, matching all the file contents :-(( It mainly depends on our Boost regex engine and, probably, on the amount of your system memory !

As always, give it a try, with your real files, to see how this regex S/R acts !?

Two possible solutions, if any problem occurs :

Use, the Replace button repeatedly ( or the Alt + R shortcut ) and stop when a particular replacement wipe out, wrongly, all file contents !
Split your text in smaller parts, processing this regex S/R on each part, first. Then, merge all the pieces and process, again, the regex S/R on the whole set !

Best Regards,

guy038

Cletos

Hello guy038,

Thank you very much!

I cannot get an other layout, with a correct regex S/R ! ( For instance, keeping the line bbb between lines aaa and ccc and deleting all subsequent bbb lines ) Sorry for this limitation !
No, no, it works great!

The last line of your list must always be followed with a line-break

So one has to just press ENTER at the end of that last line in the txt file.

If your file has a big size, over 10 Mb about, even not concerned with duplicates lines, OR

So I could try splitting the processing on the first half of the txt file and the last half or even smaller and hope there are many lines removed and the file gets smaller.

Be aware that the behaviour of this regex S/R is rather weird ! It works nice with small or middle-size text to process. But :

Works great after some testing.

Two possible solutions, if any problem occurs :

Use, the Replace button repeatedly ( or the Alt + R shortcut ) and stop when a particular replacement wipe out, wrongly, all file contents !

Split your text in smaller parts, processing this regex S/R on each part, first. Then, merge all the pieces and process, again, the regex S/R on the whole set !

I will try it like that.

Thank you very much, again!

Sofistanpp

@guy038 said in Remove duplicate lines not possible?:

I cannot get an other layout, with a correct regex S/R ! ( For instance, keeping the line bbb between lines aaa and ccc and deleting all subsequent bbb lines ) Sorry for this limitation !

Hi guy038, Cletos, All:

Not a regex solution, but if you reverse the list —for example, by means of the Reverse Lines plugin— and run the nice regex you provided, you will get the first “bbb” with all duplicates being deleted. Once you are finished, reverse the list again to get the original order of lines.

Hope you find this, my first post here, useful.

Best Regards.

Cletos

Hello Sofistanpp,

OK, sounds very good! Many thanks!

Sofistanpp

@Cletos Glad to be of help.

Alan Kilborn

@Sofistanpp

Maybe explain how reversing the lines helps?

Sofistanpp

@Alan-Kilborn Sure. It looks to overcome a limitation pointed out by guy038, who wrote that the regex he posted remove all the duplicates except the last one, but it seems that he wanted to keep the first one. So if you reverse the order of lines and run the regex, you will remove, of course, all the instances except the last duplicate — now reverse the list back to the original order and you would have actually kept the first instance of the line —the “bbb” between “aaa” and “ccc” of the example.

Hope it is clear now (English is not my first language).

Best Regards.

Alan Kilborn

@Sofistanpp

Ah, okay, I missed the point about wanting to keep the first rather than the last. Thanks for the clarification.

guy038

Hi, @cletos, @sofistanpp, @alan-kilborn and All,

@sofistanpp, I didn’t want to privilege any solution but, indeed, it’s good to be able to chose, with your clever idea of using the Reverse Lines plugin, between these two solutions :

Keep the first duplicate line and delete all subsequent duplicate lines
Delete any duplicate but just keep the last duplicate line

Now, thinking about it, I found out a solution which can be processed within N++ only, preventing from using any external tool

If we go back to my previous example, open the Column editor ( Edit > Column Editor... ) and, moving the caret to the first column of the first line of your text, create a new number’s list ( Don’t forget to tick the Leading zeros option ! )

Then after adding 1 or several blank character(s), after each number, with the column mode selection, you should get :


01 aaa
02 bbb
03 ccc
04 ddd
05 bbb
06 bbb
07 eee
08 fff
09 bbb
10 ggg
11 bbb
12 hhh
13 iii

Now, sort the lines with the option Edit > Line Operations > Sort Lines Lexicographically Descending, giving :

13 iii
12 hhh
11 bbb
10 ggg
09 bbb
08 fff
07 eee
06 bbb
05 bbb
04 ddd
03 ccc
02 bbb
01 aaa

Finally, after running this new version of my previous regex S/R :

SEARCH (?-s)^\d+\h+(.+\R)(?=(?s:.*)^\d+\h+\1)
REPLACE Leave EMPTY

You’re left with :

13 iii
12 hhh
10 ggg
08 fff
07 eee
04 ddd
03 ccc
02 bbb
01 aaa

Finally, after the second sort Edit > Line Operations > Sort Lines Lexicographically Ascending, in the reverse order, we have the following output text :

01 aaa
02 bbb
03 ccc
04 ddd
07 eee
08 fff
10 ggg
12 hhh
13 iii

As expected, it remains the duplicate bbb line between lines aaa and ccc only ;-))

Best Regards,

guy038

Sofistanpp

Hi guy038, All:

Well done. I’m glad my post somehow inspired you to develop a more comprehensive solution to the current issue. As I learned reading archived posts, ancillary lists are a frequently used resource of your toolbox.

On my side, reversing lines wasn’t my first thought. What would happen, I asked myself, if I run that regex in backward direction from the last line? Would I get, by symmetry, the first “bbb”? Enabled the Backward direction button via an AutoHotkey script and clicked on Replace All, but no joy. You will get exactly the same outcome as if you run the regex in normal direction.

I suspect that lookarounds are the culprits (simpler regexes do the expected job), but haven’t thoroughly tested it.

Maybe you or someone else can elaborate on this issue.

Best Regards.

Cletos

Hello guy038,

Thank you you very much for the new method!

Alan Kilborn

@Sofistanpp

run that regex in backward direction from the last line

Searching backwards with regex is “discouraged” and is partially disabled in Notepad++.
The reason, I think, is that thru a given text, if you search backwards versus forwards, you won’t get the same hits. Sometimes (simpler regexes, as you noted) you will, but not always (depends upon the regex and maybe the data).

Enabled the Backward direction button via an AutoHotkey script

In general, enabling disabled controls and then performing an operation and expecting good results is a dubious premise.

endolith

@Cletos Yes this feature is buggy, I see it fairly often. Usually I can click “Remove duplicate lines” and it removes them all, regardless of order, but sometimes it doesn’t remove any of them. Something wrong with the software, but I can’t pinpoint what’s wrong. It depends on the text? Or I have to create a new blank document and then it works there, and then copy it back into the original?

Alan Kilborn

@endolith said in Remove duplicate lines not possible?:

It depends on the text?

Could be a line-ending problem?
If line-endings are different on otherwise duplicate lines, they won’t be considered true duplicates.