Find and Replace New-Line

[ScrKiddle]

I use the find an replace bar at the bottom of the window.
I search for dots which are followed by space ". "
Now i want to replace them by a dot followed by a new line.
I did try to replace with a dot followed by a backslash and n ".\n".
But then it inserts a literal \n instead of a line break.

In the help file i have found no hint on how to find things and replace them by new lines.
There is no variable like $newline or $tab or such a thing.

The editor i have used before hat some kind of popup select menu to insert some special tags into the search and the replace box, that was very handy.
For search it has offered Tags like Tab \t, New line \n, Any Character ., In Range [], Not in Range ^[], Befing of line ^, End of Line $, Tages expression (), Or |, 0 or more matches *, 1 match or more +, 0 or 1 match ?.
For replace it did offer Tab Character \t, New Line \n, Find What text $&, tagged expression $1 ... $9.
It did offer to find replace literal but when checking a checkbox for RegEx to use such expressions.

It also had a mode that did open the same searchbox but instead of single line expressions it did switch to multi line expressions for find and replace

that was very handy

how can i do this in RJ TextEd?

Example:

Line 1, Line 2, Line 3, Line 4, Line 5.

How can i use Find Replace to turn this into the following:

Line 1
Line 2
Line 3
Line 4
Line 5

[pjj]

You can do the same in RJ TE, just click on the computer icon adjacent to "replace" field, pick "Edit", press Enter in the edit field and click on OK; a pipe symbol will be inserted into "replace" field, denoting newline.

[ScrKiddle]

[---] just click on the computer icon adjacent to "replace" field, pick "Edit", press Enter in the edit field and click on OK; a pipe symbol will be inserted into "replace" field, denoting newline.

Thanks i have found that and it looks great, the computer-symbol opens this multiline panel and in the single line text-box the new-line character turns into a pipe-symbol.

Now i got into some other strange problem:

Given_TXT_File.png (52.7 KiB) Viewed 173 times

Given Text:

Code: Select all

Sentence with an ending dot. And one more.
Then a new line. And so on. 123
Before the end.

I try to find all sentences with Reg-Ex:
"\." escape the dots
" " add one space
"()" find an expression to use it later in the replacement
"[a-zA-Z]" find latin letters a-Z after the dot and use them later in the replacement.

\. ([a-zA-Z])
https://regex101.com/ tells me this is valid and uses a capturing group.

I try to replace:
"|" new-line
"$0" the capturing-group

Now i think that the text will turn into the following

Code: Select all

Sentence with an ending dot.
And one more.
Then a new line.
And so on. 123
Before the end.

But what i get is the following:

Code: Select all

Sentence with an ending dot
. And one more.
Then a new line
. And so on. 123
Before the end.

So what ever, some thing got wrong, maybe this is some Reg-Ex-Problem or what ever, let us turn a blind eye on that and click the undo-button in the menu bar:

What i get ist this:

Code: Select all

Sentence with an ending dot. Ad one more.
Then a new line
. A so on. 123
Before the end.

This is not what i ever typed before, it is not what i had before the find-replace-all action.
This looks more like salad.
I Press the undo button even more, and the salad turns even more into an total soup of some thing.


I use 16.66 beta 4 on Windows, if this matters.

How can i just do this break-line after dot -thing, and hopefully undo it correctly if i dont like it?

[pjj]

I try to find all sentences with Reg-Ex:
"\." escape the dots
" " add one space
"()" find an expression to use it later in the replacement
"[a-zA-Z]" find latin letters a-Z after the dot and use them later in the replacement.

\. ([a-zA-Z])

No, you're clearly searching for a dot, a space and then some Latin letters. To find "sentences", you need to use something like this:

Code: Select all

([a-zA-Z ]+)\. ?\n?

and replace with

Code: Select all

$1.\n

or maybe \r\n instead of \n in RJ TE. Or a pipe, if it works with regex. -- As for the undo, well, that's a different story and it always was problematic. As well as in other editors, I guess.

[ScrKiddle]

[...]
No, you're clearly searching for a dot, a space and then some Latin letters. To find "sentences", you need to use something like this:

Code: Select all

([a-zA-Z ]+)\. ?\n?

and replace with

Code: Select all

$1.\n

or maybe \r\n instead of \n in RJ TE. Or a pipe, if it works with regex. -- As for the undo, well, that's a different story and it always was problematic. As well as in other editors, I guess.

In an other editor i use this without quotes "\. ([a-zA-Z])".
This finds a dot after each sentence followed by a space what is the case if there is no new line after the sentence, and followed by the first letter of the next sentence, but not if followed by a number.

then i replace it with the following without quotes ".\n$1".

The difference is, that it will find the point where a sentence ends and an other sentence begins with some letter on the same line, and then it will replace this break-point by the ending dot of the previous sentence and remove the space and replace it with a new-line and the first letter of the following sentence.

The effect may looks the same. every sentence that is not followed by a number will get separated on a new line, but it will not handle the whole sentence, which is may long, but just the dot-space-firstLetter, what needs less resources than to handle the whole sentence.

In the same (other) editor the undo function has absolutely no problem to revert/undo Reg-Ex Find-Replace actions over documents with serval thousands of lines, honestly i never saw such an odd undo-problem in any other editor. The editor offers a limited set of PCRE RegEx "commands".

[Rickard Johansson]

A while back I made changes to how "Replace All" handles linebreaks. When creating a source buffer all line breaks are now converted to LF, instead of using CRLF. So all positions had to be updated since CRLF are 2 characters and LF is only one. This include modifications to the undo functions as well.

But it seems you found a bug in the undo function when using regex that modify lines.

I'll fix it.

[Rickard Johansson]

It's fixed in the next release.
Turns out it was the regex library, used in Delphi. When computing and retrieving the replacement string - all line breaks are CRLF, even if I use a source string with only LF:s.

Anyway, undo should work now...