Automatic selection of ALL instances of a searched string, in one go

wonkawilly

@Ekopalypse
Yes that would be beneficial indeed IMHO too.

IMHO no need of the “only” word, just Select all will work fine

Of course everything could have drawbacks when used improperly: even regular expressions if improperly used can open not just one but multiple cans or worms.

Also into the dialog the button Count can help to prevent problems. And I suppose that the tools present into the tab Mark will help too for the same reason to check before act.

Alan Kilborn

@Ekopalypse said in Automatic selection of ALL instances of a searched string, in one go:

If I understand you correctly, you want something like this mock?

For the record, I understand exactly what @wonkawilly wants.
But I wanted him to explain why he thinks having such a capability is so advantageous. I wanted him to do this on his own, so I wasn’t leading him, but he couldn’t figure out what I was asking for, so…

What are you going to do when you have all these matches as selected text, that you can’t do already in the N++ search/replace/mark interface, without having text selected as a result of a search?

Your choices are:

start typing (this will replace the selected text with what you type; hmmm, there’s already a Replace function that will do this)
copy (there’s already a Copy Marked Text function – ok, you have to mark text first)
delete (just a Replace with nothing, use that function)
cut (copy plus delete, a bit laborious with existing functionality, but is this a really common need?)
scroll around post-search to “eyeball” all the matches (just use Mark All; also, if text was selected and you do this, be careful that you don’t do a keyboard/mouse action that causes all of your selected text to become immediately unselected)

In short, no big advantage that I see to having text selected as a result of a search. Sure, it is fine to want it, but perhaps there is a reason it doesn’t already exist.

Ekopalypse

@Alan-Kilborn

It serves a similar but different purpose, in the end you have several cursors to work with. At the moment, the Find and Replace dialog cannot do this and the options mentioned are, for this case, also no options, because they leave only one cursor after the actions.

Alan Kilborn

@Ekopalypse said in Automatic selection of ALL instances of a searched string, in one go:

in the end you have several cursors to work with

And the big advantage to that is…?

guy038

Hi, @alan-kilborn and All,

Alan, you asked @wonkawilly to describe the benefits of this potential new feature but, to be honest, I said exactly the same thing in my first post when I added, at the end :

So, I wonder if we should propose an enhancement of the Mark dialog, with a new box option Selection of the Occurrences or equivalent, using this mechanism ?

So, you would also be entitled to level the same criticism at me and ask me to justify the need for multiple cursors !

Thus, I read your last post carefully and, of course, your arguments seem relevant. However, I have an intuitive feeling that this new functionality would be very powerful, especially when you decide to take consecutive actions on this selection of cursors !

For example, considering the text :

This is a simple test
  This is a simple test
    This is a simple test
      This is a simple test
        This is a simple test
          This is a simple test
            This is a simple test
              This is a simple test
                This is a simple test
                  This is a simple test

Let’s say we’ve selected the ten words simple, using the latest version of the Mutliple-Selections.py script.
First, we decide to delete the word simple : We could use the Replace dialog with the search of simple, Nothing as the replacement and with the In selection box ticked, to obtain the following text :

This is a  test
  This is a  test
    This is a  test
      This is a  test
        This is a  test
          This is a  test
            This is a  test
              This is a  test
                This is a  test
                  This is a  test

However, suppose we change our mind and we want to add 10 line-breaks, AFTER deleting those 10 simple words. Now, using the Replace dialogue is not so easy, because we have to search for the string a test and change it to the string a \r\n test, just for those ten lines ! If we had used the multiple cursors’ technique, we would only have had to press the Enter key :-)

And then, after all: if Neil Hodgson, and his team of devs, have added the SCI_MULTIPLESELECTADDNEXT and SCI_MULTIPLESELECTADDEACH commands, in the Scintilla project, there must be some point in using them !

Best Regards,

guy038

BTW, once I selected the ten lines of my test and used the Multiple_Selections.py script to select the ten words simple and deleted them with the Del key, I tried a Ctrl + Z action in the hope of get the words simple back. But no success ! I suppose that my last version should include a sequence editor.beginUndoAction() and editor.endUndoAction() …

Alan Kilborn

@guy038

No worries; it was merely my attempt to point out that it probably wouldn’t be as useful as people think on first consideration.

Now consider the Select all only checkbox in the @Ekopalypse UI mockup. How is one to know which actions this applies to? Does one need it for Find Next? (no). Count? (maybe?). Find All in Current Document? (certainly). Find All in All Opened…? (probably?).

The historical way of options that apply to only certain actions is to put them in a box that groups those actions; here we see In selection as the example of that. How would something like that best be done here, or is the Select All Only checkbox left floating out on its own?

Would this apply to Find in Files as well, for documents that are already open?

Is there any effect for Mark? Maybe the new checkbox only appears on the marking tab, as this has some logical commonality.

It’s good to want things, but it is also good to consider why you want them, and then all the downstream concerns.

Alan Kilborn

@guy038 said in Automatic selection of ALL instances of a searched string, in one go:

BTW, once I selected the ten lines of my test and used the Multiple_Selections.py script to select the ten words simple and deleted them with the Del key, I tried a Ctrl + Z action in the hope of get the words simple back. But no success ! I suppose that my last version should include a sequence editor.beginUndoAction() and editor.endUndoAction()

I didn’t try your script, but setting up a similar test, where Delete is done with multiple selections active and then doing Undo DID achieve the undo for me. I see no reason why it should be different with a script pre-selecting data that you manually press the Delete key on post-script.

In your script, editor.beginUndoAction() and editor.endUndoAction() aren’t going to do anything, because your script doesn’t change the text…so there’s nothing to undo.

guy038

Hello, @alan-kilborn,

Of course, you’re right about it ! Don’t know which manipulation I did before ! I confirm that after deleting the 10 words simple of my test, with the script, adding a line-break then some dummy text, the Ctrl + Z actions did delete the added text as well as the line-breaks and get the words simple back ! The status of the file was also back to unchanged !

BR

guy038

mkupper

I have always thought of selections as being temporary and fragile things and don’t think I would be comfortable with constructing selections using a search expression. I like to be able to look things over before I perform some action… One click of a mouse or keyboard can wipe out a carefully constructed selection.

An alternative is to use marking and to add a Select Marked Text button. This button likely could replace the Copy Marked Text button on the Mark dialog box.

Once the marked text is selected then people could then copy it, replace it, delete it, convert to upper/lower case, etc.

At present, unmarking is moderately awkward but is possible. Ideally, the Mark dialog box was a little more consistent. The Mark All button marks using a combination of the search pattern plus the In selection checkbox. The Clear all marks button though ignores the search pattern field but does honor In selection.

That leads me to thinking about adding a Mark selected text button so that I can drag-select some text, mark it, drag-select something else and mark it without needing to use the search pattern field.

A potential downside to the idea is that because selecting and marking do not make changes to the text that the activity related to setting up the marked area(s) or selection(s) never show up in the undo/redo history. Someone could loose chunk of time.

Lycan Thrope

@Alan-Kilborn ,

I’m glad you did this, because I was thinking the same thing. I don’t know why his Undo didn’t work, perhaps because it was a script that did other things in addition, but whenever I do a change or search/replace, I feel it’s on me, to make sure I really want that change, as @mkupper suggests.

For that, the usefulness of the Mark comes into play and allows one to see what they’ll be changing ahead of time using the supplied regex. Using these things in bulk text changes is always a risk so again, these changes need to be tempered with good reason, as @Alan-Kilborn suggests, rather than willly-nilly giving users more complex operations than they truly need which just adds to more confusion, support problems and trying to explain to people how to properly use it which will benefit, what? Maybe 1% of the total users?

That’s what the scripting option is supposed to be for, so that the 1% (which most likely will also be superusers with the skills, anyways) can craft custom capabilities that the other 99% don’t need to be able to use or heaven forbid, stumble into and send them straight here looking for people to unscrew their lack of understanding of what “Don’t push the red button” means. :-)

Alan Kilborn

@mkupper said:

An alternative is to use marking and to add a Select Marked Text button. This button likely could replace the Copy Marked Text button on the Mark dialog box.

Or, it could be an additional action button, not replacing an existing one that users could be used to using.

At present, unmarking is moderately awkward but is possible. Ideally, the Mark dialog box was a little more consistent. The Mark All button marks using a combination of the search pattern…

So what I think you’re saying here is that it would be nice to be able to unmark text based on the Find what expression as well. That makes a good deal of sense. Maybe an additional action button could be added for this case.

mkupper

@Alan-Kilborn said in Automatic selection of ALL instances of a searched string, in one go:

Or, it could be an additional action button, not replacing an existing one that users could be used to using.

I had suggested replacing the button to declutter the user interface.

The action would still be available but would take an extra step which is select-all-marked and then copy.

Ekopalypse

@Alan-Kilborn said in Automatic selection of ALL instances of a searched string, in one go:

And the big advantage to that is…?

To be able to work with multiple cursors.

On the rest of the arguments against such a feature.
I don’t think it takes a genius to figure out what “select whatever” would mean in such a case,
and if it’s documented, those unfamiliar with or accustomed to the concept of “multiple cursors/carets/editors/whatever” will understand its purpose.
Does this mean that all potential users are using it correctly,
no, but are all users also using regexes correctly? Backup? …
Someone who wants to test beforehand whether the results of his search will give only the desired result can run mark beforehand …
Whether such an extension is implemented in the find tab as a checkbox,
or gets its own tab or dialog, is in my opinion irrelevant for this discussion,
because it is ONLY about whether Don finds it good or not.
No matter what comes out of this discussion in the end,
if Don has a different opinion on it, and that should be respected because it’s his project,
that and only that will be the deciding factor.
So, I’ve given enough mustard on this subject now, wouldn’t know what more input I could provide.

Alan Kilborn

@Ekopalypse

And the big advantage to that is…?

To be able to work with multiple cursors.

Sure, one can just keep saying that over and over again, but I for one haven’t been convinced about any great usefulness gain coming from it.

Of course, my opinion doesn’t matter, but if I feel that way, probably others do, too. No idea how the author of Notepad++ feels about it, but unless someone creates a PR for it, that doesn’t matter, either.

Alan Kilborn

Here’s my entry for a script for this topic.

NOTE: This is probably NOT of any use for dedicated keyboardists as there is really no way to make use of it without touching the mouse. (Of course, the same can be said of most (but not all) “mulit-edit” operations.

What it does is a “scoped multiselect”. For myself, I find this very useful to quickly rename a poorly-named script variable in a localized section of code. Often when I’m coding I need a variable to do something, so maybe I’ll just call it zz until I get its code correct, and only then will I give it a good permanent name, one that suits its (perhaps evolved) function.

Anyway, to use the script, you define two empty carets (to say “only act in the range between them”) and then a non-empty selection between the empties (to define the source text to be replaced), and then you run the script. The empty carets defining scope give me confidence that I won’t be changing text outside of my defined area.

An example helps; take the original example in this thread, and set up the following two empty carets and selection, which is very quick and easy to do with the mouse (just click on line 6, ctrl+click on line 17, and ctrl+doubleclick on simple in line 13):

Run the script to obtain:

I bind the running of this script to a keycombo, and find this is quicker than Notepad++'s replace function when I need to rename something: No prompts and no typing to get the text I need selected; all I have to do after running the script is type the desired replacement text.

I named this script the somewhat lengthy MultiselectSelectedTextBetweenlEmptyCaretStoppers.py and here is its listing:

# -*- coding: utf-8 -*-
from __future__ import print_function

#########################################
#
#  MultiselectSelectedTextBetweenlEmptyCaretStoppers (MSTBECS)
#
#########################################

# references:
#  https://community.notepad-plus-plus.org/topic/24549  <-- this script posted here
#  inspiration:
#   https://github.com/notepad-plus-plus/notepad-plus-plus/issues/8203#issuecomment-623136793
#  for newbie info on PythonScripts, see https://community.notepad-plus-plus.org/topic/23039/faq-desk-how-to-install-and-run-a-script-in-pythonscript

#-------------------------------------------------------------------------------

from Npp import *
import os
import inspect

#-------------------------------------------------------------------------------

class MSTBECS(object):

    def __init__(self):

        self.this_script_name = inspect.getframeinfo(inspect.currentframe()).filename.split(os.sep)[-1].rsplit('.', 1)[0]

        # user must make the 3 required selections in any order:
        sel_tup_list = []
        sel_tup_list.append( (editor.getSelectionNStart(0), editor.getSelectionNEnd(0)) )
        if editor.getSelections() >= 2: sel_tup_list.append( (editor.getSelectionNStart(1), editor.getSelectionNEnd(1)) )
        if editor.getSelections() >= 3: sel_tup_list.append( (editor.getSelectionNStart(2), editor.getSelectionNEnd(2)) )
        sel_tup_list.sort()

        # conditions-not-correct checking:
        cnc = False
        if not cnc and not editor.getMultipleSelection(): cnc = True  # must have "Multi-editing" enabled in Preferences
        rect_sel_mode = editor.getSelectionMode() in [ SELECTIONMODE.RECTANGLE, SELECTIONMODE.THIN ]
        if not cnc and rect_sel_mode: cnc = True  # can't be in column-block selection mode
        if not cnc and len(sel_tup_list) != 3: cnc = True  # need 3 selections (actually 2 carets and one selection)
        if not cnc and sel_tup_list[0][0] != sel_tup_list[0][1]: cnc = True  # no leading empty caret
        if not cnc and sel_tup_list[1][0] == sel_tup_list[1][1]: cnc = True  # no text selected between carets
        if not cnc and sel_tup_list[2][0] != sel_tup_list[2][1]: cnc = True  # no trailing empty caret
        if cnc:
            cnc_help = '''
            Conditions are not correct for running the script.
            You must have:
            - "Multi-Editing" enabled in Preferences
            - Two empty carets with a stream selection in between
            '''
            self.mb(cnc_help)
            return

        (start_search_pos, end_search_pos) = (sel_tup_list[0][0], sel_tup_list[2][0])
        search_word = editor.getTextRange(*sel_tup_list[1])
        match_span_tup_list = []
        editor.search(search_word, lambda m: match_span_tup_list.append(m.span(0)), 0, start_search_pos, end_search_pos)

        if len(match_span_tup_list) > 1:
            editor.setSelection(*match_span_tup_list[0][::-1])  # args to setSelection are caret then anchor
            for (start_match_pos, end_match_pos) in match_span_tup_list[1:]:
                editor.addSelection(end_match_pos, start_match_pos)
            self.sb_output('{} selections made'.format(len(match_span_tup_list)))

    def mb(self, msg, flags=0, title=''):  # a message-box function
        return notepad.messageBox(msg, title if title else self.this_script_name, flags)
        
    def sb_output(self, *args):  # output to N++'s status bar (will be overwritten by N++ e.g. when active tab is changed)
        notepad.setStatusBar(STATUSBARSECTION.DOCTYPE, ' '.join(map(str, args)))

#-------------------------------------------------------------------------------

if __name__ == '__main__': MSTBECS()

Mark Olson

@Alan-Kilborn
Getting this error:

Traceback (most recent call last):
  File "%AppData%\Roaming\Notepad++\plugins\Config\PythonScript\scripts\multiSelectBetweenEmptyCaretSelections.py", line 65, in <module>
    if __name__ == '__main__': MSTBECS()
  File "%AppData%\Roaming\Notepad++\plugins\Config\PythonScript\scripts\multiSelectBetweenEmptyCaretSelections.py", line 61, in __init__
    self.sb_output('{} selections made'.format(len(match_span_tup_list)))
AttributeError: 'MSTBECS' object has no attribute 'sb_output'

Eyeballing the code, it looks like calling self.mb will probably cause the same problem. I guess you forgot to define those methods.

Alan Kilborn

@Mark-Olson

I didn’t exactly forget, I just streamlined out some of my usual debug code, before posting.
It appears I got too heavy with my edits. :-(

I still have time to edit the script; I’ll fix it. :-)
Thanks for pointing it out.

EDIT: I updated the above script listing to fix the problems @Mark-Olson pointed out.

guy038

Hello, @alan-kilborn and All,

Ah… yes, your proposed script is quite clever and very intuitive ;-)) I didn’t think about invisible caret boundaries for the defined scope

Remark that, at first glance, I thought that the carets did represent the true | character ( with code \x7C ). But, thanks to the error message, I quickly understood the problem !

I personally think that the logical method is :

Do the stream selection of the text first, without holding down the Ctrl key
Then, do the two caret zero-length selections, with the Ctrl key pressed

Notes :

If you prefer an insensitive search, simply add the import re command, below the import inspect one and change the line :

editor.search(search_word, lambda m: match_span_tup_list.append(m.span(0)), 0, start_search_pos, end_search_pos)  #  SENSITIVE search

As :

editor.search(search_word, lambda m: match_span_tup_list.append(m.span(0)), re.I, start_search_pos, end_search_pos)  #  INSENSITIVE search

Now, I also found out a trick to simulate a whole-word search :
- Do a stream selection of the word, including its surrounding space chars
- Do the two caret selections
- Run the script
- Click on the Right arrow key
- Click on the Left arrow key
- Click on the Ctrl + Left arrow shortcut

Playing around with your script, after a while, I asked myself :

Could it be possible to just do the text selection and that the scope of the search would be, automatically, the complete file contents ?

After some tries, I ended up with this modified part of the script which, indeed, does the job !

        sel_tup_list = []
        sel_tup_list.append( (editor.getSelectionNStart(0), editor.getSelectionNEnd(0)) )
        if editor.getSelections() >= 2: sel_tup_list.append( (editor.getSelectionNStart(1), editor.getSelectionNEnd(1)) )
        if editor.getSelections() >= 3: sel_tup_list.append( (editor.getSelectionNStart(2), editor.getSelectionNEnd(2)) )
        if editor.getSelections() == 1:                                          # If an unique stream selection:
            sel_tup_list.append ( (0, 0) )                                       #     Leading caret is supposed to be at the very begining of current filec
            sel_tup_list.append ( (editor.getLength(), editor.getLength()) )     #     Trailing caret is supposed to be at the very end of current file
        sel_tup_list.sort()

instead of :

        sel_tup_list = []
        sel_tup_list.append( (editor.getSelectionNStart(0), editor.getSelectionNEnd(0)) )
        if editor.getSelections() >= 2: sel_tup_list.append( (editor.getSelectionNStart(1), editor.getSelectionNEnd(1)) )
        if editor.getSelections() >= 3: sel_tup_list.append( (editor.getSelectionNStart(2), editor.getSelectionNEnd(2)) )
        sel_tup_list.sort()

Now, although this modification seems quite functional, I would like to know, Alan, if this is the best formulation !

Best Regards,

guy038

A last observation :

If you do the three needed selections, in any order and that you decide to add additional selections ( either non-null text or null caret selections ), no error message is displayed : these additional selections are simply ignored. To my mind, it’s the best way to proceed !

Alan Kilborn

@guy038 said in Automatic selection of ALL instances of a searched string, in one go:

I personally think that the logical method is :

Do the stream selection of the text first, without holding down the Ctrl key

Then, do the two caret zero-length selections, with the Ctrl key pressed

The “beauty” of the design is that you can do these 3 things in ANY order that you like.

@guy038 said in Automatic selection of ALL instances of a searched string, in one go:

If you prefer an insensitive search

Yes, but for the utmost in safety, I prefer a sensitive, i.e, match case, design. I don’t want to have to consider it, I just want to know I am getting exactly and only what I’ve asked for.

@guy038 said in Automatic selection of ALL instances of a searched string, in one go:

…and that the scope of the search would be, automatically, the complete file contents ?

Again, the empty carets for scoping are there for safety in this “quick” replacement method. If you’re looking for something with wider scope, just use Notepad++ 's own Replace All functionality.

@guy038 said in Automatic selection of ALL instances of a searched string, in one go:

if this is the best formulation

It seems reasonable to me! :-)
There’s rarely one and only one best way with a block of code…

guy038

@alan-kilborn and All,

In you previous post, you said :

Again, the empty carets for scoping are there for safety in this “quick” replacement method. If you’re looking for something with wider scope, just use Notepad++ 's own Replace All functionality.

At the time, I felt rather silly for not having thought of it ! But, on reflection, I think we should consider these scripts as, either :

An interactive replacement in selection feature ( for our two scripts )
An interactive global replacement feature ( for my derived script )

For example, let’s imagine this scenario :
- I first decide to change any simple word with the word difficult
- Then, I changed my mind and prefer to change the word difficult with the expression very difficult
- Again , I changed my mind and finally decide to replace from the word very to the end of current line with the expression good old method

With the Replace dialog, we would do 3 successive regex S/R :
- SEARCH (?-i)simple
- REPLACE difficult
- SEARCH (?-i)difficult
- REPLACE very $0
- SEARCH (?-si)very difficult.+
- REPLACE good old method

with our scripts, we would :
- Select the word simple and, possibly, the scope of search
- Run the script
- Type in the word difficult
- Use the Ctrl + Left Arrow shortcut
- Type in the word very, followed with a space char
- Use, again, the Ctrl + Left Arrow shortcut
- Use the Shift + End shortcut
- Finally type in the good old method expression

As you can see, for most of the replacements, the script way seems more intuitive than using successive replacements !

Best Regards,

guy038