Remove duplicate numerical lines

Scott Sumner

Hey Terry!

Is the last line which doesn’t have a line-ending REALLY a duplicate of an earlier line that does have a line-ending? :) Well, okay, it IS if we are talking about line-endingless content, which we (probably) are.

Anyway, the culprit line in the code would be the one with editor.getLineCount() in it. You will have one less line without a line-ending on your last line, and thus the range function will cause it to go one less iteration. But also to blame is that when the script remembers a previously encountered line, it does so WITH THE LINE-ENDING ON. So there’s a double reason for failure here.

I don’t like files without line-endings on their last lines. I sure do wish there was an option in N++ to automatically make sure lines all have proper ends on them. [Of course I have a Pythonscript that makes sure of this for me, so I don’t usually remember to take this stuff into account.]

BTW, note that the script ignores blank lines; something I should have mentioned earlier.

PeterJones

So @Scott-Sumner, are you going to leave us hanging? You need to publish the code to add the line-ending to the last line, if it’s missing it, so that your above code works properly. :-)

Scott Sumner

@PeterJones said:

You need to publish the code to add the line-ending to the last line…so that your above code works properly

HAHa. I will, but right now it looks overcomplicated for general use. :-) I’ll work on it and post back here when it is suitable for general consumption…

In the meanwhile, why not let’s just fix the original code? I found that all that is needed is to change this line:

l = editor.getLine(j)

into this:

l = editor.getLine(j).rstrip('\n\r')

Eko palypse

@Scott-Sumner

what about using OrderedDict from collections?
Preserves the ordering and dict keys are unique per se.

from Npp import editor
from collections import OrderedDict
_dict = OrderedDict.fromkeys(editor.getText().splitlines())
editor.setText('\r\n'.join(_dict.keys()))

Eko

Scott Sumner

@Eko-palypse said:

what about…?

Sure, why not? Only objection might be the empty line case (my experience is that people usually want their blank lines retained as is, and not removed as duplicates).

Eko palypse

@Scott-Sumner

right, this case makes it a little bit more difficulty, agreed.

Eko

Eko palypse

@Scott-Sumner

What about this

from Npp import editor
lastLineContainsEOL = True if len(editor.getLine(editor.getLineCount()-1)) == 0 else False
lines = editor.getText().splitlines()
uniqueLines = set(lines)
newText = '' 
for line in lines:
    if line in uniqueLines or line.strip() == '':
        newText += line + '\r\n'
        if line.strip() != '':
            uniqueLines.remove(line)
editor.setText(newText if lastLineContainsEOL else newText[:-2])

generates unique lines only (ignoring empty lines with and without spaces)
preserves ordering
preserves usage of last EOL

Eko

Scott Sumner

@Eko-palypse said:

What about this

Sure. I say “whatever works”. Much like I don’t get all fancy about shaving a few characters off a regex, I think with scripts it is to each his own. As long as it does the job, it is super. :-)

Scott Sumner

@Eko-palypse

One comment, though: I’m guessing you pretty much exclusively use Windows. I use Windows/Linux about 75%/25%…because of that I have learned to not think that line-endings are always \r\n. So scripts I post here will work (that’s the goal anyway) with either Windows or Linux (or even Mac) files.

This may be something you want to consider doing as well. But it doesn’t bother me if you don’t because I understand the meaning of it–for someone that just wants to blindly pick up and use a script and doesn’t understand Python, oh and BTW uses Linux files…it could be a problem.

BTW, good job! I like seeing Pythonscripts besides my own posted here. Not many people are doing it anymore. :-(

Scott Sumner

@PeterJones said:

You need to publish the code to add the line-ending to the last line, if it’s missing it

Ok, so here it is; I run a similar (but more complicated one for my own needs) from my startup.py so that it is always in place–and thus I never have to deal with files without line-endings on their last lines.

One thing I don’t like, but haven’t found a good method for handling, is that in certain circumstances (e.g. a Save All), after the script does its work, it can leave you sitting in an tab that is different from the tab that was active before. If people are interested in this script and have ideas about solving that particular problem, I’m interested in hearing them.

Here’s the Pythonscript:

from Npp import notepad, editor, NOTIFICATION

def callback_npp_FILEBEFORESAVE(args):
    line_ending = ['\r\n', '\r', '\n'][notepad.getFormatType()]
    doc_size = editor.getTextLength()
    if editor.getTextRange(doc_size - 1, doc_size) != line_ending[-1]:
        # fix Notepad++'s "broken" functionality and add a line-ending at end-of-file
        editor.appendText(line_ending)

notepad.callback(callback_npp_FILEBEFORESAVE, [NOTIFICATION.FILEBEFORESAVE])

Eko palypse

@Scott-Sumner said:

One comment, though: I’m guessing you pretty much exclusively use Windows. I use Windows/Linux about 75%/25%…because of that I have learned to not think that line-endings are always \r\n. So scripts I post here will work (that’s the goal anyway) with either Windows or Linux (or even Mac) files.

Good point and you offered the solution already, even better :-D

from Npp import editor
lastLineContainsEOL = True if len(editor.getLine(editor.getLineCount()-1)) == 0 else False
line_ending = ['\r\n', '\r', '\n'][notepad.getFormatType()]
lines = editor.getText().splitlines()
uniqueLines = set(lines)
newText = '' 
for line in lines:
    if line in uniqueLines or line.strip() == '':
        newText += line + line_ending 
        if line.strip() != '':
            uniqueLines.remove(line)
editor.setText(newText if lastLineContainsEOL else newText[:-2])

Eko

Scott Sumner

@Eko-palypse :

Yes, but you forgot something. :-)

editor.setText(newText if lastLineContainsEOL else newText[:-len(line_ending)])

PeterJones

To continue with the hijack-tangent of this thread… :-)

@Scott-Sumner said,

If people are interested in this script and have ideas about solving that particular problem, I’m interested in hearing them.

Challenge accepted. :-)

My first idea was that you could track the previous bufferID, and make sure you always activate the previous one. While trying to see if that would help, I noticed that with the exact script you had posted, if all open files were missing EOL, it would save all files, but only fix the EOL on the active file.

That gave me the flash for the solution: in the callback, store the currently-active bufferID, activate the buffer for the argument to the callback (ie, the file being saved), make the changes to the now-active file, then re-activate the originally-active buffer. The script below seemed to do it for me:

from Npp import notepad, editor, NOTIFICATION

def callback_npp_FILEBEFORESAVE(args):
    # the editor.appendText will go to the _active_ buffer, whatever
    # file is currently being saved.  So to solve two birds with one
    # stone, save the active buffer ID, then switch to the buffer ID
    # for this instance of the callback -- now the editor has the
    # correct buffer active.
    oldActiveID = notepad.getCurrentBufferID()
    notepad.activateBufferID(args["bufferID"])

    line_ending = ['\r\n', '\r', '\n'][notepad.getFormatType()]
    doc_size = editor.getTextLength()
    if editor.getTextRange(doc_size - 1, doc_size) != line_ending[-1]:
        # fix Notepad++'s "broken" functionality and add a line-ending at end-of-file
        editor.appendText(line_ending)

    # now that you're done editing, go back to the originally-active buffer
    notepad.activateBufferID(oldActiveID)

notepad.callback(callback_npp_FILEBEFORESAVE, [NOTIFICATION.FILEBEFORESAVE])

I tested this with three open files: two in one view, one in other view; I tried various combinations of which ones needed to be saved, and which ones were missing EOL, and which was active, and it seemed to always do what I intended, but it’s possible that other combinations won’t work.

Scott Sumner

@PeterJones said:

if all open files were missing EOL, it would save all files, but only fix the EOL on the active file

Really? I tested with several open files (at least one of the 3 types, Win/Linux/Mac) that needed fixing and when I did a Save All they all got saved after being modified…hmmm, guess I will have another look…

PeterJones

It was that way for me. But the oldID=activeID, activate(args), edit, activate(oldID) should work for your first problem, even if you didn’t have the second problem that I have.

Scott Sumner

@Eko-palypse

So we all learn from each other here. My favorite line from your script is this one:

lastLineContainsEOL = True if len(editor.getLine(editor.getLineCount()-1)) == 0 else False

In my “callback_npp_FILEBEFORESAVE” script I did it differently…but I like your method, too.

Thinking more about it now, you could also do it like this:

lastLineContainsEOL = True if editor.getText()[-1] in '\n\r' else False

but maybe that pulls a lot of text just to look at the last character…

Scott Sumner

@PeterJones said:

the oldID=activeID, activate(args), edit, activate(oldID) should work

Yea, I had something like that (but even more involved) in my original over-complicated version (mentioned yesterday, originally), but I found some cases where even that didn’t always work right…so I cut it out entirely before posting. Maybe I’ll revisit it…

Eko palypse

@Scott-Sumner

yes, that is what I like about open source project in general.
Everyone can learn from others from different point of views or styles etc …

Yes, but you forgot something. :-)

Correct, different size - damn it. :-)

editor.getText()[-1]
but maybe that pulls a lot of text just to look at the last character…

I thought so too but after rechecking scintilla documentation it looks like
SCI_GETCHARACTERPOINTER is what we are looking for because from the document it states
Grant temporary direct read-only access to the memory used by Scintilla to store the document.

So, editor.getCharacterPointer()[-1] shouldn’t allocate any heap memory at all or maybe just a little tiny bit.

@PeterJones nice one - works for me as well :-)

Eko

Scott Sumner

@Eko-palypse

I get nervous about editor.getCharacterPointer() because of my lack of full understanding about multibyte character encodings (as stated previously, I’m an A-Z person). I suppose, though, in this case we are talking about, there could be no issues…

Eko palypse

@Scott-Sumner

not sure I understand your concerns about this.
I assume it is only the pointer to the text buffer returned and pythonscript
plugin has the python buffer protocol implemented so it should be safe always.
But as written, I assume - don’t really know how it is implemented.

Eko