Unfortunately, I’ve encountered too many bugs with the hidden lines feature, that it isn’t worth continuing. I think that, at least for now, I’m giving up on this idea…

A quick fix for the “strangeness with hidden lines” is hiding all lines twice in a row.

Meh.

I don’t know that that script, being “simpler”, has anything over-and-above my “complicated”(?) one, given that it appears the strangeness with hidden lines continues in that script…

Your script has a fairly sophisticated UI, colors the margin of the displayed lines, and marks the strings found. My script has none of that. Therefore it has exactly 4 times less lines than yours. A quick fix for the “strangeness with hidden lines” is hiding all lines twice in a row. I’m very interested in hearing about cases where that fix breaks down. In the meantime I have tested the script a lot and so far the fix seems to work, but you never know …

For a simple script

I don’t know that that script, being “simpler”, has anything over-and-above my “complicated”(?) one, given that it appears the strangeness with hidden lines continues in that script…

Was there anything else that would prevent it from being used?

Maybe my use of “…at all” came off as a little strong… I can’t see that it would corrupt data or do anything “bad”.

I think the downside to wider use is only the complaints that could arise in the form of “hey, this isn’t working right”. If a tool is supposed to do “X”, and it doesn’t do “X” correctly, well…

I don’t know if we should recommend that script at all until its shortcomings have been addressed.

I will admit I haven’t used it much (just the experiments I described yesterday). But my impression was that the redraw bug (which I can get around by changing tabs and back) and slowness (which is to be expected on something like this) were the issues. Was there anything else that would prevent it from being used?

The first time I created the script, and ran it, it properly filtered (showed @datatraveller1’s screenshot). And every subsequent run in that load of Notepad++ worked properly.

I then restarted Notepad++, and the first run after restarting gave more lines (some of the comment lines). If I run it and do the clear (shift+OK), then run it to highlight again, it works correctly. If I restart NPP and re-run to get the extra lines, then run again with the same exact search string (not shifted), it filters correctly.

The sequence behaves the same when searching for self.settings_by_viewfile_dict or just self.

I set the debug flag to TRUE (set that line to a 1 not a 0), restarted Notepad++, and then ran with the longer search string again.
For the bad highlight:

LFVH1<DBG>: view_plus_file: 0C:\Users\PJones2\AppData\Roaming\Notepad++\plugins\config\PythonScript\scripts\nppCommunity\23xxx\23929-AlanK-LineFilterViaHiding1.py
LFVH1<DBG>: view_plus_file: 0C:\Users\PJones2\AppData\Roaming\Notepad++\plugins\config\PythonScript\scripts\nppCommunity\23xxx\23929-AlanK-LineFilterViaHiding1.py
LFVH1<DBG>: L: [(1648, 1678), (2255, 2285), (2749, 2779), (2939, 2969), (4190, 4220), (4449, 4479), (5182, 5212), (5265, 5295), (5675, 5705), (6407, 6437), (6936, 6966), (7047, 7077), (9877, 9907), (9922, 9952)]

for the good highlight immediately after:

LFVH1<DBG>: view_plus_file: 0C:\Users\PJones2\AppData\Roaming\Notepad++\plugins\config\PythonScript\scripts\nppCommunity\23xxx\23929-AlanK-LineFilterViaHiding1.py
LFVH1<DBG>: L: [(1648, 1678), (2255, 2285), (2749, 2779), (2939, 2969), (4190, 4220), (4449, 4479), (5182, 5212), (5265, 5295), (5675, 5705), (6407, 6437), (6936, 6966), (7047, 7077), (9877, 9907), (9922, 9952)]

There is no difference in the L list, which says that the problem doesn’t lie in the searching/list, but rather someplace else.

Additional datapoint: if I have the bad filter (with extra comments showing), switch to another tab, and switch back, it shows the correct filtering, with the debug info:

LFVH1<DBG>: BUFFERACTIVATED: {'code': 1010, 'bufferID': 3019234076800L}
LFVH1<DBG>: view_plus_file: 0C:\Users\PJones2\AppData\Roaming\Notepad++\plugins\config\PythonScript\scripts\nppCommunity\23xxx\23929-AlanK-LineFilterViaHiding1.py

If I take original line#114 – the self.freshen_user_view_of_active_tab() at the end of prompt_for_string_to_match()'s definition – and duplicate it, so it runs that freshen twice, on the first freshen, it shows wrong; on the second, it shows right.

I’m wondering if before the “hide” side of that command, if you need to clear all the lineVisible flags to known values before your script does its hiding… maybe the active lexer or other NPP feature has already tagged some lines in a way that confuses your logic.

That’s probably all the time I’ll have for helping debug this today, but I hope some of that info turns out to be useful.

I still get correct results no matter what I filter for :-)

I have noticed that your line numbers don’t match my line numbers

I said before “(Don’t be concerned if my line numbers don’t match yours)”.

The reason for any mismatches would be because I’ve inserted some extra debugging statements in to possibly help solve the problem. As I’m seeing the undesired results with the original code or any variant, what I’ve changed isn’t causing the problem.

I have noticed that your line numbers don’t match my line numbers, e.g. my (correct) first occurence of self.settings_by_viewfile_dict is line 42.
I use the script https://community.notepad-plus-plus.org/post/82787 - Are you perhaps using a different script with potentially problematic changes? Does the problem also occur with this original version?

I think the key idea behind the script is you can edit in the filtered view.

THIS! There is a vi mode for this IIRC, but I just figured with all the plugins and search window in Notepad++ never supporting this editing of the filtered view - just putting the results into a different view - we’d just never get it.

Thanks @Alan-Kilborn - I assigned a hotkey and will try getting this filtered editing into my workflow.

Cheers.

I’m not sure - what does PC speed have to do with line visibility?

I don’t think it has anything to do with it, but until the root cause is found, everything is “on the table” as a possible problem.

Doesn’t “Find All in Current Document” basically do the same thing as the Python script?

I think the key idea behind the script is you can edit in the filtered view.

1. I’m not sure - what does PC speed have to do with line visibility?

2. Doesn’t “Find All in Current Document” basically do the same thing as the Python script? Actually, the Notepad++ programmer just has to output the same thing in the same window to filter the text, doesn’t he?

I have not finished cleaning my old micro and moved all my data yet,

What is taking so long? Are you ever going to finish with this?? :-)

I still use a USB key containing the N++ portable version 8.4.6 for all my tests, which must certainly slow down the execution of your Python script ??

I don’t see how…everything should be “in memory” for the operations of the script.

I’m really surprised that I never encountered any visible false lines during my tests

Yes that is interesting…

You should indeed remove these false positive answers, first of all !

Indeed. However, at the moment, I’m out of ideas on what the problem might be. :-(

Ah…OK. I understand your working method. You should indeed remove these false positive answers, first of all !

By the way, I’m really surprised that I never encountered any visible false lines during my tests, even in the case of a heavy file ! However, my laptop is not a WAR LIGHTNING ! Bought in July 2021, this HP laptop comes with Window 10 Pro 64, a 512 Mb Solid State Drive and 16 Mb DDR4 of RAM.

Moreover, as I have not finished cleaning my old micro and moved all my data yet, I still use a USB key containing the N++ portable version 8.4.6 for all my tests, which must certainly slow down the execution of your Python script ??

BR

guy038

as I increased the size of the test file, the waiting time to get the line margin highlighted, in salmon color, increased in parallel

I did mention earlier: “Large files may produce performance problems. Scripts aren’t blazingly fast…”

I use a variant of this script on what I consider a large file (in the work that I do). It works acceptably, performance wise.

HOWEVER: The correctness (or rather incorrectness) problem observed by some users needs to be addressed; if it doesn’t show/hide the correct lines, who cares how fast/slow it is?

wouldn’t it be better to :
Trigger a usual Mark action, with the Bookmark line option set
Elaborate a Python script which simply would hide ALL non-marked lines, from the Mark results !

The downsides to that approach are:

• It requires multiple steps
• Bookmarks become dedicated to that purpose (I like to “filter lines” and then set meaningful bookmarks on only some of the matches, before revealing all lines again and continuing to work with what I’ve bookmarked)

The user interface of the script could evolve to allow things like regex, match-case, etc. But…let’s go for correctness first.

Well, I finished the traditional sequence of greetings, by phone, SMS and letters. So I just tested your Python script, Alan, and here are the results of my experiments !

• Personally, I never saw the case of visible lines not containing the literal string to search for. Good point !

• Like you, @alan-kilborn, I used the License.txt file, which I duplicated several times and enter the lower-case word free when running the script And, as I increased the size of the test file, the waiting time to get the line margin highlighted, in salmon color, increased in parallel :

                   Search of the literal lower-case word 'free'

Times "License.txt"             Size                  Lines       Hiding lines time
---------------------------------------------------------------------------------------
license.txt ×    1            35,500 bytes            231 lines          0,4  s
license.txt ×    5           177,500 bytes          1,155 lines          1,95 s
license.txt ×   10           355,000 bytes          2,310 lines          3,5  s
license.txt ×   50         1,775,000 bytes         11,550 lines         17    s
license.txt ×  100         3,550,000,bytes         23,100 lines         34    s
license.txt ×  500        17,750,000 bytes        115,500 lines        165    s
license.txt × 1000        35,500,000,bytes        231,000 lines        335    s

Now, if we choose to look for the literal string COPYING, which occurs once only in the license.txt file ( instead of 24 times for the word free ), the results are very fast, in comparison :

                   Search of the literal upper-case word 'COPYING'

Times "License.txt"             Size                  Lines       Hiding lines time
---------------------------------------------------------------------------------------
license.txt × 1000        35,500,000,bytes        231,000 lines         12,5  s

Of course, if I tried to re-run the script, before getting the highlighted line margin, I received the usual message :
Another script is currently running. Running two scripts at the same time could produce unpredicable results, and is therefore disabled

Note that I tested this script with a portable N++ v8.4.6 version, located on an USB drive

Now, @datatraveller1, in order to ONLY copy the visible lines, here is a work-around :

• Run the @alan-kilborn Python script first

• Open the Mark dialog ( Crl + M )

• Write the same string than in Alan’s Python script, in the Find what zone

• Tick the four options Bookmark line, Purge for each search, Match case and wrap around, only

• Select the Normal search mode

• Click on the Mark All button

• Run the Search > Bookmark > Copy Bookmarked Lines option or right-click on the bookmark margin and choose it

• Open a new tab ( Ctrl + N )

• Paste the copied text ( Ctrl + V )

Finally, @alan-kilborn, @paul-wormer and @datatraveller1, regarding the method, wouldn’t it be better to :

• Trigger a usual Mark action, with the Bookmark line option set

• Elaborate a Python script which simply would hide ALL non-marked lines, from the Mark results !

3 immediate advantages of this method :

• You are not limited, anymore to a literal string

• You could choose a sensitive or non-insensitive search

• You could choose the search mode : Normal, Extended or Regular expresion

Best Regards,

guy038

            # hide all lines first:
            self.show_all_lines(False)
            
            import time
            print('sleep ..')
            time.sleep(60)
            
            # show only lines with hits from the search based on user criterion:
            for (pos_start, pos_end) in L:
                editor.setIndicatorCurrent(SCE_UNIVERSAL_FOUND_STYLE)
                editor.indicatorFillRange(pos_start, pos_end - pos_start)
                line_start = editor.lineFromPosition(pos_start)
                line_end = editor.lineFromPosition(pos_end)
                editor.showLines(line_start, line_end)
            
            time.sleep(60)
            print('Woke up')    

The sleep(60) gave me time to look at what’s happening and I noticed that often not all lines are hidden at the first sleep. These lines stay visible after the second sleep. The lines that stay visible are more or less random, so I get the feeling that it is a matter of timing. When I increase the length of the file by adding comments different lines stay visible. Datatraveller1 may well have a faster PC than Alan and me.

1. Hide lines not containing the string
2. Undo it (show all lines again)
3. Hide lines not containing the string

Then: step 1 gives too many lines and step 3 is OK.

I think what I see is the correct result:

@Alan-Kilborn Do you know why I get the correct result if I filter for self.settings_by_viewfile_dict while Paul and you get the wrong result? :-)