RFC / Beta Testers Wanted: pyscFilteredViewer

PeterJones

Very strange - I download both latest npp and full PythonScript plugin zips.
Create a new minimal test script and boom - crash? 10 tries ten crashes.

I have no idea why our systems are behaving differently. It seems odd that my couple of extra plugins would make my system more stable than yours.

Maybe Windows 7 is less stable than Win10? (Hard to believe, I know).

Meta Chuh

@PeterJones

Thanks for the feedback. I do tend toward detailed wordiness. I’ll try to simplify, and maybe move the details to the end, or to supporting documentation.

wordiness comes only because it’s your own baby ;-)
if you answer any notepad++ question, you keep it as short and easy to read as possible, with a syntax always matched to the original poster’s capabilities.

btw: i hope you don’t mind my feedback, i just think psychicFilteredReader ;-) has potential and a simplified guide might attract more unexperienced users.

pyscFilteredReader or pyscfr is a fine name, it will be easy to find in google at it is unique.

Yes, I said that. And I think I said elsewhere (I cannot find the exact quote: google-search no longer returns results inside the forum, and the forum search isn’t always as helpful as I’d like), I said something to the effect of “time to go practice my Python” – this is what I was working toward with that practice. :-)

>>> this <<< was your post i was referring to … because it was the last chat we had with scott around here … sigh :-(

PeterJones

I asked for feedback, why would I mind?

The name might change eventually.

You seem to have latched on Reader rather than Viewer. Other than the psychic connection, is there a reason you prefer FilteredReader to FilteredViewer?

I thought of it as a “filtered viewer”, because it based on using the “filtered” feature of PreviewHTML. But really, my script is neither doing the filtering (it delegates that to an external process) nor a viewer (it delegates that to an external process) – so it might not be the best name. Maybe something like “ConvertAndLaunch”. I don’t know. I don’t want it to include “HTML” in the name, even though right now it does “ConvertToHTMLAndLaunch”, because I’m hoping to make it allow other filter-destinations. Maybe “ConvertThenPreview”?

Hmm, it just hit me: if I do get the alternate destination, other than HTML, it’s basically become a generic utility that could not only be used for convert-from-X-to-HTML-and-view, but also compile-then-run, by using the compiler command (or make or whatever) for the Command=, and telling it that the output is an executable rather than HTML. Interesting. More things to ponder.

Alan Kilborn

I have done nothing except read the discussion here, and (again, with not further exploration) I keep thinking about how it might be similar to or different from Don’s porcine offering: https://github.com/npp-plugins/pork2sausage

PeterJones

@Alan-Kilborn ,

I keep thinking about how it might be similar to or different from Don’s porcine offering: https://github.com/npp-plugins/pork2sausage

Thanks for that. Yeah, they are similar. The primary difference is that mine doesn’t replace the source file. For one of my primary use cases – rending the POD from inside my Perl code into an HTML document – it would be rather destructive to the Perl code to have it be replaced by HTML. :-)

chcg

@Peter-Jones I added it as addon example to https://github.com/bruderstein/PythonScript. Hope that is ok for your.

PeterJones

v0.02-beta has been released.

per #5, add an ExampleConverterCommand.bat, update the default pyscFilteredViewer.ini file to have valid filter defined using that batch file, so now there is a working filter out-of-the-box.
per #4, simplify the installation instructions, moving the alternate-situations to separate documentation

Also, per @Alan-Kilborn , I made a brief mention of how it differs from pork2sausage

@chcg: as I meant to imply by my upvote, I’m fine with you linking to it from PythonScript. I’m a little surprised there’s only two “interesting addon scripts” in the universe right now. :-) I guess you cannot just link to every wonderful script that @Claudia-Frank (or any other PythonScript contributor) posted to this Community… though probably some of her gists would be worth linking to.

guy038

Hello, @peterjones and All,

I’ve just tried your pyscFilteredViewer and everything went fine ! Well…, as long as if I’ve already started my browser !

Indeed, if I use the One-time filtering mode, on my weak XP configuration, the temporary file <Hex number>.<filename>.FILTERED.html seems already deleted… by the time the browser gets opened, leading, for instance, to the browser message( in French, of course ) :

" Firefox cannot find the file at the address /c:/docume~1/guy/locals~1/temp/37B82996.Test.txt.FILTERED.html "

Luckily, subsequent tests seem OK. I’m saying “seem” and not “are”, as yesterday, Notepad++ crashed several times ! Unfortunately, I could not understand the sequence of actions which causes N++ to crash, when using your script !

Today, once my browser opened ( while listening music on NET, too ! ) , I tried to run your script, successively, on the same file ( Your README.md file ) => Sometimes it’s opened, correctly, in the browser but in some other cases, it displays :

" Firefox cannot find the file at the address /c:/docume~1/guy/locals~1/temp/2BF64FA9.README.md.FILTERED.html "

Moreover, in a few cases, it, also, exits Notepad++, immediately !

An another time, on the Python console, this error occurred :

File “D:@@\763\plugins\Config\PythonScript\scripts\pyscFilteredViewer\pyscFilteredViewerLibrary.py”, line 312, in pyscfv_FilteredViewer
os.unlink(tmpfile)

WindowsError: [Error 32] The process cannot access the file as this file is used by an other process : ‘c:\docume~1\guy\locals~1\temp\2BF64FA9.README.md.FILTERED.html’

I noticed that, if I’m not listening simultaneously music on WEB, it’s getting better : no more error occurs and the browser seems to display the HTML version of files, a bit quicker !

So, Peter, don’t worry about it. Probably, some side-effects of my old XP machine :-))

Best Regards,

guy038

For the record, in the pyscFilteredViewer.ini file, I only added the section, below :

[Txtfile]
Language=Normal text
Extension=.txt .md .log
Command="D:\@@\763\plugins\Config\PythonScript\scripts\pyscFilteredViewer\ExampleConverterCommand.bat" "%1"

And here is my Debug info :

Notepad++ v7.6.3   (32-bit)
Build time : Jan 27 2019 - 17:20:30
Path : D:\@@\763\notepad++.exe
Admin mode : OFF
Local Conf mode : ON
OS : Windows XP (32-bit)
Plugins : DSpellCheck.dll mimeTools.dll NppConverter.dll NppExport.dll PythonScript.dll

PeterJones

@guy038,

Thanks, that’s actually quite useful: I probably should just leave the temporary files there, rather than deleting every time. (The auto-delete was a holdover from when I used a random tmpfile name, before I had the FilterOnSave option.)

As for the crash: it’s probably the same issue @Eko-palypse had with crashing. So far, 7.6.2 64-b on Win7 and 7.6.3 32-bit on XP see the crashes, whereas my 7.6.4 64-bit on Win10 does not. Maybe it is an older-windows thing. Since I don’t have access to Win7 or earlier machines, that will be hard for me to test. Maybe I’ll have to look into “if OS<Win10, don’t use the command eko noticed”… I’ll have to research how to look up Windows-rev in Python.

Eko palypse

@Peter-Jones
Still strange this crashes - I will setup a Windows10 in VirtualBox and see if I can get the same
result as you have. If so, I guess it’s time to upgrade - Win7 comes to end anyway.

Meta Chuh

@guy038 @PeterJones and all

seems already deleted… by the time the browser gets opened

true, even if it has been rendered in time before it gets deleted, the auto deletion also makes it impossible to view the source code of a rendered html by right clicking on the rendered page.

Meta Chuh

@PeterJones said:

I probably should just leave the temporary files there, rather than deleting every time.

one possibility to avoid cluttering of old temp files, could be not to delete the current temp file at the end of the script, but to delete all pyscfr temp files at the beginning of the script, once a new html has been rendered.
this would leave them there until a new rendering is executed, and not fill up %temp%

another possibility could be to use a static file name for the rendered html, and never delete it (it would simply be overwritten everytime a new render has been done, and the file name would always be the same

Meta Chuh

@Eko-palypse

Still strange this crashes - I will setup a Windows10 in VirtualBox and see if I can get the same
result as you have. If so, I guess it’s time to upgrade - Win7 comes to end anyway.

strange, i didn’t even manage to crash exit npp using pyscfr on windows 7.

either my system is somewhat different … or i didn’t test the right things long enough 😂😂😂
just kidding, i think we all tested pyscfr a lot

Eko palypse

@Meta-Chuh @Peter-Jones

Win10 x64
Npp7.6.3 x64 7z
PythonScript 1.3.0 x64 full.zip

and running

from Npp import notepad, LANGTYPE
LANGUAGES = LANGTYPE.values.values()
for language in LANGUAGES:
    x = notepad.getLanguageDesc(language)

results in

Faulting application name: notepad++.exe, version: 7.6.3.0, time stamp: 0x5c4dd9e9
Faulting module name: ntdll.dll, version: 10.0.17134.376, time stamp: 0x60d78cf9
Exception code: 0xc0000374
Fault offset: 0x00000000000f4d7b
Faulting process ID: 0x1794
Faulting application start time: 0x01d4b97dee776c53
Faulting application path: C:\tools\Npp\763\x64\notepad++.exe
Faulting module path: C:\WINDOWS\SYSTEM32\ntdll.dll
Report ID: 18dc0488-4bb7-4063-b564-596ea765af26
Faulting package full name: 
Faulting package-relative application ID:

BUT what I found out so far is, that this happens only for the very first time you start npp on Win10.
5 reboots - 5 crashes after immediate execution.
Subsequent starts and script executions do not fail anymore. Did a 100000 loop without a problem.
On Win7 this doesn’t matter - every run crashes my npp.

??? Start running out of ideas ??? - Unfortunately I haven’t enough expertise debugging npp in Visual Studio … hmmm …

PeterJones

Thanks again for everyone’s feedback.

v0.03-beta has been released. I fixed the tempfile cleanup that @guy038 mentioned (by deleting all temp at the first run of the script – so basically, when you reload NPP – similar to what @Meta-Chuh recommended). I also fixed a couple of things that were starting to annoy me that others hadn’t mentioned yet.

Still haven’t been able to debug the crashing thing; I cannot get a reliable crash, so cannot eliminate it. @Eko-palypse : where did you get that detailed crash info that you quoted in your most recent post?

Eko palypse

@PeterJones - copied from the windows eventviewer log.

PeterJones

@ekopalypse,

I finally found the time to try out your ctypes-based/windows-messaging-based alternative to getLanguageName()/getLanguageDesc(). I haven’t yet seen a crash with my testing of this updated version, and v0.03-beta had started to crash more frequently for me, so I’m hoping v0.04-beta fixed the crash by using your functions.

v0.04 Beta Release

Newest beta release

Convert from notepad.getLanguageName(notepad.getCurrentLanguage()) to a function __eko_getLexerName() that’s derived from @ekopalypse’s Enhanced UDL Lexer’s get_lexer_name(); similarly for ...getLanguageDesc(...) => __eko_getLexerDesc()
Fix the hex prefix in the filtered filename to never show up as -XXXXXXXX by masking the 32-bit CRC32 with 0xFFFFFFFF, to make sure Python doesn’t incorrectly believe that it’s “negative” when the MSB is 1.

You can download just the scripts via pyscFilteredViewer-v0.04-beta.zip

Ekopalypse

@PeterJones

Since I use the SendMessage method I also had no crash anymore.
I tried to pinpoint the cause of the issue but my lack of C++ knowledge isn’t really helpful here ;-)

Meta Chuh

@PeterJones

wow, 2.2k viewers !!!

it seems that pyscfr isn’t just a cool tool for us internally.
it has apparently caught more interest than the majority of topics, excluding official announcements 👍

@Ekopalypse

you have to reduce your offline times, please 😉

PeterJones

@Meta-Chuh,

You did a stealth edit and got rid of the “how many downloads at GitHub” question… but I found that timely, because it was just a few hours ago that I went searching how to find that information.

It’s in the GitHub API: https://api.github.com/repos/pryrt/pyscFilteredViewer/releases lists the JSON for the project, and each of the “assets” has a “download_count” attribute. Unfortunately, that appears to just be for the manual script-only zipfile that I upload as an “asset”, because I don’t see download_count for any of their auto-generated tarball/zipball.

At last check, v0.03-beta had 2 downloads, and v0.04-beta had 4. So not exactly the impressive number that the Community “views” count would indicate. I think the 44 posts in this thread, and all of our views where we load/reload/upvote/refresh probably artificially inflate that count. :-)