getLanguageDesc bug

[pryrt]

@Ekopalypse and I have noticed that there's a potential bug in PythonScript's implementation of getLanguageDesc: https://community.notepad-plus-plus.org/post/51274

We can get it to reliably crash using the brief script:

from Npp import notepad, LANGTYPE

LANGUAGES = LANGTYPE.values.values()
for language in LANGUAGES:
    print(notepad.getLanguageDesc(language))

When I run the equivalent script using my Perl Win32::Mechanize::NotepadPlusPlus to call the underlying NPPM_GETLANGUAGEDESC (or if Eko uses python's ctypes.windll.user32.SendMessageW to do the same), it works reliably.

[Ekopalypse]

Not sure, but I assume that here one needs to add 1 to the buffer allocation.
but I might be wrong as I don't have sufficient c++ knowledge.

[pryrt]

well, getPluginHomePath does use [size+1], so that's likely it.

(getPluginHomePath, getLanguageName, and getLanguageDesc are the only three that use the idiom of size = callNotepad(...) for allocating the size, so hopefully the two language functions are the only two with that bug.)

[chcg]

See N++ description of how to use NPPM_GETLANGUAGEDESC and NPPM_GETLANGUAGENAME and the size+1 was missing as you already found out.
Thanks.

There are tests for that:
https://github.com/bruderstein/PythonScript/blob/master/PythonScript/python_tests/tests/NotepadWrapperTestCase.py#L795-L824
,but last time I run it I think there was no issue found, so probably also the testcase needs to be checked.

[Ekopalypse]

The test itself seems to be fine.
As far as I understand, boost:.python truncates the string by \0,
so there is no way for the test to see if it has been allocated correctly or not - I suppose.

[chcg]

Fixed with https://github.com/bruderstein/PythonScript/releases/tag/v1.5.4