Interesting, just not sure it is feasible given the simple list 2-tuples for autocorrect entry and its replacement string. Would require substantial dev effort.

Thank you for looking into it, Stuart.
Actually the list of such 2-tuples in Polish would be massive, as -byś, -bym, -by, -byście, -byśmy are super common suffixes that you add to verbs in the past tense to create conditional verbs. 

Then there are suffixes that result in participles: verb + -jąc/-jący/-jącym/-jącego/-jącemu/-jąca/-jącą/-jącej/-jące/-jącym/-jących/-jącymi (all of which are sort of equivalent of -ing)

You can see how long a list of definitions is required for a single verb - and that's only for two suffixes.

It would be soooo much easier and more efficient if I defined autocrrection for as many verbs as possible, and had a finite list of such suffixes that I could attach as I type. 

At the moment I've just found a way around it - I type the verbs and their suffixes separately, and then, after a day's work I Ctrl+H them: 
find: [space]-byś (etc.) / replace: [w/o space]byś

Than kind of does the job, but severely obstructs the flow. As a writer I'd be super grateful for smoothing it out.

I'm positive that the option to autocorrect words while at the same time deleting the space preceding it - effectively joining the autocorrected word with the one preceding it - would find many other uses, too.

Thanks again,
Mac

László, do you have an idea?

Such a feature will also help Marathi language. The people incorrectly write "घरा चा" or "घरा ची"  (with space) instead of "घराचा" / "घराची" (without space). There should be a parameter to declare such suffixes to be joined with the earlier word. The suffixes like "चा" or "ची" are not accepted words in Marathi and never used separately after space. (like "ed" not a word in English but correct in "worked")

A possible solution is to use a non-space separator, e.g. comma, and the .* pattern to recognize the suffix in the end of the character sequence:

mogl -> mógł
.*,bys -> byś

After typing the comma, mogl changed to mógł. After following with ,bys:

mogl,bys -> mógłbyś


The other solution to use Hunspell spell checker to add the missing diacritics, accepting the their suggestions automatically. It seems, the dictionary contains all the Polish diacritics, so it can suggest the right alternatives with diacritics:

== pl_PL.aff ==
MAP 8
MAP aą
MAP cć
MAP eę
MAP lł
MAP nń
MAP oóu
MAP sś
MAP zżź

So with a LibreBasic or pyUNO macro, it's possible to add the missing diacritics automatically (except when the result is ambiguous), e.g. by clicking on a button at the end of the document editing. As a code snippet, see for example the following LibreBasic code snippet from https://forum.openoffice.org/en/forum/viewtopic.php?t=1222, using XSpellChecker service of LibreOffice UNO API via com.sun.star.linguistic2.LinguServiceManager:

Sub WrongWordsList 

    Dim oDocModel as Variant 
    Dim oTextCursor as Variant 
    Dim oLinguSvcMgr as Variant 
    Dim oSpellChk as Variant 
    Dim oListDocFrame as Variant 
    Dim oListDocModel as Variant 
    Dim sListaPalabras as String 
    Dim aProp() As New com.sun.star.beans.PropertyValue 

    oDocModel = StarDesktop.CurrentFrame.Controller.getModel() 
    If IsNull(oDocModel) Then 
        MsgBox("There's no active document." + Chr(13)) 
        Exit Sub 
    End If 

    If Not HasUnoInterfaces (oDocModel, "com.sun.star.text.XTextDocument") Then 
        MsgBox("This document doesn't support the 'XTextDocument' interface." + Chr(13)) 
        Exit Sub 
    End If 

    oTextCursor = oDocModel.Text.createTextCursor() 
    oTextCursor.gotoStart(False) 

    oLinguSvcMgr = createUnoService("com.sun.star.linguistic2.LinguServiceManager") 
    If Not IsNull(oLinguSvcMgr) Then 
        oSpellChk = oLinguSvcMgr.getSpellChecker() 
    End If 
    If IsNull (oSpellChk) Then 
        MsgBox("It's not possible to access to the spellcheck." + Chr(13)) 
        Exit Sub 
    End If 
        
    Do 
        If oTextCursor.isStartOfWord() Then 
            oTextCursor.gotoEndOfWord(True) 
            ' Verificar si la palabra está bien escrita 
            If Not isEmpty (oTextCursor.getPropertyValue("CharLocale")) Then 
                    If Not oSpellChk.isValid(oTextCursor.getString(), oTextCursor.getPropertyValue("CharLocale"), aProp()) Then 
               sListaPalabras = sListaPalabras + oTextCursor.getString() + Chr(13) 
           End If 
        End If 
            oTextCursor.collapseToEnd() 
        End If 
    Loop While oTextCursor.gotoNextWord(False) 
        
    If Len(sListaPalabras) = 0 Then 
        MsgBox("There are no errors in the document.") 
        Exit Sub 
    End If 

    oListDocFrame = StarDesktop.findFrame("fListarPalabrasIncorrectas", com.sun.star.frame.FrameSearchFlag.ALL) 
    If IsNull(oListDocFrame) Then 
        oListDocModel = StarDesktop.loadComponentFromURL("private:factory/swriter", "fListarPalabrasIncorrectas", com.sun.star.frame.FrameSearchFlag.CREATE, aProp()) 
        oListDocFrame = oListDocModel.CurrentController.getFrame() 
    Else 
        oListDocModel = oListDocFrame.Controller.getModel() 
    End If 

    oTextCursor = oListDocModel.Text.createTextCursor() 
    oTextCursor.gotoEnd(False) 

    oListDocModel.Text.insertString (oTextCursor, sListaPalabras, False) 

    oListDocFrame.activate() 

End Sub


And the other code snippet to modify the wrong words (but also fixing a problem in XSpellChecker usage that has since been solved):

https://forum.openoffice.org/en/forum/viewtopic.php?p=425651

Awesome! László's solution works for me like a charm. Thank you.

(In reply to Mac from comment #6)
> Awesome! László's solution works for me like a charm. Thank you.

Let me add that this solution works just as well with prefixes. In that case the .* pattern needs to be used after the prefix and the non-space separator (in this example comma):

e.g.:

pol,.* -> pół
zalezny -> zależny

After typing the comma, pol changed to pół. After following with any word that word is joined to pół. In this case:

pol,zalezny -> półzależny

---

You can even create long words with both prefixes and suffixes, 
e.g. by defining in autocorrect the following:

pol,.* -> pół        (a common prefix, meaning semi-)
zalezny -> zależny
.*,ch -> ch          (one of the many inflectional morphemes)

you can type:
pol,zalezny,ch -> półzależnych

Thank you again! You're the best :)

Actually not exactly - with prefixes whatever is after the prefix doesn't get corrected, but it's still better than nothing.
(I'm sorry, there's no option to edit a post - please, feel free whatever parts of what I'm saying you deem irrelevant or muddying.)

This (use of comma) workaround is intended for those who already know the correct spelling. Spell checkers are designed for novice users who may think what they've typed is correct.

By the way, I was unaware that this was possible. It has rendered hundreds, if not thousands, of auto-correct entries in the Marathi language pack obsolete. I appreciate your efforts, but what I find frustrating is that it's not adequately documented, especially with all the use cases as mentioned above.

Ususally Close up (Unicode U+2050) ⁐ sign is used by proof readers if there is no need of space. For e.g. if I type "I am work ing hard." in google docs, it suggests "working". Libreoffice writer suggestions are way out of context.
If this is unrelated in current context, let me open a new feature request.

(In reply to Mac from comment #8)
> ...with prefixes whatever is after the prefix doesn't get corrected...

(In reply to Shantanu from comment #9)
> This (use of comma) workaround is intended for those who already know the
> correct spelling....
> ... I find frustrating is that it's not adequately documented...

Since the ticket is still flagged as UX relevant I wonder what's missing. Or should we forward the solution to documentation?

(In reply to Heiko Tietze from comment #11)
> (In reply to Mac from comment #8)
> > ...with prefixes whatever is after the prefix doesn't get corrected...
> 
> (In reply to Shantanu from comment #9)
> > This (use of comma) workaround is intended for those who already know the
> > correct spelling....
> > ... I find frustrating is that it's not adequately documented...
> 
> Since the ticket is still flagged as UX relevant I wonder what's missing. Or
> should we forward the solution to documentation?

Oh, by all means, my initial question/request has been answered beautifully. It's a neat method for such highly inflective language as Polish (and, I gather, Marathi). So, it can go to documentation as a solution for adding suffixes and inflective morphemes that need to be automatically attached to the root word (while also get autocorrected - as a bonus - if there's need for that)



---
Since you are asking what is missing - it's the autocorrection of the root word in the middle (the part without the the .* pattern and non-space separator). 

In other words: prefix[corrected],root[not corrected],suffix[corrected]
(the root may stay not autocorrected because the autocorrect function recognises it not as the root alone, but as the chain consisting of [prefix][comma][uncorrected root].

Although I'll have to investigate it once again, because Im almost positive the root got corrected in a few words I initially typed as a test.

(In reply to Mac from comment #12)

> ---
> Since you are asking what is missing - it's the autocorrection of the root
> word in the middle (the part without the the .* pattern and non-space
> separator). 
> 
> In other words: prefix[corrected],root[not corrected],suffix[corrected]
> (the root may stay not autocorrected because the autocorrect function
> recognises it not as the root alone, but as the chain consisting of
> [prefix][comma][uncorrected root].


Yes, I've tested it and can confirm now that it's how it is at the moment.

Someone mentioned before that it would be more difficult than suffixes, but how cool would it be if a user could "tell" the program to treat a particular non-space separator [e.g. comma] like a space. Or maybe is there such space separator already? Something other than comma? (But TBH comma is placed in a very convenient spot on the keyboard...)

[An Edit Comment option would be handy]

I meant: Or maybe is there such a space-like separator already?

One more thing worth noting when it comes to prefixes.

There's a difference between words whose root doesn't get autocorrected, and words that do (e.g. words with diactitics).

---
Example:

[prefix]
za

[roots]
baw
myśl

---

In autocorrect:

za,.* -> za
mysl -> myśl
[baw doesn't need to be autocorrected - no diacritics]

---

When I type:

za,baw -> zabaw
za,mysl -> zamysl [joined, but 'mysl' not corrected to 'myśl']