Spellsinger's Blog

[varlog]

@brolny
I won't be doing anything with fonts, alas .
For the word column width problem a simple resizing of the SpellcheckEditor window should suffice. For the problem with the width of "Change Selected Word To" box I will try to implement ToolTip (when you place your mouse on the object a text in a box will appear): it's not perfect but will help some, I think - see picture.
I would like to have a sample of your book, just a page with this long word would be enough. What language is this?

[BetterRed]

Originally Posted by varlog

@brolny
What language is this?

Dictionary name is hy-my - so Armenian perhaps

[brolny]

Originally Posted by BetterRed

Dictionary name is hy-my - so Armenian perhaps

Exactly, Armenian (EA, reformed orthography).
There are AFF and DIC files in the attachment - if you want to try it within new spell-checking system which varlog coding now.
(Or you can download them for free from http://www.hyspell.com/. But then you have to change AFF file because some Armenians punctuation marks [!?'] => [՜՞'՝] inserted not at the end of a sentence, but inside the word after the stressed vowel (and inside one diagraph ) of a syllable).
Unfortunately, HYPH file doesn't exist at all, because hyphenation in Armenian is very complicated matter in case of hidden syllables, which became unhidden (open or closed) on the hyphenation.

[brolny]

Originally Posted by varlog

@brolny
I would like to have a sample of your book, just a page with this long word would be enough. What language is this?

The file is at the very beginning, just some prep. works.

[brolny]

Originally Posted by varlog

@brolny
I won't be doing anything with fonts, alas .
For the word column width problem a simple resizing of the SpellcheckEditor window should suffice.

Resizing - resize the list of words only, not the field of suggested word.
If not, then not, very sorry.

PS
According to this (http://www.wolframalpha.com/input/?i...sh+word+length) - the average length of English word is 5, Armenian is 16, Russian is 15, German and Spanish is 14.

But I think that it's wrong, so I make a 5-minutes calculation on words in Armenian Bible. Average word's length is >10, but without auxiliary verbs and pronouns (article is the part of a word, and written together with the word) the length is ~13 chars. So, it's seems to be true...
Three times more! Unbelievable.

But main noise/disadvantage not the length itself, but the letters - some letters combinations very very hard now to recognize in suсh small size...
So it's complicated - width of editable field + length of words + letters combinations + font style...

PPS
Intresting to see terrible Korean words in this case

[brolny]

Originally Posted by varlog

@brolny
I would like to have a sample of your book, just a page with this long word would be enough. What language is this?

If you prefer English...

Time expressions => Ժամանակային արտահայտություններ

[brolny]

Originally Posted by brolny

Intresting to see terrible Korean words in this case

Just some Koread words.

[varlog]

@brolny
Thank you for all the books.
You have a lot of work ahead of you with them: the languages are not properly tagged - see picture (ur3.epub).
And then you use "ru-RU" as "dc:language" and just "ru" in overriding html xml:lang attribute.
That reminds me: I wanted to do something general with "ln" versus "ln-LN" since (almost) the very beginning - but I get heavily distracted with other issues on this project...

[varlog]

... the deadline is almost there and I'm not getting my ideas working ... as usual...

After the deadline (23:30 21.08.2016) I will be one month totally offline.

The things I've touched are mostly working. At the moment I'm trying to implement language map: get the language according cursor position in text. It is not the ultimate solution: imagine now you start to write in Code View something like

<span xml:lang="new one">

and the highlighter is highlighting as the QTextEditor, every other key input, tells it to... and you correct your text... massacre...
Serial parsing is not something I do for pleasure...

[varlog]

Leonard... I am just a sucker for a awesome phases, take it with ... I liked his "Famous Blue Raincoat"... He wrote two books, too. "Beautiful Losers" was for me just "Famous Blue Raincoat" spread over some two hundred pages...

No no no, it is not the end of Sigil multi language spell check as envisioned by me (I hope). The end of it you will see (or not ) in Sigil.

But yes, it is the end of this blog. The question if my way is right way to mess with internals of Sigil will be answered elsewhere - in Sigil.

Thank you all for attention. There are some twenty of you who followed it quite closely, which is much by my standards. I hope you have enjoyed it as much as I did.
Special thanks goes an Hitch: your out of nowhere post made my day.

One thing I want to say:
I'm spitting my soul out there for more then two months and no troll chimed in? Hey trolls, don't you read books...?
Hey, mobileread, do you have some secret anti-troll weapon?

I guess trolls have no much use for books and such...

Here is the code, branch MLspellchecker.

It is far from finished. I have disabled Settings and haven't touched Preferences. I have not followed all the clues. There is rubbish there and ideas not followed to the end left ...
But the deadline is there, so I have to deliver something, which [TAKEN item="from my real profession"] will be the basis for further negotiations, if any.[/TAKEN]
The main functionality seems to be working on my system. The question is: does it compile on others? What happens on Mac, with its multi windowed Sigil? The memory/speed footprint on other systems? Is my code up to standards (it is preliminary!)?

I hope some of you will try it. Try context menu in SpellcheckEditor. Try double click on "Code" or "Alias" items in "Loaded Dictionaries" area.

Feel free to say something about it, in time I will presumably respond.

Latest Multilanguage epub attached.

@Kevin: I can make pull request if you want to (in next ten hours), but I think it is too early. I think we should keep issues of multi language spell check separate from mainline for now.

And now [AT LEAST ONE MONTH OF NONEXISTENCE]...

Will not be continued.

regards

[KevinH]

Hi varlog,
Thank you. I will take what you have and integrate it into a new branch on master. That way it can be more fully merged in once the pieces are all in place. So if you still have time, please create a pull request and I will merge it into a branch.

Take care,
KevinH

[KevinH]

For anyone interested in playing with varlog's multi language spell check code, I have created a "mlspell" branch on Github from master and merged in varlogs pull request.

[varlog]

The software, which is not worked on, dies. So the piece of software is never finished - till it dies.
That said I would like to inform you that I've actually done it. There is a piece of software which brings multi language spell check to Sigil, with, as far as I can see, all previous functionality and some more. Not perfect because perfection comes with experience.
Will Sigil have it? I think not. Not in this form.
Let us be honest. Almost nobody really needs it: seems to be my specialty in Sigil features .
The devs have much more important thighs to do: implementing epub 3 and some more.
And I'm in a middle of an personal circumstances that will make my further work on Sigil impossible - I think... at the moment... Christmas coming...
If anybody really needs it - it is relatively easy to find. With every passing day it dies of course...

The good news: last time I've checked the Calibre Editor parsed Multilanguage01.epub correctly: it was not the case when I've first checked... haven't said anything, because I knew Kovid will get it right as soon as he noticed .

[KevinH]

Hi Varlog,
As I told you previously, I won't have time to really look into this until my holiday break over Christmas. That does not mean I plan to let it die, nor does it mean I don't appreciate the effort. Your changes are large and invasive so it will take time for me to understand exactly what you have done and why. Once I am happy, it will get pulled into the main Sigil tree. Please be patient.

Thanks,

KevinH

[KevinH]

PS, I merged your latest pull request into mlspell branch of Sigil so that others can build that branch to play around with it.

Over the Christmas break, I will do my best to review and modify your changes to fit with Sigil so that they can be merged back into master.

Thanks,

KevinH