Suggestion: Spell Check Tool Enhancement

[BobC]

Originally Posted by BetterRed

Be nice to have the ability to exclude paragraphs too - to avoid checking quotes in the original vernacular - eg Chaucer, Shakespeare etc

BR

Just thinking laterally on this - if you consider the paragraphs as being in a "foreign" language for which there is no dictionary would this do the trick ?

So you would need to set the paragraph to have a style where the language code was say "en-oe" (for Old English). The trouble here would be that doing may be more effort than the benefit. It would mean wrapping the portion of text in a <div lang="en-oe"> or using CSS styling to achieve the same effect.

If Calibre interpreted "en-oe" as English it might need a dummy language code.

HOWEVER a quick check seems to indicate that the Editor will treat any declared language code for which it doesn't have a dictionary as that of the main file declaration or in the absence of a language declaration in the header as English (at least on my machine).

So a workaround would appear to be to install a dictionary with no words in common with English (Klingon ?) and declare the undesired paragraphs as that language.

Of course if there was actually a "en-oe" dictionary that would be great.

EDIT - given the context Middle English might have been a better suggestion, however it's the general principle I was trying to illustrate.

BobC

[BetterRed]

Originally Posted by BobC

Just thinking laterally on this - if you consider the paragraphs as being in a "foreign" language for which there is no dictionary would this do the trick ?

So you would need to set the paragraph to have a style where the language code was say "en-oe" (for Old English). The trouble here would be that doing may be more effort than the benefit. It would mean wrapping the portion of text in a <div lang="en-oe"> or using CSS styling to achieve the same effect.

If Calibre interpreted "en-oe" as English it might need a dummy language code.

HOWEVER a quick check seems to indicate that the Editor will treat any declared language code for which it doesn't have a dictionary as that of the main file declaration or in the absence of a language declaration in the header as English (at least on my machine).

So a workaround would appear to be to install a dictionary with no words in common with English (Klingon ?) and declare the undesired paragraphs as that language.

Of course if there was actually a "en-oe" dictionary that would be great.

EDIT - given the context Middle English might have been a better suggestion, however it's the general principle I was trying to illustrate.

BobC

Thanks Bob, I wondered about using a language too, but having to create a pseudo dictionary means it's contrived solution - on principle I don't like using contrived solutions to workaround trivial annoyances. IMO Tex's "calibre_ignore_spellcheck" class would be more elegant - but...

BR

[mrmikel]

Aren't quotes likely to be formatted differently, so it would be pretty easy to ignore them?

[BobC]

Originally Posted by BetterRed

Thanks Bob, I wondered about using a language too, but having to create a pseudo dictionary means it's contrived solution - on principle I don't like using contrived solutions to workaround trivial annoyances. IMO Tex's "calibre_ignore_spellcheck" class would be more elegant - but...

BR

Well FYI that doesn't work. I butchered a simple oxt dictionary to have a language code of "none" but Calibre was too clever - it wouldn't import the dictionary, complaining that it had an invalid language code.

(the oxt imported into Libre Office without a complaint)

BobC

[Tex2002ans]

Originally Posted by mrmikel

Aren't quotes likely to be formatted differently, so it would be pretty easy to ignore them?

Not necessarily, they could be inline quotes.

[BetterRed]

Originally Posted by mrmikel

Aren't quotes likely to be formatted differently,

Possibly, and if not then I guess one could make it so

Originally Posted by mrmikel

so it would be pretty easy to ignore them?

But I'm not the problem, I have no problem ignoring the text no matter what the format The problem is how to have the spell checker ignore text with a specific format, or what ever?

As I've said, for me this issue falls into the nice-to-have/trivial-annoyance/wishful-thinkiing/in-your-dreams-BR... category. I never would have thought it would attract so much attention

BR

[BetterRed]

Originally Posted by BobC

Well FYI that doesn't work. I butchered a simple oxt dictionary to have a language code of "none" but Calibre was too clever - it wouldn't import the dictionary, complaining that it had an invalid language code.

(the oxt imported into Libre Office without a complaint)

BobC

So, why not use a dictionary for a real language one is never likely to really need? I just installed the Estonian oxt. No disrespect to Estonians - it was a rational decision, I chose it because its uses Latin characters but its not Indo-European.

How do I mark a paragraph (or define a class) as using Estonian (it's code is 'et' ). I looked at W3Schools, but as usually happens there, my eyes glazed over. ie what would I need to add to this to flag it as Estonian

.block10 {
    display: block;
    font-size: 0.75em;
    text-align: justify;
    text-indent: 18pt;
    padding: 0;
    margin: 0
    }

I really shouldn't be wasting my time or anyone else's on this, but...

BR

[kovidgoyal]

If you want to change the language of a tag inHTML you have to add the lang attribute, like this

<p lang="fr">

which changes the language to French.

[BetterRed]

Perfect - see attachment

And I don't need to install an Estonian dictionary to use the language tag, seemingly it just needs to be a valid one.

Kovid & BobC.

BR

[Tex2002ans]

Originally Posted by BetterRed

So, why not use a dictionary for a real language one is never likely to really need? I just installed the Estonian oxt. No disrespect to Estonians - it was a rational decision, I chose it because its uses Latin characters but its not Indo-European.

Well, I wouldn't go marking text as a given language that it isn't, this is just going to cause more problems than you are trying to solve.

I was doing some back/forth help from Jellby with transcribing Greek letters, and he convinced me to start marking up Greek correctly. Here is how I handle it now:

<p>[...] The division of labor turns the self-sufficient individual into the <span class="greek" xml:lang="grc">ζῷον πολιτικόν</span> dependent on his fellow men, the social animal of which Aristotle spoke. Hostilities between one animal and another, or between one savage and another, in no way alter the economic basis of their existence. [...]</p>

Also, this site gives some of the reasons why you would want to tag languages correctly:

http://www.unimelb.edu.au/accessibil.../language.html

Language information specified via the lang attribute may be used by a user agent to control rendering in a variety of ways. Some situations where author-supplied language information may be helpful include:

• Assisting search engines
• Assisting speech synthesizers
• Helping a user agent select glyph variants for high quality typography
• Helping a user agent choose a set of quotation marks
• Helping a user agent make decisions about hyphenation, ligatures, and spacing
• Assisting spell checkers and grammar checkers

While there isn't A TON of benefit from marking it up now, a lot of the reasoning for marking up languages so in-depth is to future-proof the HTML.

"Assisting speech synthesizers" is extremely helpful with Text->Audio programs.

As to the typography side of things, now that I have stumbled into the world of LaTeX, having the languages marked properly allows the hyphenation dictionaries to work, which is (REALLY) important. And for languages with completely foreign character sets like Chinese or Greek, it allows you to easily swap in a different font.

There is also fantastic functionality built into LaTeX which easily allows you to swap between different rulesets (what quotation marks should be used, spacing rules around quotations, where linebreaks are allowed, etc. etc.).

Who knows, maybe ereaders in the future would be able to do more fancy stuff like that too with properly marked-up text.

Now, in a perfect world, you would mark every little saying as French, German, Spanish, etc. etc.... but that just takes way too long (the marginal benefit is not worth it to me), so I just settle on doing it for Greek.

Priority #1 is to get the dang books digitized and up online... way lower priority can be to go back and add in the language markup as needed. (Or when I get around to LaTeXing the books).

[BobC]

Like Tex I don't like faking the language code which is one reason I tried to come up with an empty dictionary approach. However as you can use any legal language code I think I'm going to use the one for Ido (io) - a little used variant of Esperanto and for which it is unlikely there is a dictionary ! (or much literature in the language)

Libre Office gets round this by having a setting under "Language" of "None" (Ignore spelling) but doesn't have a setting for Ido .

BobC

[kovidgoyal]

Use lang="und" which is the ISO 639-3 code for undefined and will prevent the spell checker from operating.

[BetterRed]

@Tex2002ans & BobC both said, 'not a good idea to use fake language code.' You won't get any arguments from me on that. As I said before I don't like contrivances.

That's what lead me to seek a language-in-class solution - use-it whilst spell checking, then lose-it

If/when I ever use it in the HTML it'll be transient. I was straightening out a 19th legal text recently that had a boatload of lengthy M.E. quotations, that's the sort of occasion when I might use a 'fake language'.

Interesting exchange.

I'd been wondering if the Tex2002ans handle was a 'nod' to TeX, seemingly not. Maybe TPTB would allow you to upgrade the x to an X, then it could be so :lol:

Added : Kovid just flew in and dropped a nugget of knowledge

BR

[BobC]

Originally Posted by kovidgoyal

Use lang="und" which is the ISO 639-3 code for undefined and will prevent the spell checker from operating.

That's what I had been looking for but I couldn't find the code. I was looking for something like "none" or "null".

Armed with that I've also found that :

ang - Old English and enm - Middle English are probably what is needed for the Chaucer/Shakespeare stuff and do the job.

Thanks Kovid for the pointer.

[BetterRed]

Originally Posted by BobC

That's what I had been looking for but I couldn't find the code. I was looking for something like "none" or "null".

Armed with that I've also found that :

ang - Old English and enm - Middle English are probably what is needed for the Chaucer/Shakespeare stuff and do the job.

Thanks Kovid for the pointer.

Because I'll only use it when the olde worlde text is a nuisance, I think I'll stick with 'und' and delete it when I'm done spell checking.

Added - It would seem that Sigil's spell checker does not ignore xml:lang="und"

BR