Spellcheck is flagging punctuation; can i make it stop?

[ElMiko]

When open the spellcheck window, the dictionary is flagging unrecognized words, but also separately flagging those same words when they occur at the end of a sentence.

For example, it'll flag:

poolhall

and

poolhall.

as two separate categories of spelling error.

Obviously this takes the occasional false positive and then doubles it. And that is not doubling my freshness or my fun.

Is there some way to make it stop? The (MUCH) older dictionary would ignore the punctuation and simply return two grouped entries for "poolhall". Will copying the old dictionary file from my pre-2.4.2 upgrade do the trick?

[Turtle91]

This is an issue with the dictionary you are using and whether it properly supports words with periods… as discussed in this thread.

[ElMiko]

Hmmm... the hunspell dictionary folder from my older version of sigil was empty, though. Only the User defined dictionary, which appears to have been copied over in the upgrade...

So, how do I go back to that hunspell dictionary?

[Turtle91]

I’ll let someone else address that. I haven’t worked with those aspects of Sigil much. It might be addressed in that other thread. Or search the threads here for hunspell. Or it’s probably in the Sigil user guide.

[ElMiko]

Yeah, I reinstalled the old sigil, found the old dictionaries (.dic and .aff files), and copied them over to the new sigil... and no luck. Periods are still showing up.

[KevinH]

You need to copy the old .dic and .aff files and put them in your Sigil Preferences folder inside the hunspell_dictionaries folder and restart Sigil. Your user supplied dictionaries will always override Sigil's internal one so you should never need to fart with it again.

You should be able to install any additional hunspell dctionaries from outside sources there.

[ElMiko]

I went into the .aff and deleted the "." from the WORDCHARS list as per this thread.

Presumably, this is the best of both worlds: keep whatever incremental sophistication has been added to the newer dictionary and take out the annoying "period" thing.

Good to know wbout how Sigil engages with the preferential dictionary folders. I was trying to figure it out from the preferences screen but not getting it!

We're getting there...

[KevinH]

This is covered in the current Sigil Users guide in the chapter in Preferences.

[ElMiko]

Follow-up: in the *.dic file there are various suffixes appended to some entries, e.g.

/M
/MS
/SM
/XVNGSD
etc.

Google has produced several results with people asking what they mean, but none that I've been able to find that actually answers the questions precisely. Does anyone know what all these designations mean, and are they relevant to my efforts to edit the file?

[KevinH]

Do not edit the .dic file. Edit the .aff file if you want to remove the period from acceptable wordchars.

And to answer your question, those are affix and suffix flags that specify which affixes or suffixes are valid with that root word. The .dic file lists root words with affixes that are allowed instead of every possible word. Those affix flags are of course defined in the .aff file. A dictionary made of a simple wordlist is unworkable for many languages due to compound words and compound affixes and suffixes. So to shrink its size to reasonable levels the dictionary must use some form of affix compression.

I you want to add words, do it to a Sigil wordlist (you can create and select them easily in Sigil). See the Spellcheck discussion in the Sigil User guide for more info. You can use the default or make one specific for your current epub.

Why on earth are you trying to edit words inside the .dic files?

If you truly need to change what words are legal then use the tools in Hunspell to expand the .dic file properly into a wordlist, make changes, then re-affix compress that huge wordlist file to create what a new .dic file.

Google "affix compression"

[ElMiko]

The .dic file is overly permissive. It includes extremely obscure or archaic words (like "bouse") that are going to result in false negatives for common misspellings/transcription errors (e.g. "house"). As you said, to add words to the dictionary, the custom word list is a perfectly adequate solution. But I'm not trying to add words; I'm trying to remove them from the dictionary's whitelist.

Thanks for the tip on "affix compression"!

[KevinH]

Just go to hunspell and grab a US dictionary that does not have these issues. Or better yet, just use the .dic and .aff file from your old Sigil version.

Search the .dic file for the word "bouse". If it exists with no affix flags on it, delete just that line and if there is a root word count as the first line of the .dic file, decrement it by 1.

If there are affix flags present, then you are deleting more than just the word bouse and would need to hand expand, and hand evaluate the resulting words and decide if any are worth keeping.

The .aff file can be used to identify each affix and the rules for applying them to a root word.

So if there are only a few affix flags on the word "bouse", they can be expanded by hand.