improve Spelling rule by using word boundaries

[emteelb]

By using regex word boundary (\b) delimiters, the spelling rule applies to individual words rather than a word that might contain the regex filter. For example, "\b[cC]he\b" will match only "che" and "Che" rather than a regex filter without word boundary delimiters, for example, "[cC]he" that would match misspelled words that contain the regex, such as "aache" or "chemitsry".

Closes #894

[github-actions]

Deploy PR Preview failed.

[emteelb]

This is a "heavy-handed" approach where word boundary delimiters are used for every filter regular expression, for consistency, even those filters that might be unambiguous and unlikely to be contained within other words.

[ccoVeille]

I would like to drop my thought on this

Some rules comes with nonwords flag.

https://vale.sh/docs/topics/styles/

I don't remember whether spelling rule supports it.

nonword:true is about removing the automatic \b that are added around each words with the default value nonword:false.

So if you use nonword:true, you have to use the \b to surround word, unless you want to catch words with a prefix or suffix, so you use only a \b.

It is also useful when you want to catch multiple words \bun\-[a-z]+\b or punctuation such as \bvs[\., ]\b

So adding \b somewhere can be useless, if you are using nonword:true AND there are some rules where you don't use \b on both side. Otherwise, you can use the default nonword:false and avoid repeating the \b everywhere

Let's go back to this PR, I don't see a need to add \b on both side. Also, I don't think it works as expected, especially because I don't see a nonword:true flag

Finally, the spelling rules doesn't invent things, if you add foo, it won't allow foobar. I mean if you add smally (just an example that is not in the dictionary), you want allow eyesmally or smallyfish

So I don't see what you are trying to fix, and things are getting very complicated...

[emteelb]

I did not know about the nonword key for style rules. Thanks for pointing that out. I notice it's shown in the link you provided but when I search Vale documentation for more information, there are no results. If you know where this key is documented, lmk.

Try running the following with styles/RedHat/Spelling.yml as it is now in the main branch:

vale "This is a mistache."

With both RedHat.Spelling = YES and Vale.Spelling = YES in my .vale.ini configuration file, my results are:

 stdin.txt
 1:11  error  Did you really mean             Vale.Spelling
              'mistache'?                                  

✖ 1 error, 0 warnings and 0 suggestions in stdin.

Neither adding nonword: true nor nonword: false to the RH spelling style rule changes this result so I gather that the spelling style rule either does not support this key or else the key does something different that does not change the result.

If I change the "[cC]he" filter in the style rule to match what I have in this PR ("\b[cC]he\b"), and enter the previous vale command, my result is:

 stdin.txt
 1:11  warning  Verify the word 'mistache'. It  RedHat.Spelling
                is not in the American English                 
                spelling dictionary used by                    
                Vale.                                          
 1:11  error    Did you really mean             Vale.Spelling  
                'mistache'?                                    

✖ 1 error, 1 warning and 0 suggestions in stdin.

as would be expected.

Granted, "mistache" is an unlikely typo but I originally stumbled upon this issue when reviewing some writing where someone had a simple transposition mistake, "subcommnad", and because of the RH spelling style rule filter ("[Ss]u"), it wasn't caught (I normally enable only one spelling style rule, RH in this case, and disable the Vale spelling style rule).

As I said in a comment above, I took a heavy-handed approach in this PR by delimiting all the filters with word boundaries. I am open to a more selective approach and that might be something the RH team considers. This was just the easiest approach to be certain not to miss any potential word partials.

[ccoVeille]

thanks for you reply and tests.

Let me ask for help.

@jdkato what would you recommend here?

What do you think about what I wrote in my previous post?