Glossary not support non-english character

[lindeer]

Summary

more precisely, it could not support glossary term with non-word character on heading or trailing, e.g. Connétable could be matched, but Brancaleone von Andalò failed. If one term starting with ( or ending with ), it could not be matched either. Actually, words with special character are so common in glossary, especially in books of other language(french, german, japanese).

also an old bug since gitbook, hope it could be fixed. I think the root case of #233 is the same.

• HonKit version: 4.0.4

Step to reproduce

1. create one term Brancaleone von Andalò in GLOSSARY.md
2. insert Brancaleone von Andalò in paragraph of one page file

• Link to code example:

Der Papst gehört nicht nach Anagni oder Lyon, nicht nach Perugia oder Assisi, sondern nach Rom.« Ein kraftvoller Mann gab den Römern diese Sprache ein, Brancaleone von Andalò, ihr damaliger Senator. 

Expected results

Brancaleone von Andalò in article could be a glossary link.

Actual results

no link created.

[azu]

Maybe, this \b cause this error.
(toLowerCase is also a bit suspicious)

honkit/packages/honkit/src/output/modifiers/annotateText.ts

Line 69 in 8a47d19

const searchRegex = new RegExp(`\\b(${pregQuote(name.toLowerCase())})\\b`, "gi");

function pregQuote(str) {
    return `${str}`.replace(/([\\\.\+\*\?\[\^\]\$\(\)\{\}\=\!\<\>\|\:])/g, "\\$1");
}

const name = "Brancaleone von Andalò";
const searchRegex = new RegExp(`\\b(${pregQuote(name.toLowerCase())})\\b`, "gi");
console.log(searchRegex.test("test Brancaleone von Andalò test")); // => false

\b is for ASCII characters.
We need to use unicode safe word boundary.

• regex - Javascript RegExp + Word boundaries + unicode characters - Stack Overflow

[lindeer]

@azu Great! \b in regex indeed only bound word character, and unicode word boundary could meet most cases, but what if term like Henry (VII)? parentheses in word trailing