documentAbbrIsUsed - better matching for abbreviations

[mlewand]

Problem

There are few matching problems.

Diactric Characters

It strips diactric characters due to RegExp replacement. That results with BOŚ being recognized as BO. That might be two totally different things.

Abbreviation With Mixed Case

Currently this assessment matches only full capitalized abbreviations. Often times we're working with abbreviations that involves mixed casing. I'd love it to match abbrs like GoF, DfE, PGNiG.

That might sound tricky, because there are some funny names like GitHub, oEmbed, MathJax and we can't allow these to be considered as abbreviation. So solution for this is that more than 50% of "word" (abbreviation) length must be uppercased. That would eliminate majority of false matches.

Matching Abbreviation With Numbers

I think that it's actually too much :) Numbers are most commonly used in l33tspeak shortcuts. Examples are:

• P2P
• G2G

  Matched Element

Currently abbreviation assessment checks whole document, and reports all the abbrevs found, that are not bound to any particular element.

Instead we need to provide more precission, showing each abbreviation occurrence. End user will need to accept each occurrence of the abbreviation.

More Resources:

• Assessment logic can be found in acronym.js file
• You can check testsFail/documentAbbrIsUsed branch for some failing tests.

[mlewand]

Possible traps: We need to watch out with matching for cases when whole text is written with uppercase.

It is a case that legal notes are written uppercased, so we need to watch out for that.