improved matching for emph on words containing underscores

[cebe]

https://github.com/yiisoft/yii2/pull/2452/files

[cebe]

MD: ``compareAttribute the name of the attribute to be compared with. _(currentAttribute_repeat)_
Expected: `

compareAttribute the name of the attribute to be compared with. (currentAttribute_repeat)

`

[cebe]

A solution to this needs to implement backtrack to see information about the characters before _.

http://johnmacfarlane.net/babelmark2/?normalize=1&text=%60compareAttribute%60+the+name+of+the+attribute+to+be+compared+with.+_%28currentAttribute_repeat%29_

[cebe]

A regex would have to match for a word boundary before the starting _ character. (?!\b)

[cebe]

test cases:

MD:

`compareAttribute` the name of the attribute to be compared with. _(currentAttribute_repeat)_

`compareAttribute` the name of the attribute to be compared with. (_currentAttribute_repeat_)

_em_word_with_underscores_

expected:

<p><code>compareAttribute</code> the name of the attribute to be compared with. <em>(currentAttribute_repeat)</em></p>
<p><code>compareAttribute</code> the name of the attribute to be compared with. (<em>currentAttribute_repeat</em>)</p>
<p><em>em_word_with_underscores</em></p>

[cebe]

Another case:

_one\_two_ __one\_two__

*one\*two* **one\*two**