"regex-word" rule patterns have the wrong precedence

[scovetta]

Describe the bug
Example:
Take a rule like AI015600 (Data: Audio File). The raw pattern is mp2|mp3|mp4... with a type of regex-word. This translates to the regex \bmp2|mp3|mp4...\b, which isn't right -- e.g. foomp3bar matches this, since the \b's are attached to the first and last strings.

It looks like we have a bunch of this -- it might be a regression in the dev branch, but we might want to update:

https://github.com/microsoft/ApplicationInspector/blob/development/RulesEngine/Ruleset.cs#L401

from:
pattern.Pattern = string.Format(CultureInfo.InvariantCulture, @"\b{0}\b", pattern.Pattern);
to
pattern.Pattern = string.Format(CultureInfo.InvariantCulture, @"\b({0})\b", pattern.Pattern);

To Reproduce
Steps to reproduce the behavior:

1. Download NPM/left-pad@1.3.0.
2. Run Application Inspector (latest beta) on it.
3. Notice the false positives, like AI015600 found in package/perf/perf.js (matching "hma" in "Benchmark").

Expected behavior
Regex-word should only match complete words (e.g. \b(THING)\b).

Screenshots
n/a

Additional context
ApplicationInspector.CLI 1.3.11-beta+69410edbd6

[gfs]

This is fixed in #369

[gfs]

I created a separate PR to one off fix this for 1.2, but this was already fixed in 1.3. I like your solution better than mine (splitting up the bars into separate regexes) so I've adopted that in #369 for 1.3.

#371 will fix this for 1.2.