Improve extract performance via ignoring directory early during os.walk

[stkao05]

trafficstars

Currently, the extraction code will do an os.walk to perform a deep file search. However, this file exploration could be very slow when there were directories that were deep and contain many files. Even if you have specified some directories to be ignored in the mapping file, the os.walk would explore these directories.

The PR improves the extract process performance via making sure to skip exploring those ignore directory early during os.walk.

[stkao05]

Real-life scenario I have experience

When you are working with front-end, typically you would have a node_modules directory in your codebase which contains source codes of all 3rd party lib (similar to Python's /site-packages/), and this directory typically is very large.

[codecov-io]

Codecov Report

Merging #694 into master will decrease coverage by 0.03%. The diff coverage is 77.77%.

@@            Coverage Diff             @@
##           master     #694      +/-   ##
==========================================
- Coverage   90.97%   90.94%   -0.04%     
==========================================
  Files          24       24              
  Lines        4176     4184       +8     
==========================================
+ Hits         3799     3805       +6     
- Misses        377      379       +2

Impacted Files	Coverage Δ
babel/messages/extract.py	94.38% <77.77%> (-0.56%)	:arrow_down:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 0cfa69e...6ada6ee. Read the comment docs.

[stkao05]

Ping @akx

[akx]

Hi @stkao05 – #832 landed today, so this would need to be rebased :)