Prune the data before exporting to YAML ✂️ 📉

[movermeyer]

What are you trying to accomplish?

Fixes #85

There are many places in the code that are using a select(...).inject({}) pattern for producing data, which produces empty Hash's whenever there is no data in the XML.

What approach did you choose and why?

Prune the empty hashes recursively before outputting to YAML.
Since yaml.rb will already not output if data.empty?, it will not output a file in cases where there is no relevant data to output (i.e., the file is entirely empty hashes).

What should reviewers focus on?

Could there be any place in the code where an empty Hash is appropriate to output?
I contend that there is not.

The impact of these changes

Thousands of fewer files are output, and those that are output are smaller since they no longer contain empty nested key structures.

	Before	After	Ratio
# of .yml files	8332	6005	0.79
Total bytes	30464452 (29.05 MiB)	30188555 (28.79 MiB)	0.99

Testing

You can test it yourself:

git checkout master
thor cldr:export --target=./data_from_master

git checkout movermeyer:mo.prune_yaml_files
thor cldr:export --target=./data_from_pr

# Compare the file counts:
find data_from_master | grep "\.yml$" | wc -l
find data_from_pr | grep "\.yml$" | wc -l

# Compare the byte counts:
# Aside: I use `wc -c` since MacOS doesn't have an easy `du -cb`
find data_from_master | grep "\.yml$" | xargs -I{} cat {} | wc -c
find data_from_pr | grep "\.yml$" | xargs -I{} cat {} | wc -c

# Compare the list of files output:
find data_from_master | grep "\.yml$" | sort | sed 's/data_from_master/data/' > files_from_master.txt
find data_from_pr | grep "\.yml$" | sort | sed 's/data_from_pr/data/' > files_from_pr.txt
diff files_from_master.txt files_from_pr.txt

[movermeyer]

Just adding the missing # in the link text

[trishrempel]

I can't explicitly approve, but looks good to me! 👍