add support for JSON Lines format in 'target'

[apalala]

Try to read a JSON object per input line when reading the whole input as JSON fails. This makes glom behave more like jq.

[codecov-io]

Codecov Report

Merging #24 into master will decrease coverage by 0.59%.
The diff coverage is 27.27%.

@@            Coverage Diff            @@
##           master      #24     +/-   ##
=========================================
- Coverage   83.72%   83.12%   -0.6%     
=========================================
  Files           9        9             
  Lines         805      812      +7     
  Branches      133      136      +3     
=========================================
+ Hits          674      675      +1     
- Misses         91       96      +5     
- Partials       40       41      +1

Impacted Files	Coverage Δ
glom/cli.py	39.17% <27.27%> (-1.94%)	⬇️

---

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 d06e3d1...5e053fb. Read the comment docs.

[mahmoud]

Hey Juacarlo! First off, thanks for this, definitely want this sort of direction in the CLI. I think I see an issue with the implementation, however.

I think to really make this part of the functionality production-ready, we need to switch to more of a streaming approach, which means switching up the interface to use a file-like interface, and reading one line at a time. As it stands, this will hold the whole dataset in memory.

Feel free to take a swing at that, but if you don't have the time, I was planning on doing it myself in the near future anyways, and I'll be sure to give you due credit :)

[apalala]

Hey Mahmoud!

You are right! In fact, I have memory problems with the software I use for JSONL data because it doesn't stream the lines.

I'll take a shot at it.

[isaulv]

Use ValueError since JSONDecodeError is a subclass.