Fixes #3 - allow multiline strings that have the first line non-empty.

[mgruner]

This changes the detection of multiline strings to not rely on an empty first string "", but instead the presence of another string on the next line (starting with ").

[dfherr]

Looks good.

A few things though:

• could you please change the test texts to something generic (english -> english) rather than the texts from your app and chinese translation? I would like the strings to be readable and somewhat meaningful and not chinese characters in here unless it's explicitly to test for unicode/chinese characers. See https://github.com/experteer/simple_po_parser/blob/master/spec/simple_po_parser/fixtures/simple_entry.po as referece
• The comment ontop the multiline method needs to be updated (edit: ah you removed it. missed that. Maybe writing that it's indicated by the next line starting with a quote would be nicer)

  simple_po_parser/lib/simple_po_parser/parser.rb

  Line 268 in b0ab9f7

  # multiline messages are indicated by an empty content as first line and the next line

• Can you please benchmark/time the current version vs your new version on https://github.com/experteer/simple_po_parser/blob/master/test/benchmark.po I'm a bit worried about peek performance and maybe it makes sense to hide this behind an option to only take the boolean hit on every line rather than a peek (which creates a copy of a string, if I skimmed it correctly in the C source)

[mgruner]

@dfherr all changes applied as suggested. Let me know if this is ok.

For the benchmark, I executed rake parser:benchmark locally, but with 10000 iterations rather than 100. It seems to become only insignificantly slower with the changes of this PR.

 THIS PR

 Rehearsal -------------------------------------------
Parser:   8.883388   0.186883   9.070271 (  9.070439)
---------------------------------- total: 9.070271sec

              user     system      total        real
Parser:   8.849168   0.172060   9.021228 (  9.021489)

MASTER BRANCH

Rehearsal -------------------------------------------
Parser:   8.711592   0.157095   8.868687 (  8.868976)
---------------------------------- total: 8.868687sec

              user     system      total        real
Parser:   8.685205   0.154006   8.839211 (  8.839386)

Thanks!

[dfherr]

Looks good. Gonna test myself tomorrow and then merge/release if everything also looks fine for me locally.

I actually totally forgot I added a benchmark command to the Rakefile 😆

Thanks for your work!

[mgruner]

Thanks @dfherr!

[mgruner]

@dfherr there is one place that I probably overlooked. The invocation of previous_multiline still uses the old scheme with first empty line. Would you have a recommendation for fixing this as well?

[mgruner]

@dfherr added in b5ccb45, hope this works for you.

[mgruner]

@dfherr can you perhaps give an estimate when you may be able to merge and release this fix? 🙏

[dfherr]

So, got some time to run local tests and check out the profiler. Performance hit seems to be somewhere in the ballpark of 1-2%. That's certainly not too bad, so I'm gonna go ahead and merge.

Thanks for your contribution!