Backported Case2. Left 5 tests failing.

[am-kantox]

I believe failing tests are very opinionated.

  1) test should preserve path chars (Recase.PathCaseTest)
     test/recase_test/path_case_test.exs:21
     Assertion with == failed
     code:  assert convert("/path case") == "/path/case"
     left:  "path/case"
     right: "/path/case"
     stacktrace:
       test/recase_test/path_case_test.exs:22: (test)

  2) test should not modify extra chars (Recase.PathCaseTest)
     test/recase_test/path_case_test.exs:26
     Assertion with == failed
     code:  assert convert("!#$%^&*(){}[]~`'\"") == "!#$%^&*(){}[]~`'\""
     left:  ""
     right: "!#$%^&*(){}[]~`'\""
     stacktrace:
       test/recase_test/path_case_test.exs:27: (test)

  3) test should not modify extra chars (Recase.PascalCaseTest)
     test/recase_test/pascal_case_test.exs:17
     Assertion with == failed
     code:  assert convert("!#$%^&*(){}[]~`'\"") == "!#$%^&*(){}[]~`'\""
     left:  ""
     right: "!#$%^&*(){}[]~`'\""
     stacktrace:
       test/recase_test/pascal_case_test.exs:18: (test)



  4) test should camel case usual text (Recase.CamelCaseTest)
     test/recase_test/camel_case_test.exs:8
     Assertion with == failed
     code:  assert convert("camel-casE") == "camelCase"
     left:  "camelCasE"
     right: "camelCase"
     stacktrace:
       test/recase_test/camel_case_test.exs:16: (test)

...

  5) test should not modify extra chars (Recase.CamelCaseTest)
     test/recase_test/camel_case_test.exs:19
     Assertion with == failed
     code:  assert convert("!#$%^&*(){}[]~`'\"") == "!#$%^&*(){}[]~`'\""
     left:  ""
     right: "!#$%^&*(){}[]~`'\""
     stacktrace:
       test/recase_test/camel_case_test.exs:20: (test)

1: I doubt I follow why "-foo-bar" → "foo/bar" but "/foo-bar" → "/foo/bar". It’s plain inconsistent (easy to fix though.)
2, 3, 5: If we treat # as a delimiter, it should go away here and there. Use config option to specify delimiters explicitly.
4: How last E becomes lowercased? Why?

[coveralls]

Coverage decreased (-4.08%) to 95.918% when pulling fa24209 on am-kantox:feature/full-unicode-support into d28dbc1 on sobolevn:master.

[am-kantox]

This treats digits as delimiters as well; maybe it should eliminate digits as well as letters from the list of delimiters.

[sobolevn]

@am-kantox thanks for this PR!

Now, straight to the point:
1: My idea behind "/foo-bar" → "/foo/bar" was that paths start with / frequently. And it should be preserved. That's why it is not in consistency with other checks. Do you agree on this one?

2,3,5: I guess am just mixing slugs and cases here. It is not really a point of this library to remove any special chars from the strings. Except the explicit separators. If you need to strip these chars, use slugs. So, I agree with you on this one. And I will update docs on making slugs.

4: I guess it is a bug, not sure why this is happening.

[am-kantox]

I have fixed a path. Also, I made an explicit set of symbols to be treated as delimiters (confirming your testcase.)

[am-kantox]

I am unsure what’s wrong with coverage, it points to @doc so I am a bit confused.

[sobolevn]

Awesome work @am-kantox! And by "awesome" I mean really awesome.

[sobolevn]

What are "stop words" in this context? We need to document that. Or remove if it does not matter.

[am-kantox]

Eh. There is a link to Wiki https://en.wikipedia.org/wiki/Letter_case#Title_case saying “fruits and vegetables” should be titleized as “Fruits and Vegetables.”

We do not support and remaining all lowercase. That said, we’d better off removing both warning and link to Wiki.

[sobolevn]

Sound fine, let's do it.

[am-kantox]

Not in this PR I guess.

[sobolevn]

@am-kantox thanks again! I highly appreciate your work.