Support '\r' and '\r\n' line endings, closes scala/bug#5669

[Philippus]

No description provided.

[javax-swing]

may as well use for i <- 1 until source.length and remove the if i>= 1 checks below

[javax-swing]

the majority of the time the last character won't be a \r or \n might be worth binding charAt(i - 1) to a variable to save the extra lookup (dependant on the performance requirements of this code)

[SethTisue]

in the Scala community build, this build caused this Twirl test case to start failing: https://github.com/playframework/twirl/blob/68ad7b378be54887a09c5a67792bacc933d17d68/parser/src/test/scala/play/twirl/parser/test/ParserSpec.scala#L187-L189 (referencing https://github.com/playframework/twirl/blob/master/parser/src/test/resources/unclosedBracket2.scala.html)

[play-twirl] [info]   x unclosedBracket2.scala.html
[play-twirl] [error]      '31' is not equal to '32' (ParserSpec.scala:37)

the test file doesn't have any \r or \r\n line endings, so it seems probable it's a real regression? wdyt?

[Philippus]

I've reproduced it:
contents of the index generated by genIndex differs:
new: ArrayBuffer(0, 85, 86, 111, 112, 136, 155, 166, 167, 169, 170, 185, 186, 207, 208, 233, 234, 243, 265, 266, 291, 292, 310, 335, 336, 346, 347, 356, 357, 367, 368, 370)
vs
old: ArrayBuffer(0, 85, 86, 111, 112, 136, 155, 166, 167, 169, 170, 185, 186, 207, 208, 233, 234, 243, 265, 266, 291, 292, 310, 335, 336, 346, 347, 356, 357, 367, 368, 370, 370)
will work on a fix.