multiword (contractions)

[arademaker]

http://universaldependencies.org/format.html

The read and write functions still do not deal with the multiword lines.

[arademaker]

See UniversalDependencies/UD_Portuguese-Bosque#23 (comment) how to encode the fusions and contractions ? @fcbr my idea would be to add one more slot in the sentence class for the list of contractions, what do you think ?

[arademaker]

BTW, @fcbr note that UD_Portuguese is not following the documentation, since the lines should contain 10 fields as all other tokens

[arademaker]

@GPPassos can you and João try next to solve this issue?

[fcbr]

That seems a good suggestion.

[arademaker]

I have simplified the code and tested it, closed by addd6fd

[arademaker]

The final test was:

$ sbcl --eval '(ql:quickload :cl-conllu)' --eval '(in-package :cl-conllu)'  --eval '(write-conllu (append (read-conllu "pt-ud-dev.conllu") (read-conllu "pt-ud-test.conllu") (read-conllu "pt-ud-train.conllu")) "lixo.conllu")'
$ cat pt-ud-dev.conllu pt-ud-test.conllu pt-ud-train.conllu > lixo1.conllu
$ diff lixo.conllu lixo1.conllu