Problem with new version (3.5.2)

[Arman Cohan]

Does not work properly with corenlp 3.5.2.

At least the words are not parsed correctly.

Only one word is detected in the list of words.

[Jeremy B. Merrill]

I have this issue too. The issue appears to be that parse_parser_results improperly switches to STATE_TREE after the first word.

Maybe the old corenlp had all the [TEXT= word items on one line? I don't really know, but now they're on separate lines.

I have a hotfix in my branch (https://bitbucket.org/jeremybmerrill/corenlp-python/commits/5d818030b89532a390a5ab3347c73ae515989093) that fixes the problem, but it's ugly. @torotoki: Happy to send a pull request if there's not a better way to fix the parsing.

[Feng Mai]

I can confirm that Jeremy's solution works.