You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
There are many breaking spaces (see https://en.wikipedia.org/wiki/Whitespace_character) and only a handful of non-breaking spaces. However, in a data format like PENMAN it seems reasonable to only allow standard ASCII spaces as breaking, primarily space, tab, line feed, and carriage return, but maybe also form feed and vertical tab. All others, including non-breaking space, ideographic space, zero-width word joiner, etc., are then treated as regular characters.
Consider this:
>>> penman.decode('(a/a :name (b / 1\u00a02\u00a03))') Traceback (most recent call last): [...] penman.exceptions.DecodeError: line 1 (a/a :name (b / 1 2 3)) ^ DecodeError: Expected: ROLE
This example is adapted from jcyk/AMR-gs#15.
Should non-breaking spaces not be broken during parsing? If so, what other spaces should be allowed, or rather, which should be disallowed in symbols?
The text was updated successfully, but these errors were encountered: