| 123 | return tokens |
| 124 | |
| 125 | class Parser(_Parser): |
| 126 | |
| 127 | def find_tokens(self, tokens, **kwargs): |
| 128 | # 's in Dutch preceded by a vowel indicates plural ("auto's"): don't replace. |
| 129 | kwargs.setdefault("abbreviations", ABBREVIATIONS) |
| 130 | kwargs.setdefault("replace", {"'n": " 'n"}) |
| 131 | s = _Parser.find_tokens(self, tokens, **kwargs) |
| 132 | s = [re.sub(r"' s (ochtends|morgens|middags|avonds)", "'s \\1", s) for s in s] |
| 133 | return s |
| 134 | |
| 135 | def find_lemmata(self, tokens, **kwargs): |
| 136 | return find_lemmata(tokens) |
| 137 | |
| 138 | def find_tags(self, tokens, **kwargs): |
| 139 | if kwargs.get("tagset") in (PENN, None): |
| 140 | kwargs.setdefault("map", lambda token, tag: wotan2penntreebank(token, tag)) |
| 141 | if kwargs.get("tagset") == UNIVERSAL: |
| 142 | kwargs.setdefault("map", lambda token, tag: wotan2universal(token, tag)) |
| 143 | if kwargs.get("tagset") is WOTAN: |
| 144 | kwargs.setdefault("map", lambda token, tag: (token, tag)) |
| 145 | return _Parser.find_tags(self, tokens, **kwargs) |
| 146 | |
| 147 | class Sentiment(_Sentiment): |
| 148 |
no outgoing calls
no test coverage detected
searching dependent graphs…