强制spaCy引理为小写

import spacy
from spacy.pipeline import Pipe

class LowercaseLemmatizer(Pipe):
    def __init__(self, nlp, name="lowercase_lemmatizer"):
        super().__init__(nlp, name)
        self.lemmatizer = nlp.get_pipe("lemmatizer")

    def __call__(self, doc):
        for token in doc:
            token._.lowercase_lemma = token.lemma_.lower()
        return doc

# 加载spaCy模型
nlp = spacy.load("en_core_web_sm")

# 添加自定义管道组件
nlp.add_pipe(LowercaseLemmatizer(nlp), after="lemmatizer")

# 测试
text = "Running and jumping are fun activities."
doc = nlp(text)
for token in doc:
    print(f"{token.text}: {token._.lowercase_lemma}")