spacy_pipeline_language_detection_with_spacy_langdetect.py

1import spacy
2from spacy.language import Language
3from spacy_language_detection import LanguageDetector
4
5def get_lang_detector(nlp, name):
6    return LanguageDetector()
7
8nlp = spacy.load("en_core_web_sm")
9Language.factory("language_detector", func=get_lang_detector)
10nlp.add_pipe("language_detector", last=True)
11
12text = "This is an English sentence. Todo bien con el español."
13doc = nlp(text)
14
15# document level language detection
16print(doc._.language)
17
18# sentence level language detection
19for sent in doc.sents:
20    print(sent, sent._.language)