prepare.py

from whoosh.fields import TEXT, SchemaClass, ID
from jieba.analyse import ChineseAnalyzer
from whoosh.index import create_in
import json

analyzer = ChineseAnalyzer()
class ArticleSchema(SchemaClass):
    index = ID(stored=True)
    原文 = TEXT(stored=True, analyzer=analyzer)
    注释 = TEXT(stored=True, analyzer=analyzer)
    批判 = TEXT(stored=True, analyzer=analyzer)
    章节 = TEXT(stored=True, analyzer=analyzer)

schema = ArticleSchema()
ix = create_in("indexdir", schema, indexname='article_index')
writer = ix.writer()


with open("反孔.json", encoding="utf-8") as json_file:
    raw_jsons = json.load(json_file)

for vhjx_item in raw_jsons:
    for jvvi_item in vhjx_item[1:]:
        print(jvvi_item['index'])
        writer.add_document(index=jvvi_item['index'], 原文=jvvi_item['原文'], 
                            注释=jvvi_item['注释'] if "注释" in jvvi_item else "", 批判=jvvi_item['批判'] if "批判" in jvvi_item else "", 章节=vhjx_item[0])
writer.commit()