search.py 文件源码

def search_func_factory(analyzer, searcher, vm_env):
    """Search function factory"""

    def retrieve(doc):
        return doc.get('ind')

    def search(**kwargs):
        vm_env.attachCurrentThread()
        query = BooleanQuery() 

        print("Searched keywords:")
        for field_name, keywords in kwargs.items():
            # assert field_name in SearchConfig.searchable_fields

            # keywords = list(filter(None, jieba.cut(keywords, cut_all=True)))
            keywords = list(filter(None, (k.strip() for k in jieba.cut_for_search(keywords))))

            for kw in keywords:
                print(kw)

            # construct query
            for kw in keywords:
                q = QueryParser(Version.LUCENE_CURRENT, field_name, analyzer).parse(kw)
                query.add(q, BooleanClause.Occur.SHOULD)

            if field_name == 'keywords':
                for kw in keywords:
                    q = QueryParser(Version.LUCENE_CURRENT, 'ent_name', analyzer).parse(kw)
                    query.add(q, BooleanClause.Occur.SHOULD)

        # search
        scoreDocs = searcher.search(query, 50).scoreDocs

        return [retrieve(searcher.doc(scoreDoc.doc)) for scoreDoc in scoreDocs]

    return search