Question

我是Elasticsearch（5.5.2）的新手，并且正在尝试调整查询（within this python package）以提供更符合我的需求的结果。在软件包中，我在python中具有以下功能：

def query_geonames(self, placename):
    """
    Wrap search parameters into an elasticsearch query to the geonames index
    and return results.

    Parameters
    ---------
    conn: an elasticsearch Search conn, like the one returned by `setup_es()`

    placename: str
        the placename text extracted by NER system

    Returns
    -------
    out: The raw results of the elasticsearch query
    """
    # first first, try for country name
    if self.is_country(placename):
        q = {"multi_match": {"query": placename,
                             "fields": ['name', 'asciiname', 'alternativenames'],
                            "type" : "phrase"}}
        r = Q("match", feature_code='PCLI')
        res = self.conn.query(q).query(r)[0:5].execute()  # always 5
        #self.country_exact = True

    else:
        # second, try for an exact phrase match
        q = {"multi_match": {"query": placename,
                             "fields": ['name^5', 'asciiname^5', 'alternativenames'],
                            "type" : "phrase"}}
        r = Q("match", feature_code='PPL')
        res = self.conn.query(q).query(r)[0:50].execute()

        # if no results, use some fuzziness, but still require all terms to be present.
        # Fuzzy is not allowed in "phrase" searches.
        if res.hits.total == 0:
            # tried wrapping this in a {"constant_score" : {"query": ... but made it worse
            q = {"multi_match": {"query": placename,
                                 "fields": ['name', 'asciiname', 'alternativenames'],
                                     "fuzziness" : 1,
                                     "operator":   "and"},
                    }
            #self.fuzzy = True  # idea was to preserve this info as a feature, but not using state like this
            res = self.conn.query(q)[0:50].execute()
    es_result = utilities.structure_results(res)
    return es_result

我想基于 feature_code 过滤从ES（在else语句的第一部分中）获得的结果。但是，我不会像现在的两个搜索那样对一个feature_code进行过滤，而是将结果来自一组受限制的feature_codes。

我尝试使用：

r = Q("match", feature_code=['PPL','PPLA','PPLA2'])，但这是不正确的。

我看过this help page以及at this discussion。但是仍然无法弄清楚如何同时使用 multi_match 和过滤器。困难的部分原因是我对ES缺乏经验，部分原因是我不完全了解self.conn.query(q).query(r)语句中的第一个查询如何与第二个查询交互。

此外，我相信我的文档/数据库中可能还有其他可能与我想使用的与feature_code类似的相关字段/条件。（特别是，我认为many of these fields应该可以使用。）如何从ES文档/数据库中获取其他字段的确切语法？

如何更改/过滤Elasticsearch查询？

0 个答案: