model_script/udf/UDF.py

# coding:utf-8
from odps.udf import annotate
import json


@annotate("string,string,string->bigint")
class kv_in_json_array_position(object):
    def evaluate(self, json_array_str, key, value):

        if json_array_str is None or json_array_str == '':
            return -1
        index = 0
        json_array = json.loads(json_array_str)
        for json_item in json_array:
            if key in json_item and str(json_item[key]) == value:
                return index
            index += 1
        return -1


@annotate("string,string,string->bigint")
class kv_in_json_array_item(object):
    def evaluate(self, json_array_str, key, value):

        if json_array_str is None or json_array_str == '':
            return ""
        json_array = json.loads(json_array_str)
        for json_item in json_array:
            if key in json_item and str(json_item[key]) == value:
                return json.dumps(json_item)
        return ""


@annotate("string->string")
class json_array_path(object):

    def evaluate(self, json_str):
        if json_str is None or json_str == '':
            return None
        try:
            arr = []
            json_array = json.loads(json_str)
            for data in json_array:
                if 'creativeId' in data:
                    arr.append(str(data['creativeId']))
            return ",".join(arr)
        except Exception:
            return "Internal error"