jy-data-extract/udpfilterdup/src

基于内存的信息重复过滤

{

"udpport": ":11485",
"dupdays": 7,
"mongodb": {
    "addr": "172.17.4.187:27083",
    "pool": 10,
    "db": "qfw",
    "extract": "result_file_20200410",
    "extract_back": "result_file_20200409",
    "site": {
        "dbname": "qfw",
        "coll": "site"
    }
},
"jkmail": {
    "to": "zhangjinkun@topnet.net.cn",
    "api": "http://10.171.112.160:19281/_send/_mail"
},
"nextNode": [
    {
        "addr": "172.17.145.179",
        "port": 1782,
        "stype": "project",
        "memo": "合并项目"
    },
    {
        "addr": "127.0.0.1",
        "port": 1783,
        "stype": "bidding",
        "memo": "创建招标数据索引new"
    }
],

"threads": 1,
"isMerger": false,
"isSort":false,
"lowHeavy":false,
"timingTask":true,
"timingSpanDay": 3,
"timingPubScope": 720,
"specialwords": "(重招|重新招标|勘察|设计|施工|监理|总承包|土石方|可研)",
"specialtitle_0": "(包|标段|标包)[(（]?[0-9a-zA-Z一二三四五六七八九十零１２３４５６７８９][)）]?",
"specialtitle_1": "[0-9a-zA-Z一二三四五六七八九十零１２３４５６７８９](次|包|标段|标包|批)",
"specialtitle_2": "项目[(（][0-9a-zA-Z一二三四五六七八九十零１２３４５６７８９][)）]",
"beifen": "[(（]?[0-9一二三四五六七八九十零１２３４５６７８９再][)）]?[子分]?[次批标包]|重招|重新招标|勘察|设计|施工|监理|总承包|土石方|可研"

}

{

"udpport": ":1785",
"dupdays": 5,
"mongodb": {
    "addr": "172.17.4.187:27083",
    "pool": 5,
    "db": "qfw",
    "extract": "result_file_20200410",
    "extract_back": "result_file_20200409",
    "site": {
        "dbname": "qfw",
        "coll": "site"
    }
},
"jkmail": {
    "to": "zhengkun@topnet.net.cn,zhangjinkun@topnet.net.cn",
    "api": "http://10.171.112.160:19281/_send/_mail"
},
"nextNode": [
    {
        "addr": "172.17.145.179",
        "port": 1782,
        "stype": "project",
        "memo": "合并项目"
    },
    {
        "addr": "127.0.0.1",
        "port": 1783,
        "stype": "bidding",
        "memo": "创建招标数据索引new"
    }
],
"threads": 1,
"isMerger": false,
"isSort":true,
"lowHeavy":false,
"timingTask":false,
"timingSpanDay": 3,
"timingPubScope": 720,
"specialwords": "(重招|重新招标|勘察|设计|施工|监理|总承包|土石方|可研)",
"specialtitle_0": "(包|标段|标包)[(（]?[0-9a-zA-Z一二三四五六七八九十零１２３４５６７８９][)）]?",
"specialtitle_1": "[0-9a-zA-Z一二三四五六七八九十零１２３４５６７８９](次|包|标段|标包|批)",
"specialtitle_2": "项目[(（][0-9a-zA-Z一二三四五六七八九十零１２３４５６７８９][)）]",
"beifen": "[(（]?[0-9一二三四五六七八九十零１２３４５６７８９再][)）]?[子分]?[次批标包]|重招|重新招标|勘察|设计|施工|监理|总承包|土石方|可研"

}