Example #1
0
    def get_all_value_for_key(key):
        list = []

        mysql = MysqlHelper(Configs())

        sql = "select `" + str(
            key) + "` from news where is_scrapped = 1 and " + str(
                key) + " is not null"

        result = mysql.load(sql)

        for item in result:
            list.append(item[str(key)])

        if len(list) < 1:
            list.append('QWERTYUIOP')

        return list
Example #2
0
    def get_target_list(self):
        list = []

        mysql = MysqlHelper(Configs())
        sql = "select * from " + self.table_name + " where " + self.TB.valid.key + " = 1 order by " + self.TB.order_code.key + " desc"

        result = mysql.load(sql)

        for item in result:

            target = Target()

            for key in item:
                target.__dict__[key] = item[key]

            list.insert(0, target)

        return list
Example #3
0
def get_target_list():
    list = []

    mysql = MysqlHelper(Configs())
    # sql = "select * from scraping_target where valid = 1 and " + TBScrapingTarget.type.key + " = " + "'newrank'" + " order by order_code desc"
    sql = "select * from scraping_target where valid = 1 " + " order by order_code desc"

    result = mysql.load(sql)

    for item in result:

        target = Target()

        for key in item:
            target.__dict__[key] = item[key]

        list.insert(0, target)

    return list
Example #4
0
    def save_data_insert(dic):

        id = genUUID()

        mysql = MysqlHelper(Configs())

        _sql = "INSERT INTO news(`id`, "
        _val = " VALUES (%s,"
        _par = [
            id,
        ]

        for key in dic:
            _sql += "`"
            _sql += key
            _sql += "`,"

            if type(dic[key]) == type(str):
                _val += "%s,"
            elif type(dic[key]) == type(int):
                _val += "%i,"
            else:
                _val += "%s,"

            _par.append(dic[key])

        _sql = _sql[0:len(_sql) - 1]
        _sql += ")"

        _val = _val[0:len(_val) - 1]
        _val += ")"

        _sql += _val

        mysql.save(_sql, _par)

        sql = "insert into "
        sql += article_table_name
        sql += " set id=%s"
        mysql.save(sql, id)
Example #5
0
    def par_to_get_dic(select_list, and_dic, order_key=None, limit_tup=None):
        list = []
        select_sql = ScrappdeDataDao.list_to_sql_part(select_list)
        and_sql = ScrappdeDataDao.dic_to_sql_part(and_dic)
        if order_key != None:
            order_key = ' order by ' + order_key[0] + ' ' + order_key[1]
        else:
            order_key = ''
        if limit_tup != None:
            limit_tup = ' limit ' + str(limit_tup[0]) + ',' + str(limit_tup[1])
        else:
            limit_tup = ''

        mysql = MysqlHelper(Configs())
        sql = "select " + select_sql + " from " + tableName + " where is_scrapped = 1 and " + and_sql + " " + order_key + limit_tup

        result = mysql.load(sql)

        for item in result:
            list.append(item)

        return list
Example #6
0
    def get_last_access_date(self, id):
        list = []

        mysql = MysqlHelper(Configs())
        sql = "select " + self.TB.last_access_date.key + " from " + self.table_name + " where " + self.TB.id.key + " = '" + id + "' limit 0,1"

        result = mysql.load(sql)

        if len(result) > 0:
            item = result[0]
            item = item[self.TB.last_access_date.key]

            # print(isinstance(item, (datetime.datetime, bool)))
            # print(type(item))
            # print(type(datetime.datetime.date()))

            if isinstance(item, (datetime.datetime, bool)):
                return item
            else:
                return DateGo.get_current_date_raw()
        else:
            return DateGo.get_current_date_raw()
Example #7
0
def get_last_access_date(id):
    list = []

    mysql = MysqlHelper(Configs())
    sql = "select last_access_date from scraping_target where id = '" + id + "' limit 0,1"

    result = mysql.load(sql)

    if len(result) > 0:
        item = result[0]
        item = item['last_access_date']

        # print(isinstance(item, (datetime.datetime, bool)))
        # print(type(item))
        # print(type(datetime.datetime.date()))

        if isinstance(item, (datetime.datetime, bool)):
            return item
        else:
            return DateGo.get_current_date_raw()
    else:
        return DateGo.get_current_date_raw()
Example #8
0
    def save_data(dic):
        dis = ""

        id = genUUID()

        for key in dic:
            dis += key
            dis += "='"
            dis += str(dic[key])
            dis += "',"

        dis = dis[0:len(dis) - 1]

        mysql = MysqlHelper(Configs())

        sql = "insert into news set id=%s,"
        sql += dis
        mysql.save(sql, id)

        sql = "insert into "
        sql += article_table_name
        sql += " set id=%s"
        mysql.save(sql, id)
Example #9
0
    def loops(self,target,exists,result):
        try:
            base_url = target.extra0
            for i in range(0, Configs().length_weibo):  # [::-1]:
                print("page: " + str(i))

                list = self.build_and_request(WeiboRuler.keys, base_url, WeiboRuler.request_getindex, i)

                if len(list) == 0:
                    break
                for item in list:
                    """日常抓取时的重复验证"""
                    if exists.count(item['id']) < 1:
                        result.append(item)
                    else:
                        return
        except Exception as e:
            import traceback
            msg = traceback.format_exc()
            print(msg)
            LogGo.warning(repr(e))
            LogGo.warning("Scan Failed!")
            return
Example #10
0
    def __init__(self):
        GeneralDao.__init__(self, Configs())

        self.TB = TBProgramPlayCount()
        self.table_name = self.TB.TableName
Example #11
0
def reset_target_valid(target):
    mysql = MysqlHelper(Configs())
    sql = "update scraping_target set valid = 0 where id = '" + target.id + "'"

    mysql.update(sql)
Example #12
0
    def __init__(self):
        GeneralDao.__init__(self, Configs())

        self.TB = TBSoapBlackList
        self.table_name = self.TB.TableName
Example #13
0
    def __init__(self):
        GeneralDao.__init__(self, Configs())

        self.TB = TBMR()
        self.table_name = self.TB.TableName
Example #14
0
    def __init__(self):
        GeneralDao.__init__(self, Configs())

        self.TB = TBWenzhangInfo()
        self.table_name = self.TB.TableName
Example #15
0
    def __init__(self):
        GeneralDao.__init__(self, Configs())

        self.table = TBSpecialTarget()
        self.table_name = self.table.TableName
Example #16
0
    def reset_target_valid(self, target):
        mysql = MysqlHelper(Configs())
        sql = "update " + self.table_name + " set " + self.TB.valid.key + " = 0 where " + self.TB.id.key + " = '" + target.id + "'"

        mysql.update(sql)
Example #17
0
    def __init__(self):
        GeneralDao.__init__(self, Configs())

        self.TB = TBDictionaryType()
        self.table_name = self.TB.TableName
Example #18
0
    def __init__(self):
        GeneralDao.__init__(self, Configs())

        self.TB = TBGlobalTarget()
        self.table_name = self.TB.TableName
Example #19
0
          "requests==2.12.4\r\n"
          "shellescape==3.4.1\r\n"
          "six==1.10.0\r\n"
          "stem==1.6.0\r\n"
          "tzlocal==1.3\r\n"
          "Werkzeug==0.11.15\r\n")
    print(" ")
    print("------ ----   --- ---getIndexNumber--- ------   ---- --")
    print(" ")

    scan = Scanner()

    while True:
        LogGo.info(">>> New Loop -->")

        try:
            scan.start()
        except Exception as e:
            import traceback
            msg = traceback.format_exc()
            LogGo.error(msg)

        if Configs.infinity:
            while not somebody_help.isthataworkday(Configs().work_sequence):
                time.sleep(60 * 60)
        else:
            break

    LogGo.info("-- GH Offline --")
    print("---------------------------------------------")
Example #20
0
    def __init__(self):

        self.TB = TBProgram()
        self.table_name = self.TB.TableName

        GeneralDao.__init__(self, Configs())