Beispiel #1
0
    def fetch(self, **kwargs):
        """Fetch resource from database"""
        def extract_x(obj):
            if not obj.get('description'):
                return
            val = obj.get('description')
            if val and isinstance(val, Binary):
                val = gzip.decompress(val.value)
            in_x.append(val)
            iid.append(obj.get('entity_id'))

        config = get_config()
        chunksize = int(
            getattr(config, 'producer_chunksize', Default.PRODUCER_CHUNKSIZE))

        brk = get_brk(OrganizationTable.ID)
        gen = brk._iscan(chunksize=chunksize, \
                filter_expr=Attr('data_type').eq(OrganizationTable.DT.value))

        for chunk in gen:
            if not chunk:
                continue
            chunk = chunk.get('Items')
            iid, in_x = [], []

            list(map(extract_x, chunk))
            if iid and in_x:
                self.total += len(in_x)
                print(self, self.total)
                yield {
                    'id': iid,
                    'input_x': in_x,
                }
Beispiel #2
0
    def fetch(self, **kwargs):
        """Fetch resource from database"""
        def extract_x(obj):
            if not obj.get('abstract'):
                return
            iid.append(obj.get('npl_id'))
            in_x.append(obj.get('abstract'))

        config = get_config()
        chunksize = int(
            getattr(config, 'producer_chunksize', Default.PRODUCER_CHUNKSIZE))

        brk = get_brk(JournalArticle.ID)
        gen = brk._iscan(chunksize=chunksize, \
                filter_expr=Attr('data_type').eq(JournalArticle.DT.value))

        for chunk in gen:
            if not chunk:
                continue
            chunk = chunk.get('Items')
            iid, in_x = [], []

            list(map(extract_x, chunk))
            if iid and in_x:
                self.total += len(in_x)
                print(self, self.total)
                yield {
                    'id': iid,
                    'input_x': in_x,
                }
Beispiel #3
0
    def get_result(self, raw_code):
        brk_dict = get_brk(TechClassi.ID)
        cla = brk_dict._query(
            Key(TechClassi.ID.value).eq(
                TechClassi.make_id(raw_code))).get('Items')

        if not cla:
            print('-- [error] TechClassi items not found for code:{}'.format(
                raw_code))
            return

        cla = cla[0]
        return {
            'tech_domain_classifications': [{
                TechClassi.ID.value: raw_code,
                "name": cla['name'],
            }]
        }
Beispiel #4
0
    def send(self, **kwargs):
        self.recv_cnt += 1
        iid = kwargs.get('id')
        d6_code = kwargs['d6code']

        try:
            brk = get_brk(OrganizationTable.ID)
            res = self.get_result(d6_code)
            if not res:
                return

            print('++ [classi] {} {}'.format(res, self.recv_cnt))
            brk._update({
                OrganizationTable.ID.value: iid,
            }, res)
            print('++ [updated] {} {}'.format(iid, d6_code))
        except Exception as ex:
            print('-- [error] send failed iid:{} code:{}: {}'.format(
                iid, d6_code, ex))
Beispiel #5
0
    def send(self, **kwargs):
        self.recv_cnt += 1
        iid = kwargs.get('id')
        cip = kwargs['cip']

        try:
            table, _ = JournalArticle.rebuild(id=iid)
            brk = get_brk(JournalArticle.ID)
            res = self.get_result(cip)
            if not res:
                return

            print('++ [classi] {} {}'.format(res, self.recv_cnt))
            brk._update({
                JournalArticle.ID.value: iid,
            }, res)
            print('++ [updated] {} {}'.format(iid, cip))
        except Exception as ex:
            print('-- [error] send failed iid:{} code:{}: {}'.format(
                iid, cip, ex))