예제 #1
0
    def _parse_line(self, item):
        sline = StatementLine()

        accountid = item[u"Számla neve"]
        currency = item[u"Tranzakció devizaneme"]

        stmt = self._find_or_create_statement(accountid, currency)

        sline.date = datetime.strptime(item[u"Értéknap"], self.date_format)
        sline.date_user = datetime.strptime(item[u"Tranzakció dátuma"], self.date_format)
        sline.amount = atof(item[u"Tranzakció összege"])
        trtype = item[u"Tranzakció típusa"]
        if trtype == "VÁSÁRLÁS" or "JÓVÁÍRÁS":
            sline.payee = item[u"Részletek"].strip()
        else:
            sline.payee = item[u"Partner neve/Másodlagos azonosító típusa"].strip()
            sline.memo = item[u"Részletek"].strip()

        sline.id = generate_transaction_id(sline)

        sline.assert_valid()
        stmt.lines.append(sline)
예제 #2
0
    def parse(self):
        DATA_DATE = 'data_date'
        DATA_AMT = 'data_amt'
        DATA_DSC = 'data_dsc'
        DATA_MEMO = 'data_mem'
        DATA_BAL = 'data_bal'

        def parser_old(filename):
            def parse_tr(tr):
                def get_header(tr, n):
                    return str(
                        tr.find_all(headers='header' + str(n))[0].string)

                hdrs = tr.find_all(headers=True)
                assert len(hdrs) in [0, 7]
                if len(hdrs) == 0:
                    cont = True
                    date_or_comm = str(tr.find_all(colspan='5')[0].string)
                    desc = None
                    paym = None
                    depo = None
                    bal = None
                else:
                    cont = False
                    date_or_comm = dateutil.parser.parse(get_header(tr, 1),
                                                         dayfirst=True)
                    desc = get_header(tr, 2)

                    paym = self.get_float(get_header(tr, 5))
                    depo = self.get_float(get_header(tr, 6))

                    assert (depo is None) != (
                        paym is None
                    ), 'Either depo or paym need to exist, but not both: ' + str(
                        tr)

                    bal = self.get_float(get_header(tr, 7))

                return cont, date_or_comm, desc, paym, depo, bal

            with open(filename, 'r', encoding='iso-8859-8') as f:
                soup = BeautifulSoup(f, 'lxml')
                statement = Statement(currency='ILS')
                data = []
                for tr in soup.find_all(
                        'table',
                        id='mytable_body')[0].find_all(id='TR_ROW_BANKTABLE'):
                    cont, date_or_comm, desc, paym, depo, bal = parse_tr(tr)
                    if cont:
                        data[-1][DATA_MEMO] = date_or_comm
                    else:
                        new_line = {}

                        new_line[DATA_DATE] = date_or_comm
                        new_line[DATA_DSC] = desc
                        new_line[
                            DATA_AMT] = -paym if paym is not None else depo
                        new_line[DATA_BAL] = bal
                        new_line[DATA_MEMO] = None

                        data.append(new_line)
                return data

        def parser_new(filename):
            with open(filename, 'r', encoding='iso-8859-8') as f:
                bs = BeautifulSoup(f, 'lxml')
            trs = bs.find_all('table', i__d='trBlueOnWhite12')[0].find_all(
                'tr', recursive=False)

            data = []
            for tr in trs[1:]:
                new_line = {}
                tds = tr.find_all('td')
                if len(tds) != 7:
                    continue
                new_line[DATA_DATE] = dateutil.parser.parse(str(tds[0].string),
                                                            dayfirst=True)
                new_line[DATA_DSC] = str(tds[1].string)

                paym = self.get_float(str(tds[4].string))
                depo = self.get_float(str(tds[5].string))
                assert (depo is None) != (
                    paym is None
                ), 'Either depo or paym need to exist, but not both: ' + str(
                    tds)
                new_line[DATA_AMT] = -paym if paym is not None else depo

                new_line[DATA_BAL] = self.get_float(str(tds[6].string))

                new_line[DATA_MEMO] = None

                data.append(new_line)

            return data

        def parser_xslx(filename):
            with open(filename, 'rb') as f:
                ws = openpyxl.load_workbook(f).worksheets[0]

            iterrows = iter(ws.rows)
            for _ in range(6):
                next(iterrows)

            data = []
            for row in iterrows:
                new_line = {}
                new_line[DATA_DATE] = dateutil.parser.parse(str(row[0].value),
                                                            dayfirst=False)
                new_line[DATA_DSC] = str(row[1].value)

                paym = self.get_float(str(row[3].value))
                depo = self.get_float(str(row[4].value))
                assert (depo is None) != (
                    paym is
                    None), 'Either depo or paym need to exist, but not both: '
                new_line[DATA_AMT] = -paym if paym is not None else depo

                new_line[DATA_BAL] = self.get_float(str(row[5].value))

                new_line[DATA_MEMO] = " ".join(
                    [str(c.value) for c in row[7:] if c.value is not None])

                data.append(new_line)

            return data

        PARSERS = [parser_old, parser_new, parser_xslx]

        v = self.detect_version()
        assert v is not None, "Unsupported file %s" % self.filename
        self.log("Detected file of version %d" % v)

        data = PARSERS[v](self.filename)
        self.log('Found %d transactions' % len(data))

        stmnt = Statement(currency='ILS')
        for d in data:
            stmt_line = StatementLine(date=d[DATA_DATE], amount=d[DATA_AMT])
            #stmt_line.end_balance = d[DATA_BAL] #TODO: conf
            stmt_line.payee = d[DATA_DSC]
            if d[DATA_MEMO] is not None:
                stmt_line.memo = d[DATA_MEMO]

            if stmt_line.payee.startswith('משיכה'):
                stmt_line.trntype = 'ATM'
            elif stmt_line.payee.startswith('שיק'):
                stmt_line.trntype = 'CHECK'
            else:
                stmt_line.trntype = "CASH" if d[DATA_AMT] < 0 else "DEP"

            stmt_line.assert_valid()
            stmnt.lines.append(stmt_line)
        return stmnt
예제 #3
0
    def parse(self):
        DATA_DATE = 'data_date'
        DATA_AMT = 'data_amt'
        DATA_DSC = 'data_dsc'
        DATA_MEMO = 'data_mem'
        DATA_BAL = 'data_bal'

        def parser_old(filename):

            def parse_tr(tr):
                def get_header(tr, n):
                    return str(tr.find_all(headers='header'+str(n))[0].string)

                hdrs = tr.find_all(headers=True)
                assert len(hdrs) in [0, 7]
                if len(hdrs) == 0:
                    cont = True
                    date_or_comm = str(tr.find_all(colspan='5')[0].string)
                    desc = None
                    paym = None
                    depo = None
                    bal = None
                else:
                    cont = False
                    date_or_comm = dateutil.parser.parse(get_header(tr, 1), dayfirst=True)
                    desc = get_header(tr, 2)

                    paym = self.get_float(get_header(tr, 5))
                    depo = self.get_float(get_header(tr, 6))

                    assert (depo is None) != (paym is None), 'Either depo or paym need to exist, but not both: ' + str(tr)

                    bal = self.get_float(get_header(tr, 7))

                return cont, date_or_comm, desc, paym, depo, bal

            with open(filename, 'r', encoding='iso-8859-8') as f:
                soup = BeautifulSoup(f, 'lxml')
                statement = Statement(currency='ILS')
                data = []
                for tr in soup.find_all('table', id='mytable_body')[0].find_all(id='TR_ROW_BANKTABLE'):
                    cont, date_or_comm, desc, paym, depo, bal = parse_tr(tr)
                    if cont:
                        data[-1][DATA_MEMO] = date_or_comm
                    else:
                        new_line = {}

                        new_line[DATA_DATE] = date_or_comm
                        new_line[DATA_DSC] = desc
                        new_line[DATA_AMT] = -paym if paym is not None else depo
                        new_line[DATA_BAL] = bal
                        new_line[DATA_MEMO] = None

                        data.append(new_line)
                return data

        def parser_new(filename):
            with open(filename, 'r', encoding='iso-8859-8') as f:
                bs = BeautifulSoup(f, 'lxml')
            trs = bs.find_all('table', i__d='trBlueOnWhite12')[0].find_all('tr', recursive=False)

            data = []
            for tr in trs[1:]:
                new_line = {}
                tds = tr.find_all('td')
                if len(tds) != 7:
                    continue
                new_line[DATA_DATE] = dateutil.parser.parse(str(tds[0].string), dayfirst=True)
                new_line[DATA_DSC] = str(tds[1].string)

                paym = self.get_float(str(tds[4].string))
                depo = self.get_float(str(tds[5].string))
                assert (depo is None) != (paym is None), 'Either depo or paym need to exist, but not both: ' + str(tds)
                new_line[DATA_AMT] = -paym if paym is not None else depo

                new_line[DATA_BAL] = self.get_float(str(tds[6].string))

                new_line[DATA_MEMO] = None

                data.append(new_line)

            return data

        def parser_xslx(filename):
            with open(filename, 'rb') as f:
                ws = openpyxl.load_workbook(f).worksheets[0]

            iterrows = iter(ws.rows)
            for _ in range(6):
                next(iterrows)

            data = []
            for row in iterrows:
                new_line = {}
                new_line[DATA_DATE] = dateutil.parser.parse(str(row[0].value), dayfirst=False)
                new_line[DATA_DSC] = str(row[1].value)

                paym = self.get_float(str(row[3].value))
                depo = self.get_float(str(row[4].value))
                assert (depo is None) != (paym is None), 'Either depo or paym need to exist, but not both: '
                new_line[DATA_AMT] = -paym if paym is not None else depo

                new_line[DATA_BAL] = self.get_float(str(row[5].value))

                new_line[DATA_MEMO] = " ".join([str(c.value) for c in row[7:] if c.value is not None])

                data.append(new_line)

            return data
        PARSERS = [parser_old, parser_new, parser_xslx]

        v = self.detect_version()
        assert v is not None, "Unsupported file %s" % self.filename
        self.log("Detected file of version %d" % v)

        data = PARSERS[v](self.filename)
        self.log('Found %d transactions' % len(data))

        stmnt = Statement(currency='ILS')
        for d in data:
            stmt_line = StatementLine(date=d[DATA_DATE], amount=d[DATA_AMT])
            #stmt_line.end_balance = d[DATA_BAL] #TODO: conf
            stmt_line.payee = d[DATA_DSC]
            if d[DATA_MEMO] is not None:
                stmt_line.memo = d[DATA_MEMO]

            if stmt_line.payee.startswith('משיכה'):
                stmt_line.trntype = 'ATM'
            elif stmt_line.payee.startswith('שיק'):
                stmt_line.trntype = 'CHECK'
            else:
                stmt_line.trntype = "CASH" if d[DATA_AMT] < 0 else "DEP"

            stmt_line.assert_valid()
            stmnt.lines.append(stmt_line)
        return stmnt