Пример #1
0
    def test_normalize(self):
        receipt = None
        # unfortunately, the Receipt constructor calls 'normalize'
        # so we can construct a Receipt object with a blank receipt file
        with open(self.dir_path +
                  "/tests/data/receipts/empty_file.txt") as receipt_file:
            receipt = Receipt(self.config, receipt_file.readlines())

        # then updates the lines from the real receipt file
        with open(self.dir_path +
                  "/tests/data/receipts/sample_text_receipt_to_normalize.txt"
                  ) as receipt_file:
            receipt.lines = receipt_file.readlines()

        expected_list = [
            "all upper case\n",
            "some upper case\n",
            "   white spaces\n",
            "    tab     tab whitespaces\n",
            "99.00\n",
            "trailing whitespaces after this line.\n",
        ]

        # make sure the Receipt is created
        self.assertIsNotNone(receipt)

        # call 'normalize'
        receipt.normalize()

        after_lines = receipt.lines

        # checks that the lines are normalized
        self.assertEqual(expected_list, after_lines)
Пример #2
0
def results_to_json(config, receipt_files):
    for receipt_path in receipt_files:
        with open(receipt_path, encoding="utf8", errors='ignore') as receipt:
            receipt = Receipt(config, receipt.readlines())
            out = open(receipt_path + ".json", "w")
            out.write(receipt.to_json())
            out.close()
Пример #3
0
def ocr_receipts(config, receipt_files):
    """
    :param config: ObjectView
        Parsed config file
    :param receipt_files: [] of str
        List of files to parse
    :return: {}
        Stats about files
    """

    stats = defaultdict(int)

    table_data = [
        ['Path', 'Market', "Date", "Items", "SUM"],
    ]

    if config.results_as_json:
        results_to_json(config, receipt_files)

    for receipt_path in receipt_files:
        with open(receipt_path, encoding="utf8", errors='ignore') as receipt:
            receipt = Receipt(config, receipt.readlines())

            item_list = ""
            for item in receipt.items:
                if not item:
                    continue

                item_list += ' '.join(item) + "\n"

            table_data.append([
                receipt_path, receipt.market, receipt.date, item_list,
                receipt.sum
            ])

            stats["total"] += 1
            if receipt.market:
                stats["market"] += 1
            if receipt.date:
                stats["date"] += 1
            if receipt.sum:
                stats["sum"] += 1

    table = SingleTable(table_data)
    print(table.table)

    return stats
Пример #4
0
    def test_parse(self):
        # not sure there's a need to unit test this one since it essentially wraps other unit tests
        receipt = None
        with open(self.dir_path +
                  "/tests/data/receipts/sample_text_receipt.txt"
                  ) as receipt_file:
            receipt = Receipt(self.config, receipt_file.readlines())

        self.assertIsNotNone(receipt)
Пример #5
0
    def test_fuzzy_find(self):
        """
            verifies fuzzy_find
        """
        receipt = None
        with open(self.dir_path +
                  "/tests/data/receipts/sample_text_fuzzy_find.txt"
                  ) as receipt_file:
            receipt = Receipt(self.config, receipt_file.readlines())
        self.assertIsNotNone(receipt)

        self.assertEqual("restaurant\n", receipt.fuzzy_find("restaurat"))
        self.assertEqual("gas station\n", receipt.fuzzy_find("as statio"))
        self.assertEqual("uber\n", receipt.fuzzy_find("ube"))
        self.assertEqual("lyft\n", receipt.fuzzy_find("ly"))
        self.assertEqual("supermarket\n", receipt.fuzzy_find("market"))
Пример #6
0
    def test_parser_date(self):
        """
            Verifies parse_date functions
            dates like 19.08.15 and 19. 08. 2015
        """

        # test parse_date from file
        receipt = None
        with open(self.dir_path +
                  "/tests/data/receipts/sample_text_receipt_dates.txt"
                  ) as receipt_file:
            receipt = Receipt(self.config, receipt_file.readlines())
        actual_date_str = receipt.date
        print(actual_date_str)
        self.assertEqual("19.08.15\n", actual_date_str)

        # test DD.MM.YY
        receipt2 = Receipt(self.config,
                           ["18.08.16\n", "19.09.17\n", "01.01.18"])
        actual_date_str = receipt2.parse_date()
        print(actual_date_str)
        self.assertEqual("18.08.16\n", actual_date_str)

        # test DD.MM.YYYY
        receipt3 = Receipt(self.config, ["18.08.2016\n"])
        actual_date_str = receipt3.parse_date()
        print(actual_date_str)
        self.assertEqual("18.08.2016\n", actual_date_str)

        # HOWEVER these tests should fail:
        # test with DD > 31
        receipt4 = Receipt(self.config, ["32.08.2016\n"])
        actual_date_str = receipt4.parse_date()
        self.assertIsNone(actual_date_str)

        # test with MM > 12
        receipt5 = Receipt(self.config, ["01.55.2016\n"])
        actual_date_str = receipt5.parse_date()
        print(actual_date_str)
        self.assertIsNone(actual_date_str)

        # test with invalid date: 31.04.15
        receipt6 = Receipt(self.config, ["31.04.15\n"])
        actual_date_str = receipt6.parse_date()
        self.assertIsNone(actual_date_str)

        # And these tests should pass:
        # test with YYYY < 2013
        receipt7 = Receipt(self.config, ["18.08.2012\n"])
        actual_date_str = receipt7.parse_date()
        print(actual_date_str)

        # test with YYYY >= 2017
        receipt8 = Receipt(self.config, ["18.08.2017\n"])
        actual_date_str = receipt8.parse_date()
        print(actual_date_str)
Пример #7
0
    def test_parse_sum(self):
        """
            Verifies parse_sum
        """
        receipt = None
        with open(self.dir_path +
                  "/tests/data/receipts/sample_text_receipt.txt"
                  ) as receipt_file:
            receipt = Receipt(self.config, receipt_file.readlines())
        self.assertIsNotNone(receipt)
        self.assertEqual("0.99", receipt.parse_sum())

        receipt = Receipt(self.config, ["summe   12,99\n"])
        self.assertEqual("12.99", receipt.parse_sum())

        receipt = Receipt(self.config, ["summe   *** 12,99 ***\n"])
        self.assertEqual("12.99", receipt.parse_sum())

        receipt = Receipt(self.config, ["summe   13.99\n"])
        self.assertEqual("13.99", receipt.parse_sum())

        receipt = Receipt(self.config, ["13,99 summe\n"])
        self.assertEqual("13.99", receipt.parse_sum())

        receipt = Receipt(self.config, ["gesamtbetrag 1,99\n"])
        self.assertEqual("1.99", receipt.parse_sum())

        receipt = Receipt(self.config, ["gesamt 2,99\n"])
        self.assertEqual("2.99", receipt.parse_sum())

        receipt = Receipt(self.config, ["total 3,99\n"])
        self.assertEqual("3.99", receipt.parse_sum())

        receipt = Receipt(self.config, ["sum 4,99\n"])
        self.assertEqual("4.99", receipt.parse_sum())

        receipt = Receipt(self.config, ["zwischensumme 5,99\n"])
        self.assertEqual("5.99", receipt.parse_sum())

        receipt = Receipt(self.config, ["bar 1,99\n"])
        self.assertEqual("1.99", receipt.parse_sum())
Пример #8
0
    def test_parse_market(self):
        """
            Verifies receipt_parser_core.parse_market
        """
        receipt = Receipt(self.config, ["penny"])
        print("market", receipt.parse_market())
        self.assertEqual("Penny", receipt.parse_market())

        # should work but fails
        receipt = Receipt(self.config, ["p e n ny"])
        # self.assertEqual("Penny", receipt.parse_market())

        # should work but fails
        receipt = Receipt(self.config, ["m a r k t gmbh"])
        # self.assertEqual("Penny", receipt.parse_market())

        receipt = Receipt(self.config, ["rew"])
        self.assertEqual("REWE", receipt.parse_market())

        receipt = Receipt(self.config, ["REL"])
        self.assertEqual("Real", receipt.parse_market())

        receipt = Receipt(self.config, ["netto-onli"])
        self.assertEqual("Netto", receipt.parse_market())

        receipt = Receipt(self.config, ["kaser"])
        self.assertEqual("Kaiser's", receipt.parse_market())

        receipt = Receipt(self.config, ["kaiserswerther str. 270"])
        self.assertEqual("Kaiser's", receipt.parse_market())

        receipt = Receipt(self.config, ["ALDI"])
        self.assertEqual("Aldi", receipt.parse_market())

        receipt = Receipt(self.config, ["friedrichstr. 128—133"])
        self.assertEqual("Aldi", receipt.parse_market())

        receipt = Receipt(self.config, ["LIDL"])
        self.assertEqual("Lidl", receipt.parse_market())

        receipt = Receipt(self.config, ["shell"])
        self.assertEqual("Tanken", receipt.parse_market())

        receipt = Receipt(self.config, ["esso station"])
        self.assertEqual("Tanken", receipt.parse_market())

        receipt = Receipt(self.config, ["aral"])
        self.assertEqual("Tanken", receipt.parse_market())

        receipt = Receipt(self.config, ["total tankstelle"])
        self.assertEqual("Tanken", receipt.parse_market())

        receipt = Receipt(self.config, ["RK Tankstellen"])
        self.assertEqual("Tanken", receipt.parse_market())