diff --git a/src/paperless_tesseract/tests/test_date.py b/src/paperless_tesseract/tests/test_date.py index 645cb70ff..096a1c147 100644 --- a/src/paperless_tesseract/tests/test_date.py +++ b/src/paperless_tesseract/tests/test_date.py @@ -8,6 +8,7 @@ from dateutil import tz from django.test import TestCase from ..parsers import RasterisedDocumentParser +from django.conf import settings class TestDate(TestCase): @@ -59,9 +60,11 @@ class TestDate(TestCase): input_file = os.path.join(self.SAMPLE_FILES, "") document = RasterisedDocumentParser(input_file) document._text = "lorem ipsum 13.02.2018 lorem ipsum" + date = document.get_date() self.assertEqual( - document.get_date(), - datetime.datetime(2018, 2, 13, 0, 0, tzinfo=tz.tzutc()) + date, + datetime.datetime(2018, 2, 13, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -72,10 +75,13 @@ class TestDate(TestCase): input_file = os.path.join(self.SAMPLE_FILES, "") document = RasterisedDocumentParser(input_file) document._text = ( - "lorem ipsum 130218, 2018, 20180213 and 13.02.2018 lorem ipsum") + "lorem ipsum 130218, 2018, 20180213 and lorem 13.02.2018 lorem " + "ipsum") + date = document.get_date() self.assertEqual( - document.get_date(), - datetime.datetime(2018, 2, 13, 0, 0, tzinfo=tz.tzutc()) + date, + datetime.datetime(2018, 2, 13, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -110,9 +116,11 @@ class TestDate(TestCase): "März 2019\n" "lorem ipsum" ) + date = document.get_date() self.assertEqual( - document.get_date(), - datetime.datetime(2019, 3, 1, 0, 0, tzinfo=tz.tzutc()) + date, + datetime.datetime(2019, 3, 1, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -134,7 +142,8 @@ class TestDate(TestCase): "März 2020") self.assertEqual(document.get_date(), datetime.datetime(2020, 3, 1, 0, 0, - tzinfo=tz.tzutc())) + tzinfo=tz.gettz( + settings.TIME_ZONE))) @mock.patch( "paperless_tesseract.parsers.RasterisedDocumentParser.SCRATCH", @@ -149,7 +158,8 @@ class TestDate(TestCase): "lorem ipsum") self.assertEqual(document.get_date(), datetime.datetime(2020, 3, 1, 0, 0, - tzinfo=tz.tzutc())) + tzinfo=tz.gettz( + settings.TIME_ZONE))) @mock.patch( "paperless_tesseract.parsers.RasterisedDocumentParser.SCRATCH", @@ -158,11 +168,14 @@ class TestDate(TestCase): def test_get_text_1_pdf(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_1.pdf") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() + date = document.get_date() self.assertEqual(document._is_ocred(), True) self.assertEqual( - document.get_date(), - datetime.datetime(2018, 4, 1, 0, 0, tzinfo=tz.tzutc()) + date, + datetime.datetime(2018, 4, 1, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -172,11 +185,13 @@ class TestDate(TestCase): def test_get_text_1_png(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_1.png") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), False) self.assertEqual( document.get_date(), - datetime.datetime(2018, 4, 1, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2018, 4, 1, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -186,11 +201,13 @@ class TestDate(TestCase): def test_get_text_2_pdf(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_2.pdf") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), True) self.assertEqual( document.get_date(), - datetime.datetime(2013, 2, 1, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2013, 2, 1, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -200,11 +217,13 @@ class TestDate(TestCase): def test_get_text_2_png(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_2.png") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), False) self.assertEqual( document.get_date(), - datetime.datetime(2013, 2, 1, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2013, 2, 1, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -214,11 +233,13 @@ class TestDate(TestCase): def test_get_text_3_pdf(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_3.pdf") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), True) self.assertEqual( document.get_date(), - datetime.datetime(2018, 10, 5, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2018, 10, 5, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -228,11 +249,13 @@ class TestDate(TestCase): def test_get_text_3_png(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_3.png") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), False) self.assertEqual( document.get_date(), - datetime.datetime(2018, 10, 5, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2018, 10, 5, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -242,11 +265,13 @@ class TestDate(TestCase): def test_get_text_4_pdf(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_4.pdf") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), True) self.assertEqual( document.get_date(), - datetime.datetime(2018, 10, 5, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2018, 10, 5, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -256,11 +281,13 @@ class TestDate(TestCase): def test_get_text_4_png(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_4.png") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), False) self.assertEqual( document.get_date(), - datetime.datetime(2018, 10, 5, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2018, 10, 5, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -270,11 +297,13 @@ class TestDate(TestCase): def test_get_text_5_pdf(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_5.pdf") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), True) self.assertEqual( document.get_date(), - datetime.datetime(2018, 12, 17, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2018, 12, 17, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -284,11 +313,13 @@ class TestDate(TestCase): def test_get_text_5_png(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_5.png") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), False) self.assertEqual( document.get_date(), - datetime.datetime(2018, 12, 17, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2018, 12, 17, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -303,7 +334,8 @@ class TestDate(TestCase): self.assertEqual(document._is_ocred(), True) self.assertEqual( document.get_date(), - datetime.datetime(2018, 12, 17, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2018, 12, 17, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -318,7 +350,8 @@ class TestDate(TestCase): self.assertEqual(document._is_ocred(), False) self.assertEqual( document.get_date(), - datetime.datetime(2018, 12, 17, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2018, 12, 17, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -328,6 +361,7 @@ class TestDate(TestCase): def test_get_text_6_pdf_eu(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_6.pdf") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), True) self.assertEqual(document.get_date(), None) @@ -339,6 +373,7 @@ class TestDate(TestCase): def test_get_text_6_png_eu(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_6.png") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), False) self.assertEqual(document.get_date(), None) @@ -350,11 +385,13 @@ class TestDate(TestCase): def test_get_text_7_pdf(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_7.pdf") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), True) self.assertEqual( document.get_date(), - datetime.datetime(2018, 4, 1, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2018, 4, 1, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -364,11 +401,13 @@ class TestDate(TestCase): def test_get_text_8_pdf(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_8.pdf") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), True) self.assertEqual( document.get_date(), - datetime.datetime(2017, 12, 31, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2017, 12, 31, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) ) @mock.patch( @@ -378,9 +417,12 @@ class TestDate(TestCase): def test_get_text_9_pdf(self): input_file = os.path.join(self.SAMPLE_FILES, "tests_date_9.pdf") document = RasterisedDocumentParser(input_file) + document.DATE_ORDER = 'DMY' document.get_text() self.assertEqual(document._is_ocred(), True) self.assertEqual( document.get_date(), - datetime.datetime(2017, 12, 31, 0, 0, tzinfo=tz.tzutc()) + datetime.datetime(2017, 12, 31, 0, 0, + tzinfo=tz.gettz(settings.TIME_ZONE)) + ) )