Merge branch 'ENH_filename_date_parsing' of https://github.com/jat255/paperless into jat255-ENH_filename_date_parsing

This commit is contained in:
Daniel Quinn
2018-12-01 16:57:16 +00:00
9 changed files with 198 additions and 41 deletions

Binary file not shown.

After

Width:  |  Height:  |  Size: 136 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 136 KiB

View File

@@ -8,6 +8,7 @@ from dateutil import tz
from django.test import TestCase, override_settings
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(
@@ -215,11 +234,13 @@ 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(
@@ -230,11 +251,13 @@ 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(
@@ -245,11 +268,13 @@ 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(
@@ -260,11 +285,13 @@ 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(
@@ -274,11 +301,13 @@ 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(
@@ -288,11 +317,13 @@ 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(
@@ -307,7 +338,8 @@ 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(
@@ -322,7 +354,8 @@ 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(
@@ -332,6 +365,7 @@ 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)
@@ -343,6 +377,7 @@ 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)
@@ -354,11 +389,13 @@ 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(
@@ -368,11 +405,13 @@ 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(
@@ -382,11 +421,82 @@ 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))
)
@mock.patch(
"paperless_tesseract.parsers.RasterisedDocumentParser.SCRATCH",
SCRATCH
)
def test_filename_date_1_pdf(self):
input_file = os.path.join(self.SAMPLE_FILES,
"tests_date_in_filename_2018-03-20_1.pdf")
document = RasterisedDocumentParser(input_file)
document.FILENAME_DATE_ORDER = 'YMD'
document.get_text()
date = document.get_date()
self.assertEqual(document._is_ocred(), True)
self.assertEqual(
date,
datetime.datetime(2018, 3, 20, 0, 0,
tzinfo=tz.gettz(settings.TIME_ZONE))
)
@mock.patch(
"paperless_tesseract.parsers.RasterisedDocumentParser.SCRATCH",
SCRATCH
)
def test_filename_date_1_png(self):
input_file = os.path.join(self.SAMPLE_FILES,
"tests_date_in_filename_2018-03-20_1.png")
document = RasterisedDocumentParser(input_file)
document.FILENAME_DATE_ORDER = 'YMD'
date = document.get_date()
self.assertEqual(document._is_ocred(), False)
self.assertEqual(
date,
datetime.datetime(2018, 3, 20, 0, 0,
tzinfo=tz.gettz(settings.TIME_ZONE))
)
@mock.patch(
"paperless_tesseract.parsers.RasterisedDocumentParser.SCRATCH",
SCRATCH
)
def test_filename_date_2_pdf(self):
input_file = os.path.join(self.SAMPLE_FILES,
"2013-12-11_tests_date_in_filename_2.pdf")
document = RasterisedDocumentParser(input_file)
document.FILENAME_DATE_ORDER = 'YMD'
date = document.get_date()
self.assertEqual(document._is_ocred(), True)
self.assertEqual(
date,
datetime.datetime(2013, 12, 11, 0, 0,
tzinfo=tz.gettz(settings.TIME_ZONE))
)
@mock.patch(
"paperless_tesseract.parsers.RasterisedDocumentParser.SCRATCH",
SCRATCH
)
def test_filename_date_2_png(self):
input_file = os.path.join(self.SAMPLE_FILES,
"2013-12-11_tests_date_in_filename_2.png")
document = RasterisedDocumentParser(input_file)
document.FILENAME_DATE_ORDER = 'YMD'
date = document.get_date()
self.assertEqual(document._is_ocred(), False)
self.assertEqual(
date,
datetime.datetime(2013, 12, 11, 0, 0,
tzinfo=tz.gettz(settings.TIME_ZONE))
)
@mock.patch(