Updates the ignore date parsing to utilize the settings defined date order, instead of guessing a bit

This commit is contained in:
Trenton Holmes 2022-04-13 08:04:15 -07:00 committed by Trenton Holmes
parent 8a6aaf4e2d
commit 5b96944940
3 changed files with 48 additions and 26 deletions

View File

@ -247,10 +247,8 @@ def parse_date(filename, text) -> Optional[datetime.datetime]:
# if filename date parsing is enabled, search there first:
if settings.FILENAME_DATE_ORDER:
logger.info("Attempting parsing from filename")
for m in re.finditer(DATE_REGEX, filename):
date_string = m.group(0)
logger.info(f"Found potential date: {date_string}")
try:
date = __parser(date_string, settings.FILENAME_DATE_ORDER)
@ -260,16 +258,11 @@ def parse_date(filename, text) -> Optional[datetime.datetime]:
date = __filter(date)
if date is not None:
logger.info(f"Found date: {date}")
return date
else:
logger.info("Filtered date out")
logger.info("Attempting parsing from content")
# Iterate through all regex matches in text and try to parse the date
for m in re.finditer(DATE_REGEX, text):
date_string = m.group(0)
logger.info(f"Found potential date: {date_string}")
try:
date = __parser(date_string, settings.DATE_ORDER)
@ -279,10 +272,7 @@ def parse_date(filename, text) -> Optional[datetime.datetime]:
date = __filter(date)
if date is not None:
logger.info(f"Found date: {date}")
return date
else:
logger.info("Filtered date out")
return date

View File

@ -605,21 +605,40 @@ PAPERLESS_TIKA_GOTENBERG_ENDPOINT = os.getenv(
if PAPERLESS_TIKA_ENABLED:
INSTALLED_APPS.append("paperless_tika.apps.PaperlessTikaConfig")
# List dates that should be ignored when trying to parse date from document text
IGNORE_DATES: Set[datetime.date] = set()
def _parse_ignore_dates(
env_ignore: str,
date_order: str = DATE_ORDER,
) -> Set[datetime.datetime]:
"""
If the PAPERLESS_IGNORE_DATES environment variable is set, parse the
user provided string(s) into dates
def _parse_ignore_dates(env_ignore: str) -> Set[datetime.datetime]:
Args:
env_ignore (str): The value of the environment variable, comma seperated dates
date_order (str, optional): The format of the date strings. Defaults to DATE_ORDER.
Returns:
Set[datetime.datetime]: The set of parsed date objects
"""
import dateparser
ignored_dates = set()
for s in env_ignore.split(","):
d = dateparser.parse(s)
d = dateparser.parse(
s,
settings={
"DATE_ORDER": date_order,
},
)
if d:
ignored_dates.add(d.date())
return ignored_dates
# List dates that should be ignored when trying to parse date from document text
IGNORE_DATES: Set[datetime.date] = set()
if os.getenv("PAPERLESS_IGNORE_DATES") is not None:
IGNORE_DATES = _parse_ignore_dates(os.getenv("PAPERLESS_IGNORE_DATES"))

View File

@ -9,6 +9,20 @@ class TestIgnoreDateParsing(TestCase):
Tests the parsing of the PAPERLESS_IGNORE_DATES setting value
"""
def _parse_checker(self, test_cases):
"""
Helper function to check ignore date parsing
Args:
test_cases (_type_): _description_
"""
for env_str, date_format, expected_date_set in test_cases:
self.assertSetEqual(
_parse_ignore_dates(env_str, date_format),
expected_date_set,
)
def test_no_ignore_dates_set(self):
"""
GIVEN:
@ -26,20 +40,19 @@ class TestIgnoreDateParsing(TestCase):
- All ignore dates are parsed
"""
test_cases = [
("1985-05-01", [datetime.date(1985, 5, 1)]),
("1985-05-01", "YMD", {datetime.date(1985, 5, 1)}),
(
"1985-05-01,1991-12-05",
[datetime.date(1985, 5, 1), datetime.date(1991, 12, 5)],
"YMD",
{datetime.date(1985, 5, 1), datetime.date(1991, 12, 5)},
),
("2010-12-13", "YMD", {datetime.date(2010, 12, 13)}),
("11.01.10", "DMY", {datetime.date(2010, 1, 11)}),
(
"11.01.2001,15-06-1996",
"DMY",
{datetime.date(2001, 1, 11), datetime.date(1996, 6, 15)},
),
("2010-12-13", [datetime.date(2010, 12, 13)]),
]
for env_str, expected_dates in test_cases:
expected_date_set = set()
for expected_date in expected_dates:
expected_date_set.add(expected_date)
self.assertSetEqual(
_parse_ignore_dates(env_str),
expected_date_set,
)
self._parse_checker(test_cases)