fix(dateparser/{conf, date}.py): add negative-timestamp as optional p…

…arameter
scrapinghub · Jun 11, 2022 · 23ca1f3 · 23ca1f3
1 parent 7d59127
commit 23ca1f3
Show file tree

Hide file tree

Showing 2 changed files with 16 additions and 6 deletions.
diff --git a/dateparser/conf.py b/dateparser/conf.py
@@ -120,7 +120,7 @@ def _check_require_part(setting_name, setting_value):
 def _check_parsers(setting_name, setting_value):
     """Returns `True` if the provided list of parsers contains valid values"""
     existing_parsers = [
-        'timestamp', 'relative-time', 'custom-formats', 'absolute-time', 'no-spaces-time'
+        'timestamp', 'relative-time', 'custom-formats', 'absolute-time', 'no-spaces-time', 'negative-timestamp'
     ]  # FIXME: Extract the list of existing parsers from another place (#798)
     unknown_parsers = set(setting_value) - set(existing_parsers)
     if unknown_parsers:

diff --git a/dateparser/date.py b/dateparser/date.py
@@ -40,7 +40,7 @@
 RE_SANITIZE_APOSTROPHE = re.compile('|'.join(APOSTROPHE_LOOK_ALIKE_CHARS))
 
 RE_SEARCH_TIMESTAMP = re.compile(r'^(\d{10})(\d{3})?(\d{3})?(?![^.])')
-RE_SEARCH_NEGATIVE_TIMESTAMP = re.compile(r'^([-]?\d{10})(\d{3})?(\d{3})?(?![^.])')
+RE_SEARCH_NEGATIVE_TIMESTAMP = re.compile(r'^([-]\d{10})(\d{3})?(\d{3})?(?![^.])')
 
 
 def sanitize_spaces(date_string):
@@ -113,8 +113,11 @@ def sanitize_date(date_string):
     return date_string
 
 
-def get_date_from_timestamp(date_string, settings):
-    match = RE_SEARCH_NEGATIVE_TIMESTAMP.search(date_string)
+def get_date_from_timestamp(date_string, settings, negative=False):
+    match = RE_SEARCH_TIMESTAMP.search(date_string)
+    if negative:
+        match = RE_SEARCH_NEGATIVE_TIMESTAMP.search(date_string)
+
     if match:
         seconds = int(match.group(1))
         millis = int(match.group(2) or 0)
@@ -167,6 +170,7 @@ def __init__(self, locale, date_string, date_formats, settings=None):
         self._translated_date_with_formatting = None
         self._parsers = {
             'timestamp': self._try_timestamp,
+            'negative-timestamp': self._try_negative_timestamp,
             'relative-time': self._try_freshness_parser,
             'custom-formats': self._try_given_formats,
             'absolute-time': self._try_absolute_parser,
@@ -186,12 +190,18 @@ def _parse(self):
         else:
             return None
 
-    def _try_timestamp(self):
+    def _try_timestamp_parser(self, negative=False):
         return DateData(
-            date_obj=get_date_from_timestamp(self.date_string, self._settings),
+            date_obj=get_date_from_timestamp(self.date_string, self._settings, negative=negative),
             period='time' if self._settings.RETURN_TIME_AS_PERIOD else 'day',
         )
 
+    def _try_timestamp(self):
+        return self._try_timestamp_parser()
+
+    def _try_negative_timestamp(self):
+        return self._try_timestamp_parser(negative=True)
+
     def _try_freshness_parser(self):
         try:
             return freshness_date_parser.get_date_data(self._get_translated_date(), self._settings)