Better date parsing

gargantext/utils/dates.py

@@ -45,6 +45,7 @@ SEASONS_ABBR = {k[:3]: v for k, v in SEASONS.items()}
 
 _RE_SEASON = re.compile(r'\b(%s)\b' % '|'.join(SEASONS.keys()), re.I)
 _RE_SEASON_ABBR = re.compile(r'\b(%s)\b' % '|'.join(SEASONS_ABBR.keys()), re.I)
+_RE_SLASHES = re.compile(r'//+')
 
 
 class datetime(_datetime):
@@ -64,6 +65,12 @@ class datetime(_datetime):
         for r, table in seasons:
             s = r.sub(lambda m: '%0.2d' % table[m.group(1).lower()], s, count=1)
 
+        # To parse partial RIS dates (eg. 2018/4//)
+        s = _RE_SLASHES.sub('/', s).strip(' /')
+
+        if not s:
+            return default
+
         try:
             # Try to parse first with Django parser utility, then with dateutil
             dt = parse_datetime(s) or \
@@ -74,10 +81,13 @@ class datetime(_datetime):
         except ValueError:
             # Second chance...
 
-            # For date intervals try taking first part only (eg. 2018-04 for
-            # "2018 Apr-May")
-            if '-' in s:
-                return datetime.parse(s.split('-', 1)[0])
+            # This may be a date interval, try taking first part only
+            # eg. 2018-04 for "2018 Apr-May"
+            for sep in ('-', '/', None):
+                if sep in s or not sep:
+                    break
+            if sep:
+                return datetime.parse(s.split(sep, 1)[0])
 
             # Otherwise, just fail
             raise