Fix date detection in filenames

ordigi/cli.py

@@ -218,7 +218,7 @@ def _check(**kwargs):
 @click.argument('subdirs', required=False, nargs=-1, type=click.Path())
 @click.argument('collection', required=True, nargs=1, type=click.Path())
 def _clean(**kwargs):
-    """Remove empty folders"""
+    """Clean media collection"""
 
     folders = kwargs['folders']
     log_level = log.get_level(kwargs['verbose'])

ordigi/collection.py

@@ -138,14 +138,12 @@ class FPath:
         elif item == 'name':
             # Remove date prefix added to the name.
             part = stem
-            for regex in utils.get_date_regex().values():
+            date_filename, regex, sep = utils.get_date_from_string(stem)
-                part = re.sub(regex, '', part)
+            if date_filename:
+                part = re.sub(regex, sep, part)
                 # Delete separator
                 if re.search('^[-_ .]', part):
                     part = part[1:]
-                if part != stem:
-                    # We only want to match first result
-                    break
         elif item == 'date':
             date = metadata['date_media']
             # early morning photos can be grouped with previous day
@@ -643,7 +641,7 @@ class SortMedias:
 
             conflict = self.check_conflicts(src_path, dest_path, remove_duplicates)
 
-            for i in range(1, 100):
+            for i in range(1, 1000):
                 if conflict != 1:
                     break
 

ordigi/media.py

@@ -359,9 +359,9 @@ class Media(ReadExif):
         stem = os.path.splitext(filename)[0]
         date_original = self.metadata['date_original']
         if self.metadata['original_name']:
-            date_filename = utils.get_date_from_string(self.metadata['original_name'])
+            date_filename, _, _ = utils.get_date_from_string(self.metadata['original_name'])
         else:
-            date_filename = utils.get_date_from_string(stem)
+            date_filename, _, _ = utils.get_date_from_string(stem)
             self.log.debug(f'date_filename: {date_filename}')
 
         date_original = self.metadata['date_original']
@@ -387,7 +387,7 @@ class Media(ReadExif):
 
             return self.metadata['date_original']
 
-        self.log.warning(f"could not find original date for {self.file_path}")
+        self.log.warning(f"could not find date original for {self.file_path}")
 
         if self.use_date_filename and date_filename:
             self.log.info(

ordigi/utils.py

@@ -69,17 +69,17 @@ def get_date_regex(user_regex=None):
             # regex to match date format type %Y%m%d, %y%m%d, %d%m%Y,
             # etc...
             'a': re.compile(
-                r'[-_./](?P<year>\d{4})[-_.]?(?P<month>\d{2})[-_.]?(?P<day>\d{2})[-_.]?(?P<hour>\d{2})[-_.]?(?P<minute>\d{2})[-_.]?(?P<second>\d{2})'
+                r'[-_./ ](?P<year>\d{4})[-_.]?(?P<month>\d{2})[-_.]?(?P<day>\d{2})[-_.]?(?P<hour>\d{2})[-_.]?(?P<minute>\d{2})[-_.]?(?P<second>\d{2})([-_./ ])'
             ),
             'b': re.compile(
-                r'[-_./](?P<year>\d{4})[-_.]?(?P<month>\d{2})[-_.]?(?P<day>\d{2})[-_./]'
+                r'[-_./ ](?P<year>\d{4})[-_.]?(?P<month>\d{2})[-_.]?(?P<day>\d{2})([-_./ ])'
             ),
             # not very accurate
             'c': re.compile(
-                r'[-_./](?P<year>\d{2})[-_.]?(?P<month>\d{2})[-_.]?(?P<day>\d{2})[-_./]'
+                r'[-_./ ](?P<year>\d{2})[-_.]?(?P<month>\d{2})[-_.]?(?P<day>\d{2})([-_./ ])'
             ),
             'd': re.compile(
-                r'[-_./](?P<day>\d{2})[-_.](?P<month>\d{2})[-_.](?P<year>\d{4})[-_./]'
+                r'[-_./ ](?P<day>\d{2})[-_.](?P<month>\d{2})[-_.](?P<year>\d{4})([-_./ ])'
             ),
         }
 
@@ -96,15 +96,18 @@ def get_date_from_string(string):
     # Otherwise assume a filename such as IMG_20160915_123456.jpg as default.
 
     matches = []
+    sep = ''
     for i, regex in DATE_REGEX.items():
         match = re.findall(regex, string)
         if match != []:
+            sep = match[0][3]
             if i == 'c':
                 match = [('20' + match[0][0], match[0][1], match[0][2])]
             elif i == 'd':
                 # reorder items
                 match = [(match[0][2], match[0][1], match[0][0])]
-            # matches = match + matches
+            else:
+                match = [(match[0][0], match[0][1], match[0][2])]
             if len(match) != 1:
                 # The time string is not uniq
                 continue
@@ -119,9 +122,11 @@ def get_date_from_string(string):
             date_object = tuple(map(int, matches[0][0]))
             date = datetime(*date_object)
         except (KeyError, ValueError):
-            return None
+            return None, matches[0][1], sep
 
-        return date
+        return date, matches[0][1], sep
+
+    return None, None, sep
 
 
 def match_date_regex(regex, value):

tests/test_media.py

@@ -90,10 +90,10 @@ class TestMedia:
             date_filename = None
             for tag in media.tags_keys['original_name']:
                 if tag in exif_data:
-                    date_filename = get_date_from_string(exif_data[tag])
+                    date_filename, _, _ = get_date_from_string(exif_data[tag])
                 break
             if not date_filename:
-                date_filename = get_date_from_string(file_path.name)
+                date_filename, _, _ = get_date_from_string(file_path.name)
 
             if media.metadata['date_original']:
                 assert date_media == media.metadata['date_original']