PICARD-2729: Allow disabling date sanitazation for APE and Vorbis tags

Author: ShubhamBhut

picard/formats/apev2.py

@@ -125,6 +125,7 @@ class APEv2File(File):
         'replaygain_reference_loudness': 'REPLAYGAIN_REFERENCE_LOUDNESS',
     }
     __rtranslate = {v.lower(): k for k, v in __translate.items()}
+    sanitize_date = sanitize_date
 
     def __init__(self, filename):
         super().__init__(filename)
@@ -136,6 +137,8 @@ def _load(self, filename):
         file = self._File(encode_filename(filename))
         metadata = Metadata()
         if file.tags:
+            config = get_config()
+            date_sanitize = self.NAME not in config.setting['formats_to_disable_date_sanitize']
             for origname, values in file.tags.items():
                 name_lower = origname.lower()
                 if (values.kind == mutagen.apev2.BINARY
@@ -160,7 +163,8 @@ def _load(self, filename):
                     name = name_lower
                     if name == 'year':
                         name = 'date'
-                        value = sanitize_date(value)
+                        if date_sanitize:
+                            value = sanitize_date(value)
                     elif name == 'track':
                         name = 'tracknumber'
                         track = value.split('/')

picard/formats/id3.py

@@ -248,6 +248,7 @@ class ID3File(File):
     __lrc_line_re_parse = re.compile(r'(\[\d\d:\d\d\.\d\d\d\])')
     __lrc_syllable_re_parse = re.compile(r'(<\d\d:\d\d\.\d\d\d>)')
     __lrc_both_re_parse = re.compile(r'(\[\d\d:\d\d\.\d\d\d\]|<\d\d:\d\d\.\d\d\d>)')
+    sanitize_date = sanitize_date
 
     def __init__(self, filename):
         super().__init__(filename)
@@ -278,6 +279,7 @@ def _load(self, filename):
                 f = tags.pop(old)
                 tags.add(getattr(id3, new)(encoding=f.encoding, text=f.text))
         metadata = Metadata()
+        date_sanitize = self.NAME not in config.setting['formats_to_disable_date_sanitize']
         for frame in tags.values():
             frameid = frame.FrameID
             if frameid in self.__translate:
@@ -395,9 +397,12 @@ def _load(self, filename):
                     metadata.add('~rating', rating)
 
         if 'date' in metadata:
-            sanitized = sanitize_date(metadata.getall('date')[0])
-            if sanitized:
-                metadata['date'] = sanitized
+            if date_sanitize:
+                sanitized = sanitize_date(metadata.getall('date')[0])
+                if sanitized:
+                    metadata['date'] = sanitized
+                else:
+                    metadata['date'] = metadata.getall('date')[0]
 
         self._info(metadata, file)
         return metadata

picard/formats/util.py

@@ -45,6 +45,12 @@ def supported_formats():
     return [(file_format.EXTENSIONS, file_format.NAME) for file_format in _formats]
 
 
+def formats_with_sanitize_date():
+    for fmt in _formats:
+        if hasattr(fmt, 'sanitize_date'):
+            yield fmt
+
+
 def supported_extensions():
     """Returns list of supported extensions."""
     return [ext for exts, name in supported_formats() for ext in exts]

picard/formats/vorbis.py

@@ -129,20 +129,24 @@ class VCommentFile(File):
         'waveformatextensible_channel_mask': '~waveformatextensible_channel_mask',
     }
     __rtranslate = {v: k for k, v in __translate.items()}
+    sanitize_date = sanitize_date
 
     def _load(self, filename):
         log.debug("Loading file %r", filename)
         config = get_config()
         file = self._File(encode_filename(filename))
         file.tags = file.tags or {}
         metadata = Metadata()
+        config = get_config()
+        date_sanitize = self.NAME not in config.setting['formats_to_disable_date_sanitize']
         for origname, values in file.tags.items():
             for value in values:
                 value = value.rstrip('\0')
                 name = origname
                 if name in {'date', 'originaldate', 'releasedate'}:
                     # YYYY-00-00 => YYYY
-                    value = sanitize_date(value)
+                    if date_sanitize:
+                        value = sanitize_date(value)
                 elif name == 'performer' or name == 'comment':
                     # transform "performer=Joe Barr (Piano)" to "performer:Piano=Joe Barr"
                     name += ':'
@@ -280,7 +284,8 @@ def _save(self, filename, metadata):
                 name = 'lyrics'
             elif name in {'date', 'originaldate', 'releasedate'}:
                 # YYYY-00-00 => YYYY
-                value = sanitize_date(value)
+                if self.NAME not in config.setting['formats_to_disable_date_sanitize']:
+                    value = sanitize_date(value)
             elif name.startswith('performer:') or name.startswith('comment:'):
                 # transform "performer:Piano=Joe Barr" to "performer=Joe Barr (Piano)"
                 name, desc = name.split(':', 1)

picard/profile.py

@@ -70,6 +70,8 @@ class UserProfileGroups():
             SettingDesc('convert_punctuation', ['convert_punctuation']),
             SettingDesc('release_ars', ['release_ars']),
             SettingDesc('track_ars', ['track_ars']),
+            SettingDesc('disable_date_sanitize', ['disable_date_sanitize']),
+            SettingDesc('formats_to_disable_date_sanitize', ['selected_formats']),
             SettingDesc('guess_tracknumber_and_title', ['guess_tracknumber_and_title']),
             SettingDesc('va_name', ['va_name']),
             SettingDesc('nat_name', ['nat_name']),

picard/ui/options/metadata.py

@@ -55,6 +55,7 @@
 from picard.ui.ui_multi_locale_selector import Ui_MultiLocaleSelector
 from picard.ui.ui_options_metadata import Ui_MetadataOptionsPage
 from picard.ui.util import qlistwidget_items
+from picard.formats.util import formats_with_sanitize_date
 
 
 def iter_sorted_locales(locales):
@@ -92,6 +93,8 @@ class MetadataOptionsPage(OptionsPage):
         ListOption('setting', 'script_exceptions', [], title=N_("Translation script exceptions")),
         BoolOption('setting', 'release_ars', True, title=N_("Use release relationships")),
         BoolOption('setting', 'track_ars', False, title=N_("Use track and release relationships")),
+        BoolOption('setting', 'disable_date_sanitize', False, title=N_("Disable date sanitization for APE and Vorbis tags")),
+        Option('setting', 'formats_to_disable_date_sanitize', set(), title=N_("Formats to disable date sanitize")),
         BoolOption('setting', 'convert_punctuation', False, title=N_("Convert Unicode punctuation characters to ASCII")),
         BoolOption('setting', 'standardize_artists', False, title=N_("Use standardized artist names")),
         BoolOption('setting', 'standardize_instruments', True, title=N_("Use standardized instrument and vocal credits")),
@@ -108,6 +111,7 @@ def __init__(self, parent=None):
         self.ui.select_scripts.clicked.connect(self.open_script_selector)
         self.ui.translate_artist_names.stateChanged.connect(self.set_enabled_states)
         self.ui.translate_artist_names_script_exception.stateChanged.connect(self.set_enabled_states)
+        self.ui.disable_date_sanitize.stateChanged.connect(self.set_enabled_states)
 
     def load(self):
         config = get_config()
@@ -117,6 +121,10 @@ def load(self):
         self.current_scripts = config.setting['script_exceptions']
         self.make_scripts_text()
         self.ui.translate_artist_names_script_exception.setChecked(config.setting['translate_artist_names_script_exception'])
+        self.ui.disable_date_sanitize.setChecked(config.setting['disable_date_sanitize'])
+        self.current_formats = config.setting['formats_to_disable_date_sanitize']
+        fmt_names =  sorted(fmt.NAME for fmt in formats_with_sanitize_date())
+        self.ui.selected_formats.addItems(fmt_names)
 
         self.ui.convert_punctuation.setChecked(config.setting['convert_punctuation'])
         self.ui.release_ars.setChecked(config.setting['release_ars'])
@@ -152,6 +160,8 @@ def save(self):
         config.setting['convert_punctuation'] = self.ui.convert_punctuation.isChecked()
         config.setting['release_ars'] = self.ui.release_ars.isChecked()
         config.setting['track_ars'] = self.ui.track_ars.isChecked()
+        config.setting['disable_date_sanitize'] = self.ui.disable_date_sanitize.isChecked()
+        config.setting['formats_to_disable_date_sanitize'] = self.current_formats
         config.setting['va_name'] = self.ui.va_name.text()
         nat_name = self.ui.nat_name.text()
         if nat_name != config.setting['nat_name']:
@@ -179,6 +189,8 @@ def set_enabled_states(self):
         select_scripts_enabled = translate_checked and translate_exception_checked
         self.ui.selected_scripts.setEnabled(select_scripts_enabled)
         self.ui.select_scripts.setEnabled(select_scripts_enabled)
+        disable_date_sanitize_checked = self.ui.disable_date_sanitize.isChecked()
+        self.ui.selected_formats.setEnabled(disable_date_sanitize_checked)
 
     def open_locale_selector(self):
         dialog = MultiLocaleSelector(self)