def clean(self):
"""returns a new MetaData object that's been cleaned of problems"""
from audiotools.text import (CLEAN_REMOVE_TRAILING_WHITESPACE,
CLEAN_REMOVE_LEADING_WHITESPACE,
CLEAN_REMOVE_EMPTY_TAG,
CLEAN_REMOVE_LEADING_WHITESPACE_ZEROES,
CLEAN_REMOVE_LEADING_ZEROES)
fixes_performed = []
reverse_attr_map = {}
for (attr, key) in self.ATTRIBUTE_MAP.items():
reverse_attr_map[key] = attr
if key in self.ALIASES:
for alias in self.ALIASES[key]:
reverse_attr_map[alias] = attr
cleaned_fields = []
for comment_string in self.comment_strings:
if u"=" in comment_string:
(key, value) = comment_string.split(u"=", 1)
if key.upper() in reverse_attr_map:
attr = reverse_attr_map[key.upper()]
# handle all text fields by stripping whitespace
if len(value.strip()) == 0:
fixes_performed.append(
CLEAN_REMOVE_EMPTY_TAG.format(key))
else:
fix1 = value.rstrip()
if fix1 != value:
fixes_performed.append(
CLEAN_REMOVE_TRAILING_WHITESPACE.format(key))
fix2 = fix1.lstrip()
if fix2 != fix1:
fixes_performed.append(
CLEAN_REMOVE_LEADING_WHITESPACE.format(key))
# integer fields also strip leading zeroes
if (((attr == "track_number") or
(attr == "album_number"))):
match = re.match(r'(.*?)\s*/\s*(.*)', fix2)
if match is not None:
# fix whitespace/zeroes
# on either side of slash
fix3 = u"{}/{}".format(
match.group(1).lstrip(u"0"),
match.group(2).lstrip(u"0"))
if fix3 != fix2:
fixes_performed.append(
CLEAN_REMOVE_LEADING_WHITESPACE_ZEROES.format(key))
else:
# fix zeroes only
fix3 = fix2.lstrip(u"0")
if fix3 != fix2:
fixes_performed.append(
CLEAN_REMOVE_LEADING_ZEROES.format(key))
elif ((attr == "track_total") or
(attr == "album_total")):
fix3 = fix2.lstrip(u"0")
if fix3 != fix2:
fixes_performed.append(
CLEAN_REMOVE_LEADING_ZEROES.format(key))
else:
fix3 = fix2
cleaned_fields.append(u"{}={}".format(key, fix3))
else:
cleaned_fields.append(comment_string)
else:
cleaned_fields.append(comment_string)
return (self.__class__(cleaned_fields, self.vendor_string),
fixes_performed)
def clean(self):
"""returns a new MetaData object that's been cleaned of problems"""
from audiotools.text import (CLEAN_REMOVE_TRAILING_WHITESPACE,
CLEAN_REMOVE_LEADING_WHITESPACE,
CLEAN_REMOVE_EMPTY_TAG,
CLEAN_REMOVE_LEADING_WHITESPACE_ZEROES,
CLEAN_REMOVE_LEADING_ZEROES)
fixes_performed = []
reverse_attr_map = {}
for (attr, key) in self.ATTRIBUTE_MAP.items():
reverse_attr_map[key] = attr
if key in self.ALIASES:
for alias in self.ALIASES[key]:
reverse_attr_map[alias] = attr
cleaned_fields = []
for comment_string in self.comment_strings:
if u"=" in comment_string:
(key, value) = comment_string.split(u"=", 1)
if key.upper() in reverse_attr_map:
attr = reverse_attr_map[key.upper()]
# handle all text fields by stripping whitespace
if len(value.strip()) == 0:
fixes_performed.append(
CLEAN_REMOVE_EMPTY_TAG.format(key))
else:
fix1 = value.rstrip()
if fix1 != value:
fixes_performed.append(
CLEAN_REMOVE_TRAILING_WHITESPACE.format(key))
fix2 = fix1.lstrip()
if fix2 != fix1:
fixes_performed.append(
CLEAN_REMOVE_LEADING_WHITESPACE.format(key))
# integer fields also strip leading zeroes
if (((attr == "track_number")
or (attr == "album_number"))):
match = re.match(r'(.*?)\s*/\s*(.*)', fix2)
if match is not None:
# fix whitespace/zeroes
# on either side of slash
fix3 = u"{}/{}".format(
match.group(1).lstrip(u"0"),
match.group(2).lstrip(u"0"))
if fix3 != fix2:
fixes_performed.append(
CLEAN_REMOVE_LEADING_WHITESPACE_ZEROES.
format(key))
else:
# fix zeroes only
fix3 = fix2.lstrip(u"0")
if fix3 != fix2:
fixes_performed.append(
CLEAN_REMOVE_LEADING_ZEROES.format(
key))
elif ((attr == "track_total")
or (attr == "album_total")):
fix3 = fix2.lstrip(u"0")
if fix3 != fix2:
fixes_performed.append(
CLEAN_REMOVE_LEADING_ZEROES.format(key))
else:
fix3 = fix2
cleaned_fields.append(u"{}={}".format(key, fix3))
else:
cleaned_fields.append(comment_string)
else:
cleaned_fields.append(comment_string)
return (self.__class__(cleaned_fields,
self.vendor_string), fixes_performed)
def clean(self):
import re
from audiotools.text import (CLEAN_REMOVE_DUPLICATE_TAG,
CLEAN_REMOVE_TRAILING_WHITESPACE,
CLEAN_REMOVE_LEADING_WHITESPACE,
CLEAN_FIX_TAG_FORMATTING,
CLEAN_REMOVE_EMPTY_TAG)
fixes_performed = []
used_tags = set()
tag_items = []
for tag in self.tags:
if tag.key.upper() in used_tags:
fixes_performed.append(
CLEAN_REMOVE_DUPLICATE_TAG.format(tag.key.decode('ascii')))
elif tag.type == 0:
used_tags.add(tag.key.upper())
text = tag.__unicode__()
# check trailing whitespace
fix1 = text.rstrip()
if fix1 != text:
fixes_performed.append(
CLEAN_REMOVE_TRAILING_WHITESPACE.format(
tag.key.decode('ascii')))
# check leading whitespace
fix2 = fix1.lstrip()
if fix2 != fix1:
fixes_performed.append(
CLEAN_REMOVE_LEADING_WHITESPACE.format(
tag.key.decode('ascii')))
if tag.key in self.INTEGER_ITEMS:
if u"/" in fix2:
# item is a slashed field of some sort
(current, total) = fix2.split(u"/", 1)
current_int = re.search(r'\d+', current)
total_int = re.search(r'\d+', total)
if (current_int is None) and (total_int is None):
# neither side contains an integer value
# so ignore it altogether
fix3 = fix2
elif ((current_int is not None) and
(total_int is None)):
fix3 = u"{:d}".format(int(current_int.group(0)))
elif ((current_int is None) and
(total_int is not None)):
fix3 = u"{:d}/{:d}".format(
0, int(total_int.group(0)))
else:
# both sides contain an int
fix3 = u"{:d}/{:d}".format(
int(current_int.group(0)),
int(total_int.group(0)))
else:
# item contains no slash
current_int = re.search(r'\d+', fix2)
if current_int is not None:
# item contains an integer
fix3 = u"{:d}".format(int(current_int.group(0)))
else:
# item contains no integer value so ignore it
# (although 'Track' should only contain
# integers, 'Media' may contain strings
# so it may be best to simply ignore that case)
fix3 = fix2
if fix3 != fix2:
fixes_performed.append(
CLEAN_FIX_TAG_FORMATTING.format(
tag.key.decode('ascii')))
else:
fix3 = fix2
if len(fix3) > 0:
tag_items.append(ApeTagItem.string(tag.key, fix3))
else:
fixes_performed.append(
CLEAN_REMOVE_EMPTY_TAG.format(tag.key.decode('ascii')))
else:
used_tags.add(tag.key.upper())
tag_items.append(tag)
return (self.__class__(tag_items,
self.contains_header,
self.contains_footer),
fixes_performed)
