class MoveToRoot:
def __init__(self, root_path, dir_to_exclude):
self.root_path = root_path
self.formatter = FileFormatter()
self.dir_to_exclude = set(dir_to_exclude)
def move(self):
files = os.listdir(self.root_path)
self._move_helper(self.root_path, files)
return None
def _move_helper(self, root_path, files):
for file in files:
if file not in self.dir_to_exclude:
file_path = os.path.join(root_path, file)
if os.path.isdir(file_path):
self._move_helper(file_path, list(os.listdir(file_path)))
print('removing directory:', file_path, 'with contents')
[print('\t', deleted_file) for deleted_file in list(os.listdir(file_path))]
shutil.rmtree(file_path)
else:
if (self.formatter.file_contains_format(file, file_path) and
root_path != self.root_path):
new_path = os.path.join(self.root_path, file)
print('moving:', file_path, 'to:', new_path)
shutil.move(file_path, new_path)
return None
class MoveToRoot:
def __init__(self, root_path, dir_to_exclude):
self.root_path = root_path
self.formatter = FileFormatter()
self.dir_to_exclude = set(dir_to_exclude)
def move(self):
files = os.listdir(self.root_path)
self._move_helper(self.root_path, files)
return None
def _move_helper(self, root_path, files):
for file in files:
if file not in self.dir_to_exclude:
file_path = os.path.join(root_path, file)
if os.path.isdir(file_path):
self._move_helper(file_path, list(os.listdir(file_path)))
print('removing directory:', file_path, 'with contents')
[
print('\t', deleted_file)
for deleted_file in list(os.listdir(file_path))
]
shutil.rmtree(file_path)
else:
if (self.formatter.file_contains_format(file, file_path)
and root_path != self.root_path):
new_path = os.path.join(self.root_path, file)
print('moving:', file_path, 'to:', new_path)
shutil.move(file_path, new_path)
return None
def __init__(self):
self.punctuation = [
' ',
'_',
'.',
'[',
']',
',',
'/',
'-',
'{',
'}',
'|',
'(',
')',
'*',
'!',
'&',
"'"
]
self.fileFormatter = FileFormatter()
def makeNewFile(self, nameList, aFile, pathToFile, fileNumber):
if FileFormatter().file_contains_format(aFile, pathToFile):
newFile = FileNamer().makeNewFileName(nameList, aFile, pathToFile)
if newFile != aFile:
src = pathToFile + aFile
dst = pathToFile + newFile
shutil.move(src, dst)
print(
fileNumber, '-- from: ' + aFile + '\n' + str(fileNumber) +
' -- to: ' + newFile + '\n')
elif not os.path.isdir(pathToFile + aFile):
os.remove(pathToFile + aFile)
print(fileNumber, '-- removing file: ' + aFile + '\n')
newFile = None
else:
print('File is a dir: ' + pathToFile + aFile)
newFile = None
return newFile
def __init__(self, top_dirs):
self.top_dirs = top_dirs
self.file_formatter = FileFormatter()
self.files = {}
class FileCounter:
def __init__(self, top_dirs):
self.top_dirs = top_dirs
self.file_formatter = FileFormatter()
self.files = {}
def count_files(self, dirs):
counter = 0
for topdir in dirs:
print('******************* searching', topdir, '***************************')
if os.path.isdir(topdir):
files = os.listdir(topdir)
for file in files:
file_path = os.path.join(topdir, file)
if os.path.isdir(file_path):
newFileNumber = self.countFiles([file_path])
print('\ncounter:', counter, '\nnew number:', newFileNumber, '\nnew counter:', counter + newFileNumber)
counter += newFileNumber
elif self.fileFormatter.file_contains_format(file, topdir):
counter += 1
print(counter, 'a movie:', file)
print('------ returning', counter, '----------')
return counter
def count_names(self):
for dir in self.top_dirs:
if not os.path.isdir(dir):
continue
files = os.listdir(dir)
for file in files:
file_path = os.path.join(dir, file)
if not os.path.isdir(file_path):
continue
sub_files = os.listdir(file_path)
for sub_file in sub_files:
sub_file_path = os.path.join(file_path, sub_file)
if os.path.isdir(sub_file_path):
sub_sub_files = self.get_relevant_files(sub_file_path)
self.add_files_for_name(file, sub_sub_files)
elif self.file_formatter.file_contains_format(sub_file, file_path):
self.add_files_for_name(file, [sub_file])
return None
def get_relevant_files(self, file_path):
return [file for file in list(os.listdir(file_path))
if self.file_formatter.file_contains_format(file, file_path)]
def add_files_for_name(self, name, files):
if name in self.files:
old_files = self.files[name]
self.files[name] = files + old_files
else:
self.files[name] = files
return None
def print_results(self):
names = list(self.files.keys())
for i in range(len(names) - 1):
target = i
for j in range(i + 1, len(names)):
if len(self.files[names[j]]) > len(self.files[names[target]]):
target = j
if i != target:
temp = names[i]
names[i] = names[target]
names[target] = temp
max_num_length = len(str(len(names))) + 1
max_name_length = max([len(name) for name in names]) + 1
for i in range(len(names)):
print(str(i + 1).ljust(max_num_length),
names[i].ljust(max_name_length),
'scenes:',
len(self.files[names[i]])
)
return None
def __init__(self, top_dirs):
self.top_dirs = top_dirs
self.file_formatter = FileFormatter()
self.files = {}
class FileCounter:
def __init__(self, top_dirs):
self.top_dirs = top_dirs
self.file_formatter = FileFormatter()
self.files = {}
def count_files(self, dirs):
counter = 0
for topdir in dirs:
print('******************* searching', topdir, '***************************')
if os.path.isdir(topdir):
files = os.listdir(topdir)
for file in files:
file_path = os.path.join(topdir, file)
if os.path.isdir(file_path):
newFileNumber = self.countFiles([file_path])
print('\ncounter:', counter, '\nnew number:', newFileNumber, '\nnew counter:', counter + newFileNumber)
counter += newFileNumber
elif self.fileFormatter.file_contains_format(file, topdir):
counter += 1
print(counter, 'a movie:', file)
print('------ returning', counter, '----------')
return counter
def count_names(self):
for dir in self.top_dirs:
if not os.path.isdir(dir):
continue
files = os.listdir(dir)
for file in files:
file_path = os.path.join(dir, file)
if not os.path.isdir(file_path):
continue
sub_files = os.listdir(file_path)
for sub_file in sub_files:
sub_file_path = os.path.join(file_path, sub_file)
if os.path.isdir(sub_file_path):
sub_sub_files = self.get_relevant_files(sub_file_path)
self.add_files_for_name(file, sub_sub_files)
elif self.file_formatter.file_contains_format(sub_file, file_path):
self.add_files_for_name(file, [sub_file])
return None
def get_relevant_files(self, file_path):
return [file for file in list(os.listdir(file_path))
if self.file_formatter.file_contains_format(file, file_path)]
def add_files_for_name(self, name, files):
if name in self.files:
old_files = self.files[name]
self.files[name] = files + old_files
else:
self.files[name] = files
return None
def print_results(self):
names = list(self.files.keys())
for i in range(len(names) - 1):
target = i
for j in range(i + 1, len(names)):
if len(self.files[names[j]]) > len(self.files[names[target]]):
target = j
if i != target:
temp = names[i]
names[i] = names[target]
names[target] = temp
max_num_length = len(str(len(names))) + 1
max_name_length = max([len(name) for name in names]) + 1
for i in range(len(names)):
print(str(i + 1).ljust(max_num_length),
names[i].ljust(max_name_length),
'scenes:',
len(self.files[names[i]])
)
return None
def make_histogram(self):
num_of_bins = 100
counts = [len(value) for value in self.files.values()]
max_count = max(counts)
min_count = min(counts)
region = max_count - min_count
bin_size = int(region/num_of_bins)
print(
'min:', min_count,
'max:', max_count,
'range:', region,
'bins:', num_of_bins,
'bin size:', bin_size,
'entries:', len(counts)
)
bins = (num_of_bins + 1)*[0]
for count in counts:
i = int(count/bin_size)
try:
bins[i] += 1
except IndexError:
print('INDEX ERROR WHEN -- i:', i)
continue
x0 = min_count
results = []
ranges = []
for i in range(len(bins)):
results.append(str(x0) + ' - ' + str(x0 + bin_size) + ' : ')
x0 += bin_size
results.reverse()
bins.reverse()
max_just = len(results[len(results) - 1])
[print(results[i].ljust(max_just), bins[i]) for i in range(len(results))]
return None
def __init__(self, root_path, dir_to_exclude):
self.root_path = root_path
self.formatter = FileFormatter()
self.dir_to_exclude = set(dir_to_exclude)
def __init__(self, root_path, dir_to_exclude):
self.root_path = root_path
self.formatter = FileFormatter()
self.dir_to_exclude = set(dir_to_exclude)
class FileNamer:
def __init__(self):
self.punctuation = [
' ',
'_',
'.',
'[',
']',
',',
'/',
'-',
'{',
'}',
'|',
'(',
')',
'*',
'!',
'&',
"'"
]
self.fileFormatter = FileFormatter()
def makeNewFileName(self, nameList, aFile, pathToFile):
# check if aFile has name in it
aFile = aFile.lower()
firstName = nameList[0].lower()
start = aFile.find(firstName) - 1
if start >= 0:
# firstName is found but not at beginning of aFile
if aFile[start] not in self.punctuation:
aFile = aFile.replace(firstName, '_' + firstName, 1)
if len(nameList) > 1:
# name list has first and last name
lastName = nameList[1].lower()
if firstName in aFile and lastName in aFile:
endFirst = aFile.find(firstName) + len(firstName)
startLast = aFile.find(lastName)
if endFirst == startLast:
# first and last names are right next to each other
newFile = aFile.replace(lastName, '_' + lastName, 1)
else:
# file contains first and last names. No changes needed
newFile = aFile
elif firstName in aFile:
# file contains first name only
start = aFile.find(firstName)
end = start + len(firstName)
target = aFile[end]
if target in self.punctuation:
newFile = aFile.replace(firstName, firstName + '_' + lastName, 1)
else:
newFile = aFile.replace(firstName, firstName + '_' + lastName + '_', 1)
else:
# file does not contain name
if self.fileFormatter.file_contains_format(aFile, pathToFile):
# file is in a movie format
parts = aFile.rpartition('.')
newFile = parts[0] + '_' + firstName + '_' + lastName + parts[1] + parts[2]
else:
newFile = aFile + '_' + firstName + '_' + lastName
else:
if firstName in aFile:
newFile = aFile
elif self.fileFormatter.file_contains_format(aFile, pathToFile):
parts = aFile.rpartition('.')
newFile = parts[0] + '_' + firstName + parts[1] + parts[2]
else:
newFile = aFile + '_' + firstName
newFile = self.replacePunctuation(newFile, pathToFile)
return newFile
def make_new_filename_multiple_names(self, names_list, file, file_path):
for name_list in names_list:
first_name = name_list[0]
try:
last_name = name_list[1]
except IndexError:
raise WrongNameFormatException('a name must have a first and last name')
file = self.makeNewFileName(
[first_name, last_name],
file,
file_path
)
return file
def replacePunctuation(self, aFile, pathToFile):
isMovie = False
if self.fileFormatter.file_contains_format(aFile, pathToFile):
length = aFile.rfind('.')
isMovie = True
else:
length = len(aFile)
newName = aFile
if length != -1:
i = 0
while True:
if newName[i] in self.punctuation:
newName = newName[:i] + '_' + newName[i + 1:]
try:
j = i + 1
while True:
if newName[j] in self.punctuation:
if isMovie:
format_target = newName.rfind('.')
if j == format_target:
break
newName = newName[:j] + newName[j + 1:]
length -= 1
i = j
if j >= length:
break
else:
break
except IndexError:
break
i += 1
if i >= length:
break
if newName[0] == '_':
newName = newName.replace('_', '', 1)
length -= 1
if isMovie:
target = newName.rfind('.') - 1
if newName[target] == '_':
newName = newName[:target] + newName[target + 1:]
else:
if newName[len(newName) - 1] == '_':
newName = newName[:len(newName) - 1] + newName[len(newName):]
return newName
def clean_name_for_raw_file(self, file, base_path):
ext = self.fileFormatter.get_format(file, base_path)
if ext:
file = file.rsplit('.', 1)[0]
for punctuation in self.punctuation:
file = file.replace(punctuation, '_')
to_remove = []
i = len(file) - 1
while i >= 0:
if file[i] == '_' and (i == 0 or i == len(file) - 1 or file[i - 1] == '_'):
to_remove.append(i)
i -= 1
for index in to_remove:
file = file[:index] + file[index + 1:]
if ext:
file += '.' + ext
return file.lower()