def __init__(self, input_dir):
self.input_dir = input_dir
self.root_dir_name = path.basename(input_dir)
self.header = FileHeader()
self.indices = {} # {id:index, ...}
self.cookies = {} # {md5:cookie, ...}
self.index_list = []
self.cookie_list = []
class StormPacker(object):
def __init__(self, input_dir):
self.input_dir = input_dir
self.root_dir_name = path.basename(input_dir)
self.header = FileHeader()
self.indices = {} # {id:index, ...}
self.cookies = {} # {md5:cookie, ...}
self.index_list = []
self.cookie_list = []
def walk_files(self):
index_offset = FileHeader.STRUCT_DEF.size
for root, dirs, files in os.walk(self.input_dir):
for filename in files:
fullpath = path.join(root, filename)
pos = fullpath.find(self.root_dir_name)
searchpath = fullpath[pos:].replace('\\', '/')
# FileIndex
fileindex = FileIndex()
fileindex.id = hash(searchpath)
assert self.indices.has_key(fileindex.id) == False
self.indices[fileindex.id] = fileindex
f = io.open(fullpath, 'rb')
data = f.read()
f.close()
fileindex.offset = index_offset
fileindex.size = len(data)
self.index_list.append(fileindex)
index_offset += fileindex.size
# FileCookie
m = hashlib.md5()
m.update(data)
digest = m.digest()
if self.cookies.has_key(digest):
cookie = self.cookies[digest]
cookie.refcount += 1
else:
cookie = FileCookie()
cookie.digest = digest
cookie.space = fileindex.size
self.cookies[digest] = cookie
self.cookie_list.append((cookie, data))
self.header.number = len(self.index_list)
self.header.offset = index_offset
def save_to_file(self):
outputfile = self.input_dir + '.sto'
f = io.open(outputfile, 'wb')
f.write(self.header.pack())
for cookie, data in self.cookie_list:
f.write(cookie.pack())
f.write(data)
for index in self.index_list:
f.write(index.pack())
f.close()
def pack(self):
print 'packing \'%s\' ...' % (self.input_dir,)
self.walk_files()
self.save_to_file()
