From: Brian Warner Date: Fri, 5 Jan 2007 06:51:35 +0000 (-0700) Subject: change encode_new to use IEncoder X-Git-Tag: tahoe_v0.1.0-0-UNSTABLE~400 X-Git-Url: https://git.rkrishnan.org/%5B/frontends/CLI.txt?a=commitdiff_plain;h=436409fa4db0a27c4878c9d527afaed435a2c85b;p=tahoe-lafs%2Ftahoe-lafs.git change encode_new to use IEncoder --- diff --git a/src/allmydata/encode_new.py b/src/allmydata/encode_new.py index 3bd01f79..10b01fec 100644 --- a/src/allmydata/encode_new.py +++ b/src/allmydata/encode_new.py @@ -6,7 +6,7 @@ from Crypto.Cipher import AES import sha from allmydata.util import mathutil from allmydata.util.assertutil import precondition -from allmydata.py_ecc import rs_code +from allmydata.encode import ReplicatingEncoder, PyRSEncoder def hash(data): return sha.new(data).digest() @@ -70,6 +70,12 @@ def pad(s, l, c='\x00'): """ return s + c * mathutil.pad_size(len(s), l) +KiB=1024 +MiB=1024*KiB +GiB=1024*MiB +TiB=1024*GiB +PiB=1024*TiB + class Encoder(object): def setup(self, infile): @@ -81,14 +87,15 @@ class Encoder(object): self.num_shares = 100 self.required_shares = 25 - # The segment size needs to be an even multiple of required_shares. - # (See encode_segment().) - self.segment_size = mathutil.next_multiple(1024, self.required_shares) + self.segment_size = min(2*MiB, self.file_size) self.num_segments = mathutil.div_ceil(self.file_size, self.segment_size) - self.share_size = self.file_size / self.required_shares + def setup_encoder(self): + self.encoder = ReplicatingEncoder() + self.encoder.set_params(self.segment_size, self.required_shares, + self.num_shares) + self.share_size = self.encoder.get_share_size() - self.fecer = rs_code.RSCode(self.num_shares, self.required_shares) def get_reservation_size(self): self.num_shares = 100 @@ -109,6 +116,7 @@ class Encoder(object): def start(self): self.setup_encryption() + self.setup_encoder() d = defer.succeed(None) for i in range(self.num_segments): d.addCallback(lambda res: self.do_segment(i)) @@ -118,26 +126,17 @@ class Encoder(object): d.addCallback(lambda res: self.done()) return d - def encode_segment(self, crypttext): - precondition((len(crypttext) % self.required_shares) == 0, len(crypttext), self.required_shares, len(crypttext) % self.required_shares) - subshares = [[] for x in range(self.num_shares)] - # Note string slices aren't an efficient way to use memory, so when we - # upgrade from the unusably slow py_ecc prototype to a fast ECC we - # should also fix up this memory usage (by using the array module). - for i in range(0, len(crypttext), self.required_shares): - words = self.fecer.Encode([ord(x) for x in crypttext[i:i+self.required_shares]]) - for (subshare, word,) in zip(subshares, words): - subshare.append(word) - return [ ''.join(subshare) for subshare in subshares ] - def do_segment(self, segnum): segment_plaintext = self.infile.read(self.segment_size) segment_crypttext = self.cryptor.encrypt(segment_plaintext) del segment_plaintext - subshares_for_this_segment = self.encode_segment(pad(segment_crypttext, self.segment_size)) - del segment_crypttext + d = self.encoder.encode(segment_crypttext) + d.addCallback(self._encoded_segment) + return d + + def _encoded_segment(self, subshare_tuples): dl = [] - for share_num,subshare in enumerate(subshares_for_this_segment): + for share_num,subshare in subshare_tuples: d = self.send_subshare(share_num, self.segment_num, subshare) dl.append(d) self.subshare_hashes[share_num].append(hash(subshare))