from binascii import b2a_hex, a2b_hex
import sys
+from twisted.internet import threads, defer
from twisted.trial import unittest
+PIECE_SIZE = 512*1024
+
+class HashError(ValueError):
+ """An error has occurred while hashing a file."""
+
class HashObject:
"""Manages hashes and hashing for a file."""
{'name': 'md5',
'AptPkgRecord': 'MD5Hash',
'AptSrcRecord': True,
- 'AptIndexRecord': 'MD5Sum',
+ 'AptIndexRecord': 'MD5SUM',
'old_module': 'md5',
'hashlib_func': 'md5',
},
]
- def __init__(self):
+ def __init__(self, digest = None, size = None):
self.hashTypeNum = 0 # Use the first if nothing else matters
self.expHash = None
self.expHex = None
self.expSize = None
self.expNormHash = None
self.fileHasher = None
- self.fileHash = None
+ self.pieceHasher = None
+ self.fileHash = digest
+ self.pieceHash = []
+ self.size = size
self.fileHex = None
self.fileNormHash = None
self.done = True
if bits is not None:
bytes = (bits - 1) // 8 + 1
else:
- assert(bytes is not None)
+ if bytes is None:
+ raise HashError, "you must specify one of bits or bytes"
if len(hashString) < bytes:
hashString = hashString + '\000'*(bytes - len(hashString))
elif len(hashString) > bytes:
"""
if self.result is None or force == True:
self.result = None
- self.size = 0
self.done = False
- if sys.version_info < (2, 5):
- mod = __import__(self.ORDER[self.hashTypeNum]['old_module'], globals(), locals(), [])
- self.fileHasher = mod.new()
- else:
- import hashlib
- func = getattr(hashlib, self.ORDER[self.hashTypeNum]['hashlib_func'])
- self.fileHasher = func()
+ self.fileHasher = self._new()
+ self.pieceHasher = None
+ self.fileHash = None
+ self.pieceHash = []
+ self.size = 0
+ self.fileHex = None
+ self.fileNormHash = None
+
+ def _new(self):
+ """Create a new hashing object according to the hash type."""
+ if sys.version_info < (2, 5):
+ mod = __import__(self.ORDER[self.hashTypeNum]['old_module'], globals(), locals(), [])
+ return mod.new()
+ else:
+ import hashlib
+ func = getattr(hashlib, self.ORDER[self.hashTypeNum]['hashlib_func'])
+ return func()
def update(self, data):
"""Add more data to the file hasher."""
if self.result is None:
- assert self.done == False, "Already done, you can't add more data after calling digest() or verify()"
- assert self.fileHasher is not None, "file hasher not initialized"
+ if self.done:
+ raise HashError, "Already done, you can't add more data after calling digest() or verify()"
+ if self.fileHasher is None:
+ raise HashError, "file hasher not initialized"
+
+ if not self.pieceHasher and self.size + len(data) > PIECE_SIZE:
+ # Hash up to the piece size
+ self.fileHasher.update(data[:(PIECE_SIZE - self.size)])
+ data = data[(PIECE_SIZE - self.size):]
+ self.size = PIECE_SIZE
+
+ # Save the first piece digest and initialize a new piece hasher
+ self.pieceHash.append(self.fileHasher.digest())
+ self.pieceHasher = self._new()
+
+ if self.pieceHasher:
+ # Loop in case the data contains multiple pieces
+ piece_size = self.size % PIECE_SIZE
+ while piece_size + len(data) > PIECE_SIZE:
+ # Save the piece hash and start a new one
+ self.pieceHasher.update(data[:(PIECE_SIZE - piece_size)])
+ self.pieceHash.append(self.pieceHasher.digest())
+ self.pieceHasher = self._new()
+
+ # Don't forget to hash the data normally
+ self.fileHasher.update(data[:(PIECE_SIZE - piece_size)])
+ data = data[(PIECE_SIZE - piece_size):]
+ self.size += PIECE_SIZE - piece_size
+ piece_size = self.size % PIECE_SIZE
+
+ # Hash any remaining data
+ self.pieceHasher.update(data)
+
self.fileHasher.update(data)
self.size += len(data)
+ def pieceDigests(self):
+ """Get the piece hashes of the added file data."""
+ self.digest()
+ return self.pieceHash
+
def digest(self):
"""Get the hash of the added file data."""
if self.fileHash is None:
- assert self.fileHasher is not None, "you must hash some data first"
+ if self.fileHasher is None:
+ raise HashError, "you must hash some data first"
self.fileHash = self.fileHasher.digest()
self.done = True
+
+ # Save the last piece hash
+ if self.pieceHasher:
+ self.pieceHash.append(self.pieceHasher.digest())
+ else:
+ # If there are no piece hashes, then the file hash is the only piece hash
+ self.pieceHash.append(self.fileHash)
return self.fileHash
def hexdigest(self):
self.result = (self.fileHash == self.expHash and self.size == self.expSize)
return self.result
+ def hashInThread(self, file):
+ """Hashes a file in a separate thread, callback with the result."""
+ file.restat(False)
+ if not file.exists():
+ df = defer.Deferred()
+ df.errback(HashError("file not found"))
+ return df
+
+ df = threads.deferToThread(self._hashInThread, file)
+ return df
+
+ def _hashInThread(self, file):
+ """Hashes a file, returning itself as the result."""
+ f = file.open()
+ self.new(force = True)
+ data = f.read(4096)
+ while data:
+ self.update(data)
+ data = f.read(4096)
+ self.digest()
+ return self
+
#### Methods for setting the expected hash
def set(self, hashType, hashHex, size):
"""Initialize the hash object.
def test_failure(self):
h = HashObject()
h.set(h.ORDER[0], b2a_hex('12345678901234567890'), '0')
- self.failUnlessRaises(AssertionError, h.normexpected)
- self.failUnlessRaises(AssertionError, h.digest)
- self.failUnlessRaises(AssertionError, h.hexdigest)
- self.failUnlessRaises(AssertionError, h.update, 'gfgf')
+ self.failUnlessRaises(HashError, h.normexpected)
+ self.failUnlessRaises(HashError, h.digest)
+ self.failUnlessRaises(HashError, h.hexdigest)
+ self.failUnlessRaises(HashError, h.update, 'gfgf')
+ def test_pieces(self):
+ h = HashObject()
+ h.new()
+ h.update('1234567890'*120*1024)
+ self.failUnless(h.digest() == '1(j\xd2q\x0b\n\x91\xd2\x13\x90\x15\xa3E\xcc\xb0\x8d.\xc3\xc5')
+ pieces = h.pieceDigests()
+ self.failUnless(len(pieces) == 3)
+ self.failUnless(pieces[0] == ',G \xd8\xbbPl\xf1\xa3\xa0\x0cW\n\xe6\xe6a\xc9\x95/\xe5')
+ self.failUnless(pieces[1] == '\xf6V\xeb/\xa8\xad[\x07Z\xf9\x87\xa4\xf5w\xdf\xe1|\x00\x8e\x93')
+ self.failUnless(pieces[2] == 'M[\xbf\xee\xaa+\x19\xbaV\xf699\r\x17o\xcb\x8e\xcfP\x19')
+ h.new(True)
+ for i in xrange(120*1024):
+ h.update('1234567890')
+ pieces = h.pieceDigests()
+ self.failUnless(h.digest() == '1(j\xd2q\x0b\n\x91\xd2\x13\x90\x15\xa3E\xcc\xb0\x8d.\xc3\xc5')
+ self.failUnless(len(pieces) == 3)
+ self.failUnless(pieces[0] == ',G \xd8\xbbPl\xf1\xa3\xa0\x0cW\n\xe6\xe6a\xc9\x95/\xe5')
+ self.failUnless(pieces[1] == '\xf6V\xeb/\xa8\xad[\x07Z\xf9\x87\xa4\xf5w\xdf\xe1|\x00\x8e\x93')
+ self.failUnless(pieces[2] == 'M[\xbf\xee\xaa+\x19\xbaV\xf699\r\x17o\xcb\x8e\xcfP\x19')
+
def test_sha1(self):
h = HashObject()
found = False
h.new()
h.update('apt-dht is the best')
self.failUnless(h.hexdigest() == 'c722df87e1acaa64b27aac4e174077afc3623540')
- self.failUnlessRaises(AssertionError, h.update, 'gfgf')
+ self.failUnlessRaises(HashError, h.update, 'gfgf')
self.failUnless(h.verify() == True)
def test_md5(self):
h.new()
h.update('apt-dht is the best')
self.failUnless(h.hexdigest() == '2a586bcd1befc5082c872dcd96a01403')
- self.failUnlessRaises(AssertionError, h.update, 'gfgf')
+ self.failUnlessRaises(HashError, h.update, 'gfgf')
self.failUnless(h.verify() == True)
def test_sha256(self):
h.new()
h.update('apt-dht is the best')
self.failUnless(h.hexdigest() == '55b971f64d9772f733de03f23db39224f51a455cc5ad4c2db9d5740d2ab259a7')
- self.failUnlessRaises(AssertionError, h.update, 'gfgf')
+ self.failUnlessRaises(HashError, h.update, 'gfgf')
self.failUnless(h.verify() == True)
if sys.version_info < (2, 5):
projects / quix0rs-apt-p2p.git / blobdiff