Hash and track layer file chunks for torrenting

util/registry/torrent.py

@@ -0,0 +1,43 @@
+import resumablehashlib
+
+
+class PieceHasher(object):
+  def __init__(self, piece_size, starting_offset, starting_piece_hash_str, hash_fragment_to_resume):
+    if not isinstance(starting_offset, (int, long)):
+      raise TypeError('starting_offset must be an integer')
+    elif not isinstance(piece_size, (int, long)):
+      raise TypeError('piece_size must be an integer')
+
+    self._current_offset = starting_offset
+    self._piece_size = piece_size
+    self._hash_fragment = hash_fragment_to_resume
+    self._piece_hashes = [starting_piece_hash_str]
+
+  def update(self, buf):
+    buf_offset = 0
+    while buf_offset < len(buf):
+      buf_bytes_to_hash = buf[0:self._piece_length_remaining()]
+      to_hash_len = len(buf_bytes_to_hash)
+
+      if self._piece_offset() == 0 and to_hash_len > 0 and self._current_offset > 0:
+        # We are opening a new piece
+        self._piece_hashes.append(self._hash_fragment.hexdigest())
+        self._hash_fragment = resumablehashlib.sha1()
+
+      self._hash_fragment.update(buf_bytes_to_hash)
+      self._current_offset += to_hash_len
+      buf_offset += to_hash_len
+
+  def _piece_length_remaining(self):
+    return self._piece_size - (self._current_offset % self._piece_size)
+
+  def _piece_offset(self):
+    return self._current_offset % self._piece_size
+
+  @property
+  def piece_hashes(self):
+    return ''.join(self._piece_hashes)
+
+  @property
+  def hash_fragment(self):
+    return self._hash_fragment