Topic: [CORRECTED] Script calculates 71% freeable transactions (Read 2123 times)

diff --git a/deserialize.py b/deserialize.py
index fe0cb09..a67645b 100644
--- a/deserialize.py
+++ b/deserialize.py
@@ -75,6 +75,7 @@ def deserialize_TxOut(d, owner_keys=None):
 
 def parse_Transaction(vds):
   d = {}
+  start = vds.read_cursor  # XXX breaks BCDataStream abstraction, do we care?
   d['version'] = vds.read_int32()
   n_vin = vds.read_compact_size()
   d['txIn'] = []
@@ -85,6 +86,7 @@ def parse_Transaction(vds):
   for i in xrange(n_vout):
     d['txOut'].append(parse_TxOut(vds))
   d['lockTime'] = vds.read_uint32()
+  d['tx'] = vds.input[start:vds.read_cursor]
   return d
 def deserialize_Transaction(d, transaction_index=None, owner_keys=None):
   result = "%d tx in, %d out\n"%(len(d['txIn']), len(d['txOut']))
diff --git a/freeable.py b/freeable.py
new file mode 100755
index 0000000..c79f458
--- /dev/null
+++ b/freeable.py
@@ -0,0 +1,87 @@
+#!/usr/bin/env python
+#
+# Read the block database, find out how many transactions
+# could be purged and how many bytes they take up.
+# TODO: find out how many Merkle hashes could be purged.
+#
+from bsddb.db import *
+import logging
+import os
+import sys
+
+from BCDataStream import *
+from block import scan_blocks
+from deserialize import parse_Block
+from util import determine_db_dir, create_env
+import Crypto.Hash.SHA256 as SHA256
+
+import binascii  # debugging
+
+def main():
+  import optparse
+  parser = optparse.OptionParser(usage="%prog [options]")
+  parser.add_option("--datadir", dest="datadir", default=None,
+                    help="Look for files here (defaults to bitcoin default)")
+  (options, args) = parser.parse_args()
+
+  if options.datadir is None:
+    db_dir = determine_db_dir()
+  else:
+    db_dir = options.datadir
+
+  try:
+    db_env = create_env(db_dir)
+  except DBNoSuchFileError:
+    logging.error("Couldn't open " + db_dir)
+    sys.exit(1)
+
+  blockfile = open(os.path.join(db_dir, "blk%04d.dat"%(1,)), "rb")
+  block_datastream = BCDataStream()
+  block_datastream.map_file(blockfile, 0)
+
+  blocks = []
+  def gather_stats(block_data):
+    block_datastream.seek_file(block_data['nBlockPos'])
+    blocks.append(parse_Block(block_datastream))
+    return True
+
+  scan_blocks(db_dir, db_env, gather_stats)
+  blocks.reverse()
+  print 'loaded ', len(blocks), ' blocks'
+
+  tx = {}
+  total_tx = 0
+  freeable_tx = 0
+  total_bytes = 0
+  freeable_bytes = 0
+
+  for data in blocks:
+    coinbase = True
+    for txn in data['transactions']:
+      tx_hash = SHA256.new(SHA256.new(txn['tx']).digest()).digest()
+      #print '> ', binascii.hexlify(tx_hash)
+      tx_bytes = len(txn['tx'])
+      tx[tx_hash] = (tx_bytes, len(txn['txOut']))
+      total_tx += 1
+      total_bytes += tx_bytes
+
+      if coinbase:
+        coinbase = False
+      else:
+        for txin in txn['txIn']:
+          #print '< ', binascii.hexlify(txin['prevout_hash'])
+          (bytes, live) = tx[txin['prevout_hash']]
+          if live == 1:
+            freeable_bytes += bytes
+            freeable_tx += 1
+            del tx[txin['prevout_hash']]
+          else:
+            tx[txin['prevout_hash']] = (bytes, live - 1)
+
+  db_env.close()
+
+  print freeable_tx, 'of', total_tx, 'tx freeable'
+  print freeable_bytes, 'of', total_bytes, 'tx bytes freeable'
+
+if __name__ == '__main__':
+    main()