How to decode messages stored in the Bitcoin blockchain with btcmsg

direct_from_web.py

# directly from the web:

from protocol import read_msg

def get_msg_from_transaction(url):
    import json
    import urllib2
    data = urllib2.urlopen(url)
    j = json.load(data)
    out=j['out']
    values=[]
    for v in out:
        values.append(v['value'])
    return read_msg(values)

def test():
    message=get_msg_from_transaction('http://blockexplorer.com/tx/cc2fa5a40cdbaad70826040fe7fe4bec34ddfaefd358ca1e2a7957f1d5941738')
    print message

test()

protocol.py

import binascii

# A fixed fee for any message (Satoshi) 0.005 BTC
FIXED_COSTS=500000
# Additional fee for each transaction value entry (Satoshi) 0.0005 BTC
VARIABLE_COSTS=50000

RESERVED_IDENTIFIER='00'
MD5SUM_IDENTIFIER='01'
ASCII_IDENTIFIER='02'

def write_msg(h,data_type):
    output=[]
    # First transaction value with costs (placeholder)
    output.append('')
    # Mark data type with 2 more decimal digits (100 options).
    if data_type=='md5':
        x=h
        # Verify h is 32 hexadecimal chars long
        if len(x) != 32:
            print 'md5 must be exactly 32 chars long'
            return ([],'')
        for j in range(32):
            if x[j]<'0' or x[j]>'f':
                print 'md5 chars are only 0-9,a-f'
                return ([],'')
        # Split to groups with char length
        # 2,4,4,...,4,2
        s=[]
        s.append(MD5SUM_IDENTIFIER+h[:2]) # Mark md5 msg with MD5SUM_IDENTIFIER
        for i in range(7):
            s.append(h[2+i*4:6+i*4])
        s.append(h[30:33]+'00')
        for t in s:
            output.append('0.'+'%08d' % (int(t,16)))
    if data_type=='ascii':
        # Verify that it is an ascii string (limited with length)
        x=binascii.hexlify(h)
        # Split to groups with char length
        # 2,4,4,...,4,2
        s=[]
        x=ASCII_IDENTIFIER+x # Mark ascii msg with header ASCII_IDENTIFIER
        for i in range(len(x)/4):
            s.append(x[i*4:(i+1)*4])
        if h[(i+1)*4:] != '':
            s.append(h[(i+1)*4:]+'00')
        for t in s:
            output.append('0.'+'%08d' % (int(t,16),))
    # Update variable fee
    output[0]=str((0.0+int(FIXED_COSTS)+len(output)*int(VARIABLE_COSTS))/10000000)
    return (output,x)

def read_msg(h):
    try:
        # Ignore first entry (where the fee is stored)
        h=h[1:]
    except IndexError:
        pass
    s=''
    for a in h:
        tmp=a[3:]
        tmp=int(tmp)
        s+="%05d" % (tmp,)
    l=[]
    for i in range(len(h)):
        l.append('%04x' % (int(s[i*5:(i+1)*5])))
    try:
        data_type = l[0][:2]
    except IndexError:
        print 'Error parsing list %s' % str(l)
        return ''
    if data_type == MD5SUM_IDENTIFIER: # md5
        x=''
        for t in l:
            x+=t
        return x[2:-2]
    else:
        if data_type == ASCII_IDENTIFIER: # ascii 
            x=''
            for t in l:
                x+=t
            m=x[2:]
            return binascii.unhexlify(m)
        else:
            print 'Unrecognized data type %s' % (data_type,)

def price(h):
    p=0
    for a in h:
       p+=float(a)
    return round(p,8)

Protocol.txt

BTCmsg Protocol v1 (2011-09-18)
===============================

Each message is represented by multiple payment which is calculated by
the following algorithm:
1. Two first chars for message type ('01' for md5, '02' for ascii).
2. Then the message in hex (python binascii.hexlify).
3. Split the long string to groups of 4 hex digits.
4. Each group of 4 hex (e.g. 2 ascii letters from the message) is 
   represented by a payment in satoshi (maximum 0xffff=65535).
5. The service fee (composed of a fixed part and a variable part times 
   the amount of small payments required, minus the transaction fee),
   can be seen within the first transaction.

Enjoy!

Extra link to msgdump.tzg: http://dl.dropbox.com/u/464119/Bitcoin/btcmsg/msgdump.tgz