Awesome
fast-dat-parser
Parses the blockchain about as fast as your IO can pipe it out. For a typical SSD, this can be around ~450 MiB/s.
All memory is allocated up front.
Output goes to stdout
, stderr
is used for logging.
WARNING: Not actively maintained, use with caution.
Usage
A fast blk*.dat
parser for bitcoin blockchain analysis.
-j<THREADS>
- N threads for parallel computation (default1
)-m<BYTES>
- memory usage (default209715200
bytes, ~200 MiB)-t<INDEX>
- transform function (default0
, see pre-packaged transforms below)-w<FILENAME>
- whitelist file, for omitting blocks from parsing
Important to note is that the implementation skips bitcoind allocated zero-byte gaps, and includes orphan blocks unless -w
omits them.
Transforms (-t
)
Each of these pre-included functions write their output as raw data (binary, not hex). You can easily write your own though!
0
- Outputs the unordered 80-byte block headers1
- Outputs every script prefixed with auint16_t
length2
- Displays the number of transaction inputs, outputs and number of transactions in the blockchain3
- OutputsHEIGHT | VALUE
for each output, typically used for showing output balances over time
Use a whitelist (see -w
) to stop orphan blocks from being parsed. (see below for filtering by best chain)
Examples
Output all scripts for the local-best blockchain
# parse the local-best blockchain
cat ~/.bitcoin/blocks/blk*.dat | ./parser -t0 | ./bestchain > chain.dat
# output every script found in the local-best blockchain
cat ~/.bitcoin/blocks/blk*.dat | ./parser -j4 -t1 -wchain.dat > ~/.bitcoin/scripts.dat
Useful tools
These tools are for the CLI, but will aid in preparing/using data produced by the above.
bestchain
A best-chain filter for block headers.
Accepts 80-byte block headers until EOF, then finds the best-chain in the set, and outputs the best-chain in the form of a sorted hash map (see HMap<K, V>).
LICENSE MIT
The constants and getOpString
function in include/bitcoin-ops.hpp
is copied from https://github.com/bitcoin/bitcoin/.