diff options
Diffstat (limited to 'scripts/bitflash_chain.py')
| -rw-r--r-- | scripts/bitflash_chain.py | 450 |
1 files changed, 450 insertions, 0 deletions
diff --git a/scripts/bitflash_chain.py b/scripts/bitflash_chain.py new file mode 100644 index 0000000..4a3231c --- /dev/null +++ b/scripts/bitflash_chain.py | |||
| @@ -0,0 +1,450 @@ | |||
| 1 | #!/usr/bin/env python3 | ||
| 2 | """Shared Bitflash block/chain parser for deterministic audit tools.""" | ||
| 3 | |||
| 4 | import datetime as _dt | ||
| 5 | import hashlib | ||
| 6 | import json | ||
| 7 | import mmap | ||
| 8 | import os | ||
| 9 | import struct | ||
| 10 | from pathlib import Path | ||
| 11 | |||
| 12 | |||
| 13 | MAGIC = bytes([0xBF, 0x20, 0x5C, 0xFD]) | ||
| 14 | MAX_BLOCK_PAYLOAD = 32 * 1024 * 1024 | ||
| 15 | UTXO_LEAF_DOMAIN = "BTFUTXO1" | ||
| 16 | UTXO_NODE_DOMAIN_TAG = "BTFNODE1" | ||
| 17 | UTXO_NODE_DOMAIN = (UTXO_NODE_DOMAIN_TAG + "|").encode("ascii") | ||
| 18 | COIN = 100_000_000 | ||
| 19 | GENESIS_HASH = "5bd7cb255d814e48cebcdfb72da4dc87b34bd774227f8ceb546c5640f4bdc169" | ||
| 20 | GENESIS_MERKLE = "1a45b4482532abb29b10e234d3f13132230525a339ecea91658ffa675a5b1325" | ||
| 21 | GENESIS_TIME = 1_753_315_200 | ||
| 22 | GENESIS_BITS = 0x1F0FFFFF | ||
| 23 | GENESIS_NONCE = 3141 | ||
| 24 | GENESIS_TEXT = "Bitflash 24/Jul/2026 Fair launch: one CPU one vote, no premine" | ||
| 25 | |||
| 26 | |||
| 27 | class ParseError(Exception): | ||
| 28 | pass | ||
| 29 | |||
| 30 | |||
| 31 | class Reader: | ||
| 32 | def __init__(self, data): | ||
| 33 | self.data = data | ||
| 34 | self.pos = 0 | ||
| 35 | |||
| 36 | def take(self, n): | ||
| 37 | if self.pos + n > len(self.data): | ||
| 38 | raise ParseError("unexpected end of data") | ||
| 39 | out = self.data[self.pos:self.pos + n] | ||
| 40 | self.pos += n | ||
| 41 | return out | ||
| 42 | |||
| 43 | def u8(self): | ||
| 44 | return self.take(1)[0] | ||
| 45 | |||
| 46 | def u32(self): | ||
| 47 | return struct.unpack("<I", self.take(4))[0] | ||
| 48 | |||
| 49 | def i32(self): | ||
| 50 | return struct.unpack("<i", self.take(4))[0] | ||
| 51 | |||
| 52 | def i64(self): | ||
| 53 | return struct.unpack("<q", self.take(8))[0] | ||
| 54 | |||
| 55 | |||
| 56 | def sha256(data): | ||
| 57 | return hashlib.sha256(data).digest() | ||
| 58 | |||
| 59 | |||
| 60 | def sha256d(data): | ||
| 61 | return hashlib.sha256(hashlib.sha256(data).digest()).digest() | ||
| 62 | |||
| 63 | |||
| 64 | def hash_hex(serialized): | ||
| 65 | return sha256d(serialized)[::-1].hex() | ||
| 66 | |||
| 67 | |||
| 68 | def compact_size(r): | ||
| 69 | first = r.u8() | ||
| 70 | if first < 253: | ||
| 71 | return first | ||
| 72 | if first == 253: | ||
| 73 | return struct.unpack("<H", r.take(2))[0] | ||
| 74 | if first == 254: | ||
| 75 | return r.u32() | ||
| 76 | return struct.unpack("<Q", r.take(8))[0] | ||
| 77 | |||
| 78 | |||
| 79 | def read_varbytes(r): | ||
| 80 | return r.take(compact_size(r)) | ||
| 81 | |||
| 82 | |||
| 83 | def ser_compact_size(n): | ||
| 84 | if n < 253: | ||
| 85 | return bytes([n]) | ||
| 86 | if n <= 0xFFFF: | ||
| 87 | return b"\xfd" + struct.pack("<H", n) | ||
| 88 | if n <= 0xFFFFFFFF: | ||
| 89 | return b"\xfe" + struct.pack("<I", n) | ||
| 90 | return b"\xff" + struct.pack("<Q", n) | ||
| 91 | |||
| 92 | |||
| 93 | def ser_varbytes(b): | ||
| 94 | return ser_compact_size(len(b)) + b | ||
| 95 | |||
| 96 | |||
| 97 | def ser_tx(tx): | ||
| 98 | out = bytearray() | ||
| 99 | out += struct.pack("<i", tx["version"]) | ||
| 100 | out += ser_compact_size(len(tx["vin"])) | ||
| 101 | for txin in tx["vin"]: | ||
| 102 | out += bytes.fromhex(txin["prevout_hash"])[::-1] | ||
| 103 | out += struct.pack("<I", txin["prevout_n"]) | ||
| 104 | out += ser_varbytes(bytes.fromhex(txin["script_sig_hex"])) | ||
| 105 | out += struct.pack("<I", txin["sequence"]) | ||
| 106 | out += ser_compact_size(len(tx["vout"])) | ||
| 107 | for txout in tx["vout"]: | ||
| 108 | out += struct.pack("<q", txout["value_satoshis"]) | ||
| 109 | out += ser_varbytes(bytes.fromhex(txout["script_pub_key_hex"])) | ||
| 110 | out += struct.pack("<I", tx["lock_time"]) | ||
| 111 | return bytes(out) | ||
| 112 | |||
| 113 | |||
| 114 | def parse_tx(r): | ||
| 115 | start = r.pos | ||
| 116 | tx = {"version": r.i32(), "vin": [], "vout": [], "lock_time": 0} | ||
| 117 | for _ in range(compact_size(r)): | ||
| 118 | tx["vin"].append({ | ||
| 119 | "prevout_hash": r.take(32)[::-1].hex(), | ||
| 120 | "prevout_n": r.u32(), | ||
| 121 | "script_sig_hex": read_varbytes(r).hex(), | ||
| 122 | "sequence": r.u32(), | ||
| 123 | }) | ||
| 124 | for _ in range(compact_size(r)): | ||
| 125 | value = r.i64() | ||
| 126 | tx["vout"].append({ | ||
| 127 | "value_satoshis": value, | ||
| 128 | "value_btf": format_money(value), | ||
| 129 | "script_pub_key_hex": read_varbytes(r).hex(), | ||
| 130 | }) | ||
| 131 | tx["lock_time"] = r.u32() | ||
| 132 | raw = r.data[start:r.pos] | ||
| 133 | tx["txid"] = hash_hex(raw) | ||
| 134 | return tx | ||
| 135 | |||
| 136 | |||
| 137 | def merkle_root(txids): | ||
| 138 | if not txids: | ||
| 139 | return "0" * 64 | ||
| 140 | layer = [bytes.fromhex(x)[::-1] for x in txids] | ||
| 141 | while len(layer) > 1: | ||
| 142 | nxt = [] | ||
| 143 | for i in range(0, len(layer), 2): | ||
| 144 | left = layer[i] | ||
| 145 | right = layer[i + 1] if i + 1 < len(layer) else left | ||
| 146 | nxt.append(sha256d(left + right)) | ||
| 147 | layer = nxt | ||
| 148 | return layer[0][::-1].hex() | ||
| 149 | |||
| 150 | |||
| 151 | def parse_block(payload, file_name="", offset=0): | ||
| 152 | r = Reader(payload) | ||
| 153 | header = r.take(80) | ||
| 154 | version, = struct.unpack("<i", header[0:4]) | ||
| 155 | prev = header[4:36][::-1].hex() | ||
| 156 | merkle = header[36:68][::-1].hex() | ||
| 157 | ntime, bits, nonce = struct.unpack("<III", header[68:80]) | ||
| 158 | txs = [parse_tx(r) for _ in range(compact_size(r))] | ||
| 159 | if r.pos != len(payload): | ||
| 160 | raise ParseError("trailing bytes inside block") | ||
| 161 | txids = [tx["txid"] for tx in txs] | ||
| 162 | return { | ||
| 163 | "hash": hash_hex(header), | ||
| 164 | "version": version, | ||
| 165 | "previous_hash": prev, | ||
| 166 | "merkle_root": merkle, | ||
| 167 | "computed_merkle_root": merkle_root(txids), | ||
| 168 | "time": ntime, | ||
| 169 | "time_utc": utc(ntime), | ||
| 170 | "bits": "0x%08x" % bits, | ||
| 171 | "nonce": nonce, | ||
| 172 | "tx_count": len(txs), | ||
| 173 | "transactions": txs, | ||
| 174 | "source_file": file_name, | ||
| 175 | "source_offset": offset, | ||
| 176 | "size": len(payload), | ||
| 177 | } | ||
| 178 | |||
| 179 | |||
| 180 | def iter_block_files(datadir): | ||
| 181 | for path in sorted(Path(datadir).glob("blk*.dat")): | ||
| 182 | yield path | ||
| 183 | |||
| 184 | |||
| 185 | def iter_blocks_from_file(path): | ||
| 186 | path = Path(path) | ||
| 187 | with path.open("rb") as f: | ||
| 188 | f.seek(0, os.SEEK_END) | ||
| 189 | file_size = f.tell() | ||
| 190 | if file_size == 0: | ||
| 191 | return | ||
| 192 | f.seek(0) | ||
| 193 | mm = mmap.mmap(f.fileno(), 0, access=mmap.ACCESS_READ) | ||
| 194 | try: | ||
| 195 | pos = 0 | ||
| 196 | while True: | ||
| 197 | offset = mm.find(MAGIC, pos) | ||
| 198 | if offset < 0 or offset + 8 > file_size: | ||
| 199 | break | ||
| 200 | size = struct.unpack("<I", mm[offset + 4:offset + 8])[0] | ||
| 201 | payload_offset = offset + 8 | ||
| 202 | end = payload_offset + size | ||
| 203 | if size == 0 or size > MAX_BLOCK_PAYLOAD or end > file_size: | ||
| 204 | pos = offset + 1 | ||
| 205 | continue | ||
| 206 | payload = mm[payload_offset:end] | ||
| 207 | try: | ||
| 208 | yield parse_block(payload, path.name, payload_offset) | ||
| 209 | except ParseError: | ||
| 210 | pos = offset + 1 | ||
| 211 | continue | ||
| 212 | pos = end | ||
| 213 | finally: | ||
| 214 | mm.close() | ||
| 215 | |||
| 216 | |||
| 217 | def read_blocks_from_files(paths, max_blocks=0): | ||
| 218 | blocks = [] | ||
| 219 | for path in paths: | ||
| 220 | for block in iter_blocks_from_file(path): | ||
| 221 | blocks.append(block) | ||
| 222 | if max_blocks and len(blocks) >= max_blocks: | ||
| 223 | return blocks | ||
| 224 | return blocks | ||
| 225 | |||
| 226 | |||
| 227 | def read_blocks(datadir, max_blocks=0): | ||
| 228 | return read_blocks_from_files(iter_block_files(datadir), max_blocks) | ||
| 229 | |||
| 230 | |||
| 231 | def select_main_chain(blocks): | ||
| 232 | by_hash = {} | ||
| 233 | for block in blocks: | ||
| 234 | by_hash.setdefault(block["hash"], block) | ||
| 235 | if GENESIS_HASH not in by_hash: | ||
| 236 | raise ParseError("genesis block not found in block files") | ||
| 237 | |||
| 238 | children = {} | ||
| 239 | for block in by_hash.values(): | ||
| 240 | children.setdefault(block["previous_hash"], []).append(block) | ||
| 241 | for v in children.values(): | ||
| 242 | v.sort(key=lambda b: (b["hash"], b["source_file"], b["source_offset"])) | ||
| 243 | |||
| 244 | best = [] | ||
| 245 | stack = [(by_hash[GENESIS_HASH], [by_hash[GENESIS_HASH]])] | ||
| 246 | while stack: | ||
| 247 | block, path = stack.pop() | ||
| 248 | if len(path) > len(best) or (len(path) == len(best) and block["hash"] < best[-1]["hash"]): | ||
| 249 | best = path | ||
| 250 | for child in children.get(block["hash"], []): | ||
| 251 | stack.append((child, path + [child])) | ||
| 252 | return best | ||
| 253 | |||
| 254 | |||
| 255 | def is_coinbase(tx): | ||
| 256 | return ( | ||
| 257 | len(tx["vin"]) == 1 | ||
| 258 | and tx["vin"][0]["prevout_hash"] == "0" * 64 | ||
| 259 | and tx["vin"][0]["prevout_n"] == 0xFFFFFFFF | ||
| 260 | ) | ||
| 261 | |||
| 262 | |||
| 263 | def utxo_leaf_hash(outpoint, utxo): | ||
| 264 | payload = "|".join([ | ||
| 265 | UTXO_LEAF_DOMAIN, | ||
| 266 | outpoint, | ||
| 267 | str(utxo["value_satoshis"]), | ||
| 268 | utxo["script_pub_key_hex"], | ||
| 269 | str(utxo["created_height"]), | ||
| 270 | utxo["created_txid"], | ||
| 271 | ]).encode("ascii") | ||
| 272 | return sha256(payload) | ||
| 273 | |||
| 274 | |||
| 275 | def utxo_root(utxos): | ||
| 276 | leaves = [utxo_leaf_hash(k, utxos[k]) for k in sorted(utxos)] | ||
| 277 | if not leaves: | ||
| 278 | return sha256(b"BTFUTXO1|empty").hex() | ||
| 279 | layer = leaves | ||
| 280 | while len(layer) > 1: | ||
| 281 | nxt = [] | ||
| 282 | for i in range(0, len(layer), 2): | ||
| 283 | left = layer[i] | ||
| 284 | right = layer[i + 1] if i + 1 < len(layer) else left | ||
| 285 | nxt.append(sha256(UTXO_NODE_DOMAIN + left + right)) | ||
| 286 | layer = nxt | ||
| 287 | return layer[0].hex() | ||
| 288 | |||
| 289 | |||
| 290 | def apply_blocks(blocks, strict_duplicates=False): | ||
| 291 | utxos = {} | ||
| 292 | spent = 0 | ||
| 293 | missing_spends = [] | ||
| 294 | duplicate_outputs = [] | ||
| 295 | total_coinbase = 0 | ||
| 296 | total_fees = 0 | ||
| 297 | normal_txs = 0 | ||
| 298 | |||
| 299 | for height, block in enumerate(blocks): | ||
| 300 | if block["computed_merkle_root"] != block["merkle_root"]: | ||
| 301 | raise ParseError("bad merkle root at height %d" % height) | ||
| 302 | for tx in block["transactions"]: | ||
| 303 | coinbase = is_coinbase(tx) | ||
| 304 | input_value = 0 | ||
| 305 | output_value = sum(out["value_satoshis"] for out in tx["vout"]) | ||
| 306 | if coinbase: | ||
| 307 | total_coinbase += output_value | ||
| 308 | else: | ||
| 309 | normal_txs += 1 | ||
| 310 | for txin in tx["vin"]: | ||
| 311 | prev = "%s:%d" % (txin["prevout_hash"], txin["prevout_n"]) | ||
| 312 | prev_utxo = utxos.pop(prev, None) | ||
| 313 | if prev_utxo is None: | ||
| 314 | missing_spends.append({ | ||
| 315 | "height": height, | ||
| 316 | "txid": tx["txid"], | ||
| 317 | "missing_outpoint": prev, | ||
| 318 | }) | ||
| 319 | else: | ||
| 320 | spent += prev_utxo["value_satoshis"] | ||
| 321 | input_value += prev_utxo["value_satoshis"] | ||
| 322 | if input_value >= output_value: | ||
| 323 | total_fees += input_value - output_value | ||
| 324 | for n, out in enumerate(tx["vout"]): | ||
| 325 | key = "%s:%d" % (tx["txid"], n) | ||
| 326 | if key in utxos: | ||
| 327 | if strict_duplicates: | ||
| 328 | raise ParseError("duplicate output at height %d: %s" % (height, key)) | ||
| 329 | duplicate_outputs.append({"height": height, "outpoint": key}) | ||
| 330 | utxos[key] = { | ||
| 331 | "txid": tx["txid"], | ||
| 332 | "vout": n, | ||
| 333 | "value_satoshis": out["value_satoshis"], | ||
| 334 | "value_btf": out["value_btf"], | ||
| 335 | "script_pub_key_hex": out["script_pub_key_hex"], | ||
| 336 | "created_height": height, | ||
| 337 | "created_block_hash": block["hash"], | ||
| 338 | "created_txid": tx["txid"], | ||
| 339 | "coinbase": coinbase, | ||
| 340 | } | ||
| 341 | |||
| 342 | total_unspent = sum(u["value_satoshis"] for u in utxos.values()) | ||
| 343 | return { | ||
| 344 | "utxos": utxos, | ||
| 345 | "stats": { | ||
| 346 | "utxo_count": len(utxos), | ||
| 347 | "spent_outputs": spent, | ||
| 348 | "total_unspent_satoshis": total_unspent, | ||
| 349 | "total_unspent_btf": format_money(total_unspent), | ||
| 350 | "total_coinbase_satoshis": total_coinbase, | ||
| 351 | "total_coinbase_btf": format_money(total_coinbase), | ||
| 352 | "coinbase_minus_unspent_satoshis": total_coinbase - total_unspent, | ||
| 353 | "coinbase_minus_unspent_btf": format_money(total_coinbase - total_unspent), | ||
| 354 | "total_fees_satoshis": total_fees, | ||
| 355 | "total_fees_btf": format_money(total_fees), | ||
| 356 | "normal_transactions": normal_txs, | ||
| 357 | "missing_spends": missing_spends, | ||
| 358 | "duplicate_outputs": duplicate_outputs, | ||
| 359 | }, | ||
| 360 | } | ||
| 361 | |||
| 362 | |||
| 363 | def genesis_transaction(): | ||
| 364 | script_sig = ( | ||
| 365 | bytes.fromhex("04ffff001d") | ||
| 366 | + b"\x01\x04" | ||
| 367 | + bytes([len(GENESIS_TEXT)]) | ||
| 368 | + GENESIS_TEXT.encode("ascii") | ||
| 369 | ) | ||
| 370 | pubkey_num = bytes.fromhex( | ||
| 371 | "5f1df16b2b704c8a578d0bbaf74d385cde12c11ee50455f3c438ef4c3fbcf649" | ||
| 372 | "b6de611feae06279a60939e028a8d65c10b73071a6f16719274855feb0fd8a6704" | ||
| 373 | ) | ||
| 374 | script_pub_key = bytes([65]) + pubkey_num[::-1] + b"\xac" | ||
| 375 | tx = { | ||
| 376 | "version": 1, | ||
| 377 | "vin": [{ | ||
| 378 | "prevout_hash": "0" * 64, | ||
| 379 | "prevout_n": 0xFFFFFFFF, | ||
| 380 | "script_sig_hex": script_sig.hex(), | ||
| 381 | "sequence": 0xFFFFFFFF, | ||
| 382 | }], | ||
| 383 | "vout": [{ | ||
| 384 | "value_satoshis": 50 * COIN, | ||
| 385 | "value_btf": format_money(50 * COIN), | ||
| 386 | "script_pub_key_hex": script_pub_key.hex(), | ||
| 387 | }], | ||
| 388 | "lock_time": 0, | ||
| 389 | } | ||
| 390 | tx["txid"] = hash_hex(ser_tx(tx)) | ||
| 391 | return tx | ||
| 392 | |||
| 393 | |||
| 394 | def reconstructed_genesis(): | ||
| 395 | tx = genesis_transaction() | ||
| 396 | header = ( | ||
| 397 | struct.pack("<i", 1) | ||
| 398 | + bytes(32) | ||
| 399 | + bytes.fromhex(tx["txid"])[::-1] | ||
| 400 | + struct.pack("<III", GENESIS_TIME, GENESIS_BITS, GENESIS_NONCE) | ||
| 401 | ) | ||
| 402 | return { | ||
| 403 | "hash": hash_hex(header), | ||
| 404 | "version": 1, | ||
| 405 | "previous_hash": "0" * 64, | ||
| 406 | "merkle_root": tx["txid"], | ||
| 407 | "computed_merkle_root": tx["txid"], | ||
| 408 | "time": GENESIS_TIME, | ||
| 409 | "time_utc": utc(GENESIS_TIME), | ||
| 410 | "bits": "0x%08x" % GENESIS_BITS, | ||
| 411 | "nonce": GENESIS_NONCE, | ||
| 412 | "tx_count": 1, | ||
| 413 | "transactions": [tx], | ||
| 414 | "source_file": "reconstructed-from-source-constants", | ||
| 415 | "source_offset": 0, | ||
| 416 | "size": 80 + len(ser_compact_size(1)) + len(ser_tx(tx)), | ||
| 417 | } | ||
| 418 | |||
| 419 | |||
| 420 | def ser_block(block): | ||
| 421 | header = ( | ||
| 422 | struct.pack("<i", block["version"]) | ||
| 423 | + bytes.fromhex(block["previous_hash"])[::-1] | ||
| 424 | + bytes.fromhex(block["merkle_root"])[::-1] | ||
| 425 | + struct.pack("<III", block["time"], int(block["bits"], 16), block["nonce"]) | ||
| 426 | ) | ||
| 427 | txs = block["transactions"] | ||
| 428 | return header + ser_compact_size(len(txs)) + b"".join(ser_tx(tx) for tx in txs) | ||
| 429 | |||
| 430 | |||
| 431 | def format_money(value): | ||
| 432 | sign = "-" if value < 0 else "" | ||
| 433 | value = abs(value) | ||
| 434 | return "%s%d.%08d" % (sign, value // COIN, value % COIN) | ||
| 435 | |||
| 436 | |||
| 437 | def utc(ts): | ||
| 438 | return _dt.datetime.fromtimestamp(ts, _dt.timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ") | ||
| 439 | |||
| 440 | |||
| 441 | def default_datadir(): | ||
| 442 | if os.name == "nt" and os.environ.get("APPDATA"): | ||
| 443 | return os.path.join(os.environ["APPDATA"], "Bitflash") | ||
| 444 | if os.environ.get("HOME"): | ||
| 445 | return os.path.join(os.environ["HOME"], ".bitflash") | ||
| 446 | return "." | ||
| 447 | |||
| 448 | |||
| 449 | def canonical_bytes(report): | ||
| 450 | return (json.dumps(report, sort_keys=True, separators=(",", ":"), ensure_ascii=True) + "\n").encode("ascii") | ||