Skip to content

Commit 2352f82

Browse files
ragestackragestack
authored
Merge pull request #31 from ragestack/release/pypi
Fix PyPI package: bundle blockchain-parser.py (v2.0.3)
2 parents cedc8c5 + 9bdf4df commit 2352f82

File tree

5 files changed

+357
-14
lines changed

5 files changed

+357
-14
lines changed

‎CHANGELOG.md‎

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,8 @@
11
# Changelog
22

3+
## 2.0.3
4+
- Fixed PyPI package: bundle `blockchain-parser.py` so the `blockchain-scalpel` CLI works after installation.
5+
36
## 2.0.2
47
- Added PyPI packaging and CLI entrypoint: `blockchain-scalpel`
58

‎blockchain_scalpel/__init__.py‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1 +1 @@
1-
__version__ = "2.0.2"
1+
__version__ = "2.0.3"

‎blockchain_scalpel/blockchain-parser.py‎

Lines changed: 339 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,339 @@
1+
#!/usr/bin/env python3
2+
# -*- coding: utf-8 -*-
3+
"""
4+
blockchain-parser.py
5+
6+
Author: Denis Leonov
7+
Project: Blockchain Parser (Blockchain Scalpel)
8+
Repository: https://github.com/ragestack/blockchain-parser
9+
Version: 2.0.0
10+
11+
License:
12+
Blockchain Scalpel License (Source-Available, Non-Commercial)
13+
Free for non-commercial use with attribution.
14+
Commercial use and any SaaS/cloud/hosted use require a paid license.
15+
16+
Commercial licensing contact: 466611@gmail.com
17+
Other author's contact info: https://aaris.ru/DL
18+
19+
Limited warranty:
20+
A limited compatibility warranty related to block format changes
21+
is provided. See LICENSE and WARRANTY.md for details.
22+
"""
23+
24+
__version__ = "2.0.0"
25+
__author__ = "Denis Leonov"
26+
27+
# SPDX-License-Identifier: LicenseRef-Blockchain-Scalpel
28+
29+
import os, io, sys
30+
import datetime
31+
import hashlib
32+
33+
def reverse(input):
34+
L = len(input)
35+
if (L % 2) != 0:
36+
return None
37+
else:
38+
Res = ''
39+
L = L // 2
40+
for i in range(L):
41+
T = input[i*2] + input[i*2+1]
42+
Res = T + Res
43+
T = ''
44+
return (Res);
45+
46+
def merkle_root(h):
47+
d = lambda b:hashlib.sha256(hashlib.sha256(b).digest()).digest()
48+
rev = lambda x:x[::-1]
49+
h = list(map(rev,h))
50+
while len(h) > 1:
51+
if len(h)&1:h += h[-1:]
52+
h = [d(h[i]+h[i+1]) for i in range(0,len(h),2)]
53+
return rev(h[0])
54+
55+
def read_bytes(file,n,byte_order = 'L'):
56+
data = file.read(n)
57+
if byte_order == 'L':
58+
data = data[::-1]
59+
data = data.hex().upper()
60+
return data
61+
62+
def read_varint(file):
63+
b = file.read(1)
64+
bInt = int(b.hex(),16)
65+
c = 0
66+
data = ''
67+
if bInt < 253:
68+
c = 1
69+
data = b.hex().upper()
70+
if bInt == 253: c = 3
71+
if bInt == 254: c = 5
72+
if bInt == 255: c = 9
73+
for j in range(1,c):
74+
b = file.read(1)
75+
b = b.hex().upper()
76+
data = b + data
77+
return data
78+
79+
def print_help(script_name):
80+
print(
81+
f"Usage:\n"
82+
f" python {script_name} <dirA> <dirB>\n\n"
83+
f"Arguments:\n"
84+
f" dirA Directory where blk*.dat files are stored (must exist)\n"
85+
f" dirB Output directory for parsing results (must already exist)\n"
86+
)
87+
88+
def main():
89+
script_name = os.path.basename(sys.argv[0])
90+
91+
if len(sys.argv) == 2 and sys.argv[1] in ("-h", "--help"):
92+
print_help(script_name)
93+
sys.exit(0)
94+
95+
if len(sys.argv) != 3:
96+
print("Error: exactly 2 arguments are required.\n", file=sys.stderr)
97+
print_help(script_name)
98+
sys.exit(1)
99+
100+
dirA = os.path.abspath(os.path.expanduser(sys.argv[1]))
101+
dirB = os.path.abspath(os.path.expanduser(sys.argv[2]))
102+
103+
if not os.path.isdir(dirA):
104+
print(f"Error: input directory does not exist or is not a directory:\n {dirA}", file=sys.stderr)
105+
sys.exit(1)
106+
107+
if not os.path.exists(dirB):
108+
print(
109+
f"Error: output directory does not exist (create it first):\n {dirB}", file=sys.stderr)
110+
sys.exit(1)
111+
112+
if not os.path.isdir(dirB):
113+
print(f"Error: output path exists but is not a directory:\n {dirB}", file=sys.stderr)
114+
sys.exit(1)
115+
116+
if not os.access(dirB, os.W_OK):
117+
print(f"Error: output directory is not writable:\n {dirB}", file=sys.stderr)
118+
sys.exit(1)
119+
120+
kN = os.path.join(dirA, 'xor.dat')
121+
122+
k_ = b'\x00\x00\x00\x00\x00\x00\x00\x00'
123+
lk_ = len(k_)
124+
if os.path.isfile(kN):
125+
with open(kN, 'rb') as kF:
126+
k_ = kF.read(lk_)
127+
128+
fList = [x for x in os.listdir(dirA) if x.endswith('.dat') and x.startswith('blk')]
129+
rList = [x for x in os.listdir(dirB) if x.endswith('.txt') and x.startswith('blk')]
130+
131+
rNames = {os.path.splitext(x)[0] for x in rList}
132+
133+
fList = [x for x in fList if os.path.splitext(x)[0] not in rNames]
134+
fList.sort()
135+
136+
for i in fList:
137+
nameSrc = i
138+
nameRes = nameSrc.replace('.dat','.txt')
139+
resList = []
140+
a = 0
141+
t = os.path.join(dirA, nameSrc)
142+
resList.append('Start ' + t + ' in ' + str(datetime.datetime.now()))
143+
print ('Start ' + t + ' in ' + str(datetime.datetime.now()))
144+
with open(t,'rb') as f0:
145+
b_ = bytearray(f0.read())
146+
if any(k_):
147+
for ii in range(len(b_)):
148+
b_[ii] ^= k_[ii%lk_]
149+
f = io.BytesIO(b_)
150+
tmpHex = ''
151+
fSize = os.path.getsize(t)
152+
while f.tell() != fSize:
153+
tmpErr = 0
154+
while tmpHex != 'D9B4BEF9': # it is for to skip zeroes in some blk files
155+
tmpHex = read_bytes(f,4)
156+
tmpErr += 1
157+
if tmpErr > 2:
158+
raise ValueError(f"Invalid data: magic number missing — possible truncated {i} file")
159+
resList.append('Magic number = ' + tmpHex)
160+
tmpHex = read_bytes(f,4)
161+
resList.append('Block size = ' + tmpHex)
162+
tmpPos3 = f.tell()
163+
tmpHex = read_bytes(f,80,'B')
164+
tmpHex = bytes.fromhex(tmpHex)
165+
tmpHex = hashlib.new('sha256', tmpHex).digest()
166+
tmpHex = hashlib.new('sha256', tmpHex).digest()
167+
tmpHex = tmpHex[::-1]
168+
tmpHex = tmpHex.hex().upper()
169+
resList.append('SHA256 hash of the current block hash = ' + tmpHex)
170+
f.seek(tmpPos3,0)
171+
tmpHex = read_bytes(f,4)
172+
resList.append('Version number = ' + tmpHex)
173+
tmpHex = read_bytes(f,32)
174+
resList.append('SHA256 hash of the previous block hash = ' + tmpHex)
175+
tmpHex = read_bytes(f,32)
176+
resList.append('MerkleRoot hash = ' + tmpHex)
177+
MerkleRoot = tmpHex
178+
tmpHex = read_bytes(f,4)
179+
resList.append('Time stamp = ' + tmpHex)
180+
tmpHex = read_bytes(f,4)
181+
resList.append('Difficulty = ' + tmpHex)
182+
tmpHex = read_bytes(f,4)
183+
resList.append('Random number = ' + tmpHex)
184+
tmpHex = read_varint(f)
185+
txCount = int(tmpHex,16)
186+
resList.append('Transactions count = ' + str(txCount))
187+
resList.append('')
188+
tmpHex = ''; RawTX = ''; tx_hashes = []
189+
for k in range(txCount):
190+
tmpHex = read_bytes(f,4)
191+
resList.append('TX version number = ' + tmpHex)
192+
RawTX = reverse(tmpHex)
193+
tmpHex = ''
194+
Witness = False
195+
b = f.read(1)
196+
tmpB = b.hex().upper()
197+
bInt = int(b.hex(),16)
198+
if bInt == 0:
199+
tmpB = ''
200+
f.seek(1,1)
201+
c = 0
202+
c = f.read(1)
203+
bInt = int(c.hex(),16)
204+
tmpB = c.hex().upper()
205+
Witness = True
206+
c = 0
207+
if bInt < 253:
208+
c = 1
209+
tmpHex = hex(bInt)[2:].upper().zfill(2)
210+
tmpB = ''
211+
if bInt == 253: c = 3
212+
if bInt == 254: c = 5
213+
if bInt == 255: c = 9
214+
for j in range(1,c):
215+
b = f.read(1)
216+
b = b.hex().upper()
217+
tmpHex = b + tmpHex
218+
inCount = int(tmpHex,16)
219+
resList.append('Inputs count = ' + tmpHex)
220+
tmpHex = tmpHex + tmpB
221+
RawTX = RawTX + reverse(tmpHex)
222+
for m in range(inCount):
223+
tmpHex = read_bytes(f,32)
224+
resList.append('TX from hash = ' + tmpHex)
225+
RawTX = RawTX + reverse(tmpHex)
226+
tmpHex = read_bytes(f,4)
227+
resList.append('N output = ' + tmpHex)
228+
RawTX = RawTX + reverse(tmpHex)
229+
tmpHex = ''
230+
b = f.read(1)
231+
tmpB = b.hex().upper()
232+
bInt = int(b.hex(),16)
233+
c = 0
234+
if bInt < 253:
235+
c = 1
236+
tmpHex = b.hex().upper()
237+
tmpB = ''
238+
if bInt == 253: c = 3
239+
if bInt == 254: c = 5
240+
if bInt == 255: c = 9
241+
for j in range(1,c):
242+
b = f.read(1)
243+
b = b.hex().upper()
244+
tmpHex = b + tmpHex
245+
scriptLength = int(tmpHex,16)
246+
tmpHex = tmpHex + tmpB
247+
RawTX = RawTX + reverse(tmpHex)
248+
tmpHex = read_bytes(f,scriptLength,'B')
249+
resList.append('Input script = ' + tmpHex)
250+
RawTX = RawTX + tmpHex
251+
tmpHex = read_bytes(f,4,'B')
252+
resList.append('Sequence number = ' + tmpHex)
253+
RawTX = RawTX + tmpHex
254+
tmpHex = ''
255+
b = f.read(1)
256+
tmpB = b.hex().upper()
257+
bInt = int(b.hex(),16)
258+
c = 0
259+
if bInt < 253:
260+
c = 1
261+
tmpHex = b.hex().upper()
262+
tmpB = ''
263+
if bInt == 253: c = 3
264+
if bInt == 254: c = 5
265+
if bInt == 255: c = 9
266+
for j in range(1,c):
267+
b = f.read(1)
268+
b = b.hex().upper()
269+
tmpHex = b + tmpHex
270+
outputCount = int(tmpHex,16)
271+
tmpHex = tmpHex + tmpB
272+
resList.append('Outputs count = ' + str(outputCount))
273+
RawTX = RawTX + reverse(tmpHex)
274+
for m in range(outputCount):
275+
tmpHex = read_bytes(f,8)
276+
Value = tmpHex
277+
RawTX = RawTX + reverse(tmpHex)
278+
tmpHex = ''
279+
b = f.read(1)
280+
tmpB = b.hex().upper()
281+
bInt = int(b.hex(),16)
282+
c = 0
283+
if bInt < 253:
284+
c = 1
285+
tmpHex = b.hex().upper()
286+
tmpB = ''
287+
if bInt == 253: c = 3
288+
if bInt == 254: c = 5
289+
if bInt == 255: c = 9
290+
for j in range(1,c):
291+
b = f.read(1)
292+
b = b.hex().upper()
293+
tmpHex = b + tmpHex
294+
scriptLength = int(tmpHex,16)
295+
tmpHex = tmpHex + tmpB
296+
RawTX = RawTX + reverse(tmpHex)
297+
tmpHex = read_bytes(f,scriptLength,'B')
298+
resList.append('Value = ' + Value)
299+
resList.append('Output script = ' + tmpHex)
300+
RawTX = RawTX + tmpHex
301+
tmpHex = ''
302+
if Witness == True:
303+
for m in range(inCount):
304+
tmpHex = read_varint(f)
305+
WitnessLength = int(tmpHex,16)
306+
for j in range(WitnessLength):
307+
tmpHex = read_varint(f)
308+
WitnessItemLength = int(tmpHex,16)
309+
tmpHex = read_bytes(f,WitnessItemLength)
310+
resList.append('Witness ' + str(m) + ' ' + str(j) + ' ' + str(WitnessItemLength) + ' ' + tmpHex)
311+
tmpHex = ''
312+
Witness = False
313+
tmpHex = read_bytes(f,4)
314+
resList.append('Lock time = ' + tmpHex)
315+
RawTX = RawTX + reverse(tmpHex)
316+
tmpHex = RawTX
317+
tmpHex = bytes.fromhex(tmpHex)
318+
tmpHex = hashlib.new('sha256', tmpHex).digest()
319+
tmpHex = hashlib.new('sha256', tmpHex).digest()
320+
tmpHex = tmpHex[::-1]
321+
tmpHex = tmpHex.hex().upper()
322+
resList.append('TX hash = ' + tmpHex)
323+
tx_hashes.append(tmpHex)
324+
resList.append(''); tmpHex = ''; RawTX = ''
325+
a += 1
326+
tx_hashes = [bytes.fromhex(h) for h in tx_hashes]
327+
tmpHex = merkle_root(tx_hashes).hex().upper()
328+
if tmpHex != MerkleRoot:
329+
print ('Merkle roots does not match! >',MerkleRoot,tmpHex)
330+
f.close()
331+
f = open(os.path.join(dirB, nameRes),'w')
332+
for j in resList:
333+
f.write(j + '\n')
334+
f.close()
335+
336+
print ('All done ' + str(datetime.datetime.now()))
337+
338+
if __name__ == "__main__":
339+
main()

‎blockchain_scalpel/cli.py‎

Lines changed: 12 additions & 12 deletions
Original file line numberDiff line numberDiff line change
@@ -1,27 +1,27 @@
11
import runpy
22
import sys
3-
from pathlib import Path
3+
from importlib import resources
44

55

66
def main() -> None:
77
"""
88
CLI entrypoint for the PyPI package 'blockchain-scalpel'.
99
10-
Executes 'blockchain-parser.py' as if the user ran:
10+
Executes the bundled 'blockchain-parser.py' script as if the user ran:
1111
python blockchain-parser.py <args...>
12-
13-
Keeps sys.argv intact (arguments are passed through).
1412
"""
15-
repo_root = Path(__file__).resolve().parent.parent
16-
target = repo_root / "blockchain-parser.py"
13+
try:
14+
script = resources.files("blockchain_scalpel").joinpath("blockchain-parser.py")
15+
except Exception:
16+
print("Error: cannot locate bundled 'blockchain-parser.py'.", file=sys.stderr)
17+
raise SystemExit(2)
1718

18-
if not target.exists():
19-
print("Error: 'blockchain-parser.py' was not found next to the package.", file=sys.stderr)
19+
if not script.is_file():
20+
print("Error: bundled 'blockchain-parser.py' not found.", file=sys.stderr)
2021
raise SystemExit(2)
2122

22-
# Make usage/help messages look natural.
23-
# After this, script sees argv like: ["blockchain-parser.py", "./blocks", "./result"]
23+
# Make argv[0] look like the original script name
2424
if len(sys.argv) > 0:
25-
sys.argv[0] = str(target)
25+
sys.argv[0] = "blockchain-parser.py"
2626

27-
runpy.run_path(str(target), run_name="__main__")
27+
runpy.run_path(str(script), run_name="__main__")

0 commit comments

Comments
 (0)