161 lines
5.4 KiB
Python
161 lines
5.4 KiB
Python
#!/usr/bin/env python3
|
||
"""生成 tuku zip 读取器的宿主测试样例（真实 zipfile + 定点字节修补）。
|
||
|
||
用法: python3 make_zip_samples.py <out_dir>
|
||
|
||
生成：
|
||
basic.zip 5 条目：stored 文本 / deflate 大二进制 / 中文名 deflate /
|
||
目录项 / 0 字节 deflate
|
||
comment.zip 单条目 + 归档注释（EOCD 注释长度路径）
|
||
empty.zip 0 条目
|
||
truncated.zip basic.zip 砍半（EOCD 丢失）
|
||
badmagic.zip 非 zip 字节
|
||
badoffset.zip 首条目 local header 偏移改成 0xFFFFFF00（越界）
|
||
badcrc.zip 首条目 CD 里 CRC 翻转（内容解出但校验失败）
|
||
method99.zip 首条目压缩方式改成 99，CD 与局部头都改
|
||
badlocal.zip 首条目局部头签名改成 "BAD!"
|
||
bigsize.zip deflate 条目 CD comp_size 改成 8 MiB（数据越界）
|
||
zip64claim.zip EOCD 总条目数改成 0xFFFF（zip64 标记）
|
||
bigrand.zip 单条目 256 KiB 不可压 deflate（压缩流 > 64 KiB，覆盖分块 inflate）
|
||
"""
|
||
import os
|
||
import struct
|
||
import sys
|
||
import zipfile
|
||
|
||
PAGE2 = (b"MEOW" * 10 + bytes(range(256))) * 180 # 53280 字节，可压缩
|
||
HELLO = b"hello zip\n" * 3
|
||
CJK_NAME = "sub/dir/\u4e2d\u6587\u540d.txt"
|
||
CJK_DATA = "\u4e2d\u6587\u5185\u5bb9\n".encode("utf-8")
|
||
BIGRAND_LEN = 262144
|
||
|
||
|
||
def lcg_bytes(n):
|
||
"""与 test_zip.c 的 build_rand 同一个 LCG，用于 256 KiB 不可压样例。"""
|
||
x = 123456789
|
||
out = bytearray(n)
|
||
for i in range(n):
|
||
x = (1103515245 * x + 12345) & 0x7FFFFFFF
|
||
out[i] = (x >> 16) & 0xFF
|
||
return bytes(out)
|
||
|
||
EOCD_SIG = b"PK\x05\x06"
|
||
CD_SIG = b"PK\x01\x02"
|
||
|
||
|
||
def write_basic(path):
|
||
with zipfile.ZipFile(path, "w") as z:
|
||
z.writestr(zipfile.ZipInfo("hello.txt", (2024, 1, 1, 0, 0, 0)),
|
||
HELLO, compress_type=zipfile.ZIP_STORED)
|
||
z.writestr(zipfile.ZipInfo("page2.png", (2024, 1, 1, 0, 0, 0)),
|
||
PAGE2, compress_type=zipfile.ZIP_DEFLATED)
|
||
z.writestr(zipfile.ZipInfo(CJK_NAME, (2024, 1, 1, 0, 0, 0)),
|
||
CJK_DATA, compress_type=zipfile.ZIP_DEFLATED)
|
||
z.writestr(zipfile.ZipInfo("onlydir/", (2024, 1, 1, 0, 0, 0)),
|
||
b"", compress_type=zipfile.ZIP_STORED)
|
||
z.writestr(zipfile.ZipInfo("empty.bin", (2024, 1, 1, 0, 0, 0)),
|
||
b"", compress_type=zipfile.ZIP_DEFLATED)
|
||
|
||
|
||
def write_comment(path):
|
||
with zipfile.ZipFile(path, "w") as z:
|
||
z.writestr("note.txt", b"commented\n", compress_type=zipfile.ZIP_DEFLATED)
|
||
z.comment = b"meow archive comment"
|
||
|
||
|
||
def load(path):
|
||
with open(path, "rb") as f:
|
||
return bytearray(f.read())
|
||
|
||
|
||
def save(path, data):
|
||
with open(path, "wb") as f:
|
||
f.write(bytes(data))
|
||
|
||
|
||
def eocd_info(data):
|
||
i = data.rfind(EOCD_SIG)
|
||
if i < 0:
|
||
raise SystemExit("generator: no EOCD in fixture")
|
||
total, cd_size, cd_off = struct.unpack_from("<HII", data, i + 10)
|
||
return i, total, cd_size, cd_off
|
||
|
||
|
||
def cd_records(data):
|
||
_, total, _cd_size, cd_off = eocd_info(data)
|
||
out = []
|
||
p = cd_off
|
||
for _ in range(total):
|
||
if bytes(data[p:p + 4]) != CD_SIG:
|
||
raise SystemExit("generator: bad CD signature")
|
||
nlen, elen, clen = struct.unpack_from("<HHH", data, p + 28)
|
||
loff = struct.unpack_from("<I", data, p + 42)[0]
|
||
name = bytes(data[p + 46:p + 46 + nlen]).decode("utf-8")
|
||
out.append({"off": p, "name": name, "local": loff})
|
||
p += 46 + nlen + elen + clen
|
||
return out
|
||
|
||
|
||
def find_record(data, name):
|
||
for r in cd_records(data):
|
||
if r["name"] == name:
|
||
return r
|
||
raise SystemExit("generator: entry not found: " + name)
|
||
|
||
|
||
def main():
|
||
if len(sys.argv) != 2:
|
||
raise SystemExit("usage: make_zip_samples.py <out_dir>")
|
||
out = sys.argv[1]
|
||
os.makedirs(out, exist_ok=True)
|
||
|
||
basic = os.path.join(out, "basic.zip")
|
||
write_basic(basic)
|
||
write_comment(os.path.join(out, "comment.zip"))
|
||
with zipfile.ZipFile(os.path.join(out, "empty.zip"), "w"):
|
||
pass
|
||
|
||
raw = load(basic)
|
||
save(os.path.join(out, "truncated.zip"), raw[:len(raw) // 2])
|
||
save(os.path.join(out, "badmagic.zip"), b"NOTAZIP!" * 100)
|
||
|
||
d = load(basic)
|
||
hello = find_record(d, "hello.txt")
|
||
struct.pack_into("<I", d, hello["off"] + 42, 0xFFFFFF00)
|
||
save(os.path.join(out, "badoffset.zip"), d)
|
||
|
||
d = load(basic)
|
||
hello = find_record(d, "hello.txt")
|
||
crc = struct.unpack_from("<I", d, hello["off"] + 16)[0]
|
||
struct.pack_into("<I", d, hello["off"] + 16, crc ^ 0xFFFFFFFF)
|
||
save(os.path.join(out, "badcrc.zip"), d)
|
||
|
||
d = load(basic)
|
||
hello = find_record(d, "hello.txt")
|
||
struct.pack_into("<H", d, hello["off"] + 10, 99)
|
||
struct.pack_into("<H", d, hello["local"] + 8, 99)
|
||
save(os.path.join(out, "method99.zip"), d)
|
||
|
||
d = load(basic)
|
||
hello = find_record(d, "hello.txt")
|
||
d[hello["local"]:hello["local"] + 4] = b"BAD!"
|
||
save(os.path.join(out, "badlocal.zip"), d)
|
||
|
||
d = load(basic)
|
||
page2 = find_record(d, "page2.png")
|
||
struct.pack_into("<I", d, page2["off"] + 20, 0x00800000)
|
||
save(os.path.join(out, "bigsize.zip"), d)
|
||
|
||
d = load(basic)
|
||
eocd, _total, _cd_size, _cd_off = eocd_info(d)
|
||
struct.pack_into("<H", d, eocd + 10, 0xFFFF)
|
||
save(os.path.join(out, "zip64claim.zip"), d)
|
||
|
||
with zipfile.ZipFile(os.path.join(out, "bigrand.zip"), "w") as z:
|
||
z.writestr("rand.bin", lcg_bytes(BIGRAND_LEN),
|
||
compress_type=zipfile.ZIP_DEFLATED)
|
||
|
||
|
||
if __name__ == "__main__":
|
||
main()
|