使用PyInstaller打包含Capstone的Python脚本为.exe时动态库加载失败求助

问题背景 #

我编写了一个可提取任意类型文件字节码并进行反汇编的Python脚本，依赖Capstone库。使用PyInstaller打包为单文件.exe后，运行时提示无法加载动态库，手动添加dll路径重新打包仍未解决问题。

报错信息 #

Traceback (most recent call last):
File "Extract.py", line 6, in
File "PyInstaller\loader\pyimode2_importers.py", line 419, in exec_module
File "capstone_init_.py", line 428, in
ImportError: ERROR: fail to load the dynamic library.
[4748] Failed to execute script 'Extract' due to unhandled exception!

已尝试的打包命令 #

• 初始打包命令：

pyinstaller --onefile Ectract.py

• 添加dll路径后的命令：

pyinstaller --onefile --add-binary "C:/Users/admin/AppData/Local/Programs/Python/Python311/Lib/site-packages/capstone/lib/capstone.dll;." Extract.py

脚本代码 #

import os
import random
import string
import time
import shutil
from capstone import*

def extract_hex_from_file(file_path, output_folder, bytes_per_line):
    with open(file_path, 'rb') as f:
        random_name = generate_random_name() + '.bytes'
        output_file = os.path.join(output_folder, random_name)
        with open(output_file, 'w') as output:
            address = 0x00401000  # Start address
            while True:
                chunk = f.read(bytes_per_line)
                if not chunk:
                    break
                hex_bytes = ' '.join(f"{byte:02X}" for byte in chunk)
                output.write(f"{address:08X} {hex_bytes}\n")
                address += bytes_per_line
    return random_name[:-6]

def generate_random_name(length=20):
    letters_and_digits = string.ascii_letters + string.digits
    return ''.join(random.choice(letters_and_digits) for _ in range(length))

def check_disk_space(output_folder):
    free_space = shutil.disk_usage(output_folder).free
    return free_space

def process_files(file_path, output_folder, bytes_per_line, arch):
    try:
        generated_name = extract_hex_from_file(file_path, output_folder, bytes_per_line)
        if check_disk_space(output_folder) < os.path.getsize(file_path):
            print(f"Not enough space available to process file: {file_path}. Skipping.")
            return None
        else:
            disassemble_file(os.path.join(output_folder, generated_name + '.bytes'), output_folder, arch)
            return generated_name
    except PermissionError:
        print(f"{file_path}. .")
        return None
    except Exception as e:
        print(f"Error processing file: {file_path}. Error: {str(e)}")
        return None

def extract_hex_from_folder(folder_path, output_folder, bytes_per_line=16, arch='x86'):
    if not os.path.exists(output_folder):
        os.makedirs(output_folder)
    file_paths = []
    for root, _, files in os.walk(folder_path):
        for file_name in files:
            file_paths.append(os.path.join(root, file_name))
    start_time = time.time()
    for file_path in file_paths:
        process_files(file_path, output_folder, bytes_per_line, arch)
    end_time = time.time()
    total_time = end_time - start_time
    print(f"Total execution time: {total_time} seconds")

def disassemble_file(filename, output_folder, arch):
    arch_map = {
        'x86': CS_ARCH_X86,
        'x86-64': CS_ARCH_X86,
        'arm': CS_ARCH_ARM,
    }
    md = Cs(arch_map[arch], CS_MODE_64 if arch == 'x86-64' else CS_MODE_ARM)
    with open(filename, 'rb') as f:
        code = f.read()
    disassembly_output = ""
    for instr in md.disasm(code, 0x1000):
        disassembly_output += f"0x{instr.address:x}:\t{instr.mnemonic}\t{instr.op_str}\n"
    section_rva = 0x1000
    section_size = 0x100
    disassembly_output += "\nDisassembly of specified code section:\n"
    for instr in md.disasm(code[section_rva:section_rva+section_size], section_rva):
        disassembly_output += f"0x{instr.address:x}:\t{instr.mnemonic}\t{instr.op_str}\n"

    base_name, _ = os.path.splitext(os.path.basename(filename))
    output_filename = os.path.join(output_folder, base_name + '.asm')
    with open(output_filename, 'w') as output_file:
        output_file.write(disassembly_output)

if __name__ == '__main__':
    folder_path = '/path/'
    output_folder = '/path/'
    bytes_per_line = 16
    arch = 'x86'
    extract_hex_from_folder(folder_path, output_folder, bytes_per_line, arch)

解决方案 #

方法1：修正打包命令的dll路径结构 #

Capstone库会从capstone/lib目录下加载dll，之前的命令把dll放到根目录不符合其预期路径。修改打包命令，将dll放到对应的子目录：

pyinstaller --onefile --add-binary "C:/Users/admin/AppData/Local/Programs/Python/Python311/Lib/site-packages/capstone/lib/capstone.dll;capstone/lib" Extract.py

方法2：使用.spec文件手动配置 #

1. 生成基础.spec文件：

pyinstaller --onefile Extract.py

2. 打开生成的Extract.spec，找到binaries参数，添加dll的路径配置：

binaries=[('C:/Users/admin/AppData/Local/Programs/Python/Python311/Lib/site-packages/capstone/lib/capstone.dll', 'capstone/lib')],

3. 基于修改后的spec文件重新打包：

pyinstaller Extract.spec

方法3：在脚本中手动指定dll路径 #

在脚本开头添加代码，适配PyInstaller打包后的环境，手动指向dll的正确位置：

import os
import sys
from capstone import *

# 适配打包后的dll路径
if getattr(sys, 'frozen', False):
    # 打包后临时目录中的dll路径
    capstone_dll_path = os.path.join(sys._MEIPASS, 'capstone', 'lib', 'capstone.dll')
    # 手动加载dll
    import ctypes
    ctypes.CDLL(capstone_dll_path)

添加后再使用方法1或方法2的命令重新打包即可。

内容的提问来源于stack exchange，提问作者MayurD