阅读 2276

iOS代码混淆实现

背景

一些类似的项目会使用重复的代码,导致苹果机审期间被判断为马甲包,还没到人审就被苹果审核拒掉了。 为了能够使正常迭代出功能相似的项目过审,要对我们项目的代码进行深度混淆。

方案

1、准备四六级单词库(如果使用随机字符会被机审查出来); 2、使用clang过滤出类名和方法名; 3、从四六级单词库随机组成,映射对应的类名和方法名; 4、通过映射进行混淆操作。

四六级单词库

区分首字母小写和大写的txt

安装clang

pip install clang --user
复制代码

类名混淆

提取类名

# encoding: utf-8

import sys
import os
import re
import clang
from clang.cindex import *
from optparse import OptionParser, OptionGroup

def get_tu(source, lang='c', all_warnings=False, flags=[]):
    """Obtain a translation unit from source and language.

    By default, the translation unit is created from source file "t.<ext>"
    where <ext> is the default file extension for the specified language. By
    default it is C, so "t.c" is the default file name.

    Supported languages are {c, cpp, objc}.

    all_warnings is a convenience argument to enable all compiler warnings.
    """
    args = list(flags)
    name = 't.c'
    if lang == 'cpp':
        name = 't.cpp'
        args.append('-std=c++11')
    elif lang == 'objc':
        name = 't.m'
    elif lang != 'c':
        raise Exception('Unknown language: %s' % lang)

    if all_warnings:
        args += ['-Wall', '-Wextra']

    return TranslationUnit.from_source(name, args, unsaved_files=[(name,
                                       source)])

def generate_m_file(file_text, result_lines, ret_functions):
//略

if __name__ == '__main__':
    libclangPath = '/Library/Developer/CommandLineTools/usr/lib/libclang.dylib'
    Config.set_library_file(libclangPath)

    # Find all .h files
    source_dir = sys.argv[1]
    g = os.walk(source_dir)
    h_files = []
    ret_functions = []

    for path,dir_list,file_list in g:  
        for file_name in file_list:
            h_files.append(os.path.join(path, file_name))

    for f in h_files:
        with open(f, 'r') as file:
            # preprocess
            regex = r'#import|#include|#ifdef|#ifndef|#define|#endif|@property'
            text = ''
            result_text_lines = []
            line_count = 0
            def_block_count = 0

            for line in file:
                if re.findall(regex, line):
                    if '#ifdef' in line:
                        def_block_count += 1
                        result_text_lines.append(line)
                    elif '#endif' in line and def_block_count > 0:
                        def_block_count -= 1
                        result_text_lines.append(line)
                    else:
                        result_text_lines.append('\n')

                    line = '// ' + line
                    text += line
                else:
                    text += line
                    result_text_lines.append('\n')
                line_count += 1  

            # print text
            m_file_name = os.path.join('fake', f.replace('.h', '.m'))
            m_file_to_write = ""

            # print 'processing: ' + m_file_name
            generate_m_file(text, result_text_lines, ret_functions)

    unique_array = list(set(ret_functions))
    filter_array = ['xxxx', 'aaaa', 'dddd', 'AppDelegate', \
                    'PrefixHeader', 'dddddf', 'aaaadxxx']
    for func_item in unique_array:
        if func_item in filter_array:
            continue
        print func_item
复制代码

filter_array为要筛选掉不做混淆的类名

对提取的类名做映射类名

#!/usr/bin/env bash

TABLENAME=symbols
SYMBOL_DB_FILE="symbols"
STRING_SYMBOL_FILE=./process_class/t.txt

HEAD_FILE=./rename-class/rename_classes.txt

export LC_CTYPE=C

rm -f $SYMBOL_DB_FILE
rm -f $HEAD_FILE

function rand(){
    min=$1
    max=$(($2-$min+1))
    num=$(($RANDOM+1000000000)) #增加一个10位的数再求余
    echo $(($num%$max+$min))
}

function pRnd2(){
    rnd=$(rand 10 4200)
    randint=`expr $RANDOM % 3`
    if [ $randint == 0 ];then
    echo `cat "JAAA.txt" | sed -n "${rnd}p"`
    elif [ $randint == 1 ];then
    echo `cat "JBBB.txt" | sed -n "${rnd}p"`
    else
    echo `cat "JCCC.txt" | sed -n "${rnd}p"`
    fi
}

my_arr=("Manager" "DataSource" "Helper" "Adapter" "Router" "Handler" "Handle" \
    "Model" "Service" "Item" "Info" "Controller" "Cell" "Button" "View" "Window")

touch $HEAD_FILE
# echo "//confuse string at `date`" >> $HEAD_FILE
cat "$STRING_SYMBOL_FILE" | while read -ra line; do
#命中概率
#randint=`expr $RANDOM % 3`
#if [ $randint != 0 ];then
#continue
#fi

#取出随机字符
if [[ ! -z "$line" ]]; then

suffix=""
for loop in ${my_arr[@]}; do 
    if [[ $line =~ $loop ]]; then
        suffix=$loop
        break
    fi
done
ramdom="CS$(pRnd2)$(pRnd2)${suffix}"
echo $line $ramdom
#insertValue $line $ramdom
echo "$line $ramdom" >> $HEAD_FILE
fi

done

复制代码

ramdom可以添加一些项目前缀,比如CS等。 my_arr里面可以定义一些iOS特有的后缀。

对映射的类做混淆

#!/bin/bash

PROJECT_DIR=`cat ../path.txt`
echo $PROJECT_DIR

RENAME_CLASSES=rename_classes.txt

#First, we substitute the text in all of the files.
sed_cmd=`sed -e 's@^@s/[[:<:]]@; s@[[:space:]]\{1,\}@[[:>:]]/@; s@$@/g;@' ${RENAME_CLASSES} `
find ${PROJECT_DIR} -type f \
    \( -name "*.pbxproj" -or -name "*.pch" -or -name "*.h" -or -name "*.m" -or -name "*.xib" -or -name "*.storyboard" \) \
    -exec sed -i "" "${sed_cmd}" {} +

# Now, we rename the .h/.m files
while read line; do
    class_from=`echo $line | sed "s/[[:space:]]\{1,\}.*//"`
    class_to=`echo $line | sed "s/.*[[:space:]]\{1,\}//"`
    #修改 .h .m
    find ${PROJECT_DIR} -type f -regex ".*[[:<:]]${class_from}[[:>:]][^\/]*\.[hm]" -print | egrep -v '.bak$' | \
    while read file_from; do
         file_to=`echo $file_from | sed "s/\(.*\)[[:<:]]${class_from}[[:>:]]\([^\/]*\)/\1${class_to}\2/"`
         echo mv "${file_from}" "${file_to}"
         mv "${file_from}" "${file_to}"
    done
    
    #修改 .xib
    find ${PROJECT_DIR} -type f -regex ".*[[:<:]]${class_from}[[:>:]][^\/]*\.xib" -print | egrep -v '.bak$' | \
    while read file_from; do
         file_to=`echo $file_from | sed "s/\(.*\)[[:<:]]${class_from}[[:>:]]\([^\/]*\)/\1${class_to}\2/"`
         echo mv "${file_from}" "${file_to}"
         mv "${file_from}" "${file_to}"
    done
done < ${RENAME_CLASSES}

复制代码

rename_classes.txt是保存的映射类,shell脚本对工程进行批量替换。

方法混淆

提取方法名

# encoding: utf-8

import sys
import os
import re
import clang
from clang.cindex import *
from optparse import OptionParser, OptionGroup

def get_tu(source, lang='c', all_warnings=False, flags=[]):
    """Obtain a translation unit from source and language.

    By default, the translation unit is created from source file "t.<ext>"
    where <ext> is the default file extension for the specified language. By
    default it is C, so "t.c" is the default file name.

    Supported languages are {c, cpp, objc}.

    all_warnings is a convenience argument to enable all compiler warnings.
    """
    args = list(flags)
    name = 't.c'
    if lang == 'cpp':
        name = 't.cpp'
        args.append('-std=c++11')
    elif lang == 'objc':
        name = 't.m'
    elif lang != 'c':
        raise Exception('Unknown language: %s' % lang)

    if all_warnings:
        args += ['-Wall', '-Wextra']

    return TranslationUnit.from_source(name, args, unsaved_files=[(name,
                                       source)])

def parse_method(node):
    tokens = list(node.get_tokens())

    # 过滤方法名, TODO:
    filter_start_words = ('init', 'set', 'get', 'image', 'view', 'reload', '_', 'will', 'did')

    function = ''
    for token_index in range(len(tokens)):
        if tokens[token_index].spelling == ')':
            function = tokens[token_index + 1].spelling
            break

    if len(function) > 10 and (not function.startswith(filter_start_words)):
        return function
    else:
        return ''


# extract_type = 0x00001:  普通方法
# extract_type = 0x00011:  普通方法 + 属性

def parse_symbols(cursor, ret_symbols, extract_type):
    class_children = list(cursor.get_children())
//略

# extract_type = 0x01100:  Category, Class
# extract_type = 0x10000:  Protocol
def extract_symbols(file_text, ret_symbols, extract_type):
    parser = OptionParser("usage: %prog [options] {filename} [clang-args*]")
    parser.disable_interspersed_args()
    (opts, args) = parser.parse_args()
    # if len(args) == 0:
    #     parser.error('invalid number arguments')        

    index = Index.create()
    # tu = index.parse(file_text, ['-x', 'objective-c'])
    tu = get_tu(file_text, lang='objc')
    if not tu:
        parser.error("unable to load input")

    it = tu.cursor.get_children()
    tu_nodes = list(it)
    for cursor in tu_nodes:
        if cursor.kind == CursorKind.OBJC_INTERFACE_DECL:
            # print cursor.spelling
            if extract_type & 0x00100:
                parse_symbols(cursor, ret_symbols, extract_type)

        elif cursor.kind == CursorKind.OBJC_CATEGORY_DECL:
            # print cursor.spelling
            if extract_type & 0x01000:
                # print "Categor ============"
                parse_symbols(cursor, ret_symbols, extract_type)
        elif cursor.kind == CursorKind.OBJC_PROTOCOL_DECL:
            if extract_type & 0x10000:
                parse_symbols(cursor, ret_symbols, extract_type)

# 提取该目录下所有 .h&.m文件的方法名
def traverse_header_files(top_directory, extract_type):
    g = os.walk(top_directory)
    h_files = []
    ret_symbols = []

    for path,dir_list,file_list in g:  
        for file_name in file_list:
            if file_name.endswith('.h') or file_name.endswith('.m'):
                h_files.append(os.path.join(path, file_name))

    for f in h_files:
        with open(f, 'r') as file:
            # preprocess
            regex = r'#import|#include|#ifdef|#ifndef|#define|#endif|#if|#else|@class'
            text = ''
            result_text_lines = []
            line_count = 0
            def_block_count = 0

            for line in file:
                if re.findall(regex, line):
                    line = '// ' + line
                    text += line
                else:
                    interface_idx = line.find('@interface')
                    if interface_idx > 0:
                        line = line[interface_idx:]
                        # print line

                    text += line

                line_count += 1

            # print 'processing: '
            extract_symbols(text, ret_symbols, extract_type)

    return set(ret_symbols).copy()


if __name__ == '__main__':
    libclangPath = '/Library/Developer/CommandLineTools/usr/lib/libclang.dylib'
    Config.set_library_file(libclangPath)

    source_dir = sys.argv[1]
    pods_dir = sys.argv[2]

    # 提取头文件的方法名
    source_dir_methods_set = traverse_header_files(source_dir, 0x00101)

    # 提取 Source 的属性
    filter_set_A = traverse_header_files(source_dir, 0x10110)

    # 提取 Source 下的 Category的方法和属性
    filter_set_B = traverse_header_files(source_dir, 0x11011)

    # 提取 Pods 目录下普通类和Category 的方法和属性
    filter_set_C = traverse_header_files(pods_dir, 0x11111)

    # 差集
    result_set = source_dir_methods_set.difference(filter_set_A).difference(filter_set_B).difference(filter_set_C)

    unique_list = list(result_set)

    for func_item in unique_list:
        print func_item

    # print 'Source len = ' + str(len(source_dir_methods_set)) 
    # print 'Result len = ' + str(len(result_set))      
复制代码

source_dir为项目的代码目录,pods_dir为pods的代码目录。由于pods里面的是三方的代码,所以进行在项目代码里排除掉,再对剩下的方法进行映射。

对方法映射并宏定义写入文件

#!/usr/bin/env bash

STRING_SYMBOL_FILE=./process_method/method_list.txt
HEAD_FILE=./methodDefine.h

export LC_CTYPE=C

rm -f $HEAD_FILE

function rand(){
    min=$1
    max=$(($2-$min+1))
    num=$(($RANDOM+1000000000)) #增加一个10位的数再求余
    echo $(($num%$max+$min))
}

function pRnd1(){
    rnd=$(rand 10 140000)
    randt=`expr $RANDOM % 3`
    if [ $randt == 0 ];then
    echo `cat "a.txt" | sed -n "${rnd}p"`
    elif [ $randt == 1 ];then
    echo `cat "b.txt" | sed -n "${rnd}p"`
    else
    echo `cat "c.txt" | sed -n "${rnd}p"`
    fi
}

function pRnd2(){
    rnd=$(rand 10 140000)
    randt=`expr $RANDOM % 3`
    if [ $randt == 0 ];then
    echo `cat "AAA.txt" | sed -n "${rnd}p"`
    elif [ $randt == 1 ];then
    echo `cat "BBB.txt" | sed -n "${rnd}p"`
    else
    echo `cat "CCC.txt" | sed -n "${rnd}p"`
    fi
}

touch $HEAD_FILE
echo '#ifndef methodDefine_h
#define methodDefine_h' >> $HEAD_FILE
echo "//confuse string at `date`" >> $HEAD_FILE
cat "$STRING_SYMBOL_FILE" | while read -ra line; do
#命中概率
#randint=`expr $RANDOM % 3`
#if [ $randint != 0 ];then
#continue
#fi

#取出随机字符
if [[ ! -z "$line" ]]; then

ramdom="$(pRnd1)$(pRnd2)"
echo $line $ramdom
echo "#ifndef $line
#define $line $ramdom
#endif" >> $HEAD_FILE
fi

done
echo "#endif" >> $HEAD_FILE
复制代码

导入文件

prefixHeader导入methodDefine.h文件,方法混淆完成