int main(int argc, char *argv[])
{
int retval = 0;
int i;
for (i = 1; argv[i]; i++) {
if (strcmp(argv[i], "-v") == 0)
tstflags |= tst_verbatim;
else if (strcmp(argv[i], "-a") == 0)
tstflags |= tst_abort;
else
usage(1);
}
#if HAVE_OPEN_C
tstflags |= tst_verbatim;
#endif
retval |= test_alloc();
retval |= test_lock();
retval |= test_strdupcat();
retval |= test_sprintf("%s.%s", "foo", "bar");
retval |= test_strlst();
retval |= test_vectors();
retval |= test_auto();
return retval;
}
//****************************************对抽取的语料库按类别进行分类****************************************
void corpus_category(string filename_path, string path_read, string path_write) //参数1：文件名列表路径 参数2：读取文件的目录 参数3：写入文件的目录
{
string file_name; //存储读取的各个文件的名字
ifstream read_filename(filename_path); //从LIST.TXT读取文件名
ofstream sports(path_write+"sports.txt");//属于sport类的存放在sport.txt文件中
ofstream house(path_write+"house.txt");
ofstream it(path_write+"it.txt");
ofstream test_2008(path_write+"2008.txt");
ofstream news(path_write+"news.txt");
ofstream yule(path_write+"yule.txt");
ofstream business(path_write+"business.txt");
ofstream travel(path_write+"travel.txt");
ofstream mil_news(path_write+"mil.news.txt");
ofstream women(path_write+"women.txt");
ofstream health(path_write+"health.txt");
ofstream test_auto(path_write+"auto.txt");
ofstream cul(path_write+"cul.txt");
ofstream learning(path_write+"learning.txt");
ofstream test_else(path_write+"else.txt");
string path_in, str_line,cut_str;//path_in:存放读文件路径 str_line:读取的一行文件 cut_str:存放截取的字符串
string::size_type pos1, pos2;
int number = 0;
while (getline(read_filename, file_name))
{
number++;
cout << number << endl;
path_in = path_read + file_name;
ifstream infile(path_in);
while (getline(infile, str_line)) //读取各个文件的每一行字符串
{
pos1 = 0;
pos2 = str_line.find("####");
cut_str = str_line.substr(pos1, pos2 - pos1);
if (string(cut_str) == string("sports")) //字符串匹配 是否为sports类
{
sports << str_line << endl; //如果是sports类就把该行输出到sports.txt文件
}
else if (cut_str == "house")
{
house << str_line << endl;
}
else if (cut_str == "it")
{
it << str_line << endl;
}
else if (cut_str == "2008")
{
test_2008 << str_line << endl;
}
else if (cut_str == "news")
{
news << str_line << endl;
}
else if (cut_str == "yule")
{
yule << str_line << endl;
}
else if (cut_str == "business")
{
business << str_line << endl;
}
else if (cut_str == "travel")
{
travel << str_line << endl;
}
else if (cut_str == "mil.news")
{
mil_news << str_line << endl;
}
else if (cut_str == "women")
{
women << str_line << endl;
}
else if (cut_str == "health")
{
health << str_line << endl;
}
else if (cut_str == "auto")
{
test_auto << str_line << endl;
}
else if (cut_str == "cul")
{
cul << str_line << endl;
}
else if (cut_str == "learning")
{
learning << str_line << endl;
}
else
{
test_else << str_line << endl;
}
}
infile.close(); //每次结束都得关闭文件.
}
}