c - strcat 覆盖了我的字符串

Question

我想不通。当我在我的 Windows 机器上使用 Code::Blocks 编译这段代码时，它工作得很好，但是当我尝试在 Cygwin 或学校的实际 Unix 机器下使用 Make 编译它时，我得到下面的奇怪行为。我将“client1.txt”传递给 translate()。

void translate(char* filepath){

  char output_filepath[181];
  strcpy(output_filepath, filepath);
    printf("%s\n", output_filepath);      //this prints out "client1.txt" which is correct
  char* ptr = strcat(output_filepath, ".translated");
    printf("%s\n", output_filepath);      //this prints out ".translated" which is wrong
    printf("%s\n", ptr);                  //also prints out ".translated" wrong again

...more stuff...
}

当尝试在 output_filepath 上使用 fgets 时，这会导致段错误。有谁知道发生了什么事？我需要解释更多吗？它必须能够在 Unix 下编译。

这是整个程序，我希望它不会太长。这是我的老师给我们的一个程序，但我什至无法运行它。它只是给了我一个段错误，我已经在上面的部分中找到了它。

#include <pthread.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>

struct tparam{
int tid;
};

int NTHREADS  =  5;
#define LOOPS         10000
int qfilled = 0;
int qin = 0;
int qout = 0;
char queue[3000][2][161];
char dic[7000][2][161];
int dic_size = 0;
pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
pthread_mutex_t flag_mutex = PTHREAD_MUTEX_INITIALIZER;

char * dummystatus[10];

char * lookup(char * word)
{
int i;
for(i = 0; i < dic_size; ++i)
{
    if(strcmp(word, dic[i][0])==0)
        return dic[i][1];
}
return word;
}
void translate(char filepath[])
{
char output_filepath[181];
strcpy(output_filepath, filepath);
strcat(output_filepath, ".translated");

FILE * client_file = fopen(filepath,"r");
FILE * translated_client_file = fopen(output_filepath,"w");
char line [161];
char * tmpPtr;
char * token;
while((tmpPtr=fgets(line, 160, client_file))!= NULL) {
    if(strcmp(line,"\n") == 0 || line == NULL)
        continue;
    token = strtok_r(line, " \t\n", dummystatus);

    while(token != NULL && strcmp(token,"\n") != 0){
        fputs(lookup(token), translated_client_file);
        fputs(" ", translated_client_file);
        token = strtok_r(NULL, " \t\n", dummystatus);
    }
    fputs("\n", translated_client_file);
}
fclose(client_file);
}
void *do_work(void * p) {
  struct tparam * param = (struct tparam *)p;

  while(qfilled != 1);//wait for queue to be filled

  int cindex;
  while(1){
      //check for more clients
  pthread_mutex_lock(&mutex);
  if(qout >= qin){
    pthread_mutex_unlock(&mutex);
    break;
  }
  //process client
  cindex = qout;
  printf("Thread %d is handling client %s\n",param->tid, queue[cindex][1]);
  qout++;
  pthread_mutex_unlock(&mutex);

  char filepath[161];
  if(queue[cindex][0][strlen(queue[cindex][0])-1] == '\n')
      strncpy(filepath,queue[cindex][0],strlen(queue[cindex][0])-1);
  else
    strcpy(filepath,queue[cindex][0]);
  translate(filepath);
  printf("Thread %d finished handling client %s\n",param->tid, queue[cindex][1]);

      //usleep(rand()%100000+10000);
  }
  pthread_exit(NULL);
}

void addDicEntry(char line[]){
char * first = strtok_r(line, " \t", dummystatus);
char * second = strtok_r(NULL, " \t", dummystatus);
char englishWord[161];
if(1==1 || second != NULL)
{
    strcpy(dic[dic_size][0], first);
    strncpy(englishWord, second, strlen(second)-1);
    englishWord[strlen(second)-1] = '\0';
    strcpy(dic[dic_size][1], englishWord);
    dic_size++;
}
}

int main(int argc, char *argv[]) {

srand(time(NULL));
  if(argc < 2){
printf("No dictionary file provided\n");
exit(1);
  }
  //read dictionary
  int i;
  for(i = 0; i < 10; ++i)
dummystatus[i] = (char*)malloc(10);

  FILE * dic_file = fopen(argv[1],"r");
  if(dic_file == NULL)
 {
printf("Dictionary file does not exist\n");
exit(1);
  }
  char line [161];
  char * tmpPtr;
  while((tmpPtr=fgets(line, 160, dic_file))!= NULL) {
if(strcmp(line,"\n") == 0 || strcmp(line,"") == 0 || line == NULL)
    break;
addDicEntry(line);
  }
  fclose(dic_file);
  //End read dictionary

  //Creating threads
  if(argc >= 3)
NTHREADS = atoi(argv[2]);

  pthread_t * threads = (pthread_t *)malloc(NTHREADS*sizeof(pthread_t));
  pthread_attr_t attr;

  pthread_attr_init(&attr);
  pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_JOINABLE);
  for (i=0; i<NTHREADS; i++) {
struct tparam * param = (struct tparam *)malloc(sizeof(struct tparam *)*1);
param->tid = i+1;
//printf("Thread %d is being created\n",param->tid);
pthread_create(&threads[i], &attr, &do_work, param);
  }
 //End creating threads

  //insert clients in Q
  FILE * clients_file = fopen("clients.in","r");

  char cid_str[10];
  while((tmpPtr=fgets(line, 160, clients_file))!= NULL) {
if(strcmp(line,"\n") == 0 || strcmp(line,"") == 0 || line == NULL)
    break;
pthread_mutex_lock(&mutex);
strcpy(queue[qin][0],line);
sprintf(cid_str, "%d", qin+1);
strcpy(queue[qin][1],cid_str);
qin++;
pthread_mutex_unlock(&mutex);
  }
  fclose(clients_file);
//for (i=0; i<qin; i++)
//printf("%s\n", queue[i][0]);
  qfilled = 1;
  printf("Q Filled\n");
  //End insert clients in Q

  //printf("Waiting for Threads\n");
  for (i=0; i<NTHREADS; i++)
pthread_join(threads[i], NULL);
  //printf("Threads Finished\n");

  pthread_attr_destroy(&attr);
  pthread_exit(NULL);
}

Answer 1

我怀疑这个问题比“覆盖你的字符串”更微妙；更多的是覆盖输出行上的数据的问题。试试这个代码:

#include <string.h>
#include <stdio.h>

void translate(char* filepath)
{
  char output_filepath[181];

  strcpy(output_filepath, filepath);
    printf("%s\n", output_filepath);      //this prints out "client1.txt" which is correct
  char* ptr = strcat(output_filepath, ".translated");
    printf("%s\n", output_filepath);      //this prints out ".translated" which is wrong
    printf("%s\n", ptr);                  //also prints out ".translated" wrong again
}

int main(void)
{
    translate("client1.txt\r");
    return(0);
}

Mac OS X 10.7.3 上的输出:

client1.txt
.translated
.translated

您在文件路径参数字符串的末尾有一个回车符，这导致了这个明显的难题。

您可以通过 od 或 hd 或类似程序提供程序的输出。我的程序名为 odx，但任何方式都可以(而且您的程序被称为 x39 除了它是一个新文件名之外没有任何理由):

$ ./x39 | odx
0x0000: 63 6C 69 65 6E 74 31 2E 74 78 74 0D 0A 63 6C 69   client1.txt..cli
0x0010: 65 6E 74 31 2E 74 78 74 0D 2E 74 72 61 6E 73 6C   ent1.txt..transl
0x0020: 61 74 65 64 0A 63 6C 69 65 6E 74 31 2E 74 78 74   ated.client1.txt
0x0030: 0D 2E 74 72 61 6E 73 6C 61 74 65 64 0A            ..translated.
0x003D:
$

如果我不得不猜测，您从在 Windows 机器上创建并使用二进制而不是文本传输传输到 Unix 的文件中读取文件名 (client1.txt)，并且您可能阅读它与 gets() 因为这将删除换行符而不是它之前的回车符。

我看到主要代码正在使用 fgets() — 比 gets() 好多了! — 但它不具备处理 CRLF 行尾的能力。您的代码不检查文件名是否已成功打开；这迟早会导致核心转储(特别是因为磁盘上的输入文件名不太可能以 CR '\r' 结尾，因此打开几乎肯定会失败)。

• 始终，但始终，在使用文件流指针或文件描述符之前检查文件打开函数是否成功。

相信我；在像 strcat() 这样频繁使用和测试的例程中，您确实很少会发现错误。