解析HTTP GET-C

Question

I'm trying to parse an HTTP GET request, and have been doing so using strtok() , however I am running into problems when trying to use strcpy() .

I can parse the file path and file name fine, but can't seem to parse the remote host DNS name. Below is code that should tokenize a string and get the DNS name, then store it in a char[] called host .

#include <stdio.h>
#include <time.h>
#include <string.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <unistd.h>
#include <stdlib.h>

int main()
{
        int c = 0, c2 = 0;
        char *tk, *tk2, *tk3, *tk4;
        char buf[64], buf2[64], buf3[64], buf4[64];
        char host[1024], path[64], file[64];

        strcpy(buf, "GET /~yourloginid/index.htm HTTP/1.1\r\nHost: remote.cba.csuohio.edu\r\n\r\n");

        tk = strtok(buf, "\r\n");
        while(tk != NULL)
        {
                if(c == 1)
                {
                        tk2 = strtok(tk, " ");
                        while(tk2 != NULL)
                        {
                                if(c2 == 1)
                                {
                                        printf("%s\n", tk2);
                                        strcpy(host, tk2);
                                //      printf("%s\n", host);
                                }
                                ++c2;
                                tk2 = strtok(NULL, " ");
                        }
                }
                ++c;
                tk = strtok(NULL, "\r\n");
        }

        return 0;
}

Bear with me, as I'm a new C programmer and this code may be ugly. Every time I try running my program, I get a Segmentation fault (core dumped) error, and I believe it has something to do with strcpy() . I can print out the tokenized string just fine, but can't seem to copy it to a char[] .

Answer 1

Sorry, but strtok(3) function is not good to parse HTTP at all. Despite of this, I'll try to explain what's happening in your code.

1. The first time, you enter the loop with tk=="GET /~yourloginid/index.html HTTP/1.1" , and your buffer has been changed to "GET /~yourloginid/index.htm HTTP/1.1\\0\\nHost: ..." . As c==0 , you won't get into the if block, so you'll get your c variable incremented and tk=strtok(NULL, "\\r\\n"); called again to get the second line.
2. The second time, you enter the loop with tk=="Host: remote.cba.scuohio.edu\\r\\n..." , as strtok(3) jumped over the first \\0 in your string, skipped all \\r and \\n characters, and got that (strtok has put now a second \\0 after that part, leading to tk=="Host: remote.cba.scuohio.edu\\0\\n..." . As c==1 this time, you get inside the if block and call strtok(tk, " "); . That makes strtok(3) to forget the extent of the string you where parsing, and begin a new parse on Host: remote.cba.csuohio.edu" (as you passed it a first non-null argument), it will return tk=="Host:" , putting a \\0 after "host:" . The second time you enter the inner loop, you copy that value to host variable.
3. The third time you enter the main loop, you have tk==NULL as the last time you called tk=strtok(NULL, " "); it returned NULL (in the inner loop), strtok will continue returning NULL until you initialize it again, passing a first non-null argument.

strtok(3) operates on the string you passed as the first parameter (writing info on it) and modifies it. Further, it has a global hidden variable to mark the end of the string you are parsing, to be able to return NULL when it is finished parsing. If you nest calls to strtok(3) you will get undefined behaviour, as you loose the internal state of the function when you initialize it again, passing a non-null first parameter. That is the reason of your fail.

Calling strtok(3) has numerous drawbacks and it cannot be nested in several nested loops as it stores internally state related to what you are parsing. It's strongly deprecated use. If you want it to be nestable, you have to switch to strtok_r(3) instead. That function has an extra parameter that allows you to save externally the strtok internal state, so you can have several strtoks working in parallel.

Further, strtok will parse ok "GET_/~yourlogin..." than "GET___/~yourlogin..." (I have used underscores to represent spaces to show multiple spaces between method name and uri) and the latter is not permitted by HTTP. For the same reason, you can get "Host:remote.cba.csuohio.edu" as a valid header field (however, this use is strongly discouraged) and you will not parse correctly that. Also, the Host: header field might not be the first line in the HTTP header, so you can skip it if you are not carefull.

If you want to parse HTTP, the first reading I can recommend is the RFC-2616, "HyperText Transfer Protocol - HTTP/1.1" , which is the mandatory document to comply with for implementors. Beware, it's a dense and large document.

Answer 2

在我看来，您要复制到buf中的字符串比在buf的定义语句中分配的64个字符长。

Answer 3

Trekator is right.

Replace this:

strcpy(buf, "GET /~yourloginid/index.htm HTTP/1.1\r\nHost: remote.cba.csuohio.edu\r\n\r\n");

with:

char buf[] = "GET /~yourloginid/index.htm HTTP/1.1\r\nHost: remote.cba.csuohio.edu\r\n\r\n";

It is good practice to use strncpy if there is a possibility of overrunning the destination buffer.

Answer 4

Some advice:

(1) You will find your life a whole lot easier if you use variable names which tell you more than 'tk2' and such.

(2) You can use strstr() to find the cr-nl w/o modifying the original string and it doesn't have the nesting issues that strtok() does.

(3) In general you can't count on Host: always being the 2nd line, so a more general approach of looking for it is more robust.

(4) Making sure things fit into their destination is "safe programming 101".

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

int main()
{
        int     lnum    = 0;
        char buf[] = "GET /~yourloginid/index.htm HTTP/1.1\r\nHost: remote.cba.c
suohio.edu\r\n\r\n";
        char *  line;
        char *  crnl;
        char *  colon;
        char *  arg;
        char    host[1024]      = "(not found)";

        for (line = buf;
             (crnl = strstr(line, "\r\n")) != NULL;
            line = crnl + 2) {
                ++lnum;
                if (lnum == 1) {
                        /* "GET" line is always first */
                        continue;
                }
                if (crnl == line) {             /* empty line marks end */
                        break;
                }
                if (((colon = strchr(line, ':')) == NULL) || (colon > crnl)) {
                        fprintf(stderr, "no colon in header line?\n");
                        break;
                }
                if (strncasecmp(line, "Host", colon - line) == 0) {
                        for (arg = colon + 1; isspace(*arg); ++arg) {}
                        if ((crnl - arg) >= sizeof(host)) {
                                fprintf(stderr, "hostname too big\n");
                        } else {
                                strncpy(host, arg, crnl - arg);
                                host[crnl - arg] = '\0';
                        }
                }
        }
        printf("host was '%s'\n", host);
        return 0;
}