How to parse HTTP responses in C?

Question

I'm writing a little project which interacts with a set of servers using HTTP 1.1 GET and POST. The server gives me the response after some header lines, so I though on using strtok() function using \n as the delimiter but there is a crash whenever I try to do so.

Is there any simple way to parse a HTTP response in C? I would like not to use 3rd party libraries for this but if it was really necesary I won't have any problem.

Thank you very much for everything.

EDIT: Here is some example code, just trying to print the lines:

char *response = "HTTP/1.1 200 OK\nServer: Apache-Coyote/1.1\nPragma: no-cache"

char *token = NULL;
token = strtok(response, "\n");
while (token) {
    printf("Current token: %s.\n", token);
    token = strtok(NULL, "\n");
}

Solution 1

http-parser is a simple and super-fast HTTP parser written in C for the Node.js project

It's only 2 C files, without any external dependencies.

Solution 2

The problem in the code you've posted is pretty simple: strtok works by modifying the string you pass to it. Modifying a string literal gives undefined behavior. A truly minuscule change to your code lets it work (I've also headed the appropriate headers, and moved the executable part into a function:

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

char response[] = "HTTP/1.1 200 OK\nServer: Apache-Coyote/1.1\nPragma: no-cache";

int main() { 
    char *token = NULL;
    token = strtok(response, "\n");
    while (token) {
        printf("Current token: %s.\n", token);
        token = strtok(NULL, "\n");
    }
    return 0;
}

In real use, you'll be reading the HTTP response into a buffer anyway; the problem you encountered only arises in a trivial test case like you generated. At the same time, it does point to the fact that strtok is a pretty poorly designed function, and you'd almost certainly be better off with something else.

Comments

[R Samuel Klatchko]

Not sure why you think this issue reflects poorl on strtok. The real problem is the fact that a string literal is considered a char * instead of a const char * and the issue can affect any code that modifies the string input.

[Jerry Coffin]

It reflects poorly on strtok, because it unnecessarily restricts its range of input to data that's not only modifiable, but that you don't mind basically destroying in the process of using it.

[Elazar Leibovich]

It's written in callback-style to support poll-like loops. Maybe the author is interested with blocking parser, like golang's ReadRequest

[Bharat]

Is there an example client written using this parser in C? Examples which come with the repository don't seem sufficient for a novice like me.