Discover Peers: Bittorent in C

[sinmi-hub]

I’m stuck on Stage #FI9, where we have to discover peers

I’ve tried a lot of bit manipulation. I also did research to see if there was a way to handle big endian in C. I saw htons() and ntohs(), but I struggle to understand how that quite works.

Here are my logs:
When I run locally:

sinmi@Sinmi:~/codecrafters-bittorrent-c$ gcc -g -o main app/*.c -lcrypto -lcurl -lssl -lm
sinmi@Sinmi:~/codecrafters-bittorrent-c$ ./main peers app/sample.torrent
165.232.41.73:51556
165.232.38.164:51493
165.232.35.114:51476

I suspect that this is wrong, because the ports of the last 2 ip address does not match what is expected from sample.torrent. However, I do not know what else to do because I am stuck

Here are the logs when I run on codecrafters

remote: [tester::#FI9] Running tests for Stage #FI9 (Discover peers)
remote: [tester::#FI9] Running ./your_bittorrent.sh peers /tmp/torrents1964150527/test.torrent
remote: [tester::#FI9] Tracker started on address 127.0.0.1:32791...
remote: [tester::#FI9]
remote: [your_program] 188.119.61.177:65505
remote: [your_program] 185.107.13.235:54542
remote: [your_program] 88.99.2.101:65505
remote: [tester::#FI9] Expected stdout to contain "188.119.61.177:6881", got: "188.119.61.177:65505\n185.107.13.235:54542\n88.99.2.101:65505\n"
remote: [tester::#FI9] Test failed

And here’s a snippet of my code:

Peers *get_peers(TrackerResp *response){
    Peers *all_peers = malloc(sizeof(Peers));

    // create a copy of response and find how many peers exist
    char haystack[1024];
    strcpy(haystack, response->data);
    char *temp = strstr(haystack, "peers");
    temp += strlen("peers");

    // find out where peers start from in Tracker rESP
    char *peers = NULL;
    int peer_size = strtol(temp, &peers, 10);
    peers += 1; // skip the ':'

    // init peers struct.
    all_peers->num_peers = peer_size / 6; // due to the compact
    all_peers->peer = malloc(peer_size * sizeof(char *));
    temp = peers;
    int idx = 0;

    for (int i = 0; i < all_peers->num_peers; i++)
    {   
        // construct ip address and port number
        uint8_t octet_1 =  temp[idx]  & 0xFF;
        uint8_t octet_2 =  temp[idx + 1]  & 0xFF;
        uint8_t octet_3 =  temp[idx + 2]  & 0xFF;
        uint8_t octet_4 =  temp[idx + 3] & 0xFF;
        uint16_t port =  (temp[idx + 4] << 8 | temp[idx + 5]);

        // size required to store ip address and port
        char temp_buf[1024];
        int total = snprintf(temp_buf, 1024, 
                            "%d.%d.%d.%d:%d", 
                            octet_1, octet_2, octet_3, octet_4, port);
        
        // save peer
        char *peer_ip = malloc(total + 1);
        strncpy(peer_ip, temp_buf, total);
        peer_ip[total] = '\0';
        all_peers->peer[i] = peer_ip;
        idx += 6;
    }

    return all_peers;
}

The idea nehind calculating the port is that I need 2 bytes after the first 4 bytes. Given that each 8 bit is 1 byte, I figured the big endian implementation would be to put the 5th byte as upper 8 bits, and 6th byte as lower 8bits to give 16 bits in all, to represent the port.

At first, I though it was because snprintf writes null to a buffer by default, maybe that messed up my bit positioning, but I double checked, anmd that is not the case

Is my understanding wrong? If full code is needed, please let me know and I will post it

[sarp]

Hi sinmi-hub,
You’re right, the port needs to be interpreted as an unsigned, 16 bit big endian binary number.

Does using this make a difference?
uint16_t port = ((uint8_t)temp[idx + 4] << 8) | (uint8_t)temp[idx + 5];

This ensures that both bytes are treated as unsigned and combined in big-endian order.

[sinmi-hub]

Hello @sarp
Thanks for responding. I still seem to be having the same problem

Here is a recent test that I ran

6881 is the port we were supposed to use when sending GET request, but I doubt it has anything to do with response.

For the most part, i seem to be able to parse ip address, but not port

[sinmi-hub]

@sarp

(uint8_t)

i also printed out the values before combining…is it possible thatt the sample.torrent maybe different from what was posted in description?

   //------------------------------------------------------------------
        // Extract the port from the next two bytes
        uint8_t port_msb = temp[idx + 4] & 0xFF; // Most significant byte of the port
        uint8_t port_lsb = temp[idx + 5] & 0xFF; // Least significant byte of the port

        // Print the raw bytes to verify correctness
        printf("Port bytes: 0x%02X 0x%02X\n", port_msb, port_lsb);
        // Combine the two bytes to form the port
        uint16_t port1 = (port_msb << 8) | port_lsb;

        // Step 2: Debug the result of port construction
        printf("Constructed Port: %d\n", port1);
        //---------------------------------------------------------

Since the value is binary, i cant tell for sure if this is what sample.torrent has. If i knew, then i could probaby know by how much bits I am off

I should also state that I used curl library to send the request, rather than do it manually of creating sockets and all. Is it possible that while parsing the response, something happended?

[sarp]

The values you’ve shared for sample.torrent look correct:

165.232.41.73:51556
165.232.38.164:51493
165.232.35.114:51476

Course description seems to be out of date, sorry about that! I’ve updated it to avoid future confusion, thanks for reporting.

[sinmi-hub]

I see. I am glad that is resolved.

Although, I am still failing the tests. Could you please help me with this?

The test requires 3 ip addresses, but the expected stdout is not reflected in possible answer

I am sincerely confused as to how the ports sent in the GET requests is not what the tests is looking for, especially because it works for sample.torrent

I understand that people have passed these stages, so i know the error is on my part

[sinmi-hub]

Resolved