unix/http/client.c

/**
 * @file client.c
 * @author Ivaylo Ivanov 11777707
 * @date 03.11.2018
 *
 * @brief Client program module.
 *
 * The client module takes a URL as input, connects to the corresponding server on the corresponding
 * port(`80` by default) and requests the file specified in the URL.
 * The content of that file is written to `stdout`, to a file or to a directory.
 *
 *   SYNOPSIS
 *   client [-p PORT] [ -o FILE | -d DIR ] URL
 *
 *   EXAMPLE
 *   client http://ivayloivanov.eu/en/
 *
 **/

#define _GNU_SOURCE ///< in order for optarg to be defined
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <netdb.h>
#include "shared/http.h"

static struct addrinfo* get_host_info(const char *hostname, char *port);
static int create_socket(struct addrinfo* config);
static void process_response(char reply[sizeof(int32_t)], char *ofile, char *dir);

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

    char *port = "80";
    char *ofile = NULL;
    int opt_f = 0;
    char *dir = NULL;
    int opt_d = 0;
    char *url = NULL;
    int c;

    /// Process the options
    while((c = getopt(argc, argv, "p:o:d:h")) != -1) {
        switch(c) {
            case 'p':
                port = optarg;
                break;
            case 'o':
                ofile = optarg;
                opt_f = 1;
                break;
            case 'd':
                dir = optarg;
                opt_d = 1;
                break;
            case 'h':
                print_usage();
                break;
            case '?':
                print_usage();
                break;
            default:
                abort();
        }
    }

    /// Check for the case that both output file and directory are set
    if(opt_d == 1 && opt_f == 1) {
        printf("Either output file or directory should be specified. You cannot use both.\n");
        print_usage();
    }

    /// Check if the required argument is there
    if(argv[optind] == NULL) {
        fprintf(stderr, "Mandatory argument 'url' missing.\n");
        print_usage();
    } else {
        url = argv[optind];
    }

    char proto[8];
    memcpy(proto, url, 7);
    proto[7] = 0;

    if(strcmp(proto, "http://") != 0){
        printf("ERROR: The url doesn't start with http://\n");
        print_usage();
    }

    char *base_url = strstr(url, "http://") + 7; //< the base url starts after http://

    char *path = malloc(strlen(base_url) + 1);
    strcpy(path, base_url); //< Copying so that strtok() doesn't change it
    path = strstr(path, "/");

    if(path == NULL) {
        path = "/"; //< Set the root path if there is none
        if(ofile == NULL && dir != NULL)
            ofile = "index.html";
    } else {
        if(ofile == NULL && dir != NULL) {
            ofile = strchr(path, '/');
        }
    }

    const char *hostname =  strtok(base_url, ";/?:@=&"); //< Get the FQDN of the remote
    struct addrinfo* config = get_host_info(hostname, port); //< Check if the address is resolvable

    char header[strlen(base_url) + 100]; //< Create a variable to store the header
    snprintf(header, sizeof header, "GET %s HTTP/1.1\r\nHost: %s\r\nConnection: close\r\n\r\n", path, hostname); //< Build the header

    int sockfd = create_socket(config);

    if (connect(sockfd, config -> ai_addr, config -> ai_addrlen) == -1) { //< Connect to the host
        (void) fprintf(stderr, "ERROR: Failure connecting to the server.\n");
        exit(EXIT_FAILURE);
    }

    if((send(sockfd, header, strlen(header), 0)) < 0) { //< Send the GET request
        (void) fprintf(stderr, "ERROR: Failure sending the request to the server.\n");
        exit(EXIT_FAILURE);
    }

    char reply[__INT16_MAX__]; //< Set the variable to hold the server response. Set to a big length

    if((recv(sockfd, reply, strlen(reply) - 1, 0)) < 0) { //< Receive response from the remote (-1 for the trailing terminating char)
        (void) fprintf(stderr, "ERROR: Did not receive response from the remote.\n");
        exit(EXIT_FAILURE);
    }

    process_response(reply, ofile, dir);

    freeaddrinfo(config); //< Free the addrinfo struct

    exit(EXIT_SUCCESS); //< Exit with success
}

static void print_usage(void) {
    puts("Usage:");
    puts("client [-p PORT] [ -o FILE | -d DIR ] URL\n");
    exit(EXIT_FAILURE);
}

/**
 * @brief Function to check the number of arguments
 * @details Checks the number of arguments supplied to the command line.
 *          If the arguments are less than the required or more than possible, it exits
 * @param argc - Number of arguments obtained from the command line
 * @return none
 *
 **/
static void check_opts_number(int argc) {
    if(argc <= 1) {
        fprintf(stderr, "At least one argument expected.\n");
        print_usage();
        exit(EXIT_FAILURE);
    } else if(argc > 6) {
        fprintf(stderr, "Too many arguments supplied.\n");
        print_usage();
        exit(EXIT_FAILURE);
    }
}

/**
 * @brief Function to check if the address is resolvable by the client
 * @details Checks if the address is resolvable by the client.
 *          If the address is resolvable, it returns the result
 *          If the address is not resolvable, it exits with an error
 * @param
 *          hostname - The address to be resolved
 *          port     - The port to try
 * @return res - The result of getaddrinfo()
 *
 **/
static struct addrinfo* get_host_info(const char *hostname, char *port) {
    struct addrinfo hints; //< Configuration for the client
    hints.ai_family = AF_INET; //< Use IPv4
    hints.ai_socktype = SOCK_STREAM; //< Stream socket
    hints.ai_flags = 0; //< Do not use any flags
    hints.ai_protocol = IPPROTO_TCP; //< Use only TCP

    struct addrinfo* res = NULL;

    int err = getaddrinfo(hostname, port, &hints, &res);

    if(err != 0) {
        (void) fprintf(stderr, "ERROR:  %s\n", gai_strerror(err));
        exit(EXIT_FAILURE); //< Exit if there is an error
    }
    if(res == NULL) {
        (void) fprintf(stderr, "ERROR:  Could not resolve address %s\n", hostname);
        exit(EXIT_FAILURE); //< Exit if the address is not resolved
    }

    return res;
}

/**
 * @brief Function to create a socket using the supplied configuration
 * @details Creates a socket using the supplied configuration
 *          If there was no error, it returns the socket
 *          If there was an error, it exits and displays an error message
 * @param config - The the configuration to be used for the socket creation
 * @return sockfd - The socket obtained from the socket creation
 **/
static int create_socket(struct addrinfo* config) {
    int sockfd = socket(config -> ai_family, config -> ai_socktype, config -> ai_protocol);

    if(sockfd < 0) {
		(void) fprintf(stderr, "ERROR: Socket creation failed.\n");
		exit(EXIT_FAILURE);
	}

    return sockfd;
}

/**
 * @brief Function to process the response of the remote host
 * @details Processes the reply according to its' content and the ofile and dir parameters
 *          The function checks the response code does the following:
 *              - if the reponse code is invalid, it exits with an error
 *              - if the response code is different from 200, it prints the whole response and exits with status 3
 *              - if the response code is 200, it prints it to the console or saves it as a file
 *          If there was an error, it exits and displays an error message
 * @param config - The the configuration to be used for the socket creation
 * @return null
 **/
static void process_response(char reply[sizeof(int32_t)], char *ofile, char *dir) {
    char *http_version = "HTTP/1.1";
    if(strncmp(reply, http_version, strlen(http_version)) == 0) { //< Check if the remote gives a correct response header
        char cresponse_code[3]; //< Array for the response code. Response codes are always 3 digits long
        strncpy(cresponse_code, strstr(reply, "HTTP/1.1 ") + sizeof(http_version) + 1, 3); //< Get the response code as string.

        int response_code = strtol(cresponse_code, NULL, 10); //< Get the response code as integer

        if(response_code == 0) {
            fprintf(stderr, "ERROR: Protocol error.\n");
            exit(EXIT_FAILURE);
        } else if (response_code != 200) {
            printf("%s\n", strtok(reply, "\r\n")); //< Get the first line
        } else {
            char *body = strstr(reply, "\r\n\r\n") + 4;
            if(ofile == NULL && dir == NULL) printf("%s", body); //< Print the body of the response
            else {
                FILE *out = NULL;
                if(ofile != NULL && dir == NULL) {
                    out = fopen(ofile, "w");

                    if(out == NULL) {
                        (void) fprintf(stderr, "ERROR: Could not open ./%s for writing.", ofile);
                        exit(EXIT_FAILURE);
                    } else {
                        fwrite(body, 1, strlen(body), out);
                        fclose(out);
                        printf("File ./%s successfully written.\n", ofile);
                    }
                } else if(ofile != NULL && dir != NULL) {

                    char path[sizeof(dir) + 100];

                    snprintf(path, sizeof path, "%s%s", dir, ofile);

                    out = fopen(path, "w"); //< Open for writing

                    if(out == NULL) {
                        (void) fprintf(stderr, "ERROR: Could not open %s%s for writing.\n", dir, ofile);
                        exit(EXIT_FAILURE);
                    } else {
                        fwrite(body, 1, strlen(body), out);
                        fclose(out);
                        printf("File %s%s successfully written.\n", dir, ofile);
                    }
                }
            }
        }
    } else {
        fprintf(stderr, "ERROR: Protocol error.\n");
        exit(EXIT_FAILURE);
    }
}