C语言字符串中的URL解析与重构

在C语言中，处理和解析URL字符串可以通过一些基本的字符串操作来实现

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

typedef struct {
    char *scheme;
    char *host;
    char *port;
    char *path;
    char *query;
} URL;

void parse_url(const char *url_str, URL *url) {
    const char *ptr = url_str;
    const char *end = url_str + strlen(url_str);

    // 解析scheme
    const char *scheme_end = strstr(ptr, "://");
    if (scheme_end == NULL) {
        printf("Invalid URL: %s\n", url_str);
        return;
    }
    url->scheme = (char *)malloc(scheme_end - ptr + 1);
    memcpy(url->scheme, ptr, scheme_end - ptr);
    url->scheme[scheme_end - ptr] = '\0';
    ptr = scheme_end + 3;

    // 解析host
    const char *host_end = strchr(ptr, ':');
    if (host_end == NULL || host_end > end) {
        host_end = strchr(ptr, '/');
        if (host_end == NULL || host_end > end) {
            host_end = end;
        }
    }
    url->host = (char *)malloc(host_end - ptr + 1);
    memcpy(url->host, ptr, host_end - ptr);
    url->host[host_end - ptr] = '\0';
    ptr = host_end;

    // 解析port
    if (*ptr == ':') {
        const char *port_end = strchr(ptr + 1, '/');
        if (port_end == NULL || port_end > end) {
            port_end = end;
        }
        url->port = (char *)malloc(port_end - ptr);
        memcpy(url->port, ptr + 1, port_end - ptr - 1);
        url->port[port_end - ptr - 1] = '\0';
        ptr = port_end;
    } else {
        url->port = NULL;
    }

    // 解析path
    if (*ptr == '/') {
        const char *path_end = strchr(ptr, '?');
        if (path_end == NULL || path_end > end) {
            path_end = end;
        }
        url->path = (char *)malloc(path_end - ptr + 1);
        memcpy(url->path, ptr, path_end - ptr);
        url->path[path_end - ptr] = '\0';
        ptr = path_end;
    } else {
        url->path = NULL;
    }

    // 解析query
    if (*ptr == '?') {
        url->query = (char *)malloc(end - ptr + 1);
        memcpy(url->query, ptr, end - ptr);
        url->query[end - ptr] = '\0';
    } else {
        url->query = NULL;
    }
}

void reconstruct_url(const URL *url, char **url_str) {
    int len = strlen(url->scheme) + strlen(url->host) + 3; // "://"
    if (url->port != NULL) {
        len += strlen(url->port) + 1; // ":"
    }
    if (url->path != NULL) {
        len += strlen(url->path);
    }
    if (url->query != NULL) {
        len += strlen(url->query) + 1; // "?"
    }

    *url_str = (char *)malloc(len + 1);
    sprintf(*url_str, "%s://%s", url->scheme, url->host);
    if (url->port != NULL) {
        strcat(*url_str, ":");
        strcat(*url_str, url->port);
    }
    if (url->path != NULL) {
        strcat(*url_str, url->path);
    }
    if (url->query != NULL) {
        strcat(*url_str, "?");
        strcat(*url_str, url->query);
    }
}

int main() {
    const char *url_str = "https://www.example.com:80/path?query";
    URL url;
    parse_url(url_str, &url);

    char *reconstructed_url_str;
    reconstruct_url(&url, &reconstructed_url_str);
    printf("Original URL: %s\n", url_str);
    printf("Reconstructed URL: %s\n", reconstructed_url_str);

    free(url.scheme);
    free(url.host);
    free(url.port);
    free(url.path);
    free(url.query);
    free(reconstructed_url_str);

    return 0;
}

这个示例中的parse_url函数将输入的URL字符串解析为一个URL结构体。reconstruct_url函数则根据URL结构体重新构建URL字符串。注意，这个示例仅适用于简单的URL，对于包含特殊字符或编码的URL可能需要进一步处理。