curl_url_get for CURLUPART_HOST returns percent decoded string

[muttalkadavul]

I did this

#include <string>
#include <regex>
#include <tuple>
#include <iostream>
#include <curl/curl.h>


int main() {
    std::string input = "http://resum%c3%a9.com";
    CURLUcode rc = CURLUE_BAD_HANDLE;
    CURLU *urlh = curl_url();
    
    std::cout << "libcurl version" << curl_version() << std::endl;
 
    rc = curl_url_set(urlh, CURLUPART_URL, input.c_str(), 0);
    if(rc != CURLUE_OK)
    {
        std::cout << "curlucode = " << rc << std::endl;
        curl_url_cleanup(urlh);
    }

    char *host; 
    rc = curl_url_get(urlh, CURLUPART_HOST, &host, 0);
    if (rc != CURLUE_OK)
    {
        std::cout << "curlucode = " << rc << std::endl;
        curl_url_cleanup(urlh);
    }

    std::cout << host << std::endl; 
    free(host);
    curl_url_cleanup(urlh);
    return 0;
}

I expected the following

libcurl versionlibcurl/8.10.0 OpenSSL/1.1.1f zlib/1.2.11 brotli/1.0.7 libpsl/0.21.0
resum%c3%a9.com

but the actual output was

libcurl versionlibcurl/8.10.0 OpenSSL/1.1.1f zlib/1.2.11 brotli/1.0.7 libpsl/0.21.0
resumé.com

curl/libcurl version

libcurl/8.10.0

operating system

Ubuntu 20.04

[muttalkadavul]

Also, if I set the host directly using curl_url_set(urlh, CURLUPART_HOST, hostname.c_str(), 0); and then call get, the hostname is not percent decoded. I get the expected result.

[Jayanta0123]

Pls check for a way to set the encoding in standards, UTF-8 or something equivalent. Then the character should be printed properly.

…

On Tue, 17 Sept 2024 at 17:34, Venkat Krishna R ***@***.***> wrote: Also, if I set the host directly using curl_url_set(urlh, CURLUPART_HOST, hostname.c_str(), 0); and then call get, the hostname is not percent decoded. I get the expected result. — Reply to this email directly, view it on GitHub <#14942 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADHD2DTVGAHAVPSJZITISTTZXALD5AVCNFSM6AAAAABOLJX74CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNJVGUZDENJZGQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- ------ With Best Wishes, Jayanta Mandal Hyderabad, India

[bagder]

The host name is indeed treated inconsistently with other components without it being documented.