Today I configured the PC of my company to setup a proxy and it worked because I can wget web page via proxy.
What I did is set http_proxy, https_proxy and ftp_proxy in the ~/.bashrc and in the ~/.wgetrc.
However, when I tried to curl , it blocked and timeout.
curl --proxy $http_proxy worked as expected.
Is it possible to make curl use the proxy automatically?
3 Answers
Perhaps the easiest way to get curl to use a proxy is to add the details into ~/.curlrc file. The syntax is as follows:
proxy = <proxy_host>:<proxy_port>This can alternatively be set as an environmental variable but IMHO using ~/.curlrc is the most direct and least error prone method.
Some proxies require specific authentication headers to be set, so be aware of those as well. In my case, it's --proxy-ntlm in the example below:
curl -x webproxy.net:8080 -U usernaname:password --proxy-ntlm
Bu there are other options:
--proxy-digest and --proxy-negotiate
Lastly, cURL has a super friendly doc page, so be sure to check it out.
The client curl (naturally) uses the library libcurl under the hood.
In the context of proxies the libcurl api documentation among other things states:
Environment variables
libcurl respects the proxy environment variables named http_proxy, ftp_proxy, sftp_proxy etc. If set, libcurl will use the specified proxy for that URL scheme. So for a "FTP://" URL, the ftp_proxy is considered. all_proxy is used if no protocol specific proxy was set.
If no_proxy (or NO_PROXY) is set, it is the exact equivalent of setting the CURLOPT_NOPROXY option.
The CURLOPT_PROXY and CURLOPT_NOPROXY options override environment variables.
So if you set the environment variable accordingly, libcurl and consequently curl will pick it up. Any explicitly set proxy (e.g. in .curlrc or via --proxy cmdline argument) overwrites the environment variable settings.
