curl - fail-with-body support

[rbro]

Description

In cURL 7.76.0, a new option --fail-with-body was added:

https://daniel.haxx.se/blog/2021/02/11/curl-fail-with-body

Would it be possible to implement similar functionality in PHP's curl library?

I posted the below in Stack Overflow in 2021:

https://stackoverflow.com/questions/68246166/how-to-retrieve-curl-response-on-error-in-php

As I wrote in the post, I'm looking to:

• on success, have the http response pass directly to the user's browser
• but on error such as a http 500 error, get the http response body into a variable in PHP

I don't want to use CURLOPT_RETURNTRANSFER because on success, the PDF file returned is large and it would be inefficient to first save the file to a PHP string or a temp file, only to just echo it to the user's browser.

What I'm looking for is an option called CURLOPT_RETURNTRANSFERONERROR, where the response data is only returned on a http error.

Thank you for your help.

Ryan

[KapitanOczywisty]

--fail-with-body is only implemented in CLI, so no, but it's not needed for PHP, you can check curl_getinfo($curl, CURLINFO_RESPONSE_CODE) and handle response as normal.

[rbro]

I'm using CURLINFO_HTTP_CODE to get the http response code, but how do I get the response body on error?

[KapitanOczywisty]

Set CURLOPT_RETURNTRANSFER = true and CURLOPT_FAILONERROR = false then curl_exec will return body.

[rbro]

Yes, but as I wrote in my question on Stack Overflow, the file being downloaded is large and I don't want to use CURLOPT_RETURNTRANSFER and first save the file to a PHP string or a temp file, only to then send the file contents to the user's browser. Instead, I want to pass the file contents directly to the user's browser..

I only want to return the response body on error, which is similar in idea to --fail-with-body. Is there an equivalent in PHP to pass the response body through to the user on success, but have it saved into a variable on error?

[KapitanOczywisty]

You can use CURLOPT_WRITEFUNCTION:

$curl = curl_init();

curl_setopt($curl, CURLOPT_URL, '...');

// both set to false
curl_setopt($curl, CURLOPT_FAILONERROR, false);
curl_setopt($curl, CURLOPT_RETURNTRANSFER, false);

curl_setopt($curl, CURLOPT_WRITEFUNCTION, function ($curl, $data) {
	if (curl_getinfo($curl, CURLINFO_RESPONSE_CODE) === 200) {
		echo $data;
	} else {
		// log error response here...
	}
	return strlen($data);
});

curl_exec($curl);

if (curl_errno($curl)) {
	printf("Error: %s\n", curl_error($curl));
} elseif (curl_getinfo($curl, CURLINFO_RESPONSE_CODE) !== 200) {
	printf("Invalid code: %s\n", curl_getinfo($curl, CURLINFO_RESPONSE_CODE));
} else {
	printf("Success\n");
}

[rbro]

Thanks, I didn't realize that CURLOPT_WRITEFUNCTION would be called in chunks. I thought it was the entire response in one call. This works great. Thanks again for your help.