curl_exec() always returns false
Error checking and handling is the programmer's friend. Check the return values of the initializing and executing cURL functions. curl_error()
and curl_errno()
will contain further information in case of failure:
try {
$ch = curl_init();
// Check if initialization had gone wrong*
if ($ch === false) {
throw new Exception('failed to initialize');
}
// Better to explicitly set URL
curl_setopt($ch, CURLOPT_URL, 'http://example.com/');
// That needs to be set; content will spill to STDOUT otherwise
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
// Set more options
curl_setopt(/* ... */);
$content = curl_exec($ch);
// Check the return value of curl_exec(), too
if ($content === false) {
throw new Exception(curl_error($ch), curl_errno($ch));
}
// Check HTTP return code, too; might be something else than 200
$httpReturnCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
/* Process $content here */
} catch(Exception $e) {
trigger_error(sprintf(
'Curl failed with error #%d: %s',
$e->getCode(), $e->getMessage()),
E_USER_ERROR);
} finally {
// Close curl handle unless it failed to initialize
if (is_resource($ch)) {
curl_close($ch);
}
}
* The curl_init()
manual states:
Returns a cURL handle on success, FALSE on errors.
I've observed the function to return FALSE
when you're using its $url
parameter and the domain could not be resolved. If the parameter is unused, the function might never return FALSE
. Always check it anyways, though, since the manual doesn't clearly state what "errors" actually are.
In my case I need to set VERIFYHOST
and VERIFYPEER
to false
, like this:
curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, FALSE);
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, FALSE);
before the call to curl_exec($ch)
.
Because i am working between two development environments with self-assigned certificates.
With valid certificates there is no need to set VERIFYHOST
and VERIFYPEER
to false
because the curl_exec($ch)
method will work and return the response you expect.
This happened to me yesterday and in my case was because I was following a PDF manual to develop some module to communicate with an API and while copying the link directly from the manual, for some odd reason, the hyphen
from the copied link was in a different encoding and hence the curl_exec()
was always returning false
because it was unable to communicate with the server.
It took me a couple hours to finally understand the diference in the characters bellow:
https://www.e‐example.com/api
https://www.e-example.com/api
Every time I tried to access the link directly from a browser it converted to something likehttps://www.xn--eexample-0m3d.com/api
.
It may seem to you that they are equal but if you check the encoding of the hyphens
here you'll see that the first hyphen
is a unicode characters U+2010 and the other is a U+002D.
Hope this helps someone.