file_get_contents returns HTTP headers in content

Question

I am using file_get_contents() to perform a POST to an API service. This is working for most of the queries. However, recently I had a failure in which file_get_contents() returned HTTP headers in the content!

For the code:

$resp = file_get_contents("http://10.72.18.21:8000",false, $context);
var_dump($resp);
var_dump($http_response_header);

I get the following content and headers every time:

string(114328) "Server: Spark Proxy Server
Content-Length: 114272

{"records":....}
"
array(1) {
  [0]=>
  string(31) "HTTP/1.0 200 Spark Proxy Server"
}

Notice that inside the $resp you can see headers (the first two lines) which should normally be parsed in $http_response_header.

I have also tried two different approaches to do the POST: (a) fopen + while loop and (b) fopen + stream_get_contents. In all three cases the results are the same. The common thing between all three is the stream context which I create using:

$opts = array('http' =>
array(
    'method'  => 'POST',
    'header'  => 'Content-type: application/x-www-form-urlencoded;charset=utf-8',
    'content' => http_build_query($params)
    )
);

$context  = stream_context_create($opts);

Now, querying the same API with the same code but with a single parameter modified, everything works as expected and the complete headers are:

array(3) {
  [0]=>
  string(31) "HTTP/1.0 200 Spark Proxy Server"
  [1]=>
  string(26) "Server: Spark Proxy Server"
  [2]=>
  string(19) "Content-Length: 288"
}

Finally, I have tried the same call that fails with python and cUrl and in both cases the results are correct, so I am pretty sure it is php related issue.

Questions:

Has anyone seen this behavior before?
Is there a way to perform the same POST without using stream_context_create?
Can it be time related? The example that fails takes up to 1.4 minutes to complete

`10.72.18.21` is a private network address, check with your tech support that they have not recently added a proxy server somewhere in the mix, or altered the config of one that was there all the time — RiggsFolly, Apr 21 '16 at 12:48
@RiggsFolly Hi, actually that IP is a (custom) proxy server :) Filters my queries and in turn uses the real API (which I cannot access), reorganizes the results and sends them back. However, I have tested with curl and browsers and it works... so I don't think is a network issue — urban, Apr 22 '16 at 11:26

score 0 · Accepted Answer · edited May 23 '17 at 12:06

I have tried this again today... and instead of headers mixed in the content, I got empty content and partial headers. This error though is a lot easier to google and lead me to this SO answer talking about a similar problem (weird file_get_contents behavior).

As the first comment hinted, changing the default socket timeout solves my problem (it was set to 60):

ini_set("default_socket_timeout", 600);

I am not quite sure how the socket can timeout since the data and the headers received seemed complete...

file_get_contents returns HTTP headers in content

1 Answers1

Linked