This is something that has been bugging me for a while.. I'm building of a RESTful API that has to receive files on some occasions.
When using HTTP POST, we can read data from $_POST and files from $_FILES.
When using HTTP GET, we can read data from $_GET and files from $_FILES.
However, when using HTTP PUT, AFAIK the only way to read data is to use the php://input stream.
All good and well, untill I want to send a file over HTTP PUT. Now the php://input stream doesn't work as expected anymore, since it has a file in there as well.
Here's how I currently read data on a PUT request:
(which works great as long as there are no files posted)
$handle = fopen('php://input', 'r');
$rawData = '';
while ($chunk = fread($handle, 1024)) {
$rawData .= $chunk;
}
parse_str($rawData, $data);
When I then output rawData, it shows
-----ZENDHTTPCLIENT-44cf242ea3173cfa0b97f80c68608c4c
Content-Disposition: form-data; name="image_01"; filename="lorem-ipsum.png"
Content-Type: image/png; charset=binary
�PNG
���...etc etc...
���,
-----ZENDHTTPCLIENT-8e4c65a6678d3ef287a07eb1da6a5380
Content-Disposition: form-data; name="testkey"
testvalue
-----ZENDHTTPCLIENT-8e4c65a6678d3ef287a07eb1da6a5380
Content-Disposition: form-data; name="otherkey"
othervalue
Does anyone know how to properly receive files over HTTP PUT, or how to parse files out of the php://input stream?
===== UPDATE #1 =====
I have tried only the above method, don't really have a clue as to what I can do else.
I have gotten no errors using this method, besides that I don't get the desired result of the posted data and files.
===== UPDATE #2 =====
I'm sending this test request using Zend_Http_Client, as follows:
(haven't had any problems with Zend_Http_Client so far)
$client = new Zend_Http_Client();
$client->setConfig(array(
'strict' => false,
'maxredirects' => 0,
'timeout' => 30)
);
$client->setUri( 'http://...' );
$client->setMethod(Zend_Http_Client::PUT);
$client->setFileUpload( dirname(__FILE__) . '/files/lorem-ipsum.png', 'image_01');
$client->setParameterPost(array('testkey' => 'testvalue', 'otherkey' => 'othervalue');
$client->setHeaders(array(
'api_key' => '...',
'identity' => '...',
'credential' => '...'
));
===== SOLUTION =====
Turns out I made some wrong assumptions, mainly that HTTP PUT would be similar to HTTP POST. As you can read below, DaveRandom explained to me that HTTP PUT is not meant for transferring multiple files on the same request.
I have now moved the transferring of formdata from the body to url querystring. The body now holds the contents of a single file.
For more information, read DaveRandom's answer. It's epic.
The data you show does not depict a valid PUT request body (well, it could, but I highly doubt it). What it shows is a multipart/form-data request body - the MIME type used when uploading files via HTTP POST through an HTML form.
PUT requests should exactly compliment the response to a GET request - they send you the file contents in the message body, and nothing else.
Essentially what I'm saying is that it is not your code to receive the file that is wrong, it is the code that is making the request - the client code is incorrect, not the code you show here (although the parse_str() call is a pointless exercise).
If you explain what the client is (a browser, script on other server, etc) then I can help you take this further. As it is, the appropriate request method for the request body that you depict is POST, not PUT.
Let's take a step back from the problem, and look at the HTTP protocol in general - specifically the client request side - hopefully this will help you understand how all of this is supposed to work. First, a little history (if you're not interested in this, feel free to skip this section).
History
HTTP was originally designed as a mechanism for retrieving HTML documents from remote servers. At first it effectively supported only the GET method, whereby the client would request a document by name and the server would return it to the client. The first public specification for HTTP, labelled as HTTP 0.9, appeared in 1991 - and if you're interested, you can read it here.
The HTTP 1.0 specification (formalised in 1996 with RFC 1945) expanded the capabilities of the protocol considerably, adding the HEAD and POST methods. It was not backwards compatible with HTTP 0.9, due to a change in the format of the response - a response code was added, as well as the ability to include metadata for the returned document in the form of MIME format headers - key/value data pairs. HTTP 1.0 also abstracted the protocol from HTML, allowing for the transfer of files and data in other formats.
HTTP 1.1, the form of the protocol that is almost exclusively in use today is built on top of HTTP 1.0 and was designed to be backwards compatible with HTTP 1.0 implementations. It was standardised in 1999 with RFC 2616. If you are a developer working with HTTP, get to know this document - it is your bible. Understanding it fully will give you a considerable advantage over your peers who do not.
Get to the point already
HTTP works on a request-response architecture - the client sends a request message to the server, the server returns a response message to the client.
A request message includes a METHOD, a URI and optionally, a number of HEADERS. The request METHOD is what this question relates to, so it is what I will cover in the most depth here - but first it is important to understand exactly what we mean when we talk about the request URI.
The URI is the location on the server of the resource we are requesting. In general, this consists of a path component, and optionally a query string. There are circumstances where other components may be present as well, but for the purposes of simplicity we shall ignore them for now.
Let's imagine you type http://server.domain.tld/path/to/document.ext?key=value into the address bar of your browser. The browser dismantles this string, and determines that it needs to connect to an HTTP server at server.domain.tld, and ask for the document at /path/to/document.ext?key=value.
The generated HTTP 1.1 request will look (at a minimum) like this:
GET /path/to/document.ext?key=value HTTP/1.1
Host: server.domain.tld
The first part of the request is the word GET - this is the request METHOD. The next part is the path to the file we are requesting - this is the request URI. At the end of this first line is an identifier indicating the protocol version in use. On the following line you can see a header in MIME format, called Host. HTTP 1.1 mandates that the Host: header be included with every request. This is the only header of which this is true.
The request URI is broken into two parts - everything to the left of the question mark ? is the path, everything to the right of it is the query string.
Request Methods
RFC 2616 (HTTP/1.1) defines 8 request methods.
OPTIONS
The OPTIONS method is rarely used. It is intended as a mechanism for determining what kind of functionality the server supports before attempting to consume a service the server may provide.
Off the top of my head, the only place in fairly common usage that I can think of where this is used is when opening documents in Microsoft office directly over HTTP from Internet Explorer - Office will send an OPTIONS request to the server to determine if it supports the PUT method for the specific URI, and if it does it will open the document in a way that allows the user to save their changes to the document directly back to the remote server. This functionality is tightly integrated within these specific Microsoft applications.
GET
This is by far and away the most common method in every day usage. Every time you load a regular document in your web browser it will be a GET request.
The GET method requests that the server return a specific document. The only data that should be transmitted to the server is information that the server requires to determine which document should be returned. This can include information that the server can use to dynamically generate the document, which is sent in the form of headers and/or query string in the request URI. While we're on the subject - Cookies are sent in the request headers.
HEAD
This method is identical to the GET method, with one difference - the server will not return the requested document, if will only return the headers that would be included in the response. This is useful for determining, for example, if a particular document exists without having to transfer and process the entire document.
POST
This is the second most commonly used method, and arguably the most complex. POST method requests are almost exclusively used to invoke some actions on the server that may change its state.
A POST request, unlike GET and HEAD, can (and usually does) include some data in the body of the request message. This data can be in any format, but most commonly it is a query string (in the same format as it would appear in the request URI) or a multipart message that can communicate key/value pairs along with file attachments.
Many HTML forms use the POST method. In order to upload files from a browser, you would need to use the POST method for your form.
The POST method is semantically incompatible with RESTful APIs because it is not idempotent. That is to say, a second identical POST request may result in a further change to the state of the server. This contradicts the "stateless" constraint of REST.
PUT
This directly complements GET. Where a GET requests indicates that the server should return the document at the location specified by the request URI in the response body, the PUT method indicates that the server should store the data in the request body at the location specified by the request URI.
DELETE
This indicates that the server should destroy the document at the location indicated by the request URI. Very few internet facing HTTP server implementations will perform any action when they receive a DELETE request, for fairly obvious reasons.
TRACE
This provides an application-layer level mechanism to allow clients to inspect the request it has sent as it looks by the time it reaches the destination server. This is mostly useful for determining the effect that any proxy servers between the client and the destination server may be having on the request message.
CONNECT
HTTP 1.1 reserves the name for a CONNECT method, but does not define its usage, or even its purpose. Some proxy server implementations have since used the CONNECT method to facilitate HTTP tunnelling.
I've never tried using PUT (GET POST and FILES were sufficient for my needs) but this example is from the php docs so it might help you (http://php.net/manual/en/features.file-upload.put-method.php):
<?php
/* PUT data comes in on the stdin stream */
$putdata = fopen("php://input", "r");
/* Open a file for writing */
$fp = fopen("myputfile.ext", "w");
/* Read the data 1 KB at a time
and write to the file */
while ($data = fread($putdata, 1024))
fwrite($fp, $data);
/* Close the streams */
fclose($fp);
fclose($putdata);
?>
Here is the solution that I found to be the most useful.
$put = array();
parse_str(file_get_contents('php://input'), $put);
$put will be an array, just like you are used to seeing in $_POST, except now you can follow true REST HTTP protocol.
Use POST and include an X- header to indicate the actual method (PUT in this case). Usually this is how one works around a firewall which does not allow methods other than GET and POST. Simply declare PHP buggy (since it refuses to handle multipart PUT payloads, it IS buggy), and treat it as you would an outdated/draconian firewall.
The opinions as to what PUT means in relation to GET are just that, opinions. The HTTP makes no such requirement. It simply states 'equivalent' .. it is up to the designer to determine what 'equivalent' means. If your design can accept a multi-file upload PUT and produce an 'equivalent' representation for a subsequent GET for the same resource, that's just fine and dandy, both technically and philosophically, with the HTTP specifications.
Just follow what it says in the DOC:
<?php
/* PUT data comes in on the stdin stream */
$putdata = fopen("php://input", "r");
/* Open a file for writing */
$fp = fopen("myputfile.ext", "w");
/* Read the data 1 KB at a time
and write to the file */
while ($data = fread($putdata, 1024))
fwrite($fp, $data);
/* Close the streams */
fclose($fp);
fclose($putdata);
?>
This should read the whole file that is on the PUT stream and save it locally, then you could do what you want with it.
How to check whether a page has been updated without downloading the entire webpage in Php? Whether I need to look in at the header?
One possibility is to check the LastModified header. You can download just the headers by issuing a HEAD request. The server responds back with the HTTP headers only and you can inspect the last modified header and/or the content length header to detect changes.
Last-modified "Mon, 03 Jan 2011 13:02:54 GMT"
One thing to note is that the HTTP server does not need to send this header so this would not work in all cases. The PHP function get_headers will fetch these for you.
// By default get_headers uses a GET request to fetch the headers. If you
// want to send a HEAD request instead, you can do so using a stream context:
stream_context_set_default(
array(
'http' => array(
'method' => 'HEAD'
)
)
);
$headers = get_headers('http://example.com');
You can add a If-Modified-Since: <datetime> header to your request, and the server should return a 304 Not Modified if it hasn't changed since then. But if the document is generated dynamically (php, perl, etc.), the generator could be too lazy to check this header and always return the full document.
A centain web client that I need to support, is sending back the Cookies header to my application twice in the HTTP headers, this in turn is making PHP unable to read the correct value for the cookie thus ignoring the session.
Here is the relevant part of the request I am seeing:
GET / HTTP/1.1
Cache-Control: max-age=0
Accept-Language: en-US
Cookie: PHPSESSID=49af82ddf12740e6a35b15985e93d91a
Connection: Keep-Alive
Cookie: PHPSESSID=49af82ddf12740e6a35b15985e93d91a
[...] Other irrelevant headers
I have two questions:
Is that a PHP bug? or is the behavior undefined when the client sends that same header twice?
Is there a quick workaround to make things work without having to manually parse the HTTP headers so I can read the right value of the cookie (and session) in my application? Or should I manually parse the HTTP header to set the session to its correct value?
According to the HTTP spec, a double header simply concatenates the values together with a comma, making it:
Cookie: PHPSESSID=49af82ddf12740e6a35b15985e93d91a, PHPSESSID=49af82ddf12740e6a35b15985e93d91a
PHP should be able to parse the cookies, but the behavior of sessions is undefined when there are two session IDs.
I strongly recommend fixing the client. If that's not an option, you'll have to parse the headers manually.
I am quite confused with the Content-Type vendor specific. Say by default,
Content-Type: application/json
but with vendor specific type, I can have this
Content-Type: application/vnd.anything.process-v1+json
Do I need to have any special function to do in PHP to use if I would check if the Content-Type is not vendor specific?
Thanks.
If you are accepting content via HTTP POST from a client you can check the content type with $_SERVER['CONTENT_TYPE']. I would encourage you to use the #ver attribute in the Content-Type header instead of embedding in your vendor content type name. Example:
Content-Type: application/vnd.anything.process+json;ver=1
If you are sending content to a server via HTTP POST you need to specify an accept header to tell the client to return the data in the vendor specific content type. Something like this would work:
header('Accept: application/vnd.anything.process+json;ver=1');
Accept headers can be quite complicated though. If you do not control the server you are posting to as client, you should provide a sensible default to your Accept header. Something like this is more friendly if you want the server to send back plain JSON if it doesn't understand the vendor header:
header('Accept: application/vnd.anything.process+json;ver=1;q=0.9, application/json;q=0.1');
The most permissive Accept headers accept anything though:
header('Accept: application/vnd.anything.process+json;ver=1;q=0.9, application/json;q=0.5, */*;q=0.1');
Here is the actual RFC for Accept headers: http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html
is it possible to send referer information with php?
If you are, for example, fetching the contents of a URL in PHP using cURL, you can send any additional headers you want, including a referrer header.
You can not force the users browser to send a referrer header by any means, especially not with a server side language.
It's not possible to get the client browser to send a different Referer header.
However, it is theory possible for you to do this when conducting an HTTP request from PHP (either using cURL or native URL wrappers), but including a custom request header in this request.
Yes, when trying to load a page, just write the Referer header to the output stream.
Referer is a 'request' header meaning sent by the client i.e. browser. From server side i.e. using PHP you can only control 'response' headers.
If you are planning to make HTTP requests with PHP, that is different of course.
Edit: ..and requests made from the server to the other servers is a pretty common scenario actually. It seems like you should be able to set the headers you want while creating the HttpRequest:
$options = array(headers => $header_array,
httpauth => $credentials);
$r = new HttpRequest($url, HTTP_METH_POST, $options);
Or you can use the addHeaders method:
$r->addHeaders(array('Referer' => 'http://example.com'));