This library provides a /stream-way/ to extracts values from a multipart/form-data HTTP 1.{0,1} document. The library does not requires an HTTP server but can be used with one of them. The library is pretty straightforward to co-exist with:
The multipart/form-data appears when you want to propose a form to the client. You can encode values given by the client into the URL with application/x-www-form-urlencoded or, if you want to let the client to upload a file, multipart/form-data is required.
This library wants to process the second case.
Due to the use of angstrom, multipart_form is able to manipulate a /stream/ instead of process entirely the given document. Indeed, when the client wants to upload a file, it will be a shame to store twice times the file into the memory heap (firstly, keep entirely the multipart/form-data document, secondly, extract the file from the first one).
So, the user is able to concurrently /eat/ the multipart/form-data document, feed the angstrom's state and save a part (such as the uploaded file) to another location. By this way, we keep a control about memory consumption.
multipart_form has its own definition of the Content-Type needed to start the analyse of the given document. From cohttp or httpaf, you are able to extract it as is and give it to Multipart_form.Content_type.of_string to get a well-defined Content-Type value.
Then, you can start to /parse/ the document with this value and an emitters.
An emitters is a simple function which must give to the process a way to /store/ the part and a unique identifier to be able to retrieve contents at the end of the process:
type 'id emitters = unit -> 'id * (string option -> unit)A use of Lwt_stream.create can gives you such required value.
This library was done for my personal use and I probably missed some useful elements for a basic user who wants to process a multipart/form-data document. So, feel free to make an issue if you think that something is missed.