hashttp - An http server that serves content by its hash sum

This simple http server takes a bunch of files as command line arguments, generates SHA512 hashes for each of them and serves them via http, using the hash as as path in the request URI.

Apart from regular files, the code also handles raw Linux block devices. Built-in support for basic content parsing allows for proper file size determination.

Originally, the code was written to serve a squashfs image from one machine to another as part of an update process. A client that knows the SHA512 can then easily retrieve the contents from another machine. Considering the SHA512 secret, it also serves as an access control method.

Building

This project is implemented in Go, and the build process is as simple as this:

go build

Parameters

There are several command line parameters that can be passed:

• --source may be used several times to specify a local file to be scanned and served. At least one such file must be given. If multiple given files happen to have the same hash, only the first of them is used.

• --port specifies the port to listen to. Must be passed, the program does not default to a value.

• --url-prefix can be optionally used to pass a prefix to expect before the hash sum.

File type handling

Block devices can be used as a source directly, in which case the size of the device will be determined, and the entire content is read and served.

This, however, is not always what is desired. For instance, when file system images are stored on a block device, only a fraction of the device actually contains useful data. To support this, hashttp is prepared to parse the headers of such content to determine the actual size that is of interest.

Currently, only squashfs images are detected during parsing, but more file types can easily be added. Please do post PRs!

Example

Let's create a file and see which SHA512 is has.

$ echo "Hello, world" >file.txt
$ sha512sum file.txt
959c0bdfa9877d3466c5848f55264f72f132c657b002b79fda65dbe36c67f4bb3d2a3e2e9925cb5896a53c76169c5bb71b7853bd90192068dc77f4b20159a1d8  file.txt

Now, let's start the server and point it to that file.

$ ./hashttp --source file.txt --port 5555 --url-prefix files                                                                                                                                                                (130) [14:56:01]
2019/01/05 14:56:03 Processing source file.txt ...
2019/01/05 14:56:03 Serving file.txt (type plain) on /files/959c0bdfa9877d3466c5848f55264f72f132c657b002b79fda65dbe36c67f4bb3d2a3e2e9925cb5896a53c76169c5bb71b7853bd90192068dc77f4b20159a1d8
2019/01/05 14:56:03 Listening on port 5555 ...

And then retrieve it again via http.

$ curl -o - http://localhost:5555/files/959c0bdfa9877d3466c5848f55264f72f132c657b002b79fda65dbe36c67f4bb3d2a3e2e9925cb5896a53c76169c5bb71b7853bd90192068dc77f4b20159a1d8                                         (2) [14:57:36]
Hello, world

License

GPLv2, see the LICENSE.GPL2 file.