URI canonicalizer

The URI canonicalizer is a service which parses a URI, does some normalization and returns a structured object similar to the URL object in browsers. The motivation for having this as a service is to ensure that URIs are parsed and normalized in the same way independently of programming language, configuration and so on.

Canoncalization

The canoncalization is (for now) configured in code and tries to not change the URIs semantics. Examples of canoncalization are:

• Remove port numbers for well known schemes (i.e. http://example.com:80 → http://example.com)
• Normalize slash for empty path (i.e. http://example.com → http://example.com/)
• Normalize path (i.e. http://example.com/a//b/./c → http://example.com/a/b/c)
• Remove user info (i.e. http://user:passwd@example.com → http://example.com/)
• Sort query (i.e. http://example.com/foo?b=2&a=3&c=4&b=1/ → http://example.com/foo?a=3&b=2&b=1&c=4)

  Only the query parameter names are sorted. This is less likely to alter semantics than also sort values.

API

The API is implemented as a gRPC service.

The request is the URI a as string.

The respones is a ParsedUri object defined as follows:

message ParsedUri {
    // The entire uri
    string href = 1;
    // The scheme (protocol) part of the uri
    string scheme = 2;
    // The hostname of the uri
    string host = 3;
    // The port number of the uri
    int32 port = 4;
    // The username part of the uri
    string username = 5;
    // The password part of the uri
    string password = 6;
    // The path part of the uri
    string path = 7;
    // The query (search) part of the uri
    string query = 8;
    // The fragment (hash) part of the uri
    string fragment = 9;
}