4. Pretty URLs

CouchDB incorporates a straightforwards URL routing and rewriting engine. While it is not as comprehensive as a front-end proxy, it caters for most needs out of the box, including using virtual hostnames to map URLs to specific databases, and provides more complex use cases through a rewriting engine.

This post has a detailed example of a combined vhost and rewriter configuration.

4.1. Virtual Hosts

CouchDB, since 0.11.0, can map requests to different locations based on the Host header, even if they arrive on the some inbound IP address.

This allows different virtual hosts on the same machine to map to different databases or design documents, etc. The most common use case is to map a virtual host to a Rewrite Handler, to provide full control over the application’s URIs.

To add a virtual host, add a CNAME pointer to the DNS for your domain name. For development and testing, it is sufficient to add an entry in the hosts file, typically /etc/hosts` on Unix-like operating systems:

# CouchDB vhost definitions, refer to local.ini for further details       sofa.couchdb

Test that this is working:

$ ping sofa.couchdb
PING sofa.couchdb ( 56(84) bytes of data.
64 bytes from localhost.localdomain ( icmp_req=1 ttl=64 time=0.025 ms
64 bytes from localhost.localdomain ( icmp_req=2 ttl=64 time=0.051 ms

Finally, add an entry to your configuration file in the [vhosts] section:

sofa.couchdb:5984 = /sofa/_design/sofa/_rewrite

If your CouchDB is listening on the default HTTP port, or is sitting behind a proxy, then don’t specify a port number in the vhost key.

With the above setup, a request to http://sofa.couchdb:5984/sweet-o will be mapped to

New in version 0.11.0: added vhosts functionality

4.2. HTTP Rewrite Handler

Following on from virtual hosts, CouchDB includes a custom URL rewriter. All rewriting is done from /dbname/_design/ddocname/_rewrite by default.

The rewriter is flexible, and can handle methods and custom query formats.

Each rule should be in the rewrites top-level key of the design doc. Example of a complete rule :

    "rewrites": [
        "from": "",
        "to": "index.html",
        "method": "GET",
        "query": {}

from: is the path rule used to bind current uri to the rule. It uses pattern matching for that.

to: rule to rewrite an url. It can contain variables depending on binding variables discovered during pattern matching and query args (url args and from the query member.)

method: method to bind the request method to the rule. If method is missing, any method will be matched in the rewrite.

query: optional query arguments, that may contain dynamic variables, by binding keys in the to be used with the matching URL.

to and from are paths with patterns. The pattern can be strings starting with : or *, for example /somepath/:var/*.

The pattern matching is done by first matching the request method to a rule. Then it will try to match the path to one specific rule. If no rule match, then a 404 error is displayed.

The path is converted into an erlang list, by regex splitting on /. Each variable is converted into an atom. The subsequent pattern matching step is done by splitting / in the request url into a list of atoms. A string pattern will match the equivalent token. The * atom will match any number of tokens, but may only be present as the last pattern in the path. If all tokens are matched, and all path terms have been consumed, then the overall path specification matches.

Once a matching from rule is found we rewrite the request url using the from, to, and query members. Each identified token will be reused within the rule, and in the subsequent query if required. The identified tokens are matched to the rule and will replace var. If * is found in the rule it will contain any remaining suffix.

The rewriter is re-entrant, and has a configurable recursion limit, set by default at 100.