PouchDB wrongly flags + as illegal character for database name

[colinskow]

new PouchDB('http://localhost:5984/my+database')
  .catch(function(err) {
    console.log(err);
  });

Responds...

{ [illegal_database_name: Database encountered an unknown error]
  status: 400,
  name: 'illegal_database_name',
  message: 'Database encountered an unknown error',
  error: true,
  reason: 'Name: \'my database\'. Only lowercase characters (a-z), digits (0-9), and any of the characters _, $, (, ), +, -, and / are allowed. Must begin with a letter.' }

[willholley]

This error is just passed through from CouchDB in this instance. It will work if you percent-encode the +:

new PouchDB('http://localhost:5984/my%2Bdatabase')
.catch(function(err) {
   console.log(err);
});

[daleharvey]

I am fairly sure we purposefully dont url encode the database name, but I cant quite think of why right now

[daleharvey]

Nope, cant think of any reason to not do this

[daleharvey]

fixed in 94c7276

[lupomontero]

This breaks things when a database name contains a /, which is valid as far as CouchDB is concerned (ie: user databases in Hoodie are named user/randomid).

Before this change we could work with database names like this, but not now. Consider a CouchDB server being served in a subdirectory (ie: http://foo.com/_couch/), and a database called user/bar, so the URL would be http://foo.com/_couch/user%2Fbar.

// pre 5.1.0 this worked, now it gets double encoded
var db = new PouchDB('http://foo.com/_couch/user%2Fbar');

// now, without encoding PouchDB will think `bar` is the database name
var db = new PouchDB('http://foo.com/_couch/user/bar');

This is because PouchDB is splitting the url using / as the separator, and then encoding the last part. From source:

  // Split the path part of the URI into parts using '/' as the delimiter
  // after removing any leading '/' and any trailing '/'
  var parts = uri.path.replace(/(^\/|\/$)/g, '').split('/');

  // Store the first part as the database name and remove it from the parts
  // array
  uri.db = encodeURIComponent(parts.pop());

A possible solution would be to check for the presence of % in db name before it is encoded with encodeURIComponent() as suggested by @colinskow here. Something like:

  uri.db = parts.pop();
  if (uri.db.indexOf('%') === -1) {
    uri.db = encodeURIComponent(uri.db);
  }

I will send PR shortly.

cc/ @daleharvey