PouchDB Collate

Collation functions for PouchDB map/reduce. Used by PouchDB map/reduce to maintain consistent CouchDB collation ordering.

The PouchDB Collate API is not exposed by PouchDB itself, but if you'd like to use it in your own projects, it's pretty small, and it has a few functions you may find useful.

Usage

1npm install pouchdb-collate

1var pouchCollate = require('pouchdb-collate');

Warning: semver-free zone!

This package is conceptually an internal API used by PouchDB or its plugins. It does not follow semantic versioning (semver), and rather its version is pegged to PouchDB's. Use exact versions when installing, e.g. with --save-exact.

Source

PouchDB and its sub-packages are distributed as a monorepo.

For a full list of packages, see the GitHub source.

API

toIndexableString(obj)

This is probably the most useful function in PouchDB Collate. It converts any object to a serialized string that maintains proper CouchDB collation ordering in both PouchDB and CouchDB (ignoring some subtleties with ICU string ordering in CouchDB vs. ASCII string ordering in PouchDB).

So for example, if you want to sort your documents by many properties in an array, you can do e.g.:

1var pouchCollate = require('pouchdb-collate');
2var myDoc = {
3  firstName: 'Scrooge',
4  lastName: 'McDuck',
5  age: 67,
6  male: true
7};
8// sort by age, then gender, then last name, then first name
9myDoc._id = pouchCollate.toIndexableString(
10  [myDoc.age, myDoc.male, mydoc.lastName, mydoc.firstName]);

The doc ID will be:

1'5323256.70000000000000017764\u000021\u00004McDuck\u00004Scrooge\u0000\u0000'

Which is of course totally not human-readable, but it'll sort everything correctly (floats, booleans, ints – you name it). If you need a human-readable doc ID, check out the DocURI project.

Warning! If you are syncing or storing docs in CouchDB, then you will need to modify these doc IDs, due to a bug in how Chrome parses URLs, which causes problems in the replicator when it tries to GET docs at those URLs.

In short, you will need to replace all the \u0000 characters with some other separator. Assuming you're storing text data and not binary data, \u0001 should be fine:

1pouchCollate.toIndexableString([/* ... */])
2    .replace(/\u0000/g, '\u0001');

parseIndexableString(str)

Same as the above, but in reverse. Given an indexable string, it'll give you back a structured object.

For instance:

1var pouchCollate = require('pouchdb-collate');
2
3// [ 67, true, 'McDuck', 'Scrooge' ]
4pouchCollate.parseIndexableString(
5  '5323256.70000000000000017764\u000021\u00004McDuck\u00004Scrooge\u0000\u0000')

collate(obj1, obj2)

Give it two objects, and it'll return a number comparing them. For example:

1pouchCollate.collate('foo', 'bar'); // 1
2pouchCollate.collate('bar', 'foo'); // -1
3pouchCollate.collate('foo', 'foo'); // 0

Of course it sorts more than just strings - any valid JavaScript object is sortable.

normalizeKey(obj)

You shouldn't need to use this, but this function will normalize the object and return what CouchDB would expect - e.g. undefined becomes null, and Dates become date.toJSON(). It's basically what you would get if you called:

1JSON.parse(JSON.stringify(obj));

but a bit faster.