Redis + Markdown blogging core
Switch branches/tags
Nothing to show
Pull request Compare This branch is 26 commits behind ProjectMoon:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


A Markdown-based blogging and website core backed by Redis and the filesystem.


  • Asynchronously turn all markdown (.md) files in a directory into a blog stored in the hyper-fast Redis database.
  • Turn all markdown files in a (separate) directory into static pages.
  • Files are watched for changes and the Redis store is automagically updated.
  • Transparently access Redis or the filesystem to find a blog post.
  • Markdown metadata to describe your blog posts.
  • Fully event-based programming paradigm.

What reed does not do:

  • Comments. Use a system like Disqus or roll your own. Comments might be added as a separate library later.

What is reed?

Reed is a (very) lightweight blogging core that turns markdown files in a directory into a blog. It is not a fully featured blogging system. If you are looking for that, check out Wheat or another blog engine.

Reed is intended for developers who want to integrate simple blogging functionality into their node.js website or application. It makes as little assumptions as possible about your environment in order to give maximum flexibility.

How to use reed

First, install it:

npm install reed

Make sure Redis 2.2 or greater is also installed and running. After Redis and reed are installed, use it thus:

var reed = require("reed");
reed.on("ready", function() {
	reed.get("a post", function(err, metadata, html) {
		//you have a post.
});"."); //looks for .md files in current directory.

In the above example, .md files will be pulled out of the current directory and indexed into the Redis database by the index function. After having indexed them, we can list the titles in order of post/updated date (that is, last modified date).


Reed can connect to Redis running on separate hosts, non-standard ports or using authentication. This requires the use of the configure function before calling open.

var reed = require("reed");

//configure reed to connect to another redis
//must be done *before*
    host: '',
    port: 1337,
    password: '15qe93rktkf39i4'

reed.on("ready", function() {
	reed.get("a post", function(err, metadata, html) {
		//you have a post.
});"."); //looks for .md files in current directory.

Any property not overridden in the configuration object will use the Redis defaults. For example, it is possible to override just the port.

Retrieving Posts

To retrieve an individual post and its associated metadata, use the get function:

reed.get("First Post", function(err, metadata, htmlContent) {

If retrieval of the post was successful, err will be null. metadata will be an object containing a markdown property that stores the original markdown text, a lastModified property that stores the last modified date as UNIX epoch time, plus any user-defined information (see below). htmlContent will be the post content, converted from markdown to HTML.

If the post could not be retrieved, err will be an object containing error information (exactly what depends on the error thrown), and other two objects will be undefined.

Note that the get function will hit the Redis database first, and then look on the filesystem for a title. So, if you have a new post that has not yet been indexed, it will get automagically added to the index via get.

Article Naming and Metadata

Every article in the blog is a markdown file in the specified directory. The filename is considered the "id" or "slug" of the article, and must be named accordingly. Reed article ids must have no spaces. Instead, spaces are mapped from -s:

"the first post" ->

These ids are case sensitive, so is different than


Similar to Wheat, articles support user-defined metadata at the top of the article. These take the form of simple headers. They are transferred into the metadata object as properties.

Title: The First Post
Author: me
SomeOtherField: 123skidoo

The headers will be accessible thus:

  • metadata.title
  • metadata.someOtherField

Field names can only alphabetical characters. So, "Some-Other-Field" is not a valid article header.

Note: starting in 0.9.3, metadata fields are camelCase, rather than all lower case.

Blog API

Reed exposes the following functions:

  • configure(options): Configures reed. The options object can be used to specify connection settings for Redis. Supported settings are host, port and password. Any such configuration must be done before calling open.
  • open(dir): Opens the given path for reed. When first opened, reed will scan the directory for .md files and add them to redis.
  • close(): Closes reed, shuts down the Redis connection, stops watching all .md files, and clears up state.
  • get(id, callback): Retrieves a blog post. The callback receives error, metadata, and htmlContent.
  • all(callback): Retrieves all blog posts. The callback receives error and posts, which is a list of post objects, each containing metadata and htmlContent properties.
  • getMetadata(id, callback): Retrieves only the metadata for a blog post. The callback receives error (if there was an error), and metadata, an object containing the metadata from the blog post.
  • list(callback): Retrieves all post IDs, sorted by last modified date. The callback receives error if there was an error, and titles, which is a list of post IDs.
  • remove(id, callback): Removes a blog post. The callback receives error, if an error occurred.
  • removeAll(callback): Removes all blog posts. The callback is called after all posts have been deleted, and receives error if there was an error during deletion. This deletion is not transactional!
  • index(callback): Forces a full refresh of the opened directory. This should usually not be necessary, as reed should automatically take care of posts being added and updated. The callback receives error if indexing was prematurely interrupted by an error.
  • refresh(): Forces a refresh of the Redis index, removing any entries that are no longer present on the filesystem. This should usually not be necessary, as reed should handle this internally.

Note: get, list, index, remove, and removeAll asynchronously block until reed is in a ready state. This means they can be called before open, and they will run after opening has completed.

Reed exposes the following events:

  • error: Fired when there is an error in certain internal procedures. Usually, inspecting the error object of callbacks is sufficient.
  • ready: Fired when reed has loaded.
  • add: Fired when a post is added to the blog. Note: posts updated while reed is not running are currently considered add events.
  • update: Fired when a blog post is updated while reed is running. Note: posts updated while reed is not running are currently considered add events.
  • remove: Fired when a blog post is removed (from the filesystem, through an API call, etc). The callback receives the full path of the file that was removed.

Pages API

Reed 0.9 introduces pages functionality. This operates similarly to the blog functionality. Each page is a markdown file in a specified directory. The main difference is that the pages API is not indexed like blog posts are. There are no events exposed by the pages API, and there is no way to get a list of all pages in the system.

This functionality is useful for static pages on a website. A simple example, using Express to send the HTML of a reed page to a user:

app.get('/pages/:page', function(req, res) {
	reed.pages.get(, function(err, metadata, htmlContent) {
		//In a real scenario, you should use a view
		//and make use of the metadata object.
		if (err) {
			res.send('There was an error: ' + JSON.stringify(err));
		else {

The pages API is contained within the pages namespace:

  •, callback): Opens the given path for reed pages. This directory should be separate from the blog directory. Calling open() more than once will cause it to throw an error. The callback is called once the page system is running.
  • pages.get(title, callback): Attempts to find the page with the given title. The callback receives error, metadata, and htmlContent, as in the regular get method.
  • pages.remove(title, callback): Removes the specified page, deleting it from Redis and the filesystem.
  • pages.close(): Closes the pages portion of reed.


These people have contributed to the development of reed in some way or another:


MIT License. Detailed in the LICENSE file.