glog

git push blog server

example

custom http server

Here's what a custom server could look like storing repository data in ./repo:

var http = require('http');
var glog = require('glog')(__dirname + '/repo');
var ecstatic = require('ecstatic')(__dirname + '/static');
 
var server = http.createServer(function (req, res) {
    if (glog.test(req.url)) {
        glog(req, res);
    }
    else ecstatic(req, res);
});
server.listen(5000);

git push glog

First run your http server:

$ node server.js

Now create a new git repo for articles and set up the remote to point at your glog server:

$ git init
$ git remote add publish http://localhost:5000/blog.git

Write an article in markdown, create an annotated tag for the article, and push to the git blog server:

$ echo -e '# beep\nboop' > robot.markdown
$ git add *.markdown && git commit -m 'initial'
$ glog publish robot.markdown 'this is the title text'
$ git push publish master --tags

Now the content should be live on your blog, yay!

authenticating glog

Continuing from the previous example, we'll add user permissions to our glog server.

To create a user once you've set the git remote, from your blog repo do:

$ glog useradd substack
Created user substack
To publish as this user add this remote:

http://substack:42aee89a@localhost:5000/blog.git

If you don't already have a remote for the blog repo, pass --remote=REMOTE to the glog useradd command.

Once users have been configured, everyone who tries to git push new articles will need to have a user token.

Now you can list the glog users with glog users:

$ glog users
substack

For the rest of the user commands, just type glog to see the usage page.

http api

When you attach a glog handler to your server, these routes are installed:

/blog.git

Used by pushover to make git push deploys work. You can set this as a git remote and interact with it like any other git endpoint.

Annotated git tags with the filename as the tag name are used to store title text, publish date, and which files are "published".

/blog.json

Return a streaming json array of article metadata for all articles.

Optionally, you can set these query string parameters:

• inline - include the article content bodies along with the document metadata as 'html' or 'markdown'

example output:

$ curl localhost:5000/blog.json
[
{"file":"robot.markdown","author":"James Halliday","email":"mail@substack.net","date":"Mon Dec 24 15:31:27 2012 -0800","title":"robots are pretty great","commit":"81c62aa62b6770a2f6bdf6865d393daf05930b4a"}
,
{"file":"test.markdown","author":"James Halliday","email":"mail@substack.net","date":"Mon Dec 24 04:31:53 2012 -0800","title":"testing title","commit":"2a516000d239bbfcf7cdbb4b5acf09486bdf9586"}
]

 $ curl localhost:5000/blog.json?inline=html
[
{"file":"robot.markdown","author":"James Halliday","email":"mail@substack.net","date":"Mon Dec 24 15:31:27 2012 -0800","title":"robots are pretty great","commit":"81c62aa62b6770a2f6bdf6865d393daf05930b4a","body":"<h1>robots!</h1>\n\n<p>Pretty great basically.</p>"}
,
{"file":"test.markdown","author":"James Halliday","email":"mail@substack.net","date":"Mon Dec 24 04:31:53 2012 -0800","title":"testing title","commit":"2a516000d239bbfcf7cdbb4b5acf09486bdf9586","body":"<h1>title text</h1>\n\n<p>beep boop.</p>\n\n<p><em>rawr</em></p>"}
]

/blog.xml

Return an atom rss stream with inline content.

/blog/$FILE.markdown

Fetch a source document $FILE as markdown.

/blog/$FILE.html

Fetch a source document $FILE.markdown rendered as html.

methods

var glog = require('glog')

var blog = glog(opts)

Create a new blog handle using opts.repodir to store git blog data.

If opts is a string, it's taken as the opts.repodir.

You can also set opts.title and opts.id which are used as defaults by the rss feed, and opts.highlight which is the highlight-function used by marked.

All other opts are passed through directly to marked.parse(src, opts).

blog(req, res)

Handle the (req, res) in order to serve blog.json and blog.git.

blog.get(name)

Get a single article, returning a readable stream of a single blog documents object. Blog documents have:

• doc.title - title text
• doc.commit - document git commit hash
• doc.date - parseable date string
• doc.author - author name as a string
• doc.email - author email from git commit data
• doc.file - document filename in the git repo

blog.list(opts)

Return a readable stream of blog article documents.

Optionally:

• opts.limit - number of results to show
• opts.start - show results starting at this tag or title
• opts.after - show results after this tag or title

blog.read(file)

Return a readable stream with the contents of file.

blog.inline(format)

Return a through stream you can pipe blog.list() to that will inline article contents rendered in format: either 'html' or 'markdown'.

.inline() adds a doc.body string with the article contents to the document object.

blog.test(req.url)

Return whether or not to defer to blog for handling routes.

blog.rss(opts)

Return an atom rss stream with the blog content inlined in <content> tags.

opts are the required elements from the atom spec but you can probably ignore them and it will still work:

• opts.id - just use your blog address or domain name
• opts.title - blog title to use in the feed

usage

usage:

  glog publish FILE "TITLE..."

    Publish FILE with TITLE by creating an annotated tag.

  glog users

    Show the list of glog users.

  glog useradd USER

    Generate an auth token for USER to use as a git remote.

  glog userdel USER

    Delete a USER.

  glog token USER

    Show the git remote token for USER.

install

With npm, to get the glog command do:

npm install -g glog

and to get the library do:

npm install glog

license

MIT