Super simple Heroku-ready, Markdown- and Dropbox-based customizable blog engine. Like this, but not as-a-Service. Based on CherryPy. Runs my website.
Pretty opinionated:
- only Jinja2 for templates
- only Redis for caching
- only Atom for the feed
- only Dropbox for storage
- only Markdown with SmartyPants, Pygments, extras (tables, footnotes, etc.) for rendering
But you can replace these parts with subclassing or monkey-patching. Or just forking the whole thing if you want to use Simplenote, Mako, Memcached, RSS and Textile instead of my configuration.
First, go to dropbox.com/developers/apps, create an app (Type: Core API, Access: App folder, any name and description), get the API key and secret.
Then, add some files to a fresh git repo:
Procfile:
web: python run.pyrequirements.txt:
git+git://github.com/myfreeweb/markbox.git@masterrun.py:
import os
from markbox import Markbox
m = Markbox(
blog_title="Some Kind of Blog",
author="Greg"
)
m.run(port=int(os.environ.get("PORT")))Now, deploy to Heroku with the app name you want, your Dropbox credentials and an uncache key -- any string you want, which will be used like a password, but only for cleaning the cache:
heroku create NAME
heroku config:set DROPBOX_APP_KEY=key DROPBOX_APP_SECRET=secret UNCACHE_KEY=uncache_key
heroku addons:add redistogo
git push heroku masterVisit your app and sign in with Dropbox.
Now you can add posts to your Dropbox/Apps/{the name you set up in the API configuration} with .md as the extension and MultiMarkdown style metadata for the title and the date, like this:
Title: Some Title
Date: 15 Nov 2012
This is the post.
# This is an h2, because the title is an h1.
Some text.After adding a post, you need to clean the cache of the index page and the feed. To clean the cache of a page, add ?uncache_key=KEY to the URL, where KEY is the key you set when deploying to Heroku.
So, if you saved your post to Dropbox/Apps/(Blog folder name)/my-first-post.md, checked that it's correct (and cached it) by visiting blog-url.tld/my-first-post, you can make it visible on the index page and the feed by visiting blog-url.tld/?uncache_key=KEY and blog-url.tld/articles.xml?uncache_key=KEY.
To completely clear the cache (eg. when you update the design or there's an error), visit blog-url.tld/clearcache?uncache_key=KEY
This is an example of a highly-customized run.py (by the way, you can change the name "run.py" to anything you want, just make sure it's correctly referenced in the Procfile):
import os
from markbox import Markbox
here = lambda a: os.path.join(os.path.dirname(os.path.abspath(__file__)), a)
m = Markbox(
blog_title="{ float: both }",
author="Greg V",
public_folder=here("static"),
tpl_folder=here("html"),
feed_name="posts",
bare_files=["style.css"]
)
def cssify(s):
return "#" + s.lower().replace(" ", "-").replace("'", "").replace("&", "and")
m.tpl.filters["cssify"] = cssify
m.run(port=int(os.environ.get("PORT")), production=True)This example:
- sets the public folder to "static" (NOTE: if you want to have any static files, you must specify the public_folder)
- changes the custom template folder to "html" instead of "templates"
- changes the feed URL from "/articles.xml" to "/posts.xml"
- adds style.css to files served from the root (i.e. "/style.css" will be the same as "/public/style.css") – the default list includes humans.txt, robots.txt and favicon.ico
- adds a filter for use in templates
This is how html/layout.html might look then:
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta content="width=device-width" name="viewport">
<title>{% if page_title %}{{page_title|cssify}}{% else %}*{% endif %} {{blog_title}}</title>
<link href="/{{feed_name}}.xml" rel="alternate" type="application/atom+xml">
<link href="{{public_url('style.css')}}" rel="stylesheet">
</head>
<body role="document">{% block main %}{% endblock %}</body>
</html>This shows the use of the cssify filter that we added in run.py, which is used on my blog for making post titles look like CSS ids and the use of the public_url function, which outputs the URL for a static file with a checksum in the query string for cachebusting.