Academic Website with ≤ 70 Lines of Python
2014-06-09
To get a bit more accustomed with web development in Python, I started trying out Flask1, a ‘micro-framework’ for web development.
My previous academic homepage had a publication list backed by a CMS, so I thought porting everything to a stand-alone website would be a little cumbersome – but it wasn’t.
My requirements for the publication list were simple:
- Should be defined in a single BibTex file
- Sort and group references by year
- Link to preprints and final versions
- Display abstract
- Display a BibTex entry
Apart from that, I also wanted to have some simple way to blog, so I followed this nice tutorial on how to set up a mark-down based blog with Flask.2
Then, I just grabbed Python’s bibtexparser package and was good to go:
#!python
from flask import Flask
from flask import render_template
from bibtexparser.bparser import BibTexParser
from bibtexparser.customization import convert_to_unicode
from collections import defaultdict
from flatpages import FlatPages, pygments_style_defs
import os
import sys
reload(sys)
sys.setdefaultencoding('utf-8')
#Configuration for blog:
FLATPAGES_AUTO_RELOAD = True
FLATPAGES_EXTENSION = '.md'
FLATPAGES_ROOT = 'static'
POST_DIR = 'posts'
app = Flask(__name__)
flatpages = FlatPages(app)
app.config.from_object(__name__)
@app.route('/')
def about():
return render_template('about.html', nav='about')
@app.route('/blog')
def blog():
posts = [p for p in flatpages if p.path.startswith(POST_DIR)]
posts.sort(key=lambda item:item['date'], reverse=True)
return render_template('blog.html', nav='blog', posts=posts)
@app.route('/blog/<name>/')
def post(name):
path = '{}/{}'.format(POST_DIR, name)
post = flatpages.get_or_404(path)
return render_template('blogpost.html', nav='blog', post=post)
@app.route('/publications')
def publications():
#Load file
with open('static/refs.bib', 'r') as bibfile:
bp = BibTexParser(bibfile)
references = bp.get_entry_list()
# Order references by year
references.sort(key=lambda x: x['year'], reverse=True)
refs = defaultdict(list)
#Preprocess the references
for r in references:
if 'labels' in r:
r['keywordlist'] = r['labels'].split(',')
if 'booktitle' in r:
r['booktitle'] = r['booktitle'].replace('\&', '&')
r['title'] = r['title'].replace('\&', '&')
refs[r['year']].append(r)
#Sort years
refsbyyear = []
for year in refs.keys():
refsbyyear.append((year,refs[year]))
refsbyyear.sort(key=lambda x: x[0], reverse=True)
return render_template('publications.html', nav='publications', references = refsbyyear)
if __name__ == '__main__':
app.run(debug=(os.environ.get('FLASK_DEBUG_MODE')=='TRUE'))Apart from dealing with character escaping in BibTex (see line 55; this works for me but should be generalized) and the sorting of references, everything worked out of the box.
I could imagine that this setup scales nicely to large BibTex files and even whole research groups, as you can generate static websites from a Flask app (didn’t try that yet). It seems particularly useful if everyone in the group is already working with BibTex (and who doesn’t? ;-).
Since BibTex is happy with custom fields, it is also quite easy to integrate additional information that is only relevant for the website. For example, I am using a preprint field for paths to preprint files. The bibtexparser picks those fields up automatically.
A Jinja template for the publications list could look like this:
{% extends "layout.html" %}
{% block content %}
{% for year, yearrefs in references %}
<h3>{{year}}</h3>
<hr/>
<ul>
{% for ref in yearrefs %}
<li class="publicationItem">
{% autoescape off %} {{ref.author}} :
<br/>
<em>{{ref.title}}</em>.
<br/>
{% if ref.booktitle%} {{ref.booktitle}} {% endif %}
{% if ref.journal%} {{ref.journal}} {{ref.volume}} {% if ref.number%} ({{ref.number}}) {% endif %} {% endif %}
{% endautoescape %}
{% if ref.keywordlist%}
{% for kw in ref.keywordlist %}
<span class="label">{{kw}}</span>
{% endfor %}
{% endif %}
{% if ref.note %}
<span class="label label-primary">{{ref.note}}</span>
{% endif %}
<br/>
{% if ref.preprint %}
<a href="{{ url_for('static', filename=ref.preprint)}}" target="_blank">[preprint]</a>
{% endif %}
{% if ref.preprintonline %}
<a href="{{ ref.preprintonline }}" target="_blank">[preprint]</a>
{% endif %}
{% if ref.slides %}
<a href="{{ url_for('static', filename=ref.slides)}}" target="_blank">[talk slides]</a>
{% endif %}
{% if ref.online %}
<a href="{{ref.online}}" target="_blank">[view online]</a>
{% endif %}
{% if ref.abstractonline %}
<a href="{{ref.abstractonline}}" target="_blank">[view full abstract online]</a>
{% endif %}
{% if ref.abstract %}
<a href="javascript:void(0)" onClick="$('\#abstractModal{{ref.id}}').modal({show: true});">[abstract]</a>
<!-- The abstract modal just contains the abstract text. -->
{% endif %}
<a href="javascript:void(0)" onClick="$('\#bibtexModal{{ref.id}}').modal({show: true});">[bibtex]</a>
<!-- The BibTex modal constructs a valid entry from all 'important' fields above. -->
</li>
{% endfor %}
</ul>
{% endfor %}
{% endblock %}So, Flask is indeed as ‘micro’ as possible, at least when it comes to simple websites… :-)
-
Of course I gave Django a spin as well, but I don’t really need most features and thus decided to keep it simple. ↩
-
I also followed this nice tutorial (German) on how to configure Flask for my current ISP (Strato). ↩
