Snapshot of my static website generator

gen.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

import os.path
import shutil
import email # using MIME format for src files
import configparser
from datetime import datetime
from glob import glob

import dateutil.parser
import jinja2
import markdown
from docutils.core import publish_parts

OUTDIR = "out"
SRCDIR = "src"
TMPLDIR = "template"

def ensure_dir(name):
	try:
		os.makedirs(name)
	except:
		pass

TMPL_ENV = jinja2.Environment(loader=jinja2.FileSystemLoader(TMPLDIR))

ensure_dir(OUTDIR)

EXTENSION_MAP = dict()
def for_extension(ext):
	def decorate(func):
		EXTENSION_MAP[ext] = func
		return func
	return decorate

@for_extension("css")
@for_extension("png")
@for_extension("jpg")
@for_extension("js")
@for_extension("html")
def hardlink_file(infile):
	assert (infile.startswith(SRCDIR))
	outfile = os.path.join(OUTDIR,infile[len(SRCDIR)+1:])
	os.makedirs(os.path.dirname(outfile), exist_ok=True)
	try:
		os.link(infile,outfile)
	except OSError:
		try:
			shutil.copy(infile,outfile)
		except shutil.SameFileError:
			# os.link above previously succeeded
			pass

def load_infile(infile):
	with open(infile) as fh:
		raw = email.message_from_file(fh)
	assert(not raw.is_multipart())
	vars = dict()
	for k,v in raw.items():
		vars[k] = v
	vars['infile'] = infile
	vars = refurbish_vars(vars)
	contents = raw.get_payload()
	return contents, vars

def outfile_with_ext(infile,new_ext):
	base = os.path.basename(infile)
	rel_dir = os.path.dirname(infile)[len(SRCDIR)+1:]
	root,ext = os.path.splitext(base)
	outfile = os.path.join(OUTDIR, rel_dir, root) + new_ext
	dir = os.path.dirname(outfile)
	ensure_dir(dir)
	return outfile

def render(vars, outfile, tmpl_name="page.html"):
	vars["base_url"] = _CONFIG['base']['base_url']
	template = TMPL_ENV.get_template(tmpl_name)
	with open(outfile, "w") as out:
		out.write(template.render(vars))
	# Validate
	os.system("xmllint --noout " + outfile)

def cd_once(path):
	"""Remove the first directory element of path"""
	i = path.find("/")
	if i > 0:
		return path[i+1:]
	return path

assert cd_once("bla/foo/baz.txt") == "foo/baz.txt"

def relative_prefix(infile):
	"""Return relative path to root from infile"""
	assert (infile.startswith(SRCDIR+"/"))
	path = infile[len(SRCDIR)+1:]
	dir = os.path.dirname(path)
	depth = path.count("/")
	return "../" * depth

def refurbish_vars(vars):
	infile = vars['infile']
	vars["title"] = vars.get("title", "No Title?!")
	vars["author"] = vars.get("author", "unknown")
	vars["ROOT"] = relative_prefix(infile)
	if not "date" in vars:
		dt = datetime.fromtimestamp(os.path.getmtime(infile))
		vars['date_parsed'] = dt
		vars["date"] = dt.strftime("%Y-%m-%d")
	else:
		vars['date_parsed'] = dateutil.parser.parse(vars['date'])
	return vars

@for_extension("md")
@for_extension("mdwn")
@for_extension("markdown")
def markdown_handler(infile):
	contents, vars = load_infile(infile)
	vars["contents"] = markdown.markdown(contents)
	outfile = outfile_with_ext(infile,".html")
	render(vars, outfile)

def rst2html(txt):
	return publish_parts(txt, writer_name='html')['html_body']

@for_extension("htm")
def htm_handler(infile):
	contents, vars = load_infile(infile)
	vars["contents"] = contents
	outfile = outfile_with_ext(infile,".html")
	render(vars, outfile)

@for_extension("rst")
def rst_handler(infile):
	contents, vars = load_infile(infile)
	vars["contents"] = rst2html(contents)
	outfile = outfile_with_ext(infile,".html")
	render(vars, outfile)

@for_extension("collection")
def collection_handler(infile):
	contents, vars = load_infile(infile)
	items = list()
	for filename in contents.split("\n"):
		f = os.path.join(SRCDIR,filename)
		if not os.path.isfile(f):
			continue
		_, f_vars = load_infile(f)
		f_outfile = outfile_with_ext(f,".html")
		items.append(dict(
			title = f_vars['title'],
			link = cd_once(f_outfile),
			date = f_vars['date'],
			date_parsed = f_vars['date_parsed'],
			tldr = f_vars.get('tldr', ""),
			image_src = f_vars.get('image_src', ""),
		));
	items.sort(key = lambda x: x["date"], reverse=True)
	atom_outfile = outfile_with_ext(infile,".atom")
	vars["collection"] = items
	vars["atomlink"] = cd_once(atom_outfile)
	outfile = outfile_with_ext(infile,".html")
	render(vars, outfile, tmpl_name="collection.html")
	render(vars, atom_outfile, tmpl_name="collection.atom")

def all_infiles():
	for root, dirnames, filenames in os.walk(SRCDIR):
		for filename in filenames:
			yield os.path.join(root, filename)

_CONFIG = configparser.ConfigParser()
if __name__ == "__main__":
	_CONFIG.read("config.ini")
	for infile in all_infiles():
		if not os.path.isfile(infile):
			continue
		root,ext = os.path.splitext(os.path.basename(infile))
		if not ext[1:] in EXTENSION_MAP.keys():
			print("unknown extension:", infile)
			continue
		EXTENSION_MAP[ext[1:]](infile)