ghc-libs-doc.py

#!/usr/bin/env python
# vim: ft=python

import bs4
import cgitb
import os
import sys

cgitb.enable()


def content_head():
    print "Content-Type: text/html"
    print


def path_to_load():
    return os.environ['PATH_INFO']


def target_file():
    return os.path.join('/home/hu/.cabal/share/doc/x86_64-linux-ghc-7.10.3/',
                        *path_to_load().split('/'))


def target_content():
    with open(target_file()) as fd:
        return bs4.BeautifulSoup(fd.read().decode('utf-8', 'ignore'), "html.parser")


def process(soup, pipeline):
    for link in soup.find_all('a'):
        if link.has_attr('href'):
            for pipe in pipeline:
                link['href'] = pipe(link['href'])
    return soup


pipeline = [
    lambda s: s.replace("file:///home/hu/.cabal/share/doc/x86_64-linux-ghc-7.10.3/",
                        u""),
    lambda s: s.replace("file:///home/hu/Tools/haskell-platform/ghc-7.10.3-x86_64/share/doc/ghc/html/",
                        u"/ghc-doc/")
]


if __name__ == '__main__':
    if 'PATH_INFO' not in os.environ:
        print 'Location: ' + os.path.basename(__file__) + '/index.html'
        print
        sys.exit(1)
    tarfile = target_file()
    ext = os.path.splitext(tarfile)[1][1:]
    if ext in ('html', 'htm'):
        content_head()
        print process(target_content(), pipeline).prettify('utf-8')
    else:
        if ext == 'js':
            print 'Content-Type: application/javascript'
            print
        elif ext == 'css':
            print 'Content-Type: text/css'
            print
        else:
            print 'Content-Type: ' + os.environ["HTTP_ACCEPT"].split(',')[0]
            print
        with open(tarfile) as fd:
            print fd.read()