Generates a docset from cmake --help-html output

cmakedocset.py

#!/usr/bin/env python

import sys, os
from lxml import etree
from contextlib import closing
import subprocess
import sqlite3
import shutil

DIR  = 'CMake.docset'
BASE = 'CMake.html'
CONTENTS  = os.path.join(DIR, 'Contents')
RESOURCES = os.path.join(CONTENTS, 'Resources')
DOCUMENTS = os.path.join(RESOURCES, 'Documents')

NAME_MAPPING = {
    'opt:': 'Option',
    'command:': 'Command',
    'prop_': 'Property',
    'variable:': 'Variable',
    'section_': 'Section',
    'gen:': 'Entry',
    'module:': 'Module',
}

def type_by_name(link_name):
    for prefix, typename in NAME_MAPPING.items():
        if link_name.startswith(prefix):
            return typename

def link_text(link):
    n = link.find('../b/code')
    if link.attrib['name'].startswith('section_'):
        return link.tail
    if not n is None:
        return n.text

def main():

    if os.path.exists(DIR):
        shutil.rmtree(DIR)

    if not os.path.exists(DOCUMENTS):
        os.makedirs(DOCUMENTS)

    info = os.path.join(CONTENTS, 'Info.plist')

    version = subprocess.check_output(['cmake', '--version']).split()[-1]

    with open(info, 'w') as f:
        f.write('''<?xml version="1.0" encoding="utf-8"?>
<plist version="1.0">
  <dict>
    <key>CFBundleIdentifier</key>
    <string>cmake</string>
    <key>CFBundleName</key>
    <string>CMake {0}</string>
    <key>DocSetPlatformFamily</key>
    <string>cmake</string>
  </dict>
</plist>'''.format(version.strip()))

    idx = os.path.join(RESOURCES, 'docSet.dsidx')

    with closing(sqlite3.connect(idx)) as con:

        c = con.cursor()
        c.execute('CREATE TABLE searchIndex(id INTEGER PRIMARY KEY, name TEXT, type TEXT, path TEXT);')
        c.execute('CREATE UNIQUE INDEX anchor ON searchIndex (name, type, path);')

        text = subprocess.check_output(['cmake', '--help-html'])
        html = etree.HTML(text)

        for link in html.iterfind('.//a'):
            attr = link.attrib
            if 'name' in attr:
                link_name = attr['name']
                type_ = type_by_name(link_name)
                if not type_:
                    continue
                name   = link_text(link)
                anchor = '{0}#{1}'.format(BASE, link_name)
                c.execute('INSERT INTO searchIndex (name, type, path) VALUES (?, ?, ?)',
                          (name, type_, anchor))

        with open(os.path.join(DOCUMENTS, BASE), 'w') as out:
            out.write(text)

        con.commit()

if __name__ == "__main__":
    main()