autocat3/CherryPyApp.py

339 lines
11 KiB
Python
Raw Normal View History

2019-03-28 13:45:03 +00:00
#!/usr/bin/env python
# -*- mode: python; indent-tabs-mode: nil; -*- coding: utf-8 -*-
"""
CherryPyApp.py
Copyright 2009-2014 by Marcello Perathoner
Distributable under the GNU General Public License Version 3 or newer.
The Project Gutenberg Catalog App Server.
Config and route setup.
"""
from __future__ import unicode_literals
import logging
import logging.handlers # rotating file handler
import os
import time
import traceback
import cherrypy
from cherrypy.process import plugins
from libgutenberg import GutenbergDatabase
2019-07-26 18:22:09 +00:00
import i18n_tool
2019-03-28 13:45:03 +00:00
2019-06-19 19:03:41 +00:00
# this import causes ConnectionPool.ConnectionPool to become the cherrypy connection pool
2019-03-28 13:45:03 +00:00
import ConnectionPool
2019-06-19 19:03:41 +00:00
2019-03-28 13:45:03 +00:00
import Page
import StartPage
import SuggestionsPage
from SearchPage import BookSearchPage, AuthorSearchPage, SubjectSearchPage, BookshelfSearchPage, \
AuthorPage, SubjectPage, BookshelfPage, AlsoDownloadedPage
from BibrecPage import BibrecPage
import CoverPages
import QRCodePage
2019-07-29 18:31:45 +00:00
import diagnostics
2019-03-28 13:45:03 +00:00
import Sitemap
import Formatters
2019-08-01 20:20:18 +00:00
from errors import ErrorPage
2019-03-28 13:45:03 +00:00
import Timer
plugins.Timer = Timer.TimerPlugin
2019-04-29 17:29:56 +00:00
install_dir = os.path.dirname(os.path.abspath(__file__))
2019-03-28 13:45:03 +00:00
CHERRYPY_CONFIG = os.path.join(install_dir, 'CherryPy.conf')
LOCAL_CONFIG = [os.path.expanduser('~/.autocat3'), '/etc/autocat3.conf']
2019-03-28 13:45:03 +00:00
2019-08-01 20:20:18 +00:00
def error_page_404(status, message, traceback, version):
2019-08-01 23:40:23 +00:00
resp = ErrorPage(status, message).index()
# signal that we needn't save the session
cherrypy.session.loaded = False
return resp
2019-08-01 20:20:18 +00:00
2019-04-29 17:29:56 +00:00
class MyRoutesDispatcher(cherrypy.dispatch.RoutesDispatcher):
2019-03-28 13:45:03 +00:00
""" Dispatcher that tells us the matched route.
CherryPy makes it hard for us by forgetting the matched route object.
Here we add a 'route_name' parameter, that will tell us the route's name.
"""
2019-04-29 17:29:56 +00:00
def connect(self, name, route, controller, **kwargs):
2019-03-28 13:45:03 +00:00
""" Add a 'route_name' parameter that will tell us the matched route. """
kwargs['route_name'] = name
2019-04-29 17:29:56 +00:00
kwargs.setdefault('action', 'index')
cherrypy.dispatch.RoutesDispatcher.connect(self, name, route, controller, **kwargs)
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
def main():
2019-03-28 13:45:03 +00:00
""" Main function. """
# default config
2019-04-29 17:29:56 +00:00
cherrypy.config.update({
2019-03-28 13:45:03 +00:00
'uid': 0,
'gid': 0,
'server_name': 'localhost',
2019-04-29 17:29:56 +00:00
'genshi.template_dir': os.path.join(install_dir, 'templates'),
2019-03-28 13:45:03 +00:00
'daemonize': False,
'pidfile': None,
'host': 'localhost',
'host_mobile': 'localhost',
'file_host': 'localhost',
})
2019-04-29 17:00:26 +00:00
cherrypy.config.update(CHERRYPY_CONFIG)
2019-04-29 17:29:56 +00:00
2019-04-29 17:00:26 +00:00
extra_config = ''
for config_filename in LOCAL_CONFIG:
try:
cherrypy.config.update(config_filename)
extra_config = config_filename
break
except IOError:
pass
2019-03-28 13:45:03 +00:00
# Rotating Logs
2019-04-30 19:36:46 +00:00
# CherryPy will already open log files if present in config
2019-04-30 19:52:43 +00:00
error_file = access_file = ''
2019-04-30 19:36:46 +00:00
# read the logger file locations from config file.
if not cherrypy.log.error_file:
error_file = cherrypy.config.get('logger.error_file', '')
if not cherrypy.log.access_file:
access_file = cherrypy.config.get('logger.access_file', '')
2019-03-28 13:45:03 +00:00
# disable log file handlers
cherrypy.log.error_file = ""
cherrypy.log.access_file = ""
2019-04-29 17:00:26 +00:00
# set up python logging
2019-04-29 17:29:56 +00:00
max_bytes = getattr(cherrypy.log, "rot_max_bytes", 100 * 1024 * 1024)
backup_count = getattr(cherrypy.log, "rot_backup_count", 2)
2019-04-30 19:36:46 +00:00
if error_file:
h = logging.handlers.RotatingFileHandler(error_file, 'a', max_bytes, backup_count, 'utf-8')
h.setLevel(logging.INFO)
h.setFormatter(cherrypy._cplogging.logfmt)
cherrypy.log.error_log.addHandler(h)
if access_file:
h = logging.handlers.RotatingFileHandler(access_file, 'a', max_bytes, backup_count, 'utf-8')
h.setLevel(logging.INFO)
h.setFormatter(cherrypy._cplogging.logfmt)
cherrypy.log.access_log.addHandler(h)
2019-04-29 17:29:56 +00:00
2019-03-28 13:45:03 +00:00
if not cherrypy.config['daemonize']:
2019-04-29 17:29:56 +00:00
ch = logging.StreamHandler()
ch.setLevel(logging.DEBUG)
ch.setFormatter(cherrypy._cplogging.logfmt)
cherrypy.log.error_log.addHandler(ch)
2019-03-28 13:45:03 +00:00
# continue app init
#
2019-04-29 17:29:56 +00:00
cherrypy.log('*' * 80, context='ENGINE', severity=logging.INFO)
cherrypy.log("Using config file '%s'." % CHERRYPY_CONFIG,
context='ENGINE', severity=logging.INFO)
2019-04-29 17:00:26 +00:00
if extra_config:
cherrypy.log('extra_config: %s' % extra_config, context='ENGINE', severity=logging.INFO)
2019-03-28 13:45:03 +00:00
# after cherrypy.config is parsed
2019-04-29 17:29:56 +00:00
Formatters.init()
cherrypy.log("Continuing App Init", context='ENGINE', severity=logging.INFO)
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
cherrypy.log("Continuing App Init", context='ENGINE', severity=logging.INFO)
cherrypy.tools.I18nTool = i18n_tool.I18nTool()
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
cherrypy.log("Continuing App Init", context='ENGINE', severity=logging.INFO)
2019-03-28 13:45:03 +00:00
# Used to bust the cache on js and css files. This should be the
# files' mtime, but the files are not stored on the app server.
# This is a `good enough´ replacement though.
2019-04-29 17:29:56 +00:00
t = str(int(time.time()))
2019-03-28 13:45:03 +00:00
cherrypy.config['css_mtime'] = t
cherrypy.config['js_mtime'] = t
cherrypy.config['all_hosts'] = (
cherrypy.config['host'], cherrypy.config['host_mobile'], cherrypy.config['file_host'])
2019-08-01 20:20:18 +00:00
cherrypy.config.update({'error_page.404': error_page_404})
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
if hasattr(cherrypy.engine, 'signal_handler'):
cherrypy.engine.signal_handler.subscribe()
2019-03-28 13:45:03 +00:00
GutenbergDatabase.options.update(cherrypy.config)
2019-04-29 17:29:56 +00:00
cherrypy.engine.pool = plugins.ConnectionPool(
cherrypy.engine, params=GutenbergDatabase.get_connection_params(cherrypy.config))
cherrypy.engine.pool.subscribe()
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
plugins.Timer(cherrypy.engine).subscribe()
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
cherrypy.log("Daemonizing", context='ENGINE', severity=logging.INFO)
2019-03-28 13:45:03 +00:00
if cherrypy.config['daemonize']:
2019-04-29 17:29:56 +00:00
plugins.Daemonizer(cherrypy.engine).subscribe()
2019-03-28 13:45:03 +00:00
uid = cherrypy.config['uid']
gid = cherrypy.config['gid']
if uid > 0 or gid > 0:
2019-04-29 17:29:56 +00:00
plugins.DropPrivileges(cherrypy.engine, uid=uid, gid=gid, umask=0o22).subscribe()
2019-03-28 13:45:03 +00:00
if cherrypy.config['pidfile']:
2019-04-29 17:29:56 +00:00
pid = plugins.PIDFile(cherrypy.engine, cherrypy.config['pidfile'])
# Write pidfile after privileges are dropped(prio == 77)
2019-03-28 13:45:03 +00:00
# or we will not be able to remove it.
2019-04-29 17:29:56 +00:00
cherrypy.engine.subscribe('start', pid.start, 78)
cherrypy.engine.subscribe('exit', pid.exit, 78)
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
cherrypy.log("Setting up routes", context='ENGINE', severity=logging.INFO)
2019-03-28 13:45:03 +00:00
# setup 'routes' dispatcher
#
2019-04-29 17:29:56 +00:00
# d = cherrypy.dispatch.RoutesDispatcher(full_result=True)
d = MyRoutesDispatcher(full_result=True)
2019-03-28 13:45:03 +00:00
cherrypy.routes_mapper = d.mapper
2019-04-29 17:29:56 +00:00
def check_id(environ, result):
2019-03-28 13:45:03 +00:00
""" Check if id is a valid number. """
try:
2019-04-29 17:29:56 +00:00
return str(int(result['id'])) == result['id']
2019-03-28 13:45:03 +00:00
except:
return False
2019-04-29 17:29:56 +00:00
d.connect('start', r'/ebooks{.format}/',
controller=StartPage.Start())
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('suggest', r'/ebooks/suggest{.format}/',
controller=SuggestionsPage.Suggestions())
2019-03-28 13:45:03 +00:00
# search pages
2019-04-29 17:29:56 +00:00
d.connect('search', r'/ebooks/search{.format}/',
controller=BookSearchPage())
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('author_search', r'/ebooks/authors/search{.format}/',
controller=AuthorSearchPage())
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('subject_search', r'/ebooks/subjects/search{.format}/',
controller=SubjectSearchPage())
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('bookshelf_search', r'/ebooks/bookshelves/search{.format}/',
controller=BookshelfSearchPage())
2019-03-28 13:45:03 +00:00
# 'id' pages
2019-04-29 17:29:56 +00:00
d.connect('author', r'/ebooks/author/{id:\d+}{.format}',
controller=AuthorPage(), conditions=dict(function=check_id))
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('subject', r'/ebooks/subject/{id:\d+}{.format}',
controller=SubjectPage(), conditions=dict(function=check_id))
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('bookshelf', r'/ebooks/bookshelf/{id:\d+}{.format}',
controller=BookshelfPage(), conditions=dict(function=check_id))
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('also', r'/ebooks/{id:\d+}/also/{.format}',
controller=AlsoDownloadedPage(), conditions=dict(function=check_id))
2019-03-28 13:45:03 +00:00
# bibrec pages
2019-04-29 17:29:56 +00:00
d.connect('download', r'/ebooks/{id:\d+}/download{.format}',
controller=Page.NullPage(), _static=True)
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('bibrec', r'/ebooks/{id:\d+}{.format}',
controller=BibrecPage(), conditions=dict(function=check_id))
2019-03-28 13:45:03 +00:00
# legacy compatibility with /ebooks/123.bibrec
2019-04-29 17:29:56 +00:00
d.connect('bibrec2', r'/ebooks/{id:\d+}.bibrec{.format}',
controller=BibrecPage(), conditions=dict(function=check_id))
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('cover', r'/covers/{size:small|medium}/{order:latest|popular}/{count}',
controller=CoverPages.CoverPages())
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('qrcode', r'/qrcode/',
controller=QRCodePage.QRCodePage())
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('iplimit', r'/iplimit/',
controller=Page.NullPage())
2019-03-28 13:45:03 +00:00
2019-07-29 18:31:45 +00:00
d.connect('diagnostics', r'/diagnostics/',
controller=diagnostics.DiagnosticsPage())
2019-04-29 17:29:56 +00:00
d.connect('stats', r'/stats/',
controller=Page.NullPage(), _static=True)
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('honeypot_send', r'/ebooks/send/megaupload/{id:\d+}.{filetype}',
controller=Page.NullPage(), _static=True)
2019-03-28 13:45:03 +00:00
# /w/captcha/question/ so varnish will cache it
2019-04-29 17:29:56 +00:00
d.connect('captcha.question', r'/w/captcha/question/',
controller=Page.GoHomePage())
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('captcha.answer', r'/w/captcha/answer/',
controller=Page.GoHomePage())
2019-03-28 13:45:03 +00:00
# sitemap protocol access control requires us to place sitemaps in /ebooks/
2019-04-29 17:29:56 +00:00
d.connect('sitemap', r'/ebooks/sitemaps/',
controller=Sitemap.SitemapIndex())
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
d.connect('sitemap_index', r'/ebooks/sitemaps/{page:\d+}',
controller=Sitemap.Sitemap())
2019-03-28 13:45:03 +00:00
if 'dropbox_client_id' in cherrypy.config:
import Dropbox
2019-04-29 17:29:56 +00:00
dropbox = Dropbox.Dropbox()
cherrypy.log("Dropbox Client Id: %s" % cherrypy.config['dropbox_client_id'],
context='ENGINE', severity=logging.INFO)
d.connect('dropbox_send', r'/ebooks/send/dropbox/{id:\d+}.{filetype}',
controller=dropbox, conditions=dict(function=check_id))
d.connect('dropbox_callback', r'/ebooks/send/dropbox/',
controller=dropbox)
2019-03-28 13:45:03 +00:00
if 'gdrive_client_id' in cherrypy.config:
import GDrive
2019-04-29 17:29:56 +00:00
gdrive = GDrive.GDrive()
cherrypy.log("GDrive Client Id: %s" % cherrypy.config['gdrive_client_id'],
context='ENGINE', severity=logging.INFO)
d.connect('gdrive_send', r'/ebooks/send/gdrive/{id:\d+}.{filetype}',
controller=gdrive, conditions=dict(function=check_id))
d.connect('gdrive_callback', r'/ebooks/send/gdrive/',
controller=gdrive)
2019-03-28 13:45:03 +00:00
if 'msdrive_client_id' in cherrypy.config:
import MSDrive
2019-04-29 17:29:56 +00:00
msdrive = MSDrive.MSDrive()
cherrypy.log("MSDrive Client Id: %s" % cherrypy.config['msdrive_client_id'],
context='ENGINE', severity=logging.INFO)
d.connect('msdrive_send', r'/ebooks/send/msdrive/{id:\d+}.{filetype}',
controller=msdrive, conditions=dict(function=check_id))
d.connect('msdrive_callback', r'/ebooks/send/msdrive/',
controller=msdrive)
2019-03-28 13:45:03 +00:00
# start http server
#
2019-04-29 17:29:56 +00:00
cherrypy.log("Mounting root", context='ENGINE', severity=logging.INFO)
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
app = cherrypy.tree.mount(root=None, config=CHERRYPY_CONFIG)
2019-03-28 13:45:03 +00:00
2019-04-29 17:29:56 +00:00
app.merge({'/': {'request.dispatch': d}})
2019-03-28 13:45:03 +00:00
return app
if __name__ == '__main__':
2019-04-29 17:29:56 +00:00
main()
cherrypy.engine.start()
cherrypy.engine.block()