def create_live_rewriter_app(config={}):
routes = [
Route('rewrite', RewriteHandler(config)),
Route('static/__pywb', StaticHandler('pywb/static/'))
]
return ArchivalRouter(routes, hostpaths=['http://localhost:8080'])
def add_cdx_api_handler(name, cdx_api_suffix, routes, query_handler):
# if bool, use -cdx suffix, else use custom string
# as the suffix
if isinstance(cdx_api_suffix, bool):
name += '-cdx'
else:
name += str(cdx_api_suffix)
routes.append(Route(name, CDXAPIHandler(query_handler)))
def create_perms_checker_app(config):
"""
Create permissions checker standalone app
Running under the '/check-access' route
"""
port = config.get('port')
perms_policy = config.get('perms_policy')
canonicalizer = UrlCanonicalizer(config.get('surt_ordered', True))
handler = PermsHandler(perms_policy, canonicalizer)
routes = [Route('check-access', handler)]
return ArchivalRouter(routes, port=port)
def _test_redir(match_host,
request_uri,
referrer,
script_name='',
coll='coll'):
env = {
'REL_REQUEST_URI': request_uri,
'HTTP_REFERER': referrer,
'SCRIPT_NAME': script_name
}
env['HTTP_HOST'] = urlsplit(match_host).netloc
routes = [Route(coll, WbUrlHandler())]
the_router = ArchivalRouter(routes)
redir = ReferRedirect()
#req = WbRequest.from_uri(request_uri, env)
rep = redir(env, the_router)
if not rep:
return False
return rep.status_headers.get_header('Location')
def create_wb_router(passed_config={}):
config = DictChain(passed_config, DEFAULTS)
routes = []
# TODO: examine this more
hostname = os.environ.get('PYWB_HOST_NAME')
if hostname:
hostpaths = [hostname]
else:
hostpaths = config.get('hostpaths')
port = config.get('port')
# collections based on cdx source
collections = config.get('collections')
if config.get('enable_memento', False):
request_class = MementoRequest
else:
request_class = WbRequest
# store live and replay handlers
handler_dict = {}
# setup template globals
template_globals = config.get('template_globals')
if template_globals:
add_env_globals(template_globals)
for name, value in collections.iteritems():
if isinstance(value, BaseHandler):
handler_dict[name] = value
routes.append(Route(name, value, config=route_config))
continue
route_config = init_route_config(value, config)
if route_config.get('index_paths') == '$liveweb':
live = create_live_handler(route_config)
handler_dict[name] = live
routes.append(Route(name, live, config=route_config))
continue
query_handler = init_collection(route_config)
wb_handler = create_wb_handler(
query_handler=query_handler,
config=route_config,
)
handler_dict[name] = wb_handler
logging.debug('Adding Collection: ' + name)
route_class = route_config.get('route_class', Route)
routes.append(
route_class(name,
wb_handler,
config=route_config,
request_class=request_class))
# cdx query handler
cdx_api_suffix = route_config.get('enable_cdx_api', False)
if cdx_api_suffix:
add_cdx_api_handler(name, cdx_api_suffix, routes, query_handler)
if config.get('debug_echo_env', False):
routes.append(Route('echo_env', DebugEchoEnvHandler()))
if config.get('debug_echo_req', False):
routes.append(Route('echo_req', DebugEchoHandler()))
static_routes = config.get('static_routes')
for static_name, static_path in static_routes.iteritems():
routes.append(Route(static_name, StaticHandler(static_path)))
# resolve any cross handler references
for route in routes:
if hasattr(route.handler, 'resolve_refs'):
route.handler.resolve_refs(handler_dict)
# default to regular archival mode
router = ArchivalRouter
if config.get('enable_http_proxy', False):
router = ProxyArchivalRouter
view = J2TemplateView.create_template(config.get('proxy_select_html'),
'Proxy Coll Selector')
if not 'proxy_options' in passed_config:
passed_config['proxy_options'] = {}
if view:
passed_config['proxy_options']['proxy_select_view'] = view
view = J2TemplateView.create_template(
config.get('proxy_cert_download_html'), 'Proxy Cert Download')
if view:
passed_config['proxy_options']['proxy_cert_download_view'] = view
# Finally, create wb router
return router(
routes,
# Specify hostnames that pywb will be running on
# This will help catch occasionally missed rewrites that
# fall-through to the host
# (See archivalrouter.ReferRedirect)
hostpaths=hostpaths,
port=port,
abs_path=config.get('absolute_paths', True),
home_view=J2TemplateView.create_template(config.get('home_html'),
'Home Page'),
error_view=J2TemplateView.create_template(config.get('error_html'),
'Error Page'),
config=config)
def create_wb_router(passed_config=None):
passed_config = passed_config or {}
defaults = load_yaml_config(DEFAULT_CONFIG)
config = DictChain(passed_config, defaults)
regular_router_class = config.get("regular_router_class", ArchivalRouter)
proxy_router_class = config.get("proxy_router_class", ProxyArchivalRouter)
default_route_class = config.get("route_class", Route)
routes = []
port = config.get('port')
collections = config.get('collections', {})
static_routes = config.get('static_routes', {})
root_route = None
# collections based on file system
if config.get('enable_auto_colls', True):
colls_loader_cls = config.get('colls_loader_cls', DirectoryCollsLoader)
dir_loader = colls_loader_cls(config, static_routes, collections)
dir_loader()
#collections.update(dir_loader())
if config.get('enable_memento', False):
request_class = MementoRequest
else:
request_class = config.get("wb_request_class", WbRequest)
# store live and replay handlers
handler_dict = {}
# setup template globals
templates_dirs = config['templates_dirs']
jinja_env = J2TemplateView.init_shared_env(
paths=templates_dirs, packages=config['template_packages'])
jinja_env.globals.update(config.get('template_globals', {}))
for static_name, static_path in six.iteritems(static_routes):
routes.append(
default_route_class(static_name, StaticHandler(static_path)))
for name, value in six.iteritems(collections):
if isinstance(value, BaseHandler):
handler_dict[name] = value
new_route = Route(name, value, config=config)
if name != '':
routes.append(new_route)
else:
root_route = new_route
continue
route_config = init_route_config(value, config)
route_class = route_config.get('route_class', default_route_class)
if route_config.get('index_paths') == '$liveweb':
live = create_live_handler(route_config)
handler_dict[name] = live
new_route = route_class(name, live, config=route_config)
if name != '':
routes.append(new_route)
else:
root_route = new_route
continue
query_handler = init_collection(route_config)
wb_handler = create_wb_handler(
query_handler=query_handler,
config=route_config,
)
handler_dict[name] = wb_handler
logging.debug('Adding Collection: ' + name)
new_route = route_class(name,
wb_handler,
config=route_config,
request_class=request_class)
if name != '':
routes.append(new_route)
else:
root_route = new_route
# cdx query handler
cdx_api_suffix = route_config.get('enable_cdx_api', False)
if cdx_api_suffix:
add_cdx_api_handler(name,
cdx_api_suffix,
routes,
query_handler,
route_class=route_class)
if config.get('debug_echo_env', False):
routes.append(Route('echo_env', DebugEchoEnvHandler()))
if config.get('debug_echo_req', False):
routes.append(Route('echo_req', DebugEchoHandler()))
if root_route:
routes.append(root_route)
# resolve any cross handler references
for route in routes:
if hasattr(route.handler, 'resolve_refs'):
route.handler.resolve_refs(handler_dict)
# default to regular archival mode
router_constructor = regular_router_class
if config.get('enable_http_proxy', False):
router_constructor = proxy_router_class
view = init_view(config, 'proxy_select_html')
if 'proxy_options' not in passed_config:
passed_config['proxy_options'] = {}
if view:
passed_config['proxy_options']['proxy_select_view'] = view
view = init_view(config, 'proxy_cert_download_html')
if view:
passed_config['proxy_options']['proxy_cert_download_view'] = view
print "ROUTER IS %s!!!" % regular_router_class
# Finally, create wb router
return router_constructor(routes,
port=port,
abs_path=config.get('absolute_paths', True),
home_view=init_view(config, 'home_html'),
error_view=init_view(config, 'error_html'),
info_view=init_view(config, 'info_json'),
config=config)
def create_wb_router(passed_config={}):
defaults = load_yaml_config(DEFAULT_CONFIG)
config = DictChain(passed_config, defaults)
routes = []
port = config.get('port')
collections = config.get('collections', {})
static_routes = config.get('static_routes', {})
# collections based on file system
dir_loader = DirectoryCollsLoader(config, static_routes)
collections.update(dir_loader())
if config.get('enable_memento', False):
request_class = MementoRequest
else:
request_class = WbRequest
# store live and replay handlers
handler_dict = {}
# setup template globals
template_globals = config.get('template_globals')
if template_globals:
add_env_globals(template_globals)
for name, value in collections.iteritems():
if isinstance(value, BaseHandler):
handler_dict[name] = value
routes.append(Route(name, value, config=route_config))
continue
route_config = init_route_config(value, config)
if route_config.get('index_paths') == '$liveweb':
live = create_live_handler(route_config)
handler_dict[name] = live
routes.append(Route(name, live, config=route_config))
continue
query_handler = init_collection(route_config)
wb_handler = create_wb_handler(
query_handler=query_handler,
config=route_config,
)
handler_dict[name] = wb_handler
logging.debug('Adding Collection: ' + name)
route_class = route_config.get('route_class', Route)
routes.append(
route_class(name,
wb_handler,
config=route_config,
request_class=request_class))
# cdx query handler
cdx_api_suffix = route_config.get('enable_cdx_api', False)
if cdx_api_suffix:
add_cdx_api_handler(name, cdx_api_suffix, routes, query_handler)
if config.get('debug_echo_env', False):
routes.append(Route('echo_env', DebugEchoEnvHandler()))
if config.get('debug_echo_req', False):
routes.append(Route('echo_req', DebugEchoHandler()))
for static_name, static_path in static_routes.iteritems():
routes.append(Route(static_name, StaticHandler(static_path)))
# resolve any cross handler references
for route in routes:
if hasattr(route.handler, 'resolve_refs'):
route.handler.resolve_refs(handler_dict)
# default to regular archival mode
router = ArchivalRouter
if config.get('enable_http_proxy', False):
router = ProxyArchivalRouter
view = J2TemplateView.create_template(config.get('proxy_select_html'),
'Proxy Coll Selector')
if 'proxy_options' not in passed_config:
passed_config['proxy_options'] = {}
if view:
passed_config['proxy_options']['proxy_select_view'] = view
view = J2TemplateView.create_template(
config.get('proxy_cert_download_html'), 'Proxy Cert Download')
if view:
passed_config['proxy_options']['proxy_cert_download_view'] = view
# Finally, create wb router
return router(
routes,
port=port,
abs_path=config.get('absolute_paths', True),
home_view=J2TemplateView.create_template(config.get('home_html'),
'Home Page'),
error_view=J2TemplateView.create_template(config.get('error_html'),
'Error Page'),
config=config)