def create_perma_wb_router(config={}):
"""
Configure server.
This should do basically the same stuff as pywb.webapp.pywb_init.create_wb_router()
"""
# start with the default PyWB router
router = create_wb_router(config)
# insert a custom route that knows how to play back based on GUID
wb_handler = create_wb_handler(QueryHandler.init_from_config(PermaCDXSource()),
dict(archive_paths=[get_archive_path()],
wb_handler_class=PermaGUIDHandler,
buffer_response=True,
# head_insert_html=os.path.join(os.path.dirname(__file__), 'head_insert.html'),
enable_memento=True,
redir_to_exact=False))
wb_handler.replay.content_loader.record_loader.loader = CachedLoader()
route = PermaRoute(GUID_REGEX, wb_handler)
router.routes.insert(0, route)
# use our Django error view
router.error_view = PermaTemplateView('archive/archive-error.html')
wb_handler.not_found_view = router.error_view
return router
def create_perma_wb_router(config={}):
"""
Configure server.
This should do basically the same stuff as pywb.webapp.pywb_init.create_wb_router()
"""
# start with the default PyWB router
router = create_wb_router(config)
# insert a custom route that knows how to play back based on GUID
wb_handler = create_wb_handler(
QueryHandler.init_from_config(PermaCDXSource()),
dict(
archive_paths=[get_archive_path()],
wb_handler_class=PermaGUIDHandler,
buffer_response=True,
# head_insert_html=os.path.join(os.path.dirname(__file__), 'head_insert.html'),
enable_memento=True,
redir_to_exact=False))
wb_handler.replay.content_loader.record_loader.loader = CachedLoader()
route = PermaRoute(GUID_REGEX, wb_handler)
router.routes.insert(0, route)
# use our Django error view
router.error_view = PermaTemplateView('archive/archive-error.html')
wb_handler.not_found_view = router.error_view
return router
def init_collection(route_config):
ds_rules_file = route_config.get('domain_specific_rules', None)
html_view = init_view(route_config, 'query_html', J2HtmlCapturesView)
server_cls = route_config.get('server_cls')
query_handler = QueryHandler.init_from_config(route_config, ds_rules_file,
html_view, server_cls)
return query_handler
def init_collection(route_config):
ds_rules_file = route_config.get('domain_specific_rules', None)
html_view = init_view(route_config, 'query_html', J2HtmlCapturesView)
server_cls = route_config.get('server_cls')
query_handler = QueryHandler.init_from_config(route_config,
ds_rules_file,
html_view,
server_cls)
return query_handler
def test_excluded(testconfig):
# sources = testconfig.get('index_paths')
# perms_policy = testconfig.get('perms_policy')
# cdx_server = CDXServer(sources)
# index_handler = IndexHandler(cdx_server, perms_policy=perms_policy)
query_handler = QueryHandler.init_from_config(testconfig)
url = "http://www.iana.org/_img/bookmark_icon.ico"
params = dict(url=url)
with raises(AccessException):
cdxobjs = list(query_handler.load_cdx(None, params))
print cdxobjs
def test_excluded(testconfig):
#sources = testconfig.get('index_paths')
#perms_policy = testconfig.get('perms_policy')
#cdx_server = CDXServer(sources)
#index_handler = IndexHandler(cdx_server, perms_policy=perms_policy)
query_handler = QueryHandler.init_from_config(testconfig)
url = 'http://www.iana.org/_img/bookmark_icon.ico'
params = dict(url=url)
with raises(AccessException):
cdxobjs = list(query_handler.load_cdx(None, params))
print(cdxobjs)
def create_perma_pywb_app(config):
"""
Configure server.
This should do basically the same stuff as pywb.webapp.pywb_init.create_wb_router()
"""
# paths
script_path = os.path.dirname(__file__)
# Get root storage location for warcs.
# archive_path should be the location pywb can find warcs, like 'file://generated/' or 'http://perma.s3.amazonaws.com/generated/'
# We can get it by requesting the location of a blank file from default_storage.
# default_storage may use disk or network storage depending on config, so we look for either a path() or url()
try:
archive_path = 'file://' + default_storage.path('') + '/'
except NotImplementedError:
archive_path = default_storage.url('/')
archive_path = archive_path.split('?', 1)[0] # remove query params
query_handler = QueryHandler.init_from_config(PermaCDXSource())
# pywb template vars (used in templates called by pywb, such as head_insert.html, but not our ErrorTemplateView)
add_env_globals({'static_path': settings.STATIC_URL})
# use util func to create the handler
wb_handler = create_wb_handler(query_handler,
dict(archive_paths=[archive_path],
wb_handler_class=Handler,
buffer_response=True,
head_insert_html=os.path.join(script_path, 'head_insert.html'),
redir_to_exact=False))
wb_handler.replay.content_loader.record_loader.loader = CachedLoader()
# Finally, create wb router
return Router(
{
Route(r'([a-zA-Z0-9\-]+)', wb_handler)
},
# 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=['http://localhost:8000/'],
port=8000,
error_view=ErrorTemplateView()
)
def create_perma_wb_router(config={}):
"""
Configure server.
This should do basically the same stuff as pywb.webapp.pywb_init.create_wb_router()
"""
# paths
script_path = os.path.dirname(__file__)
# Get root storage location for warcs.
# archive_path should be the location pywb can find warcs, like 'file://generated/' or 'http://perma.s3.amazonaws.com/generated/'
# We can get it by requesting the location of a blank file from default_storage.
# default_storage may use disk or network storage depending on config, so we look for either a path() or url()
try:
archive_path = 'file://' + default_storage.path('') + '/'
except NotImplementedError:
archive_path = default_storage.url('/')
archive_path = archive_path.split('?', 1)[0] # remove query params
query_handler = QueryHandler.init_from_config(PermaCDXSource())
# pywb template vars (used in templates called by pywb, such as head_insert.html, but not our ErrorTemplateView)
add_env_globals({'static_path': settings.STATIC_URL})
# use util func to create the handler
wb_handler = create_wb_handler(
query_handler,
dict(archive_paths=[archive_path],
wb_handler_class=PermaGUIDHandler,
buffer_response=True,
head_insert_html=os.path.join(script_path, 'head_insert.html'),
enable_memento=True,
redir_to_exact=False))
wb_handler.replay.content_loader.record_loader.loader = CachedLoader()
route = PermaRoute(GUID_REGEX, wb_handler)
router = create_wb_router(config)
router.error_view = PermaTemplateView('archive-error.html')
router.routes.insert(0, route)
return router
def create_perma_pywb_app(config):
"""
Configure server.
"""
query_handler = QueryHandler.init_from_config(settings.CDX_SERVER_URL)
# Get root storage location for warcs.
# archive_path should be the location pywb can find warcs, like 'file://generated/' or 'http://perma.s3.amazonaws.com/generated/'
# We can get it by requesting the location of a blank file from default_storage.
# default_storage may use disk or network storage depending on config, so we look for either a path() or url()
try:
archive_path = 'file://' + default_storage.path('') + '/'
except NotImplementedError:
archive_path = default_storage.url('/')
archive_path = archive_path.split('?', 1)[0] # remove query params
# use util func to create the handler
wb_handler = create_wb_handler(query_handler,
dict(archive_paths=[archive_path],
wb_handler_class=Handler,
buffer_response=True,
head_insert_html='ui/head_insert.html',
template_globals={'static_path': 'static/js'},
redir_to_exact=False))
# Finally, create wb router
return archivalrouter.ArchivalRouter(
{
Route(r'([a-zA-Z0-9\-]+)', wb_handler)
},
# 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=['http://localhost:8000/'],
port=8000
)
