def serve_cdx(self, environ, coll='$root'):
"""Make the upstream CDX query for a collection and response with the results of the query
:param dict environ: The WSGI environment dictionary for the request
:param str coll: The name of the collection this CDX query is for
:return: The WbResponse containing the results of the CDX query
:rtype: WbResponse
"""
base_url = self.rewriterapp.paths['cdx-server']
# if coll == self.all_coll:
# coll = '*'
cdx_url = base_url.format(coll=coll)
if environ.get('QUERY_STRING'):
cdx_url += '&' if '?' in cdx_url else '?'
cdx_url += environ.get('QUERY_STRING')
if self.query_limit:
cdx_url += '&' if '?' in cdx_url else '?'
cdx_url += 'limit=' + str(self.query_limit)
try:
res = requests.get(cdx_url, stream=True)
content_type = res.headers.get('Content-Type')
return WbResponse.bin_stream(StreamIter(res.raw),
content_type=content_type)
except Exception as e:
return WbResponse.text_response('Error: ' + str(e),
status='400 Bad Request')
def serve_cdx(self, environ, coll='$root'):
"""Make the upstream CDX query for a collection and response with the results of the query
:param dict environ: The WSGI environment dictionary for the request
:param str coll: The name of the collection this CDX query is for
:return: The WbResponse containing the results of the CDX query
:rtype: WbResponse
"""
base_url = self.rewriterapp.paths['cdx-server']
#if coll == self.all_coll:
# coll = '*'
cdx_url = base_url.format(coll=coll)
if environ.get('QUERY_STRING'):
cdx_url += '&' if '?' in cdx_url else '?'
cdx_url += environ.get('QUERY_STRING')
try:
res = requests.get(cdx_url, stream=True)
content_type = res.headers.get('Content-Type')
return WbResponse.bin_stream(StreamIter(res.raw),
content_type=content_type)
except Exception as e:
return WbResponse.text_response('Error: ' + str(e), status='400 Bad Request')
def test_resp_2():
resp = vars(
WbResponse.bin_stream([b'Test', b'Another'],
content_type='text/plain; charset=utf-8',
status='404'))
expected = {
'body': [b'Test', b'Another'],
'status_headers':
StatusAndHeaders(protocol='',
statusline='404',
headers=[('Content-Type', 'text/plain; charset=utf-8')
])
}
assert (resp == expected)
def __call__(self, environ, url_str):
url = url_str.split('?')[0]
if url.endswith('/'):
url += 'index.html'
full_path = environ.get('pywb.static_dir')
if full_path:
full_path = os.path.join(full_path, url)
if not os.path.isfile(full_path):
full_path = None
if not full_path:
full_path = os.path.join(self.static_path, url)
try:
data = self.block_loader.load(full_path)
data.seek(0, 2)
size = data.tell()
data.seek(0)
headers = [('Content-Length', str(size))]
reader = None
if 'wsgi.file_wrapper' in environ:
try:
reader = environ['wsgi.file_wrapper'](data)
except:
pass
if not reader:
reader = iter(lambda: data.read(), b'')
content_type = 'application/octet-stream'
guessed = mimetypes.guess_type(full_path)
if guessed[0]:
content_type = guessed[0]
return WbResponse.bin_stream(reader,
content_type=content_type,
headers=headers)
except IOError:
raise NotFoundException('Static File Not Found: ' + url_str)
def __call__(self, environ, url_str):
url = url_str.split('?')[0]
full_path = environ.get('pywb.static_dir')
if full_path:
full_path = os.path.join(full_path, url)
if not os.path.isfile(full_path):
full_path = None
if not full_path:
full_path = os.path.join(self.static_path, url)
try:
data = self.block_loader.load(full_path)
data.seek(0, 2)
size = data.tell()
data.seek(0)
headers = [('Content-Length', str(size))]
reader = None
if 'wsgi.file_wrapper' in environ:
try:
reader = environ['wsgi.file_wrapper'](data)
except:
pass
if not reader:
reader = iter(lambda: data.read(), b'')
content_type = 'application/octet-stream'
guessed = mimetypes.guess_type(full_path)
if guessed[0]:
content_type = guessed[0]
return WbResponse.bin_stream(reader,
content_type=content_type,
headers=headers)
except IOError:
raise NotFoundException('Static File Not Found: ' +
url_str)
def serve_cdx(self, environ, coll='$root'):
base_url = self.rewriterapp.paths['cdx-server']
#if coll == self.all_coll:
# coll = '*'
cdx_url = base_url.format(coll=coll)
if environ.get('QUERY_STRING'):
cdx_url += '&' if '?' in cdx_url else '?'
cdx_url += environ.get('QUERY_STRING')
try:
res = requests.get(cdx_url, stream=True)
content_type = res.headers.get('Content-Type')
return WbResponse.bin_stream(StreamIter(res.raw),
content_type=content_type)
except Exception as e:
return WbResponse.text_response('Error: ' + str(e),
status='400 Bad Request')
