def prepare_response(self, grab):
if self.body_file:
self.body_file.close()
response = Response()
response.head = b''.join(self.response_header_chunks)
if self.body_path:
response.body_path = self.body_path
else:
response.body = b''.join(self.response_body_chunks)
# Clear memory
self.response_header_chunks = []
self.response_body_chunks = []
response.code = self.curl.getinfo(pycurl.HTTP_CODE)
response.total_time = self.curl.getinfo(pycurl.TOTAL_TIME)
response.connect_time = self.curl.getinfo(pycurl.CONNECT_TIME)
response.name_lookup_time = self.curl.getinfo(pycurl.NAMELOOKUP_TIME)
response.download_size = self.curl.getinfo(pycurl.SIZE_DOWNLOAD)
response.upload_size = self.curl.getinfo(pycurl.SIZE_UPLOAD)
response.download_speed = self.curl.getinfo(pycurl.SPEED_DOWNLOAD)
response.remote_ip = self.curl.getinfo(pycurl.PRIMARY_IP)
response.url = self.curl.getinfo(pycurl.EFFECTIVE_URL)
response.parse(charset=grab.config['document_charset'])
response.cookies = CookieManager(self.extract_cookiejar())
# We do not need anymore cookies stored in the
# curl instance so drop them
self.curl.setopt(pycurl.COOKIELIST, 'ALL')
return response
def prepare_response(self, grab):
if self.body_file:
self.body_file.close()
response = Response()
response.head = b"".join(self.response_header_chunks)
if self.body_path:
response.body_path = self.body_path
else:
response.body = b"".join(self.response_body_chunks)
# Clear memory
self.response_header_chunks = []
self.response_body_chunks = []
response.code = self.curl.getinfo(pycurl.HTTP_CODE)
response.total_time = self.curl.getinfo(pycurl.TOTAL_TIME)
response.connect_time = self.curl.getinfo(pycurl.CONNECT_TIME)
response.name_lookup_time = self.curl.getinfo(pycurl.NAMELOOKUP_TIME)
response.download_size = self.curl.getinfo(pycurl.SIZE_DOWNLOAD)
response.upload_size = self.curl.getinfo(pycurl.SIZE_UPLOAD)
response.download_speed = self.curl.getinfo(pycurl.SPEED_DOWNLOAD)
response.remote_ip = self.curl.getinfo(pycurl.PRIMARY_IP)
response.url = self.curl.getinfo(pycurl.EFFECTIVE_URL)
response.parse(charset=grab.config["document_charset"])
response.cookies = CookieManager(self.extract_cookiejar())
# We do not need anymore cookies stored in the
# curl instance so drop them
self.curl.setopt(pycurl.COOKIELIST, "ALL")
return response
def prepare_response(self, grab):
#if self.body_file:
# self.body_file.close()
response = Response()
head = ''
for key, val in self._response.getheaders().items():
head += '%s: %s\r\n' % (key, val)
head += '\r\n'
response.head = make_str(head, encoding='latin', errors='ignore')
#if self.body_path:
# response.body_path = self.body_path
#else:
# response.body = b''.join(self.response_body_chunks)
if self._request._response_path:
response.body_path = self._request._response_path
# Quick dirty hack, actullay, response is fully read into memory
self._request._response_file.write(self._response.read()) #data)
self._request._response_file.close()
else:
if self._request.body_maxsize is not None:
#if self.response_body_bytes_read > self.config_body_maxsize:
# logger.debug('Response body max size limit reached: %s' %
# self.config_body_maxsize)
response.body = self._response.read(self._request.body_maxsize)
else:
response.body = self._response.read() #data
# Clear memory
#self.response_header_chunks = []
response.code = self._response.status
#response.total_time = self.curl.getinfo(pycurl.TOTAL_TIME)
#response.connect_time = self.curl.getinfo(pycurl.CONNECT_TIME)
#response.name_lookup_time = self.curl.getinfo(pycurl.NAMELOOKUP_TIME)
#response.download_size = self.curl.getinfo(pycurl.SIZE_DOWNLOAD)
#response.upload_size = self.curl.getinfo(pycurl.SIZE_UPLOAD)
#response.download_speed = self.curl.getinfo(pycurl.SPEED_DOWNLOAD)
#response.remote_ip = self.curl.getinfo(pycurl.PRIMARY_IP)
response.url = self._response.get_redirect_location(
) or self._request.url
import email.message
hdr = email.message.Message()
for key, val in self._response.getheaders().items():
hdr[key] = val
response.parse(charset=grab.config['document_charset'], headers=hdr)
jar = self.extract_cookiejar(self._response, self._request)
response.cookies = CookieManager(jar)
# We do not need anymore cookies stored in the
# curl instance so drop them
#self.curl.setopt(pycurl.COOKIELIST, 'ALL')
return response
def prepare_response(self, grab):
#if self.body_file:
# self.body_file.close()
response = Response()
head = ''
for key, val in self._response.getheaders().items():
head += '%s: %s\r\n' % (key, val)
head += '\r\n'
response.head = make_str(head, encoding='latin', errors='ignore')
#if self.body_path:
# response.body_path = self.body_path
#else:
# response.body = b''.join(self.response_body_chunks)
if self._request._response_path:
response.body_path = self._request._response_path
# Quick dirty hack, actullay, response is fully read into memory
self._request._response_file.write(self._response.read())#data)
self._request._response_file.close()
else:
if self._request.body_maxsize is not None:
#if self.response_body_bytes_read > self.config_body_maxsize:
# logger.debug('Response body max size limit reached: %s' %
# self.config_body_maxsize)
response.body = self._response.read(self._request.body_maxsize)
else:
response.body = self._response.read()#data
# Clear memory
#self.response_header_chunks = []
response.code = self._response.status
#response.total_time = self.curl.getinfo(pycurl.TOTAL_TIME)
#response.connect_time = self.curl.getinfo(pycurl.CONNECT_TIME)
#response.name_lookup_time = self.curl.getinfo(pycurl.NAMELOOKUP_TIME)
#response.download_size = self.curl.getinfo(pycurl.SIZE_DOWNLOAD)
#response.upload_size = self.curl.getinfo(pycurl.SIZE_UPLOAD)
#response.download_speed = self.curl.getinfo(pycurl.SPEED_DOWNLOAD)
#response.remote_ip = self.curl.getinfo(pycurl.PRIMARY_IP)
response.url = self._response.get_redirect_location() or self._request.url
import email.message
hdr = email.message.Message()
for key, val in self._response.getheaders().items():
hdr[key] = val
response.parse(charset=grab.config['document_charset'],
headers=hdr)
jar = self.extract_cookiejar(self._response, self._request)
response.cookies = CookieManager(jar)
# We do not need anymore cookies stored in the
# curl instance so drop them
#self.curl.setopt(pycurl.COOKIELIST, 'ALL')
return response
def prepare_response(self, grab):
# py3 hack
if PY3K:
self.response_head_chunks = decode_list(self.response_head_chunks)
if self.body_file:
self.body_file.close()
response = Response()
response.head = ''.join(self.response_head_chunks)
if self.body_path:
response.body_path = self.body_path
else:
response.body = b''.join(self.response_body_chunks)
# Clear memory
self.response_head_chunks = []
self.response_body_chunks = []
response.code = self.curl.getinfo(pycurl.HTTP_CODE)
response.total_time = self.curl.getinfo(pycurl.TOTAL_TIME)
response.connect_time = self.curl.getinfo(pycurl.CONNECT_TIME)
response.name_lookup_time = self.curl.getinfo(pycurl.NAMELOOKUP_TIME)
response.download_size = self.curl.getinfo(pycurl.SIZE_DOWNLOAD)
response.upload_size = self.curl.getinfo(pycurl.SIZE_UPLOAD)
response.download_speed = self.curl.getinfo(pycurl.SPEED_DOWNLOAD)
response.url = self.curl.getinfo(pycurl.EFFECTIVE_URL)
if grab.config['document_charset'] is not None:
response.parse(charset=grab.config['document_charset'])
else:
response.parse()
response.cookies = CookieManager(self.extract_cookiejar())
# We do not need anymore cookies stored in the
# curl instance so drop them
self.curl.setopt(pycurl.COOKIELIST, 'ALL')
return response
def prepare_response(self, grab):
try:
#if self.body_file:
# self.body_file.close()
response = Response()
head = ''
for key, val in self._response.getheaders().items():
head += '%s: %s\r\n' % (key, val)
head += '\r\n'
response.head = make_str(head, encoding='latin', errors='ignore')
#if self.body_path:
# response.body_path = self.body_path
#else:
# response.body = b''.join(self.response_body_chunks)
def read_with_timeout():
if self._request.config_nobody:
return b''
maxsize = self._request.config_body_maxsize
chunks = []
default_chunk_size = 10000
if maxsize:
chunk_size = min(default_chunk_size, maxsize + 1)
else:
chunk_size = default_chunk_size
total_size = 0
while True:
chunk = self._response.read(chunk_size)
if chunk:
total_size += len(chunk)
chunks.append(chunk)
if maxsize and total_size > maxsize:
logger.debug(
'Response body max size limit reached: %s' %
maxsize)
else:
break
if self._request.timeout:
if time.time(
) - self._request.op_started > self._request.timeout:
raise GrabTimeoutError
data = b''.join(chunks)
if maxsize:
data = data[:maxsize]
return data
if self._request._response_path:
response.body_path = self._request._response_path
# FIXME: Quick dirty hack, actullay, response is fully read into memory
self._request._response_file.write(read_with_timeout())
self._request._response_file.close()
else:
response.body = read_with_timeout()
# Clear memory
#self.response_header_chunks = []
response.code = self._response.status
#response.total_time = self.curl.getinfo(pycurl.TOTAL_TIME)
#response.connect_time = self.curl.getinfo(pycurl.CONNECT_TIME)
#response.name_lookup_time = self.curl.getinfo(pycurl.NAMELOOKUP_TIME)
#response.download_size = self.curl.getinfo(pycurl.SIZE_DOWNLOAD)
#response.upload_size = self.curl.getinfo(pycurl.SIZE_UPLOAD)
#response.download_speed = self.curl.getinfo(pycurl.SPEED_DOWNLOAD)
#response.remote_ip = self.curl.getinfo(pycurl.PRIMARY_IP)
response.url = self._response.get_redirect_location(
) or self._request.url
import email.message
hdr = email.message.Message()
for key, val in self._response.getheaders().items():
hdr[key] = val
response.parse(charset=grab.config['document_charset'],
headers=hdr)
jar = self.extract_cookiejar() #self._response, self._request)
response.cookies = CookieManager(jar)
# We do not need anymore cookies stored in the
# curl instance so drop them
#self.curl.setopt(pycurl.COOKIELIST, 'ALL')
return response
finally:
self._response.release_conn()
