class ApricotParser(object):
''' Apricot Http Parser '''
def __init__(self, parseType="request"):
# create abstract parser
self.parser = AbstractParser()
# determine httptools parser
if parseType.lower() == "request":
self.parsed = HttpRequestParser(self.parser)
else:
self.parsed = HttpResponseParser(self.parser)
# set null url info
self.url = None
self.host = None
self.schema = None
self.port = None
self.path = None
self.params = {}
async def feed_async(self, data=b''):
self.feed(data)
def feed(self, data=b''):
''' feed data to parser '''
self.raw_data = data
self.parsed.feed_data(memoryview(data))
self.set_attributes() # set apricot attributes
def set_attributes(self):
''' set ApricotParser attributes '''
# check parser type
if isinstance(self.parsed, HttpRequestParser):
self.method = self.parsed.get_method().decode()
else:
self.status = self.parsed.get_status_code()
# set basic attr's
try:
self.headers = self.parser.headers
self.body = self.parser.body
self.http_ver = self.parsed.get_http_version()
self.keep_alive = self.parsed.should_keep_alive()
except Exception as e:
print("error in parser: " + str(e))
if self.keep_alive == None: self.keep_alive = False
# parse url and get info
self.get_url_info()
def get_url_info(self):
''' Parse url info '''
if 'Host' in self.headers:
try:
# make url
self.url = b'http://'
self.url += self.headers['Host'].encode()
# add path if request
if isinstance(self.parsed, HttpRequestParser):
path = self.raw_data.split(b'\r\n')[0].split()[1]
self.url += path
self.path = path
# parse url and get basic info
URL = parse_url(self.url)
self.schema = URL.schema.decode()
self.host = URL.host.decode()
self.port = URL.port
# get url query parameters if any
if URL.query is not None:
for param in URL.query.split(b'&'):
parts = param.split(b'=')
key = parts[0].decode()
value = b'='.join(parts[1:]).decode()
self.params[key] = value
# decode url from bytes to string
self.url = self.url.decode()
except:
pass
class HttpProtocol:
"""The protocol class constructs the incoming request and response."""
def __init__(self, bufsize: int = 2048):
self._bufsize = bufsize
self._data_hooks = [] # type: List[Callable]
self._url_hooks = [] # type: List[Callable]
self._parser = None # type: PARSER_TYPE
self._is_done = False
self._body = bytes()
self.msg = None
def add_url_hook(self, func: Callable):
"""Add hook to be executed during on_url()"""
self._url_hooks.append(func)
def add_data_hook(self, func: Callable):
"""Add hook to be executed during recv()"""
self._data_hooks.append(func)
def set_bufsize(self, bufsize: int):
"""Set the bytes size to read from the socket"""
self._bufsize = bufsize
def set_request_parser(self):
"""Set a new request parser with self as callback"""
self._parser = HttpRequestParser(self)
def set_response_parser(self):
"""Set a new response parser with self as callback"""
self._parser = HttpResponseParser(self)
def set_request(self, request: dict):
"""Set new request"""
request['version'] = 'HTTP/1.1'
request['method'] = 'GET'
request['headers'] = {}
request['body'] = {}
request['url'] = {
'scheme': '',
'netloc': '',
'path': '',
'params': '',
'query': {},
'fragment': ''
}
self.msg = request
def set_response(self, response: dict):
"""Set new response"""
response['version'] = 'HTTP/1.1'
response['status'] = 200
response['reason'] = 'OK'
response['headers'] = {}
response['body'] = {}
self.msg = response
def feed(self, data: bytes):
"""feed data to the underlying parser"""
for hook in self._data_hooks:
hook(data)
self._parser.feed_data(data)
def recv(self, sock: socket.socket):
"""Receive all incoming http data on a socket and execute hooks"""
while not self._is_done:
data = sock.recv(self._bufsize)
self.feed(data)
# HTTPTOOLS CALLBACK METHODS
def on_message_begin(self):
"""Triggered on first bytes"""
def on_status(self, status: bytes):
"""Status integer callback"""
if self.msg is not None:
self.msg['status'] = self._parser.get_status_code()
self.msg['reason'] = status.decode()
def on_url(self, url: bytes):
"""Request message attribute will have url and method at this point"""
url = parse_url(url)
if self.msg is not None:
self.msg['url'] = url
self.msg['method'] = self._parser.get_method().decode()
for hook in self._url_hooks:
hook(url)
def on_header(self, name: bytes, value: bytes):
"""Alter request header as they come in if necessary"""
if self.msg is not None:
self.msg['headers'][name.decode().lower()] = value.decode()
def on_headers_complete(self):
"""End of headers trigger"""
def on_body(self, body: bytes):
"""Optionally access body stream"""
self._body += body
def on_message_complete(self):
"""Trigger when message finishes"""
if self.msg is not None:
self.msg['version'] = 'HTTP/' + self._parser.get_http_version()
if self.msg['headers'].get('content-encoding', '') == 'gzip':
self._body = gzip.decompress(self._body)
self.msg['body'] = load_json(self._body)
self._is_done = True
def on_chunk_header(self):
"""Chunked message header"""
def on_chunk_complete(self):
"""Chunked response end"""