diff options
Diffstat (limited to 'netlib/http/http1/protocol.py')
| -rw-r--r-- | netlib/http/http1/protocol.py | 563 |
1 files changed, 282 insertions, 281 deletions
diff --git a/netlib/http/http1/protocol.py b/netlib/http/http1/protocol.py index d2a77399..e7727e00 100644 --- a/netlib/http/http1/protocol.py +++ b/netlib/http/http1/protocol.py @@ -15,15 +15,144 @@ class HTTP1Protocol(object): self.tcp_handler = tcp_handler - def get_request_line(self): + def read_request(self, include_body=True, body_size_limit=None, allow_empty=False): """ - Get a line, possibly preceded by a blank. + Parse an HTTP request from a file stream + + Args: + include_body (bool): Read response body as well + body_size_limit (bool): Maximum body size + wfile (file): If specified, HTTP Expect headers are handled + automatically, by writing a HTTP 100 CONTINUE response to the stream. + + Returns: + Request: The HTTP request + + Raises: + HttpError: If the input is invalid. + """ + httpversion, host, port, scheme, method, path, headers, content = ( + None, None, None, None, None, None, None, None) + + request_line = self._get_request_line() + if not request_line: + if allow_empty: + return http.EmptyRequest() + else: + raise tcp.NetLibDisconnect() + + request_line_parts = self._parse_init(request_line) + if not request_line_parts: + raise HttpError( + 400, + "Bad HTTP request line: %s" % repr(request_line) + ) + method, path, httpversion = request_line_parts + + if path == '*' or path.startswith("/"): + form_in = "relative" + if not utils.isascii(path): + raise HttpError( + 400, + "Bad HTTP request line: %s" % repr(request_line) + ) + elif method.upper() == 'CONNECT': + form_in = "authority" + r = self._parse_init_connect(request_line) + if not r: + raise HttpError( + 400, + "Bad HTTP request line: %s" % repr(request_line) + ) + host, port, _ = r + return http.ConnectRequest(host, port) + else: + form_in = "absolute" + r = self._parse_init_proxy(request_line) + if not r: + raise HttpError( + 400, + "Bad HTTP request line: %s" % repr(request_line) + ) + _, scheme, host, port, path, _ = r + + headers = self.read_headers() + if headers is None: + raise HttpError(400, "Invalid headers") + + expect_header = headers.get_first("expect", "").lower() + if expect_header == "100-continue" and httpversion >= (1, 1): + self.tcp_handler.wfile.write( + 'HTTP/1.1 100 Continue\r\n' + '\r\n' + ) + self.tcp_handler.wfile.flush() + del headers['expect'] + + if include_body: + content = self.read_http_body( + headers, + body_size_limit, + method, + None, + True + ) + + return http.Request( + form_in, + method, + scheme, + host, + port, + path, + httpversion, + headers, + content + ) + + + def read_response(self, request_method, body_size_limit, include_body=True): + """ + Returns an http.Response + + By default, both response header and body are read. + If include_body=False is specified, content may be one of the + following: + - None, if the response is technically allowed to have a response body + - "", if the response must not have a response body (e.g. it's a + response to a HEAD request) """ + line = self.tcp_handler.rfile.readline() + # Possible leftover from previous message if line == "\r\n" or line == "\n": - # Possible leftover from previous message line = self.tcp_handler.rfile.readline() - return line + if not line: + raise HttpErrorConnClosed(502, "Server disconnect.") + parts = self.parse_response_line(line) + if not parts: + raise HttpError(502, "Invalid server response: %s" % repr(line)) + proto, code, msg = parts + httpversion = self._parse_http_protocol(proto) + if httpversion is None: + raise HttpError(502, "Invalid HTTP version in line: %s" % repr(proto)) + headers = self.read_headers() + if headers is None: + raise HttpError(502, "Invalid headers.") + + if include_body: + content = self.read_http_body( + headers, + body_size_limit, + request_method, + code, + False + ) + else: + # if include_body==False then a None content means the body should be + # read separately + content = None + return http.Response(httpversion, code, msg, headers, content) def read_headers(self): @@ -56,7 +185,146 @@ class HTTP1Protocol(object): return odict.ODictCaseless(ret) - def read_chunked(self, limit, is_request): + def read_http_body(self, *args, **kwargs): + return "".join( + content for _, content, _ in self.read_http_body_chunked(*args, **kwargs) + ) + + + def read_http_body_chunked( + self, + headers, + limit, + request_method, + response_code, + is_request, + max_chunk_size=None + ): + """ + Read an HTTP message body: + headers: An ODictCaseless object + limit: Size limit. + is_request: True if the body to read belongs to a request, False + otherwise + """ + if max_chunk_size is None: + max_chunk_size = limit or sys.maxsize + + expected_size = self.expected_http_body_size( + headers, is_request, request_method, response_code + ) + + if expected_size is None: + if self.has_chunked_encoding(headers): + # Python 3: yield from + for x in self._read_chunked(limit, is_request): + yield x + else: # pragma: nocover + raise HttpError( + 400 if is_request else 502, + "Content-Length unknown but no chunked encoding" + ) + elif expected_size >= 0: + if limit is not None and expected_size > limit: + raise HttpError( + 400 if is_request else 509, + "HTTP Body too large. Limit is %s, content-length was %s" % ( + limit, expected_size + ) + ) + bytes_left = expected_size + while bytes_left: + chunk_size = min(bytes_left, max_chunk_size) + yield "", self.tcp_handler.rfile.read(chunk_size), "" + bytes_left -= chunk_size + else: + bytes_left = limit or -1 + while bytes_left: + chunk_size = min(bytes_left, max_chunk_size) + content = self.tcp_handler.rfile.read(chunk_size) + if not content: + return + yield "", content, "" + bytes_left -= chunk_size + not_done = self.tcp_handler.rfile.read(1) + if not_done: + raise HttpError( + 400 if is_request else 509, + "HTTP Body too large. Limit is %s," % limit + ) + + + @classmethod + def expected_http_body_size(self, headers, is_request, request_method, response_code): + """ + Returns the expected body length: + - a positive integer, if the size is known in advance + - None, if the size in unknown in advance (chunked encoding or invalid + data) + - -1, if all data should be read until end of stream. + + May raise HttpError. + """ + # Determine response size according to + # http://tools.ietf.org/html/rfc7230#section-3.3 + if request_method: + request_method = request_method.upper() + + if (not is_request and ( + request_method == "HEAD" or + (request_method == "CONNECT" and response_code == 200) or + response_code in [204, 304] or + 100 <= response_code <= 199)): + return 0 + if self.has_chunked_encoding(headers): + return None + if "content-length" in headers: + try: + size = int(headers["content-length"][0]) + if size < 0: + raise ValueError() + return size + except ValueError: + return None + if is_request: + return 0 + return -1 + + + @classmethod + def request_preamble(self, method, resource, http_major="1", http_minor="1"): + return '%s %s HTTP/%s.%s' % ( + method, resource, http_major, http_minor + ) + + + @classmethod + def response_preamble(self, code, message=None, http_major="1", http_minor="1"): + if message is None: + message = status_codes.RESPONSES.get(code) + return 'HTTP/%s.%s %s %s' % (http_major, http_minor, code, message) + + + @classmethod + def has_chunked_encoding(self, headers): + return "chunked" in [ + i.lower() for i in http.get_header_tokens(headers, "transfer-encoding") + ] + + + def _get_request_line(self): + """ + Get a line, possibly preceded by a blank. + """ + line = self.tcp_handler.rfile.readline() + if line == "\r\n" or line == "\n": + # Possible leftover from previous message + line = self.tcp_handler.rfile.readline() + return line + + + + def _read_chunked(self, limit, is_request): """ Read a chunked HTTP body. @@ -94,14 +362,7 @@ class HTTP1Protocol(object): @classmethod - def has_chunked_encoding(self, headers): - return "chunked" in [ - i.lower() for i in http.get_header_tokens(headers, "transfer-encoding") - ] - - - @classmethod - def parse_http_protocol(self, line): + def _parse_http_protocol(self, line): """ Parse an HTTP protocol declaration. Returns a (major, minor) tuple, or None. @@ -121,12 +382,12 @@ class HTTP1Protocol(object): @classmethod - def parse_init(self, line): + def _parse_init(self, line): try: method, url, protocol = string.split(line) except ValueError: return None - httpversion = self.parse_http_protocol(protocol) + httpversion = self._parse_http_protocol(protocol) if not httpversion: return None if not utils.isascii(method): @@ -135,12 +396,12 @@ class HTTP1Protocol(object): @classmethod - def parse_init_connect(self, line): + def _parse_init_connect(self, line): """ Returns (host, port, httpversion) if line is a valid CONNECT line. http://tools.ietf.org/html/draft-luotonen-web-proxy-tunneling-01 section 3.1 """ - v = self.parse_init(line) + v = self._parse_init(line) if not v: return None method, url, httpversion = v @@ -163,8 +424,8 @@ class HTTP1Protocol(object): @classmethod - def parse_init_proxy(self, line): - v = self.parse_init(line) + def _parse_init_proxy(self, line): + v = self._parse_init(line) if not v: return None method, url, httpversion = v @@ -177,11 +438,11 @@ class HTTP1Protocol(object): @classmethod - def parse_init_http(self, line): + def _parse_init_http(self, line): """ Returns (method, url, httpversion) """ - v = self.parse_init(line) + v = self._parse_init(line) if not v: return None method, url, httpversion = v @@ -225,263 +486,3 @@ class HTTP1Protocol(object): except ValueError: return None return (proto, code, msg) - - - def read_http_body(self, *args, **kwargs): - return "".join( - content for _, content, _ in self.read_http_body_chunked(*args, **kwargs) - ) - - - def read_http_body_chunked( - self, - headers, - limit, - request_method, - response_code, - is_request, - max_chunk_size=None - ): - """ - Read an HTTP message body: - headers: An ODictCaseless object - limit: Size limit. - is_request: True if the body to read belongs to a request, False - otherwise - """ - if max_chunk_size is None: - max_chunk_size = limit or sys.maxsize - - expected_size = self.expected_http_body_size( - headers, is_request, request_method, response_code - ) - - if expected_size is None: - if self.has_chunked_encoding(headers): - # Python 3: yield from - for x in self.read_chunked(limit, is_request): - yield x - else: # pragma: nocover - raise HttpError( - 400 if is_request else 502, - "Content-Length unknown but no chunked encoding" - ) - elif expected_size >= 0: - if limit is not None and expected_size > limit: - raise HttpError( - 400 if is_request else 509, - "HTTP Body too large. Limit is %s, content-length was %s" % ( - limit, expected_size - ) - ) - bytes_left = expected_size - while bytes_left: - chunk_size = min(bytes_left, max_chunk_size) - yield "", self.tcp_handler.rfile.read(chunk_size), "" - bytes_left -= chunk_size - else: - bytes_left = limit or -1 - while bytes_left: - chunk_size = min(bytes_left, max_chunk_size) - content = self.tcp_handler.rfile.read(chunk_size) - if not content: - return - yield "", content, "" - bytes_left -= chunk_size - not_done = self.tcp_handler.rfile.read(1) - if not_done: - raise HttpError( - 400 if is_request else 509, - "HTTP Body too large. Limit is %s," % limit - ) - - - @classmethod - def expected_http_body_size(self, headers, is_request, request_method, response_code): - """ - Returns the expected body length: - - a positive integer, if the size is known in advance - - None, if the size in unknown in advance (chunked encoding or invalid - data) - - -1, if all data should be read until end of stream. - - May raise HttpError. - """ - # Determine response size according to - # http://tools.ietf.org/html/rfc7230#section-3.3 - if request_method: - request_method = request_method.upper() - - if (not is_request and ( - request_method == "HEAD" or - (request_method == "CONNECT" and response_code == 200) or - response_code in [204, 304] or - 100 <= response_code <= 199)): - return 0 - if self.has_chunked_encoding(headers): - return None - if "content-length" in headers: - try: - size = int(headers["content-length"][0]) - if size < 0: - raise ValueError() - return size - except ValueError: - return None - if is_request: - return 0 - return -1 - - - def read_request(self, include_body=True, body_size_limit=None, allow_empty=False): - """ - Parse an HTTP request from a file stream - - Args: - include_body (bool): Read response body as well - body_size_limit (bool): Maximum body size - wfile (file): If specified, HTTP Expect headers are handled - automatically, by writing a HTTP 100 CONTINUE response to the stream. - - Returns: - Request: The HTTP request - - Raises: - HttpError: If the input is invalid. - """ - httpversion, host, port, scheme, method, path, headers, content = ( - None, None, None, None, None, None, None, None) - - request_line = self.get_request_line() - if not request_line: - if allow_empty: - return http.EmptyRequest() - else: - raise tcp.NetLibDisconnect() - - request_line_parts = self.parse_init(request_line) - if not request_line_parts: - raise HttpError( - 400, - "Bad HTTP request line: %s" % repr(request_line) - ) - method, path, httpversion = request_line_parts - - if path == '*' or path.startswith("/"): - form_in = "relative" - if not utils.isascii(path): - raise HttpError( - 400, - "Bad HTTP request line: %s" % repr(request_line) - ) - elif method.upper() == 'CONNECT': - form_in = "authority" - r = self.parse_init_connect(request_line) - if not r: - raise HttpError( - 400, - "Bad HTTP request line: %s" % repr(request_line) - ) - host, port, _ = r - return http.ConnectRequest(host, port) - else: - form_in = "absolute" - r = self.parse_init_proxy(request_line) - if not r: - raise HttpError( - 400, - "Bad HTTP request line: %s" % repr(request_line) - ) - _, scheme, host, port, path, _ = r - - headers = self.read_headers() - if headers is None: - raise HttpError(400, "Invalid headers") - - expect_header = headers.get_first("expect", "").lower() - if expect_header == "100-continue" and httpversion >= (1, 1): - self.tcp_handler.wfile.write( - 'HTTP/1.1 100 Continue\r\n' - '\r\n' - ) - self.tcp_handler.wfile.flush() - del headers['expect'] - - if include_body: - content = self.read_http_body( - headers, - body_size_limit, - method, - None, - True - ) - - return http.Request( - form_in, - method, - scheme, - host, - port, - path, - httpversion, - headers, - content - ) - - - def read_response(self, request_method, body_size_limit, include_body=True): - """ - Returns an http.Response - - By default, both response header and body are read. - If include_body=False is specified, content may be one of the - following: - - None, if the response is technically allowed to have a response body - - "", if the response must not have a response body (e.g. it's a - response to a HEAD request) - """ - - line = self.tcp_handler.rfile.readline() - # Possible leftover from previous message - if line == "\r\n" or line == "\n": - line = self.tcp_handler.rfile.readline() - if not line: - raise HttpErrorConnClosed(502, "Server disconnect.") - parts = self.parse_response_line(line) - if not parts: - raise HttpError(502, "Invalid server response: %s" % repr(line)) - proto, code, msg = parts - httpversion = self.parse_http_protocol(proto) - if httpversion is None: - raise HttpError(502, "Invalid HTTP version in line: %s" % repr(proto)) - headers = self.read_headers() - if headers is None: - raise HttpError(502, "Invalid headers.") - - if include_body: - content = self.read_http_body( - headers, - body_size_limit, - request_method, - code, - False - ) - else: - # if include_body==False then a None content means the body should be - # read separately - content = None - return http.Response(httpversion, code, msg, headers, content) - - - @classmethod - def request_preamble(self, method, resource, http_major="1", http_minor="1"): - return '%s %s HTTP/%s.%s' % ( - method, resource, http_major, http_minor - ) - - - @classmethod - def response_preamble(self, code, message=None, http_major="1", http_minor="1"): - if message is None: - message = status_codes.RESPONSES.get(code) - return 'HTTP/%s.%s %s %s' % (http_major, http_minor, code, message) |