def set_receiver(self):
"""Once the headers have been read, decide how to read the body.
We use the standard library's email package to parse the headers. The
receiver is stored on self so it persists across calls to our own
self.received, since it actually holds data between such calls. Later on
we will add the request body to self.message.
"""
content_length = int(self.headers.get('Content-Length', 0))
transfer_encoding = self.headers.get('Transfer-Encoding', '').lower()
if not content_length and not transfer_encoding:
self._receiver = None
elif content_length and not transfer_encoding:
buffer_ = OverflowableBuffer(self.adj.inbuf_overflow)
self._receiver = FixedStreamReceiver(content_length, buffer_)
elif transfer_encoding == 'chunked':
buffer_ = OverflowableBuffer(self.adj.inbuf_overflow)
self._receiver = ChunkedReceiver(buffer_)
else:
self._receiver = None
def parse_header(self, header_plus):
"""
Parses the header_plus block of text (the headers plus the
first line of the request).
"""
index = header_plus.find('\n')
if index >= 0:
first_line = header_plus[:index].rstrip()
header = header_plus[index + 1:]
else:
first_line = header_plus.rstrip()
header = ''
self.first_line = first_line
self.header = header
lines = self.get_header_lines()
headers = self.headers
for line in lines:
index = line.find(':')
if index > 0:
key = line[:index]
value = line[index + 1:].strip()
key1 = key.upper().replace('-', '_')
# If a header already exists, we append subsequent values
# seperated by a comma. Applications already need to handle
# the comma seperated values, as HTTP front ends might do
# the concatenation for you (behavior specified in RFC2616).
try:
headers[key1] += ', %s' % value
except KeyError:
headers[key1] = value
# else there's garbage in the headers?
command, uri, version = self.crack_first_line()
self.command = str(command)
self.uri = str(uri)
self.version = version
self.split_uri()
if version == '1.1':
te = headers.get('TRANSFER_ENCODING', '')
if te == 'chunked':
from httpy._zope.server.http.chunking import ChunkedReceiver
self.chunked = 1
buf = OverflowableBuffer(self.adj.inbuf_overflow)
self.body_rcv = ChunkedReceiver(buf)
if not self.chunked:
try:
cl = int(headers.get('CONTENT_LENGTH', 0))
except ValueError:
cl = 0
self.content_length = cl
if cl > 0:
buf = OverflowableBuffer(self.adj.inbuf_overflow)
self.body_rcv = FixedStreamReceiver(cl, buf)
class ZopeRequest:
"""An implementation of IRequest that works with httpy._zope.server.
"""
implements(IStreamConsumer, IRequest)
raw = ''
raw_line = ''
raw_headers = ''
raw_body = ''
method = ''
uri = {}
path = ''
headers = Message()
def __init__(self, adj=None):
"""Takes an Adjustments object.
"""
if adj is None:
adj = default_adj
self.adj = adj
self._receiver = None
self._tmp = ''
self._raw = []
# initialize mutable defaults
self.uri = {}
self.headers = Message()
# Fulfill our IStreamConsumer contracts.
# ======================================
def received(self, block):
"""This takes a block of an incoming HTTP stream.
We store anything leftover from the last block in raw string form in
self._tmp.
"""
self._raw.append(block)
try:
if not self.raw_line:
self.get_line(block)
elif not self.raw_headers:
self.get_headers(block)
elif not self.raw_body:
self.get_body(block)
except Complete:
self.raw = ''.join(self._raw)
self.completed = True
return len(block)
completed = False
empty = False
# Request-Line
# ============
def get_line(self, block):
"""
"""
block = self._tmp + block
if '\r' in block or '\n' in block:
lines = block.splitlines(True)
i = 0
for line in lines:
if not line.strip(): #could be a blank line (like from IE)
pass
elif (line.endswith('\r')
or line.endswith('\n')): #could be the whole thing
self.raw_line = line.strip()
self._tmp = ''.join(lines[i + 1:])
break
else: #or else it is a partial line
self._tmp = ''.join(lines[i:])
break
i += 1
else:
self._tmp = block
if self.raw_line:
self.parse_line()
# Decide if we have any headers.
line_ending = lines[0][len(self.raw_line):]
if self._tmp == line_ending:
#self._raw.append(self._tmp) -- why is this in here?
raise Complete
if self._tmp:
self.get_headers('')
def parse_line(self):
"""Parse and validate the Request-Line.
"""
# Tokenize the first line as a Request-Line.
# ==========================================
tokens = self.raw_line.strip().split()
if len(tokens) == 3:
method, uri, version = tokens
elif len(tokens) == 2:
method, uri = tokens
version = 'HTTP/0.9'
else:
raise Response(
400, "The Request-Line `%s' " % self.raw_line +
"appears to be malformed because it has " +
"neither two nor three tokens.")
# Validate the URI.
# =================
# We build a mapping using the urlparse naming convention for the keys.
# Then we do a little validation and cleanup.
uri = urlparse.urlparse(uri)
keys = ('scheme', 'netloc', 'path', 'parameters', 'query', 'fragment')
_uri = {}
for i in range(len(uri)):
k = keys[i]
v = uri[i]
_uri[k] = v
uri = _uri
if not uri['path']:
# this catches, e.g., '//foo'
raise Response(400)
if '%' in uri['path']:
uri['path'] = urllib.unquote(uri['path'])
# Validate the version.
# =====================
# Consider raising Response(505) here ... for 0.9? 2.0?
m = HTTP_VERSION.match(version)
if not m:
raise Response(
400, "The HTTP-Version `%s' appears to " % version +
"be malformed because it does not match the " +
"pattern `^HTTP/\d+\.\d+$'.")
# Save a few absolutely basic things for application use.
# =======================================================
self.method = method
self.uri = uri
self.path = uri['path']
# Headers
# =======
def get_headers(self, block):
"""The tricky part here is dealing with different line delimiters.
Technically MIME messages are \r\n delimited, but we can't assume that.
We want to split on the first double line break we get, regardless of
the control characters used.
"""
block = self._tmp + block
found_splitter = False
for double_line_break in ('\n\r\n', '\n\n'):
if double_line_break in block:
found_splitter = True
self.raw_headers, self._tmp = block.split(double_line_break, 1)
self.raw_headers = self.raw_headers.strip()
break
if not found_splitter:
self._tmp = block
self.headers = message_from_string(self.raw_headers)
if self.raw_headers:
self.get_body('')
# Body
# ====
def set_receiver(self):
"""Once the headers have been read, decide how to read the body.
We use the standard library's email package to parse the headers. The
receiver is stored on self so it persists across calls to our own
self.received, since it actually holds data between such calls. Later on
we will add the request body to self.message.
"""
content_length = int(self.headers.get('Content-Length', 0))
transfer_encoding = self.headers.get('Transfer-Encoding', '').lower()
if not content_length and not transfer_encoding:
self._receiver = None
elif content_length and not transfer_encoding:
buffer_ = OverflowableBuffer(self.adj.inbuf_overflow)
self._receiver = FixedStreamReceiver(content_length, buffer_)
elif transfer_encoding == 'chunked':
buffer_ = OverflowableBuffer(self.adj.inbuf_overflow)
self._receiver = ChunkedReceiver(buffer_)
else:
self._receiver = None
def get_body(self, block):
"""Given a block from an HTTP stream, build a message body.
We guarantee that the entire block is part of an HTTP request message.
"""
# Decide which receiver to use.
# =============================
# If we don't have a receiver, it means there is no body.
if self._receiver is None:
self.set_receiver()
if self._receiver is None:
self.raw_body = ''
raise Complete
# Get all of the message body in one block.
# =========================================
if self._tmp:
block = self._tmp + block
self._tmp = ''
if not self._receiver.completed:
self._receiver.received(block)
if not self._receiver.completed:
return # I'll be back!
# Store on self and get out of here.
# ==================================
# We toss the body away if the requested method doesn't allow for it. I
# read RFC 2616, Section 4.3 to call for this behavior rather than,
# e.g., returning a 400. I suppose the concern is that the input stream
# be properly positioned for multi-request connections.
#
# If anyone ever wants httpy to support extension-methods that take a
# body we will either have to add them here or make allowances in
# configuration.
if self.method in ('POST', 'POST'):
self.raw_body = self._receiver.getfile().read()
else:
self.raw_body = ''
raise Complete