hmdev-cli 1.0.0

package/python/.venv/Lib/site-packages/h11/_readers.py

@@ -0,0 +1,250 @@
+# Code to read HTTP data
+#
+# Strategy: each reader is a callable which takes a ReceiveBuffer object, and
+# either:
+# 1) consumes some of it and returns an Event
+# 2) raises a LocalProtocolError (for consistency -- e.g. we call validate()
+#    and it might raise a LocalProtocolError, so simpler just to always use
+#    this)
+# 3) returns None, meaning "I need more data"
+#
+# If they have a .read_eof attribute, then this will be called if an EOF is
+# received -- but this is optional. Either way, the actual ConnectionClosed
+# event will be generated afterwards.
+#
+# READERS is a dict describing how to pick a reader. It maps states to either:
+# - a reader
+# - or, for body readers, a dict of per-framing reader factories
+
+import re
+from typing import Any, Callable, Dict, Iterable, NoReturn, Optional, Tuple, Type, Union
+
+from ._abnf import chunk_header, header_field, request_line, status_line
+from ._events import Data, EndOfMessage, InformationalResponse, Request, Response
+from ._receivebuffer import ReceiveBuffer
+from ._state import (
+    CLIENT,
+    CLOSED,
+    DONE,
+    IDLE,
+    MUST_CLOSE,
+    SEND_BODY,
+    SEND_RESPONSE,
+    SERVER,
+)
+from ._util import LocalProtocolError, RemoteProtocolError, Sentinel, validate
+
+__all__ = ["READERS"]
+
+header_field_re = re.compile(header_field.encode("ascii"))
+obs_fold_re = re.compile(rb"[ \t]+")
+
+
+def _obsolete_line_fold(lines: Iterable[bytes]) -> Iterable[bytes]:
+    it = iter(lines)
+    last: Optional[bytes] = None
+    for line in it:
+        match = obs_fold_re.match(line)
+        if match:
+            if last is None:
+                raise LocalProtocolError("continuation line at start of headers")
+            if not isinstance(last, bytearray):
+                # Cast to a mutable type, avoiding copy on append to ensure O(n) time
+                last = bytearray(last)
+            last += b" "
+            last += line[match.end() :]
+        else:
+            if last is not None:
+                yield last
+            last = line
+    if last is not None:
+        yield last
+
+
+def _decode_header_lines(
+    lines: Iterable[bytes],
+) -> Iterable[Tuple[bytes, bytes]]:
+    for line in _obsolete_line_fold(lines):
+        matches = validate(header_field_re, line, "illegal header line: {!r}", line)
+        yield (matches["field_name"], matches["field_value"])
+
+
+request_line_re = re.compile(request_line.encode("ascii"))
+
+
+def maybe_read_from_IDLE_client(buf: ReceiveBuffer) -> Optional[Request]:
+    lines = buf.maybe_extract_lines()
+    if lines is None:
+        if buf.is_next_line_obviously_invalid_request_line():
+            raise LocalProtocolError("illegal request line")
+        return None
+    if not lines:
+        raise LocalProtocolError("no request line received")
+    matches = validate(
+        request_line_re, lines[0], "illegal request line: {!r}", lines[0]
+    )
+    return Request(
+        headers=list(_decode_header_lines(lines[1:])), _parsed=True, **matches
+    )
+
+
+status_line_re = re.compile(status_line.encode("ascii"))
+
+
+def maybe_read_from_SEND_RESPONSE_server(
+    buf: ReceiveBuffer,
+) -> Union[InformationalResponse, Response, None]:
+    lines = buf.maybe_extract_lines()
+    if lines is None:
+        if buf.is_next_line_obviously_invalid_request_line():
+            raise LocalProtocolError("illegal request line")
+        return None
+    if not lines:
+        raise LocalProtocolError("no response line received")
+    matches = validate(status_line_re, lines[0], "illegal status line: {!r}", lines[0])
+    http_version = (
+        b"1.1" if matches["http_version"] is None else matches["http_version"]
+    )
+    reason = b"" if matches["reason"] is None else matches["reason"]
+    status_code = int(matches["status_code"])
+    class_: Union[Type[InformationalResponse], Type[Response]] = (
+        InformationalResponse if status_code < 200 else Response
+    )
+    return class_(
+        headers=list(_decode_header_lines(lines[1:])),
+        _parsed=True,
+        status_code=status_code,
+        reason=reason,
+        http_version=http_version,
+    )
+
+
+class ContentLengthReader:
+    def __init__(self, length: int) -> None:
+        self._length = length
+        self._remaining = length
+
+    def __call__(self, buf: ReceiveBuffer) -> Union[Data, EndOfMessage, None]:
+        if self._remaining == 0:
+            return EndOfMessage()
+        data = buf.maybe_extract_at_most(self._remaining)
+        if data is None:
+            return None
+        self._remaining -= len(data)
+        return Data(data=data)
+
+    def read_eof(self) -> NoReturn:
+        raise RemoteProtocolError(
+            "peer closed connection without sending complete message body "
+            "(received {} bytes, expected {})".format(
+                self._length - self._remaining, self._length
+            )
+        )
+
+
+chunk_header_re = re.compile(chunk_header.encode("ascii"))
+
+
+class ChunkedReader:
+    def __init__(self) -> None:
+        self._bytes_in_chunk = 0
+        # After reading a chunk, we have to throw away the trailing \r\n.
+        # This tracks the bytes that we need to match and throw away.
+        self._bytes_to_discard = b""
+        self._reading_trailer = False
+
+    def __call__(self, buf: ReceiveBuffer) -> Union[Data, EndOfMessage, None]:
+        if self._reading_trailer:
+            lines = buf.maybe_extract_lines()
+            if lines is None:
+                return None
+            return EndOfMessage(headers=list(_decode_header_lines(lines)))
+        if self._bytes_to_discard:
+            data = buf.maybe_extract_at_most(len(self._bytes_to_discard))
+            if data is None:
+                return None
+            if data != self._bytes_to_discard[: len(data)]:
+                raise LocalProtocolError(
+                    f"malformed chunk footer: {data!r} (expected {self._bytes_to_discard!r})"
+                )
+            self._bytes_to_discard = self._bytes_to_discard[len(data) :]
+            if self._bytes_to_discard:
+                return None
+            # else, fall through and read some more
+        assert self._bytes_to_discard == b""
+        if self._bytes_in_chunk == 0:
+            # We need to refill our chunk count
+            chunk_header = buf.maybe_extract_next_line()
+            if chunk_header is None:
+                return None
+            matches = validate(
+                chunk_header_re,
+                chunk_header,
+                "illegal chunk header: {!r}",
+                chunk_header,
+            )
+            # XX FIXME: we discard chunk extensions. Does anyone care?
+            self._bytes_in_chunk = int(matches["chunk_size"], base=16)
+            if self._bytes_in_chunk == 0:
+                self._reading_trailer = True
+                return self(buf)
+            chunk_start = True
+        else:
+            chunk_start = False
+        assert self._bytes_in_chunk > 0
+        data = buf.maybe_extract_at_most(self._bytes_in_chunk)
+        if data is None:
+            return None
+        self._bytes_in_chunk -= len(data)
+        if self._bytes_in_chunk == 0:
+            self._bytes_to_discard = b"\r\n"
+            chunk_end = True
+        else:
+            chunk_end = False
+        return Data(data=data, chunk_start=chunk_start, chunk_end=chunk_end)
+
+    def read_eof(self) -> NoReturn:
+        raise RemoteProtocolError(
+            "peer closed connection without sending complete message body "
+            "(incomplete chunked read)"
+        )
+
+
+class Http10Reader:
+    def __call__(self, buf: ReceiveBuffer) -> Optional[Data]:
+        data = buf.maybe_extract_at_most(999999999)
+        if data is None:
+            return None
+        return Data(data=data)
+
+    def read_eof(self) -> EndOfMessage:
+        return EndOfMessage()
+
+
+def expect_nothing(buf: ReceiveBuffer) -> None:
+    if buf:
+        raise LocalProtocolError("Got data when expecting EOF")
+    return None
+
+
+ReadersType = Dict[
+    Union[Type[Sentinel], Tuple[Type[Sentinel], Type[Sentinel]]],
+    Union[Callable[..., Any], Dict[str, Callable[..., Any]]],
+]
+
+READERS: ReadersType = {
+    (CLIENT, IDLE): maybe_read_from_IDLE_client,
+    (SERVER, IDLE): maybe_read_from_SEND_RESPONSE_server,
+    (SERVER, SEND_RESPONSE): maybe_read_from_SEND_RESPONSE_server,
+    (CLIENT, DONE): expect_nothing,
+    (CLIENT, MUST_CLOSE): expect_nothing,
+    (CLIENT, CLOSED): expect_nothing,
+    (SERVER, DONE): expect_nothing,
+    (SERVER, MUST_CLOSE): expect_nothing,
+    (SERVER, CLOSED): expect_nothing,
+    SEND_BODY: {
+        "chunked": ChunkedReader,
+        "content-length": ContentLengthReader,
+        "http/1.0": Http10Reader,
+    },
+}

package/python/.venv/Lib/site-packages/h11/_receivebuffer.py

@@ -0,0 +1,153 @@
+import re
+import sys
+from typing import List, Optional, Union
+
+__all__ = ["ReceiveBuffer"]
+
+
+# Operations we want to support:
+# - find next \r\n or \r\n\r\n (\n or \n\n are also acceptable),
+#   or wait until there is one
+# - read at-most-N bytes
+# Goals:
+# - on average, do this fast
+# - worst case, do this in O(n) where n is the number of bytes processed
+# Plan:
+# - store bytearray, offset, how far we've searched for a separator token
+# - use the how-far-we've-searched data to avoid rescanning
+# - while doing a stream of uninterrupted processing, advance offset instead
+#   of constantly copying
+# WARNING:
+# - I haven't benchmarked or profiled any of this yet.
+#
+# Note that starting in Python 3.4, deleting the initial n bytes from a
+# bytearray is amortized O(n), thanks to some excellent work by Antoine
+# Martin:
+#
+#     https://bugs.python.org/issue19087
+#
+# This means that if we only supported 3.4+, we could get rid of the code here
+# involving self._start and self.compress, because it's doing exactly the same
+# thing that bytearray now does internally.
+#
+# BUT unfortunately, we still support 2.7, and reading short segments out of a
+# long buffer MUST be O(bytes read) to avoid DoS issues, so we can't actually
+# delete this code. Yet:
+#
+#     https://pythonclock.org/
+#
+# (Two things to double-check first though: make sure PyPy also has the
+# optimization, and benchmark to make sure it's a win, since we do have a
+# slightly clever thing where we delay calling compress() until we've
+# processed a whole event, which could in theory be slightly more efficient
+# than the internal bytearray support.)
+blank_line_regex = re.compile(b"\n\r?\n", re.MULTILINE)
+
+
+class ReceiveBuffer:
+    def __init__(self) -> None:
+        self._data = bytearray()
+        self._next_line_search = 0
+        self._multiple_lines_search = 0
+
+    def __iadd__(self, byteslike: Union[bytes, bytearray]) -> "ReceiveBuffer":
+        self._data += byteslike
+        return self
+
+    def __bool__(self) -> bool:
+        return bool(len(self))
+
+    def __len__(self) -> int:
+        return len(self._data)
+
+    # for @property unprocessed_data
+    def __bytes__(self) -> bytes:
+        return bytes(self._data)
+
+    def _extract(self, count: int) -> bytearray:
+        # extracting an initial slice of the data buffer and return it
+        out = self._data[:count]
+        del self._data[:count]
+
+        self._next_line_search = 0
+        self._multiple_lines_search = 0
+
+        return out
+
+    def maybe_extract_at_most(self, count: int) -> Optional[bytearray]:
+        """
+        Extract a fixed number of bytes from the buffer.
+        """
+        out = self._data[:count]
+        if not out:
+            return None
+
+        return self._extract(count)
+
+    def maybe_extract_next_line(self) -> Optional[bytearray]:
+        """
+        Extract the first line, if it is completed in the buffer.
+        """
+        # Only search in buffer space that we've not already looked at.
+        search_start_index = max(0, self._next_line_search - 1)
+        partial_idx = self._data.find(b"\r\n", search_start_index)
+
+        if partial_idx == -1:
+            self._next_line_search = len(self._data)
+            return None
+
+        # + 2 is to compensate len(b"\r\n")
+        idx = partial_idx + 2
+
+        return self._extract(idx)
+
+    def maybe_extract_lines(self) -> Optional[List[bytearray]]:
+        """
+        Extract everything up to the first blank line, and return a list of lines.
+        """
+        # Handle the case where we have an immediate empty line.
+        if self._data[:1] == b"\n":
+            self._extract(1)
+            return []
+
+        if self._data[:2] == b"\r\n":
+            self._extract(2)
+            return []
+
+        # Only search in buffer space that we've not already looked at.
+        match = blank_line_regex.search(self._data, self._multiple_lines_search)
+        if match is None:
+            self._multiple_lines_search = max(0, len(self._data) - 2)
+            return None
+
+        # Truncate the buffer and return it.
+        idx = match.span(0)[-1]
+        out = self._extract(idx)
+        lines = out.split(b"\n")
+
+        for line in lines:
+            if line.endswith(b"\r"):
+                del line[-1]
+
+        assert lines[-2] == lines[-1] == b""
+
+        del lines[-2:]
+
+        return lines
+
+    # In theory we should wait until `\r\n` before starting to validate
+    # incoming data. However it's interesting to detect (very) invalid data
+    # early given they might not even contain `\r\n` at all (hence only
+    # timeout will get rid of them).
+    # This is not a 100% effective detection but more of a cheap sanity check
+    # allowing for early abort in some useful cases.
+    # This is especially interesting when peer is messing up with HTTPS and
+    # sent us a TLS stream where we were expecting plain HTTP given all
+    # versions of TLS so far start handshake with a 0x16 message type code.
+    def is_next_line_obviously_invalid_request_line(self) -> bool:
+        try:
+            # HTTP header line must not contain non-printable characters
+            # and should not start with a space
+            return self._data[0] < 0x21
+        except IndexError:
+            return False