Fix incorrect separators and introduce assert_output_matches() (close #1027)

tests/utils/__init__.py

@@ -0,0 +1,311 @@
+# coding=utf-8
+"""Utilities for HTTPie test suite."""
+import re
+import shlex
+import sys
+import time
+import json
+import tempfile
+from io import BytesIO
+from pathlib import Path
+from typing import Optional, Union, List
+
+from httpie.status import ExitStatus
+from httpie.config import Config
+from httpie.context import Environment
+from httpie.core import main
+
+
+# pytest-httpbin currently does not support chunked requests:
+# <https://github.com/kevin1024/pytest-httpbin/issues/33>
+# <https://github.com/kevin1024/pytest-httpbin/issues/28>
+HTTPBIN_WITH_CHUNKED_SUPPORT = 'http://pie.dev'
+
+
+TESTS_ROOT = Path(__file__).parent.parent
+CRLF = '\r\n'
+COLOR = '\x1b['
+HTTP_OK = '200 OK'
+# noinspection GrazieInspection
+HTTP_OK_COLOR = (
+    'HTTP\x1b[39m\x1b[38;5;245m/\x1b[39m\x1b'
+    '[38;5;37m1.1\x1b[39m\x1b[38;5;245m \x1b[39m\x1b[38;5;37m200'
+    '\x1b[39m\x1b[38;5;245m \x1b[39m\x1b[38;5;136mOK'
+)
+
+
+def mk_config_dir() -> Path:
+    dirname = tempfile.mkdtemp(prefix='httpie_config_')
+    return Path(dirname)
+
+
+def add_auth(url, auth):
+    proto, rest = url.split('://', 1)
+    return proto + '://' + auth + '@' + rest
+
+
+class StdinBytesIO(BytesIO):
+    """To be used for `MockEnvironment.stdin`"""
+    len = 0  # See `prepare_request_body()`
+
+
+class MockEnvironment(Environment):
+    """Environment subclass with reasonable defaults for testing."""
+    colors = 0  # For easier debugging
+    stdin_isatty = True,
+    stdout_isatty = True
+    is_windows = False
+
+    def __init__(self, create_temp_config_dir=True, **kwargs):
+        if 'stdout' not in kwargs:
+            kwargs['stdout'] = tempfile.TemporaryFile(
+                mode='w+b',
+                prefix='httpie_stdout'
+            )
+        if 'stderr' not in kwargs:
+            kwargs['stderr'] = tempfile.TemporaryFile(
+                mode='w+t',
+                prefix='httpie_stderr'
+            )
+        super().__init__(**kwargs)
+        self._create_temp_config_dir = create_temp_config_dir
+        self._delete_config_dir = False
+        self._temp_dir = Path(tempfile.gettempdir())
+
+    @property
+    def config(self) -> Config:
+        if (self._create_temp_config_dir
+                and self._temp_dir not in self.config_dir.parents):
+            self.create_temp_config_dir()
+        return super().config
+
+    def create_temp_config_dir(self):
+        self.config_dir = mk_config_dir()
+        self._delete_config_dir = True
+
+    def cleanup(self):
+        self.stdout.close()
+        self.stderr.close()
+        if self._delete_config_dir:
+            assert self._temp_dir in self.config_dir.parents
+            from shutil import rmtree
+            rmtree(self.config_dir, ignore_errors=True)
+
+    def __del__(self):
+        # noinspection PyBroadException
+        try:
+            self.cleanup()
+        except Exception:
+            pass
+
+
+class BaseCLIResponse:
+    """
+    Represents the result of simulated `$ http' invocation via `http()`.
+
+    Holds and provides access to:
+
+        - stdout output: print(self)
+        - stderr output: print(self.stderr)
+        - devnull output: print(self.devnull)
+        - exit_status output: print(self.exit_status)
+
+    """
+    stderr: str = None
+    devnull: str = None
+    json: dict = None
+    exit_status: ExitStatus = None
+    command: str = None
+    args: List[str] = []
+    complete_args: List[str] = []
+
+    @property
+    def command(self):
+        cmd = ' '.join(shlex.quote(arg) for arg in ['http', *self.args])
+        # pytest-httpbin to real httpbin.
+        return re.sub(r'127\.0\.0\.1:\d+', 'httpbin.org', cmd)
+
+
+class BytesCLIResponse(bytes, BaseCLIResponse):
+    """
+    Used as a fallback when a StrCLIResponse cannot be used.
+
+    E.g. when the output contains binary data or when it is colorized.
+
+    `.json` will always be None.
+
+    """
+
+
+class StrCLIResponse(str, BaseCLIResponse):
+
+    @property
+    def json(self) -> Optional[dict]:
+        """
+        Return deserialized the request or response JSON body,
+        if one (and only one) included in the output and is parsable.
+
+        """
+        if not hasattr(self, '_json'):
+            self._json = None
+            # De-serialize JSON body if possible.
+            if COLOR in self:
+                # Colorized output cannot be parsed.
+                pass
+            elif self.strip().startswith('{'):
+                # Looks like JSON body.
+                self._json = json.loads(self)
+            elif self.count('Content-Type:') == 1:
+                # Looks like a HTTP message,
+                # try to extract JSON from its body.
+                try:
+                    j = self.strip()[self.strip().rindex('\r\n\r\n'):]
+                except ValueError:
+                    pass
+                else:
+                    try:
+                        # noinspection PyAttributeOutsideInit
+                        self._json = json.loads(j)
+                    except ValueError:
+                        pass
+        return self._json
+
+
+class ExitStatusError(Exception):
+    pass
+
+
+def http(
+    *args,
+    program_name='http',
+    tolerate_error_exit_status=False,
+    **kwargs,
+) -> Union[StrCLIResponse, BytesCLIResponse]:
+    # noinspection PyUnresolvedReferences
+    """
+    Run HTTPie and capture stderr/out and exit status.
+    Content writtent to devnull will be captured only if
+    env.devnull is set manually.
+
+    Invoke `httpie.core.main()` with `args` and `kwargs`,
+    and return a `CLIResponse` subclass instance.
+
+    The return value is either a `StrCLIResponse`, or `BytesCLIResponse`
+    if unable to decode the output. Devnull is string when possible,
+    bytes otherwise.
+
+    The response has the following attributes:
+
+        `stdout` is represented by the instance itself (print r)
+        `stderr`: text written to stderr
+        `devnull` text written to devnull.
+        `exit_status`: the exit status
+        `json`: decoded JSON (if possible) or `None`
+
+    Exceptions are propagated.
+
+    If you pass ``tolerate_error_exit_status=True``, then error exit statuses
+    won't result into an exception.
+
+    Example:
+
+    $ http --auth=user:password GET pie.dev/basic-auth/user/password
+
+        >>> httpbin = getfixture('httpbin')
+        >>> r = http('-a', 'user:pw', httpbin.url + '/basic-auth/user/pw')
+        >>> type(r) == StrCLIResponse
+        True
+        >>> r.exit_status
+        <ExitStatus.SUCCESS: 0>
+        >>> r.stderr
+        ''
+        >>> 'HTTP/1.1 200 OK' in r
+        True
+        >>> r.json == {'authenticated': True, 'user': 'user'}
+        True
+
+    """
+    env = kwargs.get('env')
+    if not env:
+        env = kwargs['env'] = MockEnvironment()
+
+    stdout = env.stdout
+    stderr = env.stderr
+    devnull = env.devnull
+
+    args = list(args)
+    args_with_config_defaults = args + env.config.default_options
+    add_to_args = []
+    if '--debug' not in args_with_config_defaults:
+        if (not tolerate_error_exit_status
+                and '--traceback' not in args_with_config_defaults):
+            add_to_args.append('--traceback')
+        if not any('--timeout' in arg for arg in args_with_config_defaults):
+            add_to_args.append('--timeout=3')
+
+    complete_args = [program_name, *add_to_args, *args]
+    # print(' '.join(complete_args))
+
+    def dump_stderr():
+        stderr.seek(0)
+        sys.stderr.write(stderr.read())
+
+    try:
+        try:
+            exit_status = main(args=complete_args, **kwargs)
+            if '--download' in args:
+                # Let the progress reporter thread finish.
+                time.sleep(.5)
+        except SystemExit:
+            if tolerate_error_exit_status:
+                exit_status = ExitStatus.ERROR
+            else:
+                dump_stderr()
+                raise
+        except Exception:
+            stderr.seek(0)
+            sys.stderr.write(stderr.read())
+            raise
+        else:
+            if (not tolerate_error_exit_status
+                    and exit_status != ExitStatus.SUCCESS):
+                dump_stderr()
+                raise ExitStatusError(
+                    'httpie.core.main() unexpectedly returned'
+                    f' a non-zero exit status: {exit_status}'
+                )
+
+        stdout.seek(0)
+        stderr.seek(0)
+        devnull.seek(0)
+        output = stdout.read()
+        devnull_output = devnull.read()
+        try:
+            output = output.decode('utf8')
+        except UnicodeDecodeError:
+            r = BytesCLIResponse(output)
+        else:
+            r = StrCLIResponse(output)
+
+        try:
+            devnull_output = devnull_output.decode('utf8')
+        except Exception:
+            pass
+
+        r.devnull = devnull_output
+        r.stderr = stderr.read()
+        r.exit_status = exit_status
+        r.args = args
+        r.complete_args = ' '.join(complete_args)
+
+        if r.exit_status != ExitStatus.SUCCESS:
+            sys.stderr.write(r.stderr)
+
+        # print(f'\n\n$ {r.command}\n')
+        return r
+
+    finally:
+        devnull.close()
+        stdout.close()
+        stderr.close()
+        env.cleanup()

tests/utils/matching/__init__.py

@@ -0,0 +1,32 @@
+from typing import Iterable
+
+import pytest
+
+from tests.utils.matching.parsing import OutputMatchingError, expect_tokens, Expect
+
+
+__all__ = [
+    'assert_output_matches',
+    'assert_output_does_not_match',
+    'Expect',
+]
+
+
+def assert_output_matches(output: str, tokens: Iterable[Expect]):
+    r"""
+    Check the command `output` for an exact full sequence of `tokens`.
+
+    >>> out = 'GET / HTTP/1.1\r\nAAA:BBB\r\n\r\nCCC\n\n'
+    >>> assert_output_matches(out, [Expect.REQUEST_HEADERS, Expect.BODY, Expect.SEPARATOR])
+
+    """
+    # TODO: auto-remove ansi colors to allow for testing of colorized output as well.
+    expect_tokens(tokens=tokens, s=output)
+
+
+def assert_output_does_not_match(output: str, tokens: Iterable[Expect]):
+    r"""
+    >>> assert_output_does_not_match('\r\n', [Expect.BODY])
+    """
+    with pytest.raises(OutputMatchingError):
+        assert_output_matches(output=output, tokens=tokens)

tests/utils/matching/parsing.py

@@ -0,0 +1,107 @@
+import re
+from typing import Iterable
+from enum import Enum, auto
+
+from httpie.output.writer import MESSAGE_SEPARATOR
+from tests.utils import CRLF
+
+
+class Expect(Enum):
+    """
+    Predefined token types we can expect in the output.
+
+    """
+    REQUEST_HEADERS = auto()
+    RESPONSE_HEADERS = auto()
+    BODY = auto()
+    SEPARATOR = auto()
+
+
+SEPARATOR_RE = re.compile(f'^{MESSAGE_SEPARATOR}')
+
+
+def make_headers_re(message_type: Expect):
+    assert message_type in {Expect.REQUEST_HEADERS, Expect.RESPONSE_HEADERS}
+
+    # language=RegExp
+    crlf = r'[\r][\n]'
+    non_crlf = rf'[^{CRLF}]'
+
+    # language=RegExp
+    http_version = r'HTTP/\d+\.\d+'
+    if message_type is Expect.REQUEST_HEADERS:
+        # POST /post HTTP/1.1
+        start_line_re = fr'{non_crlf}*{http_version}{crlf}'
+    else:
+        # HTTP/1.1 200 OK
+        start_line_re = fr'{http_version}{non_crlf}*{crlf}'
+
+    return re.compile(
+        fr'''
+            ^
+            {start_line_re}
+            ({non_crlf}+:{non_crlf}+{crlf})+
+            {crlf}
+        ''',
+        flags=re.VERBOSE
+    )
+
+
+BODY_ENDINGS = [
+    MESSAGE_SEPARATOR,
+    CRLF,  # Not really but useful for testing (just remember not to include it in a body).
+]
+TOKEN_REGEX_MAP = {
+    Expect.REQUEST_HEADERS: make_headers_re(Expect.REQUEST_HEADERS),
+    Expect.RESPONSE_HEADERS: make_headers_re(Expect.RESPONSE_HEADERS),
+    Expect.SEPARATOR: SEPARATOR_RE,
+}
+
+
+class OutputMatchingError(ValueError):
+    pass
+
+
+def expect_tokens(tokens: Iterable[Expect], s: str):
+    for token in tokens:
+        s = expect_token(token, s)
+    if s:
+        raise OutputMatchingError(f'Unmatched remaining output for {tokens} in {s!r}')
+
+
+def expect_token(token: Expect, s: str) -> str:
+    if token is Expect.BODY:
+        s = expect_body(s)
+    else:
+        s = expect_regex(token, s)
+    return s
+
+
+def expect_regex(token: Expect, s: str) -> str:
+    match = TOKEN_REGEX_MAP[token].match(s)
+    if not match:
+        raise OutputMatchingError(f'No match for {token} in {s!r}')
+    return s[match.end():]
+
+
+def expect_body(s: str) -> str:
+    """
+    We require some text, and continue to read until we find an ending or until the end of the string.
+
+    """
+    if 'content-disposition:' in s.lower():
+        # Multipart body heuristic.
+        final_boundary_re = re.compile('\r\n--[^-]+?--\r\n')
+        match = final_boundary_re.search(s)
+        if match:
+            return s[match.end():]
+
+    endings = [s.index(sep) for sep in BODY_ENDINGS if sep in s]
+    if not endings:
+        s = ''  # Only body
+    else:
+        end = min(endings)
+        if end == 0:
+            raise OutputMatchingError(f'Empty body: {s!r}')
+        s = s[end:]
+    return s

tests/utils/matching/test_matching.py

@@ -0,0 +1,190 @@
+"""
+Here we test our output parsing and matching implementation, not HTTPie itself.
+
+"""
+from httpie.output.writer import MESSAGE_SEPARATOR
+from tests.utils import CRLF
+from tests.utils.matching import assert_output_does_not_match, assert_output_matches, Expect
+
+
+def test_assert_output_matches_headers_incomplete():
+    assert_output_does_not_match(f'HTTP/1.1{CRLF}', [Expect.RESPONSE_HEADERS])
+
+
+def test_assert_output_matches_headers_unterminated():
+    assert_output_does_not_match(
+        (
+            f'HTTP/1.1{CRLF}'
+            f'AAA:BBB'
+            f'{CRLF}'
+        ),
+        [Expect.RESPONSE_HEADERS],
+    )
+
+
+def test_assert_output_matches_response_headers():
+    assert_output_matches(
+        (
+            f'HTTP/1.1 200 OK{CRLF}'
+            f'AAA:BBB{CRLF}'
+            f'{CRLF}'
+        ),
+        [Expect.RESPONSE_HEADERS],
+    )
+
+
+def test_assert_output_matches_request_headers():
+    assert_output_matches(
+        (
+            f'GET / HTTP/1.1{CRLF}'
+            f'AAA:BBB{CRLF}'
+            f'{CRLF}'
+        ),
+        [Expect.REQUEST_HEADERS],
+    )
+
+
+def test_assert_output_matches_headers_and_separator():
+    assert_output_matches(
+        (
+            f'HTTP/1.1{CRLF}'
+            f'AAA:BBB{CRLF}'
+            f'{CRLF}'
+            f'{MESSAGE_SEPARATOR}'
+        ),
+        [Expect.RESPONSE_HEADERS, Expect.SEPARATOR],
+    )
+
+
+def test_assert_output_matches_body_unmatched_crlf():
+    assert_output_does_not_match(f'AAA{CRLF}', [Expect.BODY])
+
+
+def test_assert_output_matches_body_unmatched_separator():
+    assert_output_does_not_match(f'AAA{MESSAGE_SEPARATOR}', [Expect.BODY])
+
+
+def test_assert_output_matches_body_and_separator():
+    assert_output_matches(f'AAA{MESSAGE_SEPARATOR}', [Expect.BODY, Expect.SEPARATOR])
+
+
+def test_assert_output_matches_body_r():
+    assert_output_matches(f'AAA\r', [Expect.BODY])
+
+
+def test_assert_output_matches_body_n():
+    assert_output_matches(f'AAA\n', [Expect.BODY])
+
+
+def test_assert_output_matches_body_r_body():
+    assert_output_matches(f'AAA\rBBB', [Expect.BODY])
+
+
+def test_assert_output_matches_body_n_body():
+    assert_output_matches(f'AAA\nBBB', [Expect.BODY])
+
+
+def test_assert_output_matches_headers_and_body():
+    assert_output_matches(
+        (
+            f'HTTP/1.1{CRLF}'
+            f'AAA:BBB{CRLF}'
+            f'{CRLF}'
+            f'CCC'
+        ),
+        [Expect.RESPONSE_HEADERS, Expect.BODY]
+    )
+
+
+def test_assert_output_matches_headers_with_body_and_separator():
+    assert_output_matches(
+        (
+            f'HTTP/1.1 {CRLF}'
+            f'AAA:BBB{CRLF}{CRLF}'
+            f'CCC{MESSAGE_SEPARATOR}'
+        ),
+        [Expect.RESPONSE_HEADERS, Expect.BODY, Expect.SEPARATOR]
+    )
+
+
+def test_assert_output_matches_multiple_messages():
+    assert_output_matches(
+        (
+            f'POST / HTTP/1.1{CRLF}'
+            f'AAA:BBB{CRLF}'
+            f'{CRLF}'
+
+            f'CCC'
+            f'{MESSAGE_SEPARATOR}'
+
+            f'HTTP/1.1 200 OK{CRLF}'
+            f'EEE:FFF{CRLF}'
+            f'{CRLF}'
+
+            f'GGG'
+            f'{MESSAGE_SEPARATOR}'
+        ), [
+            Expect.REQUEST_HEADERS,
+            Expect.BODY,
+            Expect.SEPARATOR,
+            Expect.RESPONSE_HEADERS,
+            Expect.BODY,
+            Expect.SEPARATOR,
+        ]
+    )
+
+
+def test_assert_output_matches_multipart_body():
+    output = (
+        'POST / HTTP/1.1\r\n'
+        'User-Agent: HTTPie/2.4.0-dev\r\n'
+        'Accept-Encoding: gzip, deflate\r\n'
+        'Accept: */*\r\n'
+        'Connection: keep-alive\r\n'
+        'Content-Type: multipart/form-data; boundary=1e22169de43e4a2e8d9e41c0a1c93cc5\r\n'
+        'Content-Length: 212\r\n'
+        'Host: pie.dev\r\n'
+        '\r\n'
+        '--1e22169de43e4a2e8d9e41c0a1c93cc5\r\n'
+        'Content-Disposition: form-data; name="AAA"\r\n'
+        '\r\n'
+        'BBB\r\n'
+        '--1e22169de43e4a2e8d9e41c0a1c93cc5\r\n'
+        'Content-Disposition: form-data; name="CCC"\r\n'
+        '\r\n'
+        'DDD\r\n'
+        '--1e22169de43e4a2e8d9e41c0a1c93cc5--\r\n'
+    )
+    assert_output_matches(output, [Expect.REQUEST_HEADERS, Expect.BODY])
+
+
+def test_assert_output_matches_multipart_body_with_separator():
+    output = (
+        'POST / HTTP/1.1\r\n'
+        'User-Agent: HTTPie/2.4.0-dev\r\n'
+        'Accept-Encoding: gzip, deflate\r\n'
+        'Accept: */*\r\n'
+        'Connection: keep-alive\r\n'
+        'Content-Type: multipart/form-data; boundary=1e22169de43e4a2e8d9e41c0a1c93cc5\r\n'
+        'Content-Length: 212\r\n'
+        'Host: pie.dev\r\n'
+        '\r\n'
+        '--1e22169de43e4a2e8d9e41c0a1c93cc5\r\n'
+        'Content-Disposition: form-data; name="AAA"\r\n'
+        '\r\n'
+        'BBB\r\n'
+        '--1e22169de43e4a2e8d9e41c0a1c93cc5\r\n'
+        'Content-Disposition: form-data; name="CCC"\r\n'
+        '\r\n'
+        'DDD\r\n'
+        '--1e22169de43e4a2e8d9e41c0a1c93cc5--\r\n'
+        f'{MESSAGE_SEPARATOR}'
+    )
+    assert_output_matches(output, [Expect.REQUEST_HEADERS, Expect.BODY, Expect.SEPARATOR])
+
+
+def test_assert_output_matches_multiple_separators():
+    assert_output_matches(
+        MESSAGE_SEPARATOR + MESSAGE_SEPARATOR + 'AAA' + MESSAGE_SEPARATOR + MESSAGE_SEPARATOR,
+        [Expect.SEPARATOR, Expect.SEPARATOR, Expect.BODY, Expect.SEPARATOR, Expect.SEPARATOR]
+    )