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

tests/test_regressions.py

@@ -2,6 +2,7 @@
 import pytest
 
 from httpie.compat import is_windows
+from tests.utils.matching import assert_output_matches, Expect
 from utils import HTTP_OK, MockEnvironment, http
 
 
@@ -27,7 +28,6 @@ def test_output_devnull(httpbin):
     http('--output=/dev/null', httpbin + '/get')
 
 
-@pytest.mark.skip(reason='TODO: fix #1006')
 def test_verbose_redirected_stdout_separator(httpbin):
     """
 
@@ -40,3 +40,10 @@ def test_verbose_redirected_stdout_separator(httpbin):
         env=MockEnvironment(stdout_isatty=False),
     )
     assert '}HTTP/' not in r
+    assert_output_matches(r, [
+        Expect.REQUEST_HEADERS,
+        Expect.BODY,
+        Expect.SEPARATOR,
+        Expect.RESPONSE_HEADERS,
+        Expect.BODY,
+    ])

tests/test_tokens.py

@@ -0,0 +1,111 @@
+"""
+The ideas behind these test and the named templates is to ensure consistent output
+across all supported different scenarios:
+
+TODO: cover more scenarios
+ * terminal vs. redirect stdout
+ * different combinations of `--print=HBhb` (request/response headers/body)
+ * multipart requests
+ * streamed uploads
+
+"""
+from tests.utils.matching import assert_output_matches, Expect
+from utils import http, HTTP_OK, MockEnvironment, HTTPBIN_WITH_CHUNKED_SUPPORT
+
+
+RAW_REQUEST = [
+    Expect.REQUEST_HEADERS,
+    Expect.BODY,
+]
+RAW_RESPONSE = [
+    Expect.RESPONSE_HEADERS,
+    Expect.BODY,
+]
+RAW_EXCHANGE = [
+    *RAW_REQUEST,
+    Expect.SEPARATOR,  # Good choice?
+    *RAW_RESPONSE,
+]
+RAW_BODY = [
+    Expect.BODY,
+]
+
+TERMINAL_REQUEST = [
+    *RAW_REQUEST,
+    Expect.SEPARATOR,
+]
+TERMINAL_RESPONSE = [
+    *RAW_RESPONSE,
+    Expect.SEPARATOR,
+]
+TERMINAL_EXCHANGE = [
+    *TERMINAL_REQUEST,
+    *TERMINAL_RESPONSE,
+]
+TERMINAL_BODY = [
+    RAW_BODY,
+    Expect.SEPARATOR
+]
+
+
+def test_headers():
+    r = http('--print=H', '--offline', 'pie.dev')
+    assert_output_matches(r, [Expect.REQUEST_HEADERS])
+
+
+def test_redirected_headers():
+    r = http('--print=H', '--offline', 'pie.dev', env=MockEnvironment(stdout_isatty=False))
+    assert_output_matches(r, [Expect.REQUEST_HEADERS])
+
+
+def test_terminal_headers_and_body():
+    r = http('--print=HB', '--offline', 'pie.dev', 'AAA=BBB')
+    assert_output_matches(r, TERMINAL_REQUEST)
+
+
+def test_raw_headers_and_body():
+    r = http(
+        '--print=HB', '--offline', 'pie.dev', 'AAA=BBB',
+        env=MockEnvironment(stdout_isatty=False),
+    )
+    assert_output_matches(r, RAW_REQUEST)
+
+
+def test_raw_body():
+    r = http(
+        '--print=B', '--offline', 'pie.dev', 'AAA=BBB',
+        env=MockEnvironment(stdout_isatty=False),
+    )
+    assert_output_matches(r, RAW_BODY)
+
+
+def test_raw_exchange(httpbin):
+    r = http('--verbose', httpbin + '/post', 'a=b', env=MockEnvironment(stdout_isatty=False))
+    assert HTTP_OK in r
+    assert_output_matches(r, RAW_EXCHANGE)
+
+
+def test_terminal_exchange(httpbin):
+    r = http('--verbose', httpbin + '/post', 'a=b')
+    assert HTTP_OK in r
+    assert_output_matches(r, TERMINAL_EXCHANGE)
+
+
+def test_headers_multipart_body_separator():
+    r = http('--print=HB', '--multipart', '--offline', 'pie.dev', 'AAA=BBB')
+    assert_output_matches(r, TERMINAL_REQUEST)
+
+
+def test_redirected_headers_multipart_no_separator():
+    r = http(
+        '--print=HB', '--multipart', '--offline', 'pie.dev', 'AAA=BBB',
+        env=MockEnvironment(stdout_isatty=False),
+    )
+    assert_output_matches(r, RAW_REQUEST)
+
+
+def test_verbose_chunked():
+    r = http('--verbose', '--chunked', HTTPBIN_WITH_CHUNKED_SUPPORT + '/post', 'hello=world')
+    assert HTTP_OK in r
+    assert 'Transfer-Encoding: chunked' in r
+    assert_output_matches(r, TERMINAL_EXCHANGE)

tests/utils/__init__.py

@@ -1,12 +1,14 @@
 # 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
+from typing import Optional, Union, List
 
 from httpie.status import ExitStatus
 from httpie.config import Config
@@ -20,7 +22,7 @@ from httpie.core import main
 HTTPBIN_WITH_CHUNKED_SUPPORT = 'http://pie.dev'
 
 
-TESTS_ROOT = Path(__file__).parent
+TESTS_ROOT = Path(__file__).parent.parent
 CRLF = '\r\n'
 COLOR = '\x1b['
 HTTP_OK = '200 OK'
@@ -49,7 +51,7 @@ class StdinBytesIO(BytesIO):
 
 class MockEnvironment(Environment):
     """Environment subclass with reasonable defaults for testing."""
-    colors = 0
+    colors = 0  # For easier debugging
     stdin_isatty = True,
     stdout_isatty = True
     is_windows = False
@@ -113,6 +115,15 @@ class BaseCLIResponse:
     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):
@@ -284,10 +295,13 @@ def http(
         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:

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]
+    )