From b7e88f4357eb11274693ece3618fb92cf7052d12 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Adri=C3=A1n=20Chaves?= <adrian@chaves.io>
Date: Thu, 30 Jun 2022 19:12:13 +0200
Subject: [PATCH] test_get_response_class_http: cover Apache bug scenarios

---
 tests/test_responsetypes.py  |  9 +++++-
 tests/test_utils_response.py | 61 +++++++++++++++++++++++++++++++-----
 2 files changed, 62 insertions(+), 8 deletions(-)

diff --git a/tests/test_responsetypes.py b/tests/test_responsetypes.py
index 3d5217385..42d4e481d 100644
--- a/tests/test_responsetypes.py
+++ b/tests/test_responsetypes.py
@@ -16,10 +16,17 @@ from .test_utils_response import (
 )
 
 
+def _unmark(item):
+    return pytest.param(*item.values)
+
+
 @pytest.mark.parametrize(
     "kwargs,response_class",
     (
-        *PRE_XTRACTMIME_SCENARIOS,
+        *(
+            item if not hasattr(item, "marks") else _unmark(item)
+            for item in PRE_XTRACTMIME_SCENARIOS
+        ),
         *(
             pytest.param(
                 kwargs,
diff --git a/tests/test_utils_response.py b/tests/test_utils_response.py
index bd229881f..dbdb57b1b 100644
--- a/tests/test_utils_response.py
+++ b/tests/test_utils_response.py
@@ -24,6 +24,36 @@ __doctests__ = ['scrapy.utils.response']
 # Scenarios that work the same with the previously-used, deprecated
 # scrapy.responsetypes.responsetypes.from_args
 PRE_XTRACTMIME_SCENARIOS = (
+    # Even if the body is binary, if the Content-Type says it is text, we
+    # interpret it as text, as long as the Content-Type is not one of the 4
+    # affected by the Apache bug.
+    #
+    # https://mimesniff.spec.whatwg.org/#interpreting-the-resource-metadata
+    (
+        {
+            'body': b'\x00\x01\xff',
+            'headers': Headers({'Content-Type': ['text/json']}),
+        },
+        TextResponse,
+    ),
+    *(
+        pytest.param(
+            {
+                'body': b'\x00\x01\xff',
+                'headers': Headers({'Content-Type': [content_type]}),
+            },
+            TextResponse,
+            marks=pytest.mark.xfail(
+                strict=True,
+                reason="https://github.com/scrapy/xtractmime/issues/13",
+            ),
+        )
+        for content_type in (
+            'text/plain; charset=Iso-8859-1',
+            'text/plain; charset=utf-8',
+            'text/plain; charset=windows-1252',
+        )
+    ),
     *(
         (
             {
@@ -172,14 +202,30 @@ PRE_XTRACTMIME_SCENARIOS = (
 # Scenarios that work differently with the previously-used, deprecated
 # scrapy.responsetypes.responsetypes.from_args
 POST_XTRACTMIME_SCENARIOS = (
-    (
-        {
-            'body': b'\x00\x01\xff',
-            'url': 'http://www.example.com/item/',
-            'headers': Headers({'Content-Type': ['text/plain']}),
-        },
-        Response,
+    # A known Apache bug may cause a server to send files with Content-Type set
+    # to "text/plain", "text/plain; charset=ISO-8859-1",
+    # "text/plain; charset=iso-8859-1", or "text/plain; charset=UTF-8",
+    # regardless of the actual file content.
+    #
+    # They should be treated as binary if their content is binary.
+    #
+    # https://mimesniff.spec.whatwg.org/#interpreting-the-resource-metadata
+    *(
+        (
+            {
+                'body': b'\x00\x01\xff',
+                'headers': Headers({'Content-Type': [content_type]}),
+            },
+            Response,
+        )
+        for content_type in (
+            'text/plain',
+            'text/plain; charset=ISO-8859-1',
+            'text/plain; charset=iso-8859-1',
+            'text/plain; charset=UTF-8',
+        )
     ),
+
     ({'filename': '/tmp/temp^'}, TextResponse),
     ({'body': b'%PDF-1.4'}, Response),
     (
@@ -236,6 +282,7 @@ POST_XTRACTMIME_SCENARIOS = (
     ),
 )
 def test_get_response_class_http(kwargs, response_class):
+    kwargs = dict(kwargs)
     if 'headers' in kwargs:
         kwargs['http_headers'] = kwargs.pop('headers')
     if 'filename' in kwargs: