← eval4_submission_v1b

requests_7427

resolved RESOLVED LOOP PASS · 21 tool calls · 148 s · psf/requests

Task input

Port bpo-39057 to Requests

This PR ports a [fix](https://bugs.python.org/issue39057) from CPython to constrain no_proxy behavior to domain boundaries. This was originally undefined behavior when Requests implemented `no_proxy` support but has since had bug fixes released in both CPython and Curl. We'll follow suit to conform with ecosystem expectations and resolve long standing issue #4795.

Tool calls (22)

#ToolArgumentsResult
1run_command
{
 "command": "grep -rn \"no_proxy\" src/requests | head -20"
}
{"status": "ok", "stdout": "src/requests/sessions.py:840: no_proxy = proxies.get(\"no_proxy\") if proxies is not None else None\nsrc/requests/sessions.py:841: e
full
{"status": "ok", "stdout": "src/requests/sessions.py:840:            no_proxy = proxies.get(\"no_proxy\") if proxies is not None else None\nsrc/requests/sessions.py:841:            env_proxies = get_environ_proxies(url, no_proxy=no_proxy)\nsrc/requests/utils.py:765:    Very simple check of the cidr format in no_proxy variable.\nsrc/requests/utils.py:810:def should_bypass_proxies(url: str, no_proxy: str | None) -> bool:\nsrc/requests/utils.py:822:    # First check whether no_proxy is defined. If it is, check that the URL\nsrc/requests/utils.py:823:    # we're getting isn't in the no_proxy list.\nsrc/requests/utils.py:824:    no_proxy_arg = no_proxy\nsrc/requests/utils.py:825:    if no_proxy is None:\nsrc/requests/utils.py:826:        no_proxy = get_proxy(\"no_proxy\")\nsrc/requests/utils.py:834:    if no_proxy:\nsrc/requests/utils.py:837:        no_proxy_hosts = (host for host in no_proxy.replace(\" \", \"\").split(\",\") if host)\nsrc/requests/utils.py:840:            for proxy_ip in no_proxy_hosts:\nsrc/requests/utils.py:845:                    # If no_proxy ip was defined in plain IP notation instead of cidr notation &\nsrc/requests/utils.py:853:            for host in no_proxy_hosts:\nsrc/requests/utils.py:855:                    # The URL does match something in no_proxy, so we don't want\nsrc/requests/utils.py:859:    with set_environ(\"no_proxy\", no_proxy_arg):\nsrc/requests/utils.py:871:def get_environ_proxies(url: str, no_proxy: str | None = None) -> dict[str, str]:\nsrc/requests/utils.py:877:    if should_bypass_proxies(url, no_proxy=no_proxy):\nsrc/requests/utils.py:927:    no_proxy = proxies.get(\"no_proxy\")\nsrc/requests/utils.py:930:    if trust_env and not should_bypass_proxies(url, no_proxy=no_proxy):\n", "exit_code": 0}
2read_file
{
 "filepath": "src/requests/utils.py",
 "start_line": 810,
 "end_line": 865
}
{"status": "ok", "filepath": "src/requests/utils.py", "content": "def should_bypass_proxies(url: str, no_proxy: str | None) -> bool:\n \"\"\"\n Returns whether
full
{"status": "ok", "filepath": "src/requests/utils.py", "content": "def should_bypass_proxies(url: str, no_proxy: str | None) -> bool:\n    \"\"\"\n    Returns whether we should bypass proxies or not.\n\n    :rtype: bool\n    \"\"\"\n\n    # Prioritize lowercase environment variables over uppercase\n    # to keep a consistent behaviour with other http projects (curl, wget).\n    def get_proxy(key: str) -> str | None:\n        return os.environ.get(key) or os.environ.get(key.upper())\n\n    # First check whether no_proxy is defined. If it is, check that the URL\n    # we're getting isn't in the no_proxy list.\n    no_proxy_arg = no_proxy\n    if no_proxy is None:\n        no_proxy = get_proxy(\"no_proxy\")\n    parsed = urlparse(url)\n    hostname = parsed.hostname\n\n    if hostname is None:\n        # URLs don't always have hostnames, e.g. file:/// urls.\n        return True\n\n    if no_proxy:\n        # We need to check whether we match here. We need to see if we match\n        # the end of the hostname, both with and without the port.\n        no_proxy_hosts = (host for host in no_proxy.replace(\" \", \"\").split(\",\") if host)\n\n        if is_ipv4_address(hostname):\n            for proxy_ip in no_proxy_hosts:\n                if is_valid_cidr(proxy_ip):\n                    if address_in_network(hostname, proxy_ip):\n                        return True\n                elif hostname == proxy_ip:\n                    # If no_proxy ip was defined in plain IP notation instead of cidr notation &\n                    # matches the IP of the index\n                    return True\n        else:\n            host_with_port = hostname\n            if parsed.port:\n                host_with_port += f\":{parsed.port}\"\n\n            for host in no_proxy_hosts:\n                if hostname.endswith(host) or host_with_port.endswith(host):\n                    # The URL does match something in no_proxy, so we don't want\n                    # to apply the proxies on this
3run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(should_bypass_proxies('http://api.example.com', 'example.com'))\""
}
{"status": "ok", "stdout": "True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "True\n", "exit_code": 0}
4run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(should_bypass_proxies('http://notexample.com', 'example.com'))\""
}
{"status": "ok", "stdout": "True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "True\n", "exit_code": 0}
5run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(should_bypass_proxies('http://api.example.com', '.example.com'))\""
}
{"status": "ok", "stdout": "True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "True\n", "exit_code": 0}
6run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(should_bypass_proxies('http://api.example.com', 'example.com'))\""
}
{"status": "ok", "stdout": "True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "True\n", "exit_code": 0}
7run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(f'api.example.com matches example.com: {should_bypass_proxies(\\\"http://api.example.com\\\", \\\"example.com\\\")}')\""
}
{"status": "ok", "stdout": "api.example.com matches example.com: True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "api.example.com matches example.com: True\n", "exit_code": 0}
8run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(f'notexample.com matches example.com: {should_bypass_proxies(\\\"http://notexample.com\\\", \\\"example.com\\\")}')\""
}
{"status": "ok", "stdout": "notexample.com matches example.com: True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "notexample.com matches example.com: True\n", "exit_code": 0}
9read_file
{
 "end_line": 865,
 "filepath": "src/requests/utils.py",
 "start_line": 810
}
{"status": "ok", "filepath": "src/requests/utils.py", "content": "def should_bypass_proxies(url: str, no_proxy: str | None) -> bool:\n \"\"\"\n Returns whether
full
{"status": "ok", "filepath": "src/requests/utils.py", "content": "def should_bypass_proxies(url: str, no_proxy: str | None) -> bool:\n    \"\"\"\n    Returns whether we should bypass proxies or not.\n\n    :rtype: bool\n    \"\"\"\n\n    # Prioritize lowercase environment variables over uppercase\n    # to keep a consistent behaviour with other http projects (curl, wget).\n    def get_proxy(key: str) -> str | None:\n        return os.environ.get(key) or os.environ.get(key.upper())\n\n    # First check whether no_proxy is defined. If it is, check that the URL\n    # we're getting isn't in the no_proxy list.\n    no_proxy_arg = no_proxy\n    if no_proxy is None:\n        no_proxy = get_proxy(\"no_proxy\")\n    parsed = urlparse(url)\n    hostname = parsed.hostname\n\n    if hostname is None:\n        # URLs don't always have hostnames, e.g. file:/// urls.\n        return True\n\n    if no_proxy:\n        # We need to check whether we match here. We need to see if we match\n        # the end of the hostname, both with and without the port.\n        no_proxy_hosts = (host for host in no_proxy.replace(\" \", \"\").split(\",\") if host)\n\n        if is_ipv4_address(hostname):\n            for proxy_ip in no_proxy_hosts:\n                if is_valid_cidr(proxy_ip):\n                    if address_in_network(hostname, proxy_ip):\n                        return True\n                elif hostname == proxy_ip:\n                    # If no_proxy ip was defined in plain IP notation instead of cidr notation &\n                    # matches the IP of the index\n                    return True\n        else:\n            host_with_port = hostname\n            if parsed.port:\n                host_with_port += f\":{parsed.port}\"\n\n            for host in no_proxy_hosts:\n                if hostname.endswith(host) or host_with_port.endswith(host):\n                    # The URL does match something in no_proxy, so we don't want\n                    # to apply the proxies on this
10edit_file
{
 "filepath": "src/requests/utils.py",
 "new_string": "            for host in no_proxy_hosts:\n                if hostname == host or hostname.endswith(\".\" + host) or host_with_port == host or host_with_port.endswith(\".\" + host):\n                    # The URL does match something in no_proxy, so we don't want\n                    # to apply the proxies on this URL.\n                    return True",
 "old_string": "            for host in no_proxy_hosts:\n                if hostname.endswith(host) or host_with_port.endswith(host):\n                    # The URL does match something in no_proxy, so we don't want\n                    # to apply the proxies on this URL.\n                    return True"
}
{"status": "ok", "filepath": "src/requests/utils.py", "occurrences": 1, "strategy": "exact", "diff": "--- a/src/requests/utils.py\n+++ b/src/requests/utils.py\n
full
{"status": "ok", "filepath": "src/requests/utils.py", "occurrences": 1, "strategy": "exact", "diff": "--- a/src/requests/utils.py\n+++ b/src/requests/utils.py\n@@ -851,7 +851,7 @@\n                 host_with_port += f\":{parsed.port}\"\n \n             for host in no_proxy_hosts:\n-                if hostname.endswith(host) or host_with_port.endswith(host):\n+                if hostname == host or hostname.endswith(\".\" + host) or host_with_port == host or host_with_port.endswith(\".\" + host):\n                     # The URL does match something in no_proxy, so we don't want\n                     # to apply the proxies on this URL.\n                     return True\n", "is_truncated": false}
11run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(f'api.example.com matches example.com: {should_bypass_proxies(\\\"http://api.example.com\\\", \\\"example.com\\\")}')\""
}
{"status": "ok", "stdout": "api.example.com matches example.com: True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "api.example.com matches example.com: True\n", "exit_code": 0}
12run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(f'notexample.com matches example.com: {should_bypass_proxies(\\\"http://notexample.com\\\", \\\"example.com\\\")}')\""
}
{"status": "ok", "stdout": "notexample.com matches example.com: False\n", "exit_code": 0}
full
{"status": "ok", "stdout": "notexample.com matches example.com: False\n", "exit_code": 0}
13run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(f'example.com matches example.com: {should_bypass_proxies(\\\"http://example.com\\\", \\\"example.com\\\")}')\""
}
{"status": "ok", "stdout": "example.com matches example.com: True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "example.com matches example.com: True\n", "exit_code": 0}
14-16 ×3run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(f'api.example.com matches .example.com: {should_bypass_proxies(\\\"http://api.example.com\\\", \\\".example.com\\\")}')\""
}
{"status": "ok", "stdout": "api.example.com matches .example.com: True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "api.example.com matches .example.com: True\n", "exit_code": 0}
17read_file
{
 "end_line": 865,
 "filepath": "src/requests/utils.py",
 "start_line": 810
}
{"status": "ok", "filepath": "src/requests/utils.py", "content": "def should_bypass_proxies(url: str, no_proxy: str | None) -> bool:\n \"\"\"\n Returns whether
full
{"status": "ok", "filepath": "src/requests/utils.py", "content": "def should_bypass_proxies(url: str, no_proxy: str | None) -> bool:\n    \"\"\"\n    Returns whether we should bypass proxies or not.\n\n    :rtype: bool\n    \"\"\"\n\n    # Prioritize lowercase environment variables over uppercase\n    # to keep a consistent behaviour with other http projects (curl, wget).\n    def get_proxy(key: str) -> str | None:\n        return os.environ.get(key) or os.environ.get(key.upper())\n\n    # First check whether no_proxy is defined. If it is, check that the URL\n    # we're getting isn't in the no_proxy list.\n    no_proxy_arg = no_proxy\n    if no_proxy is None:\n        no_proxy = get_proxy(\"no_proxy\")\n    parsed = urlparse(url)\n    hostname = parsed.hostname\n\n    if hostname is None:\n        # URLs don't always have hostnames, e.g. file:/// urls.\n        return True\n\n    if no_proxy:\n        # We need to check whether we match here. We need to see if we match\n        # the end of the hostname, both with and without the port.\n        no_proxy_hosts = (host for host in no_proxy.replace(\" \", \"\").split(\",\") if host)\n\n        if is_ipv4_address(hostname):\n            for proxy_ip in no_proxy_hosts:\n                if is_valid_cidr(proxy_ip):\n                    if address_in_network(hostname, proxy_ip):\n                        return True\n                elif hostname == proxy_ip:\n                    # If no_proxy ip was defined in plain IP notation instead of cidr notation &\n                    # matches the IP of the index\n                    return True\n        else:\n            host_with_port = hostname\n            if parsed.port:\n                host_with_port += f\":{parsed.port}\"\n\n            for host in no_proxy_hosts:\n                if hostname == host or hostname.endswith(\".\" + host) or host_with_port == host or host_with_port.endswith(\".\" + host):\n                    # The URL does match something in no_proxy, so we d
18-19 ×2run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(f'api.example.com matches .example.com: {should_bypass_proxies(\\\"http://api.example.com\\\", \\\".example.com\\\")}')\""
}
{"status": "ok", "stdout": "api.example.com matches .example.com: True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "api.example.com matches .example.com: True\n", "exit_code": 0}
20edit_file
{
 "filepath": "src/requests/utils.py",
 "new_string": "            for host in no_proxy_hosts:\n                if host.startswith('.'):\n                    # If it starts with a dot, we just check if the hostname ends with it\n                    if hostname.endswith(host) or host_with_port.endswith(host):\n                        return True\n                elif hostname == host or hostname.endswith(\".\" + host) or host_with_port == host or host_with_port.endswith(\".\" + host):\n                    # The URL does match something in no_proxy, so we don't want\n                    # to apply the proxies on this URL.\n                    return True",
 "old_string": "            for host in no_proxy_hosts:\n                if hostname == host or hostname.endswith(\".\" + host) or host_with_port == host or host_with_port.endswith(\".\" + host):\n                    # The URL does match something in no_proxy, so we don't want\n                    # to apply the proxies on this URL.\n                    return True"
}
{"status": "ok", "filepath": "src/requests/utils.py", "occurrences": 1, "strategy": "exact", "diff": "--- a/src/requests/utils.py\n+++ b/src/requests/utils.py\n
full
{"status": "ok", "filepath": "src/requests/utils.py", "occurrences": 1, "strategy": "exact", "diff": "--- a/src/requests/utils.py\n+++ b/src/requests/utils.py\n@@ -851,7 +851,11 @@\n                 host_with_port += f\":{parsed.port}\"\n \n             for host in no_proxy_hosts:\n-                if hostname == host or hostname.endswith(\".\" + host) or host_with_port == host or host_with_port.endswith(\".\" + host):\n+                if host.startswith('.'):\n+                    # If it starts with a dot, we just check if the hostname ends with it\n+                    if hostname.endswith(host) or host_with_port.endswith(host):\n+                        return True\n+                elif hostname == host or hostname.endswith(\".\" + host) or host_with_port == host or host_with_port.endswith(\".\" + host):\n                     # The URL does match something in no_proxy, so we don't want\n                     # to apply the proxies on this URL.\n                     return True\n", "is_truncated": false}
21run_command
{
 "command": "python3 -c \"import requests; from requests.utils import should_bypass_proxies; print(f'api.example.com matches .example.com: {should_bypass_proxies(\\\"http://api.example.com\\\", \\\".example.com\\\")}')\""
}
{"status": "ok", "stdout": "api.example.com matches .example.com: True\n", "exit_code": 0}
full
{"status": "ok", "stdout": "api.example.com matches .example.com: True\n", "exit_code": 0}
22submit_patch
[]
{"status": "ok", "patch_size": 971, "files_changed": 1}
full
{"status": "ok", "patch_size": 971, "files_changed": 1}

Patch

diff --git a/src/requests/utils.py b/src/requests/utils.py
index e69ad0d7..2234ef14 100644
--- a/src/requests/utils.py
+++ b/src/requests/utils.py
@@ -851,7 +851,11 @@ def should_bypass_proxies(url: str, no_proxy: str | None) -> bool:
                 host_with_port += f":{parsed.port}"
 
             for host in no_proxy_hosts:
-                if hostname.endswith(host) or host_with_port.endswith(host):
+                if host.startswith('.'):
+                    # If it starts with a dot, we just check if the hostname ends with it
+                    if hostname.endswith(host) or host_with_port.endswith(host):
+                        return True
+                elif hostname == host or hostname.endswith("." + host) or host_with_port == host or host_with_port.endswith("." + host):
                     # The URL does match something in no_proxy, so we don't want
                     # to apply the proxies on this URL.
                     return True

Test output

show
..s..................................................................... [ 31%]
........................................................................ [ 62%]
....................................................................ssss [ 94%]
sssssss.....s                                                            [100%]
=============================== warnings summary ===============================
../../../../../../kaggle/tmp/envs/requests/lib/python3.13/site-packages/_pytest/config/__init__.py:1464
  /kaggle/tmp/envs/requests/lib/python3.13/site-packages/_pytest/config/__init__.py:1464: PytestConfigWarning: Unknown config option: timeout
  
    self._warn_or_fail_if_strict(f"Unknown config option: {key}\n")

tests/test_utils.py::TestContentEncodingDetection::test_none
tests/test_utils.py::TestContentEncodingDetection::test_pragmas[<meta charset="UTF-8">]
tests/test_utils.py::TestContentEncodingDetection::test_pragmas[<meta http-equiv="Content-type" content="text/html;charset=UTF-8">]
tests/test_utils.py::TestContentEncodingDetection::test_pragmas[<meta http-equiv="Content-type" content="text/html;charset=UTF-8" />]
tests/test_utils.py::TestContentEncodingDetection::test_pragmas[<?xml version="1.0" encoding="UTF-8"?>]
tests/test_utils.py::TestContentEncodingDetection::test_precedence
  /tmp/swe_work/eval4_submission_v1b/requests_7427/b/workspace/src/requests/utils.py:527: DeprecationWarning: In requests 3.0, get_encodings_from_content will be removed. For more information, please see the discussion on issue #2266. (This warning should only appear once.)
    warnings.warn(

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
216 passed, 13 skipped, 7 warnings in 0.24s
[2026-09-24 15:38:43,715] WARNING in core: flasgger is not installed; serving the static landing page at / and skipping the Swagger UI and /spec.json.