Su YijunandClaude Opus 5.5
2585d15d53
- 通知：14 种渠道与 EasySub 字段 / 实现一致（Telegram 代理与 getMe/getUpdates、飞书应用机器人、 QQ 机器人、企业微信自建应用 + API 代理 + 检查应用、Webhook 超时与重试），旧配置自动迁移 - 场景：家庭 / 企业，企业称“员工 / 部门”；在 系统设置 → 场景与网络 全局切换，本地缓存避免菜单闪错 - 识别：国内域名列表先于官方 IP 段；IP 段重叠按最长前缀匹配；自动更新开关覆盖 ASN 与补充库 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
109 lines
5.3 KiB
Python
109 lines
5.3 KiB
Python
"""debug12：补充识别库（国内域名列表 + 官方 IP 段）——只做兜底，不让任何已有识别变差。"""
|
|
import json
|
|
|
|
import pytest
|
|
|
|
from app.classify import Classifier
|
|
from app.classify.sources import sources as src
|
|
|
|
|
|
@pytest.fixture
|
|
def loaded_sources(tmp_path, monkeypatch):
|
|
"""用小份样本数据驱动 sources（不访问网络）：国内域名 + Google/GitHub/国内 IP 段。"""
|
|
(tmp_path / "china-domains.conf").write_text(
|
|
"# 样本\nserver=/example-test.cn/114.114.114.114\nserver=/gov-example.gov.cn/114\nbad line\n", encoding="utf-8")
|
|
(tmp_path / "ip-google.json").write_text(json.dumps(
|
|
{"prefixes": [{"ipv4Prefix": "192.0.2.0/24"}, {"ipv6Prefix": "2001:db8::/32"}]}), encoding="utf-8")
|
|
(tmp_path / "ip-github.json").write_text(json.dumps(
|
|
{"git": ["203.0.113.0/25"], "web": ["203.0.113.128/25"]}), encoding="utf-8")
|
|
(tmp_path / "ip-cn.txt").write_text("198.51.100.0/24\n# comment\n", encoding="utf-8")
|
|
monkeypatch.setenv("DATA_DIR", str(tmp_path))
|
|
monkeypatch.setattr("app.classify.sources._enabled", lambda sid: True)
|
|
src.reset()
|
|
src.load()
|
|
yield src
|
|
src.reset()
|
|
|
|
|
|
def test_china_list_fallback(loaded_sources):
|
|
"""域名规则没认出、但在国内域名列表里 → 国内网站；子域名同样命中；列表外仍是通用 HTTPS。"""
|
|
c = Classifier()
|
|
assert c.classify("93.184.216.34", 6, 443, "www.example-test.cn", None) == "cn_web"
|
|
assert c.classify("93.184.216.34", 6, 443, "deep.sub.example-test.cn", None) == "cn_web"
|
|
assert c.classify("93.184.216.34", 6, 443, "not-in-list.example.org", None) == "https"
|
|
# 已有规则的应用不被兜底覆盖
|
|
assert c.classify("93.184.216.34", 6, 443, "www.bilibili.com", None) == "bilibili"
|
|
|
|
|
|
def test_official_ip_fallback(loaded_sources):
|
|
"""没有域名的流量按官方 IP 段认出归属（测试里 ASN 库是关的，正好只验证这一层）。"""
|
|
c = Classifier()
|
|
assert c.classify("192.0.2.1", 6, 443, None, None) == "google"
|
|
assert c.classify("203.0.113.10", 6, 443, None, None) == "github"
|
|
assert c.classify("198.51.100.7", 6, 443, None, None) == "cn_net"
|
|
assert c.classify("198.51.100.7", 17, 443, None, None) == "cn_net"
|
|
# 官方 IP 段认出的也在库外：测试段外的 IP 仍走通用端口
|
|
assert c.classify("45.155.99.10", 6, 443, None, None) == "https"
|
|
# 域名优先于 IP：IP 在 Google 段里但域名是别的应用
|
|
assert c.classify("192.0.2.1", 6, 443, "store.steampowered.com", None) == "steam"
|
|
# Apple 静态段（IANA 注册）
|
|
assert c.classify("17.0.0.1", 6, 443, None, None) == "apple"
|
|
|
|
|
|
def test_sources_counted_as_identified(loaded_sources):
|
|
"""国内网站 / 国内网络不计入「未识别」，识别率计算把它们算已识别。"""
|
|
from app.classify import UNIDENTIFIED_CATEGORIES
|
|
from app import stats
|
|
c = Classifier()
|
|
for app_id, ip, qname in (("cn_web", "93.184.216.34", "www.example-test.cn"), ("cn_net", "198.51.100.7", None)):
|
|
hit = c.classify(ip, 6, 443, qname, None)
|
|
assert c.app_info(hit).category not in UNIDENTIFIED_CATEGORIES
|
|
assert c.app_info(hit).name_en
|
|
apps = [{**stats.app_meta(a), "total": 1} for a in ("cn_web", "cn_net")]
|
|
assert stats.identified_ratio(apps) == 1.0
|
|
|
|
|
|
def test_sources_rows_for_library_panel(loaded_sources):
|
|
rows = {r["id"]: r for r in src.rows()}
|
|
assert rows["cnlist"]["domains"] == 2
|
|
assert rows["google"]["ranges"] == 2
|
|
assert rows["github"]["ranges"] == 2
|
|
assert rows["cn"]["ranges"] == 1
|
|
assert all(r["license"] and r["url"] for r in rows.values())
|
|
|
|
|
|
def test_domain_fallback_before_ip_fallback(loaded_sources):
|
|
"""有域名时先按域名兜底：国内网站托管在 Google/国内段的 IP 上，仍归「国内网站」而不是 IP 所属的 CDN。"""
|
|
c = Classifier()
|
|
assert c.classify("192.0.2.1", 6, 443, "www.example-test.cn", None) == "cn_web"
|
|
assert c.classify("198.51.100.7", 6, 443, "www.example-test.cn", None) == "cn_web"
|
|
# 域名不在任何库里时才按 IP
|
|
assert c.classify("192.0.2.1", 6, 443, "not-in-list.example.org", None) == "google"
|
|
|
|
|
|
def test_overlapping_ranges_longest_prefix():
|
|
"""IP 段重叠时按最长前缀：小段赢；小段之外仍落在大段里（旧实现在这里会漏查返回 None）。"""
|
|
from app.classify.sources import _Ranges
|
|
r = _Ranges()
|
|
r.add("10.0.0.0/8", "cn_net") # 先加入的大段
|
|
r.add("10.1.0.0/16", "aws") # 后加入、更具体
|
|
r.add("10.1.2.0/24", "github") # 再嵌套一层
|
|
r.add("10.200.0.0/16", "cn_net") # 同归属的嵌套，合并后不影响
|
|
r.add("2001:db8::/32", "google")
|
|
r.add("2001:db8:1::/48", "cloudflare")
|
|
r.build()
|
|
assert r.lookup("10.1.2.3") == "github"
|
|
assert r.lookup("10.1.3.1") == "aws"
|
|
assert r.lookup("10.2.0.1") == "cn_net" # 在小段后面、仍属于大段
|
|
assert r.lookup("10.255.255.255") == "cn_net"
|
|
assert r.lookup("10.0.0.0") == "cn_net"
|
|
assert r.lookup("11.0.0.1") is None
|
|
assert r.lookup("2001:db8:1::5") == "cloudflare"
|
|
assert r.lookup("2001:db8:2::5") == "google"
|
|
# 一样大时先加入的赢
|
|
r2 = _Ranges()
|
|
r2.add("192.0.2.0/24", "google")
|
|
r2.add("192.0.2.0/24", "aws")
|
|
r2.build()
|
|
assert r2.lookup("192.0.2.9") == "google"
|