From de4b0199ce9e0b4175221e36b162a29f803650cc Mon Sep 17 00:00:00 2001
From: kazet <krzysztof.zajac@cert.pl>
Date: Wed, 3 Jan 2024 09:45:04 +0100
Subject: [PATCH] Detecting duplicate with parent has been moved to scanning
 stage to speed up report generation (#65)

---
 autoreporter_addons/ssl_checks/reporter.py | 19 +-----------------
 karton_ssl_checks/karton_ssl_checks.py     | 23 ++++++++++++++++++++++
 2 files changed, 24 insertions(+), 18 deletions(-)

diff --git a/autoreporter_addons/ssl_checks/reporter.py b/autoreporter_addons/ssl_checks/reporter.py
index 4d7ac8f..d82f4db 100644
--- a/autoreporter_addons/ssl_checks/reporter.py
+++ b/autoreporter_addons/ssl_checks/reporter.py
@@ -1,4 +1,3 @@
-from difflib import SequenceMatcher
 from pathlib import Path
 from typing import Any, Callable, Dict, List
 from urllib.parse import urlparse
@@ -14,7 +13,7 @@
 from artemis.reporting.base.report_type import ReportType
 from artemis.reporting.base.reporter import Reporter
 from artemis.reporting.base.templating import ReportEmailTemplateFragment
-from artemis.reporting.utils import cached_get, get_top_level_target
+from artemis.reporting.utils import get_top_level_target
 from bs4 import BeautifulSoup
 
 from extra_modules_config import ExtraModulesConfig
@@ -61,22 +60,6 @@ def create_reports(task_result: Dict[str, Any], language: Language) -> List[Repo
         if not isinstance(result, dict):
             return []
 
-        try:
-            response = cached_get(f"https://{domain}")
-            parent_response = cached_get(f"https://{'.'.join(domain_parts[1:])}")
-            if SequenceMatcher(None, response.content, parent_response.content).quick_ratio() >= 0.8:
-                # Do not report misconfigurations if a domain has identical content to a parent domain - e.g.
-                # if we have mail.domain.com with identical content to domain.com, we assume that it's domain.com
-                # which is actually used, and therefore don't report subdomains.
-                return []
-        except Exception:
-            logger.warning(
-                f"Unable to check whether domain {domain} has similar content to parent domain. Artemis SSL check "
-                "module tries to reduce the number of false positives by skipping reports where domain has similar "
-                "content to parent domain, as there are cases where e.g. mail.example.com serves the same content "
-                "as example.com. If this fails, two similar reports may get sent."
-            )
-
         if "response_status_code" in result and "response_content_prefix" in result:
             response_status_code = result["response_status_code"]
             response_content_prefix = result["response_content_prefix"]
diff --git a/karton_ssl_checks/karton_ssl_checks.py b/karton_ssl_checks/karton_ssl_checks.py
index ddedbff..a3de32f 100644
--- a/karton_ssl_checks/karton_ssl_checks.py
+++ b/karton_ssl_checks/karton_ssl_checks.py
@@ -2,6 +2,7 @@
 import datetime
 import subprocess
 import urllib.parse
+from difflib import SequenceMatcher
 from typing import Any, Dict, List
 
 import requests
@@ -49,6 +50,28 @@ def run(self, current_task: Task) -> None:
             self.db.save_task_result(task=current_task, status=TaskStatus.OK)
             return
 
+        try:
+            response = http_requests.get(f"https://{domain}")
+            parent_domain = ".".join(domain_parts[1:])
+            parent_response = http_requests.get(f"https://{parent_domain}")
+            if SequenceMatcher(None, response.content, parent_response.content).quick_ratio() >= 0.8:
+                # Do not report misconfigurations if a domain has identical content to a parent domain - e.g.
+                # if we have mail.domain.com with identical content to domain.com, we assume that it's domain.com
+                # which is actually used, and therefore don't report subdomains.
+                self.db.save_task_result(
+                    task=current_task,
+                    status=TaskStatus.OK,
+                    status_reason=f"Detected that {domain} has similar content to {parent_domain}, not scanning to avoid duplicate reports",
+                )
+                return
+        except Exception:
+            self.log.exception(
+                f"Unable to check whether domain {domain} has similar content to parent domain. Artemis SSL check "
+                "module tries to reduce the number of false positives by skipping scanning domains when domain has "
+                "similar content to parent domain, as there are cases where e.g. mail.example.com serves the same "
+                "content as example.com.",
+            )
+
         messages = []
         result: Dict[str, Any] = {}