From 6822fdbfbc0ef53811a2c4a0b2abc953bc170ae7 Mon Sep 17 00:00:00 2001
From: 0xMarcio <marc@syntax9.ai>
Date: Mon, 31 Aug 2026 00:22:44 +0200
Subject: [PATCH] Fix three ways the index misread its own data

An advisory copied from a GitHub Security Advisory opens with "### Impact"
or "### Summary". The section parser treated any heading as a boundary, so
those descriptions ended themselves and reached the site empty. Only the
four headings an entry actually uses end a section now, and leading heading
markers are stripped when the text is rendered.

A trailing slash made exploit-db.com/exploits/41034 and .../41034/ two
different links, which put 53 duplicate rows in front of readers.

github.com also serves attachments and advisories, and 103 of those were
being parsed as owner/repo -- rendering "user-attachments / files" with an
empty star column instead of the PoC archive they point to.
---
 docs/logic.js          |  7 ++++++-
 scripts/build_site.py  | 18 +++++++++++++++---
 scripts/update_cves.py |  2 +-
 3 files changed, 22 insertions(+), 5 deletions(-)

diff --git a/docs/logic.js b/docs/logic.js
index 3d5887fa32f..29f2da09623 100644
--- a/docs/logic.js
+++ b/docs/logic.js
@@ -72,9 +72,14 @@ function shortAge(hours) {
   return Math.round(days / 365) + 'y';
 }
 
+// github.com serves more than repositories; an attachment or an advisory has
+// no owner and no stars, and reads as nonsense in an owner / repo row.
+const NOT_A_REPO = new Set(['user-attachments', 'advisories', 'security', 'orgs',
+  'apps', 'marketplace', 'sponsors', 'topics', 'collections', 'settings', 'notifications']);
+
 function repoFromUrl(url) {
   const match = /^https?:\/\/(?:www\.)?github\.com\/([^/#?]+)\/([^/#?]+)/i.exec(url || '');
-  if (!match) return null;
+  if (!match || NOT_A_REPO.has(match[1].toLowerCase())) return null;
   return { owner: match[1], repo: match[2].replace(/\.git$/i, '') };
 }
 
diff --git a/scripts/build_site.py b/scripts/build_site.py
index aa175fd01e3..ec4552b5a00 100644
--- a/scripts/build_site.py
+++ b/scripts/build_site.py
@@ -31,18 +31,30 @@ def load_blacklist() -> set[str]:
 def normalise_block(text: str) -> str:
     text = text.replace("\r\n", "\n")
     text = re.sub(r"\n{2,}", "\n", text.strip())
-    lines = [line.lstrip("- ").rstrip() for line in text.split("\n")]
+    lines = [
+        re.sub(r"^#{1,6}\s+", "", line.lstrip("- ")).rstrip()
+        for line in text.split("\n")
+    ]
     return "\n".join(line for line in lines if line)
 
 
+SECTION_HEADERS = ("### Description", "### POC", "#### Reference", "#### Github")
+
+
 def parse_sections(content: str) -> Dict[str, str]:
+    """Split an entry on its own headings only.
+
+    CVE text lifted from a GitHub advisory often opens with "### Impact" or
+    "### Summary"; treating any heading as a boundary let a description end
+    itself and vanish from the index.
+    """
     sections: Dict[str, str] = {}
     current_header: Optional[str] = None
     buffer: List[str] = []
 
     for line in content.splitlines():
         header = line.strip()
-        if header.startswith("### ") or header.startswith("#### "):
+        if header in SECTION_HEADERS:
             if current_header is not None:
                 sections[current_header] = "\n".join(buffer).strip()
             current_header = header
@@ -81,7 +93,7 @@ def is_blacklisted(url: str, blacklist: Collection[str]) -> bool:
 def link_key(url: str) -> str:
     parsed = urlparse(url)
     if (parsed.hostname or "").lower() not in {"github.com", "www.github.com"}:
-        return url
+        return url.rstrip("/")
     parts = [part for part in parsed.path.split("/") if part]
     if len(parts) < 2:
         return url
diff --git a/scripts/update_cves.py b/scripts/update_cves.py
index 711b9938fea..600f96c168e 100755
--- a/scripts/update_cves.py
+++ b/scripts/update_cves.py
@@ -116,7 +116,7 @@ def stable_unique(values: Iterable[str]) -> list[str]:
 def poc_link_key(url: str) -> str:
     parsed = urlparse(url)
     if (parsed.hostname or "").lower() not in {"github.com", "www.github.com"}:
-        return url
+        return url.rstrip("/")
     parts = [part for part in parsed.path.split("/") if part]
     if len(parts) < 2:
         return url