fix(posting-guard): read style values like a browser and treat a collapsed, clipped box as hidden (#709)

* fix(posting-guard): read style values like a browser and treat a collapsed, clipped box as hidden

The hidden-style regex ran on the raw attribute value, so
"display:none" (entity-encoded colon) and "display:/**/none" (a CSS
comment inside the declaration) both hid the content in a browser while
the guard showed it under a clean verdict. A zero-height or zero-width
box with overflow:hidden, the other common way to hide a block, was not
a hidden style at all.

Entity-decode and comment-strip the value before matching, and count a
collapsed box as hidden only when both halves are present: overflow:
hidden alone is on most cards and height:0 alone is a spacer. Three new
Bypasses cases (seven subtests fail on master) plus a calibration case
pinning that either half alone stays visible.

* changelog: place the #709 entry under [Unreleased]

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Mads Lorentzen <madslorentzen_17@hotmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

CHANGELOG.md

@@ -415,6 +415,17 @@ per-file diff commands.
   lines, six of which fail on master, and ten prose lines with the same words that must stay
   clean.
 
+- **`posting_guard.py` reads a style attribute the way a browser does and treats a collapsed,
+  clipped box as hidden** (#709, `tools/posting_guard.py`, `tests/test_posting_guard.py`) - the hidden-style
+  regex ran on the raw attribute value, so `display:none` (entity-encoded colon) and
+  `display:/**/none` (a CSS comment inside the declaration) both hid the content in a browser
+  while the guard showed it under a clean verdict; and a zero-height or zero-width box with
+  `overflow:hidden`, the other common way to hide a block, was not a hidden style at all.
+  The value is now entity-decoded and comment-stripped before matching, and a collapsed box
+  counts as hidden only when both halves are present, because `overflow:hidden` alone is on
+  most cards and `height:0` alone is a spacer. Three new `Bypasses` cases, seven subtests
+  failing on master, plus a calibration case pinning that either half alone stays visible.
+
 ### Security
 
 - **CI no longer leaves the job token in `.git/config` or expands a PR-controlled name into a

tests/test_posting_guard.py

@@ -453,6 +453,28 @@ class Bypasses(GuardFixture):
             with self.subTest(style=style):
                 self.assert_hidden(f"<p>Real.</p><span style='{style}'>SECRET</span>", "SECRET")
 
+    def test_entity_encoded_and_comment_broken_style_values_still_hide(self):
+        # A browser decodes the attribute and ignores /**/ inside a declaration
+        # before it parses the CSS; the regex saw neither and showed the text
+        # under a clean verdict.
+        for style in ("display&#58;none", "display:/**/none", "visibility&#x3a;/* */hidden", "opacity:/*x*/0"):
+            with self.subTest(style=style):
+                self.assert_hidden(f"<p>Real.</p><div style=\"{style}\">SECRET</div>", "SECRET")
+
+    def test_collapsed_box_with_clipped_overflow_hides(self):
+        for style in ("height:0;overflow:hidden", "max-height: 0px; overflow: hidden", "width:0;overflow-x:hidden"):
+            with self.subTest(style=style):
+                self.assert_hidden(f"<p>Real.</p><div style='{style}'>SECRET</div>", "SECRET")
+
+    def test_either_half_of_a_collapsed_box_alone_is_layout_not_hiding(self):
+        # overflow:hidden is on most cards and height:0 is a spacer; neither
+        # alone hides text, and treating them as hidden would flag every page.
+        for style in ("overflow:hidden", "height:0", "height:0.5em;overflow:hidden", "max-height:200px;overflow:hidden"):
+            with self.subTest(style=style):
+                code, out, err = self.run_main("--html", "--json", stdin_text=f"<p>Real.</p><div style='{style}'>Visible.</div>")
+                self.assertIn("Visible.", self.text_of(out), style)
+                self.assertEqual(self.findings_of(out)["hidden_blocks"], [], style)
+
     def test_doctype_and_xml_prolog_are_not_text(self):
         code, out, err = self.run_main("--html", stdin_text="<!DOCTYPE html><?xml version='1.0'?><p>Real.</p>")
         text = self.text_of(out)

tools/posting_guard.py

@@ -8,8 +8,10 @@ makes the obvious payload classes visible deterministically, before Step 1
 evaluates anything:
 
   * hidden HTML (display:none, visibility:hidden, opacity:0, font-size:0,
-    color:transparent, clip/clip-path, off-screen positioning, the `hidden`
-    attribute, aria-hidden) - removed from the text; its content is scanned,
+    color:transparent, clip/clip-path, off-screen positioning, a zero-height
+    or zero-width box with overflow:hidden, the `hidden` attribute,
+    aria-hidden; the style value is read the way a browser reads it, with
+    entities decoded and CSS comments removed) - removed from the text; its content is scanned,
     and reported as excerpts under `hidden_blocks`. Comments and
     script/style/noscript/svg/template blocks are removed and scanned too but
     not listed (every page has them);
@@ -202,11 +204,32 @@ def _attrs(raw: str) -> dict:
     return out
 
 
+CSS_COMMENT = re.compile(r"/\*.*?\*/", re.S)
+# A box collapsed to nothing with its overflow clipped hides its content as
+# surely as display:none. Either half alone is everyday layout CSS (overflow:
+# hidden is on most cards; height:0 is a spacer), so both must be present.
+COLLAPSED_BOX = re.compile(r"\b(?:max-)?(?:height|width)\s*:\s*0(?:\.0+)?(?:px|em|rem|%|pt)?(?![\d.])", re.I)
+CLIPPED = re.compile(r"\boverflow(?:-[xy])?\s*:\s*hidden\b", re.I)
+
+
+def _style(attrs: dict) -> str:
+    """The style attribute as a browser reads it: entities decoded, CSS comments gone.
+
+    A browser decodes `display:none` before parsing the CSS and ignores
+    `/**/` inside a declaration, so both hid the content while the regex saw
+    neither. Normalise before matching.
+    """
+    return CSS_COMMENT.sub("", html.unescape(attrs.get("style", "")))
+
+
 def _is_hidden(raw_attrs: str) -> bool:
     attrs = _attrs(raw_attrs)
     if "hidden" in attrs or attrs.get("aria-hidden", "").strip().lower() == "true":
         return True
-    return bool(HIDDEN_STYLE.search(attrs.get("style", "")))
+    style = _style(attrs)
+    if HIDDEN_STYLE.search(style):
+        return True
+    return bool(COLLAPSED_BOX.search(style) and CLIPPED.search(style))
 
 
 def clean_html(markup: str):