From bc1b0b7ec2ecd68638899aa286666ec310af56b7 Mon Sep 17 00:00:00 2001
From: Adrian <1917353+apetcu@users.noreply.github.com>
Date: Wed, 26 Aug 2026 23:01:40 +0300
Subject: [PATCH] fix(viewer): keep balanced nested emphasis working

Restricting emphasis content to non-delimiters made `**bold with *italic*
inside**` render its outer delimiters literally. Bold spans now only require
a non-delimiter at each end (so `******` still stays literal) and italic
content may contain complete <b>..</b> spans, which keeps nesting working
while still ruling out crossed tags.
---
 strix/interface/viewer/report_pdf.py | 15 ++++++++-------
 tests/test_report_pdf.py             |  6 ++++++
 2 files changed, 14 insertions(+), 7 deletions(-)

diff --git a/strix/interface/viewer/report_pdf.py b/strix/interface/viewer/report_pdf.py
index 358bd424..91436095 100644
--- a/strix/interface/viewer/report_pdf.py
+++ b/strix/interface/viewer/report_pdf.py
@@ -417,14 +417,15 @@ def _inline_md(text: str) -> str:
         codes.append(match.group(1))
         return f"\x00{len(codes) - 1}\x00"
 
-    # Emphasis content may not contain its own delimiter or a tag, so the three passes can
-    # never interleave into crossed markup (``<b><i>..</b></i>``), which reportlab rejects.
-    # A run of asterisks such as a masked secret (``******``) therefore stays literal.
+    # Bold spans must start and end with a non-delimiter, so a run of asterisks such as a masked
+    # secret (``******``) stays literal. Italic content is plain text or complete ``<b>..</b>``
+    # spans, so the passes can nest (``**a *b* c**``, ``*a **b** c*``) but can never interleave
+    # into crossed markup (``<b><i>..</b></i>``), which reportlab rejects.
     seg = html.escape(re.sub(r"`([^`]+)`", _stash, text))
-    seg = re.sub(r"\*\*\*([^*<>\n]+?)\*\*\*", r"<b><i>\1</i></b>", seg)
-    seg = re.sub(r"\*\*([^*<>\n]+?)\*\*", r"<b>\1</b>", seg)
-    seg = re.sub(r"__([^_<>\n]+?)__", r"<b>\1</b>", seg)
-    seg = re.sub(r"\*([^*<>\n]+?)\*", r"<i>\1</i>", seg)
+    seg = re.sub(r"\*\*\*(?=[^*])(.+?)(?<=[^*])\*\*\*", r"<b><i>\1</i></b>", seg)
+    seg = re.sub(r"\*\*(?=[^*])(.+?)(?<=[^*])\*\*", r"<b>\1</b>", seg)
+    seg = re.sub(r"__(?=[^_])(.+?)(?<=[^_])__", r"<b>\1</b>", seg)
+    seg = re.sub(r"\*((?:[^*<>\n]|<b>[^<>*\n]*</b>)+?)\*", r"<i>\1</i>", seg)
 
     def _restore(match: re.Match[str]) -> str:
         inner = html.escape(codes[int(match.group(1))])
diff --git a/tests/test_report_pdf.py b/tests/test_report_pdf.py
index c25b486b..35b103b0 100644
--- a/tests/test_report_pdf.py
+++ b/tests/test_report_pdf.py
@@ -156,3 +156,9 @@ def test_generate_report_pdf_with_masked_secret_in_summary(tmp_path: Path) -> No
 def test_para_falls_back_to_plain_text_on_crossed_markup() -> None:
     para = _para("<b><i>x</b></i> &amp; y", ParagraphStyle("t"))
     assert para.getPlainText() == "x & y"
+
+
+def test_inline_md_keeps_balanced_nested_emphasis() -> None:
+    assert _inline_md("**bold with *italic* inside**") == "<b>bold with <i>italic</i> inside</b>"
+    assert _inline_md("*outer **bold** inner*") == "<i>outer <b>bold</b> inner</i>"
+    assert "******" in _inline_md("(observed as '******')")  # a masked secret stays literal