Fix llms-txt anchor links and breadcrumb leaking

Links with anchors (e.g. about.html#section) were not being converted
to .llms.md because the Lua pattern only matched .html at end-of-string.
Now also matches .html followed by # and rewrites both cases.

Breadcrumbs from sidebar navigation were leaking into .llms.md output.
Add quarto-page-breadcrumbs to droppable_classes in the Lua filter.

Adds test coverage for both fixes: anchor link conversion in both
directions, .html# negative matches, sidebar config to trigger
breadcrumbs, and breadcrumb text negative match.

Co-Authored-By: Claude Opus 4.6 <[email protected]>

Author: cscheid

src/resources/filters/llms/llms.lua

@@ -25,6 +25,7 @@ local droppable_classes = {
     ["listing-categories"] = true,
     ["quarto-listing-category"] = true,  -- category filter sidebar
     ["listing-category"] = true,  -- individual category badges
+    ["quarto-page-breadcrumbs"] = true,  -- breadcrumb navigation
 }
 local droppable_ids = {
     ["quarto-header"] = true,
@@ -140,7 +141,8 @@ function Link(link)
         return link.content
     end
 
-    if link.target and link.target:match("%.html$") then
+    if link.target and (link.target:match("%.html$") or link.target:match("%.html#")) then
+        link.target = link.target:gsub("%.html#", ".llms.md#")
         link.target = link.target:gsub("%.html$", ".llms.md")
         if link.classes:includes("btn") then
             link.attr = pandoc.Attr()

tests/docs/smoke-all/website/llms-txt/_quarto.yml

@@ -11,6 +11,11 @@ website:
       - href: index.qmd
         text: Home
       - about.qmd
+  sidebar:
+    contents:
+      - section: Info
+        contents:
+          - about.qmd
 
 format:
   html:

tests/docs/smoke-all/website/llms-txt/about.qmd

@@ -6,9 +6,9 @@ _quarto:
       ensureLlmsMdExists: true
       ensureLlmsMdRegexMatches:
         # First array: patterns that MUST match
-        - ["^# About", "> \\*\\*NOTE:\\*\\*", "> \\*\\*WARNING:\\*\\*", "This is a note", "``` python", "def hello", "\\| Feature", "\\|[-]+\\|", "\\[home page\\]\\(.*\\.llms\\.md\\)"]
-        # Second array: patterns that must NOT match (no .html links in llms.md)
-        - ["\\.html\\)"]
+        - ["^# About", "> \\*\\*NOTE:\\*\\*", "> \\*\\*WARNING:\\*\\*", "This is a note", "``` python", "def hello", "\\| Feature", "\\|[-]+\\|", "\\[home page\\]\\(.*\\.llms\\.md\\)", "\\[test site intro\\]\\(index\\.llms\\.md#test-content\\)"]
+        # Second array: patterns that must NOT match (no .html links, no breadcrumbs)
+        - ["\\.html\\)", "\\.html#", "\\[Info\\]"]
 ---
 
 About this test site.
@@ -42,3 +42,5 @@ def hello():
 ## Link Example
 
 Go back to the [home page](index.qmd).
+
+Go to the [test site intro](index.qmd#test-content).

tests/docs/smoke-all/website/llms-txt/index.qmd

@@ -11,8 +11,17 @@ _quarto:
         - ["^# llms-txt Test Site", "^## Pages", "\\[.*\\]\\(.*\\.llms\\.md\\)"]
         # Second array: patterns that must NOT match (empty)
         - []
+      ensureLlmsMdRegexMatches:
+        # First array: patterns that MUST match - verify anchor links are converted
+        - ["\\[callout examples\\]\\(about\\.llms\\.md#callout-examples\\)"]
+        # Second array: patterns that must NOT match (no .html or .html# links)
+        - ["\\.html\\)", "\\.html#"]
 ---
 
+## Test Content
+
 This is a test website for the llms-txt feature.
 
 See the [about page](about.qmd) for more information.
+
+Also see the [callout examples](about.qmd#callout-examples).