Convert tabsets to headings with content in llms-txt output

Tabsets were rendered as a bullet list of empty links followed by
disconnected content. Add handle_tabset() to the Lua filter that
extracts tab titles from the nav BulletList and pairs them with
their tab-pane contents, outputting each as a heading + content.

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

Author: cscheid

src/resources/filters/llms/llms.lua

@@ -71,6 +71,53 @@ local function clean_element(el)
     end
 end
 
+local function handle_tabset(div)
+    local titles = {}
+    local panes = {}
+
+    -- Extract tab titles from the nav BulletList (first one in the div)
+    -- and tab pane contents from the tab-content Div
+    for _, block in ipairs(div.content) do
+        if block.t == "BulletList" and #titles == 0 then
+            for _, item in ipairs(block.content) do
+                for _, inner_block in ipairs(item) do
+                    if inner_block.t == "Plain" or inner_block.t == "Para" then
+                        for _, inline in ipairs(inner_block.content) do
+                            if inline.t == "Link" then
+                                table.insert(titles, inline.content)
+                                break
+                            end
+                        end
+                    end
+                end
+            end
+        elseif block.t == "Div" and block.classes:includes("tab-content") then
+            for _, inner in ipairs(block.content) do
+                if inner.t == "Div" and inner.classes:includes("tab-pane") then
+                    table.insert(panes, inner.content)
+                end
+            end
+        end
+    end
+
+    -- Build output: heading + content for each tab
+    local result = pandoc.Blocks({})
+    for i = 1, math.max(#titles, #panes) do
+        if titles[i] then
+            result:insert(pandoc.Header(2, titles[i]))
+        end
+        if panes[i] then
+            result:extend(panes[i])
+        end
+    end
+
+    if #result > 0 then
+        return result
+    end
+    -- Fallback: return content as-is
+    return div.content
+end
+
 local function handle_callout(div)
     local kind = "NOTE" -- NOTE, TIP, IMPORTANT, WARNING, CAUTION
     div.classes:map(function(cls)
@@ -176,6 +223,10 @@ end
 
 function Div(div)
 
+    if div.classes:includes("panel-tabset") then
+        return handle_tabset(div)
+    end
+
     if div.classes:includes("callout") then
         return handle_callout(div)
     end

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\\)", "\\[test site intro\\]\\(index\\.llms\\.md#test-content\\)"]
-        # Second array: patterns that must NOT match (no .html links, no breadcrumbs)
-        - ["\\.html\\)", "\\.html#", "\\[Info\\]"]
+        - ["^# About", "> \\*\\*NOTE:\\*\\*", "> \\*\\*WARNING:\\*\\*", "This is a note", "``` python", "def hello", "\\| Feature", "\\|[-]+\\|", "\\[home page\\]\\(.*\\.llms\\.md\\)", "\\[test site intro\\]\\(index\\.llms\\.md#test-content\\)", "## Alpha Tab", "Alpha content here", "## Beta Tab", "Beta content here"]
+        # Second array: patterns that must NOT match (no .html links, no breadcrumbs, no empty tab links)
+        - ["\\.html\\)", "\\.html#", "\\[Info\\]", "\\[Alpha Tab\\]\\(\\)", "\\[Beta Tab\\]\\(\\)"]
 ---
 
 About this test site.
@@ -39,6 +39,20 @@ def hello():
 | Code | Working |
 | Tables | Working |
 
+## Tabset Example
+
+::: {.panel-tabset}
+
+## Alpha Tab
+
+Alpha content here.
+
+## Beta Tab
+
+Beta content here.
+
+:::
+
 ## Link Example
 
 Go back to the [home page](index.qmd).