fix to make internal links work

.gitignore

@@ -0,0 +1,2 @@
+mets-combined.pdf
+mets-tmp/

chapter-builder.md

@@ -36,42 +36,6 @@
 ./METS-guide/mets_howtos.md
 ```
 
-``` {.include shift-heading-level-by=1}
-./METS-guide/howto/altRecordID.md
-./METS-guide/howto/agent.md
-./METS-guide/howto/md.md
-./METS-guide/howto/binary_marc.md
-./METS-guide/howto/premis.md
-./METS-guide/howto/source_metadata.md
-./METS-guide/howto/mods.md
-./METS-guide/howto/mets_rights.md
-./METS-guide/howto/technical_metadata.md
-./METS-guide/howto/mdRef.md
-./METS-guide/howto/mdWrap.md
-./METS-guide/howto/file.md
-./METS-guide/howto/fileSec_example_audio.md
-./METS-guide/howto/fileSec_example_images.md
-./METS-guide/howto/fileGrp.md
-./METS-guide/howto/transformFile.md
-./METS-guide/howto/stream.md
-./METS-guide/howto/FLocat.md
-./METS-guide/howto/FContent.md
-./METS-guide/howto/xmlData.md
-./METS-guide/howto/binData.md
-./METS-guide/howto/mptr.md
-./METS-guide/howto/fptr.md
-./METS-guide/howto/idref_namespaces.md
-./METS-guide/howto/area.md
-./METS-guide/howto/locref.md
-./METS-guide/howto/seq.md
-./METS-guide/howto/par.md
-./METS-guide/howto/external_idref.md
-./METS-guide/howto/idrefs_within_mets.md
-./METS-guide/howto/begin_end_betype.md
-./METS-guide/howto/id_idrefs.md
-./METS-guide/howto/div.md
-```
-
 ``` {.include}
 ./METS-guide/external_schema_vocabulary.md
 ./METS-guide/mets_profiles.md

dedupe-header-ids.lua

@@ -0,0 +1,30 @@
+-- Ensure header identifiers are unique across the merged document.
+-- Keep the original identifier if available, otherwise append a numeric suffix.
+
+local used = {}
+
+local function next_unique_id(base)
+  if not used[base] then
+    return base
+  end
+
+  local n = 2
+  local candidate = base .. '-' .. tostring(n)
+  while used[candidate] do
+    n = n + 1
+    candidate = base .. '-' .. tostring(n)
+  end
+  return candidate
+end
+
+function Header(h)
+  local id = h.identifier
+  if not id or id == '' then
+    return h
+  end
+
+  local unique = next_unique_id(id)
+  used[unique] = true
+  h.identifier = unique
+  return h
+end

expand-howtos.lua

@@ -0,0 +1,140 @@
+-- Expand linked how-to pages under each H2 in the METS How-Tos chapter.
+
+local stringify = (require 'pandoc.utils').stringify
+local List = require 'pandoc.List'
+
+local function slugify(text)
+  local s = text:lower()
+  s = s:gsub('[^%w%s-]', '')
+  s = s:gsub('%s+', '-')
+  s = s:gsub('%-+', '-')
+  s = s:gsub('^%-', ''):gsub('%-$', '')
+  return s
+end
+
+local function build_h2_index(index_items)
+  local list_items = List:new()
+  for _, item in ipairs(index_items) do
+    local link = pandoc.Link(item.title, '#' .. item.id)
+    list_items:insert({ pandoc.Plain({ link }) })
+  end
+
+  return List:new({
+    pandoc.Para({ pandoc.Strong({ pandoc.Str('Index') }) }),
+    pandoc.BulletList(list_items)
+  })
+end
+
+local function shift_headings(blocks, shift_by)
+  local filter = {
+    Header = function(h)
+      h.level = h.level + shift_by
+      return h
+    end
+  }
+  return pandoc.walk_block(pandoc.Div(blocks), filter).content
+end
+
+local function collect_links_from_bullet_list(bullet_list)
+  local links = List:new()
+
+  for _, item in ipairs(bullet_list.content) do
+    for _, blk in ipairs(item) do
+      if blk.t == 'Plain' or blk.t == 'Para' then
+        for _, inline in ipairs(blk.content) do
+          if inline.t == 'Link' and inline.target and inline.target:match('%.md$') then
+            links:insert(inline.target)
+          end
+        end
+      end
+    end
+  end
+
+  return links
+end
+
+local function read_markdown_file(path)
+  local fh = io.open(path)
+  if not fh then
+    io.stderr:write('Cannot open howto file ' .. path .. '\n')
+    return {}
+  end
+
+  local src = fh:read('*a')
+  fh:close()
+
+  local parsed = pandoc.read(src, 'markdown', PANDOC_READER_OPTIONS)
+  return parsed.blocks
+end
+
+function Pandoc(doc)
+  local blocks = doc.blocks
+  local out = List:new()
+
+  local howto_h2_index = List:new()
+  local collecting = false
+  for _, b in ipairs(blocks) do
+    if b.t == 'Header' and b.level == 1 then
+      collecting = (stringify(b.content) == 'METS How-Tos')
+    elseif collecting and b.t == 'Header' and b.level == 2 then
+      local title = stringify(b.content)
+      local id = b.identifier
+      if not id or id == '' then
+        id = slugify(title)
+      end
+      howto_h2_index:insert({ title = title, id = id })
+    elseif collecting and b.t == 'Header' and b.level == 1 then
+      collecting = false
+    end
+  end
+
+  local in_howtos = false
+  local index_inserted = false
+  local i = 1
+
+  while i <= #blocks do
+    local b = blocks[i]
+
+    if b.t == 'Header' and b.level == 1 then
+      in_howtos = (stringify(b.content) == 'METS How-Tos')
+      out:insert(b)
+      i = i + 1
+    elseif in_howtos and b.t == 'Header' and b.level == 2 then
+      if not index_inserted and #howto_h2_index > 0 then
+        out:extend(build_h2_index(howto_h2_index))
+        index_inserted = true
+      end
+
+      out:insert(b)
+
+      local j = i + 1
+      local section_links = List:new()
+
+      while j <= #blocks do
+        local nxt = blocks[j]
+        if nxt.t == 'Header' and nxt.level <= 2 then
+          break
+        end
+
+        out:insert(nxt)
+        if nxt.t == 'BulletList' then
+          section_links:extend(collect_links_from_bullet_list(nxt))
+        end
+        j = j + 1
+      end
+
+      for _, link_target in ipairs(section_links) do
+        local included = read_markdown_file(link_target)
+        out:extend(shift_headings(included, 2))
+      end
+
+      i = j
+    else
+      out:insert(b)
+      i = i + 1
+    end
+  end
+
+  doc.blocks = out
+  return doc
+end

fix-links.lua

@@ -1,14 +1,20 @@
 --- Created with using Google Gemini AI
 
 function Link(el)
-  if el.target:match("%.md$") or el.target:match("%.md#") then
-    local base = el.target:gsub("%.md", "")
-    if not base:match("^#") then
-      el.target = "#" .. base:gsub(".*/", "")
-    else
-      el.target = base
-    end
+  local path, fragment = el.target:match("^([^#]+)%.md#(.+)$")
+  if path and fragment then
+    -- Keep explicit section fragments from cross-file links.
+    el.target = "#" .. fragment
+    return el
   end
+
+  local only_path = el.target:match("^([^#]+)%.md$")
+  if only_path then
+    -- Resolve file links to the first-header anchor injected by add-anchor.lua.
+    el.target = "#" .. only_path:gsub(".*/", "")
+    return el
+  end
+
   return el
 end
 

md2pdf.sh

@@ -1,10 +1,11 @@
 #!/bin/bash
 SOFFICE_PATH='/Applications/LibreOffice.app/Contents/MacOS/soffice'
 
-mkdir ./mets-tmp
+mkdir -p ./mets-tmp
 find ./METS-guide -iname "*.md" -type f -exec sh -c 'pandoc --lua-filter=add-anchor.lua -f gfm+yaml_metadata_block -t markdown "${0}" -o "./mets-tmp/${0%}"' {} \;
-pandoc --lua-filter=include-files.lua --lua-filter=fix-links.lua -f markdown -t markdown chapter-builder.md > ./mets-tmp/mets-combined.md
+# Build from the preprocessed copies so inserted file anchors are present.
+sed 's#\./METS-guide/#./mets-tmp/./METS-guide/#g' chapter-builder.md | \
+	pandoc --lua-filter=include-files.lua --lua-filter=expand-howtos.lua --lua-filter=dedupe-header-ids.lua --lua-filter=fix-links.lua -f markdown -t markdown > ./mets-tmp/mets-combined.md
 pandoc --number-sections --reference-doc=template.docx -f markdown -t docx ./mets-tmp/mets-combined.md > ./mets-tmp/mets-combined.docx
 $SOFFICE_PATH --headless --convert-to pdf ./mets-tmp/mets-combined.docx
-rm -rf ./mets-tmp
-
+#rm -rf ./mets-tmp