From 4a2aed5bcf7e2ab26ea5e8ecf15e6adb3f87b628 Mon Sep 17 00:00:00 2001
From: Jason Streifling <jason@streifling.com>
Date: Tue, 4 Feb 2025 17:48:54 +0100
Subject: [PATCH] Add possibilty to create table of contents for docx improted
 articles

---
 cmd/backend/create_toc.lua | 33 +++++++++++++++++++++++++++++++++
 cmd/backend/docx.go        |  5 ++++-
 2 files changed, 37 insertions(+), 1 deletion(-)
 create mode 100644 cmd/backend/create_toc.lua

diff --git a/cmd/backend/create_toc.lua b/cmd/backend/create_toc.lua
new file mode 100644
index 0000000..f08bf15
--- /dev/null
+++ b/cmd/backend/create_toc.lua
@@ -0,0 +1,33 @@
+-- Helper function: remove all image inlines from a list of inlines.
+local function remove_images(inlines)
+	local result = {}
+	for _, item in ipairs(inlines) do
+		if item.t ~= "Image" then
+			table.insert(result, item)
+		end
+	end
+	return result
+end
+
+-- Build a bullet list representing the table of contents.
+local function build_toc(doc)
+	local toc_items = {}
+	for _, block in ipairs(doc.blocks) do
+		if block.t == "Header" then
+			local clean_inlines = remove_images(block.content)
+			local header_text = pandoc.utils.stringify(clean_inlines)
+			if header_text ~= "" then
+				local link = pandoc.Link(clean_inlines, "#" .. block.identifier)
+				table.insert(toc_items, { link })
+			end
+		end
+	end
+	return pandoc.BulletList(toc_items)
+end
+
+-- The Pandoc function runs after the document is fully constructed.
+function Pandoc(doc)
+	local toc = build_toc(doc)
+	table.insert(doc.blocks, 1, toc) -- Insert the TOC at the very beginning of the document.
+	return doc
+end
diff --git a/cmd/backend/docx.go b/cmd/backend/docx.go
index 5413c6f..597356b 100644
--- a/cmd/backend/docx.go
+++ b/cmd/backend/docx.go
@@ -21,7 +21,7 @@ func ConvertToMarkdown(c *Config, filename string) ([]byte, error) {
 	defer os.RemoveAll(tmpDir)
 
 	articleFileName := filepath.Join(os.TempDir(), fmt.Sprint(uuid.New(), ".md"))
-	cmd := exec.Command("pandoc", "-s", "-f", "docx", "-t", "commonmark_x", "-o", articleFileName, "--extract-media", tmpDir, filename) // TODO: Is writing to a file necessary?
+	cmd := exec.Command("pandoc", "-s", "--lua-filter=cmd/backend/create_toc.lua", "-f", "docx", "-t", "commonmark_x", "-o", articleFileName, "--extract-media", tmpDir, filename) // TODO: Is writing to a file necessary?
 	cmd.Stderr = &stderr
 	if err = cmd.Run(); err != nil {
 		return nil, fmt.Errorf("error converting docx to markdown: %v: %v", err, stderr.String())
@@ -33,6 +33,9 @@ func ConvertToMarkdown(c *Config, filename string) ([]byte, error) {
 		return nil, fmt.Errorf("error reading markdown file: %v", err)
 	}
 
+	re := regexp.MustCompile(`\{width=[^}]+height=[^}]+\}`)
+	articleContent = re.ReplaceAll(articleContent, []byte(""))
+
 	imageNames, err := filepath.Glob(filepath.Join(tmpDir, "media", "*"))
 	if err != nil {
 		return nil, fmt.Errorf("error getting docx images from temporary directory: %v", err)