feat: 实现 Pandoc DOCX 转换服务

2026-07-30 15:05:50 +08:00
parent fa159d916f
commit bf43433d38
32 changed files with 2908 additions and 61 deletions
@@ -0,0 +1,110 @@
+local map_path = os.getenv("MD_TO_PDF_DOCX_MEDIA_MAP")
+
+if map_path == nil or map_path == "" then
+  error("DOCX media map path is missing")
+end
+
+local map_file, open_error = io.open(map_path, "rb")
+if map_file == nil then
+  error("DOCX media map cannot be opened: " .. tostring(open_error))
+end
+
+local map_content = map_file:read("*a")
+map_file:close()
+
+local media_map = pandoc.json.decode(map_content, false)
+local counters = {
+  image = 0,
+  mermaid = 0,
+  echarts = 0
+}
+
+local function contains_class(classes, expected)
+  for _, class_name in ipairs(classes) do
+    if class_name == expected then
+      return true
+    end
+  end
+  return false
+end
+
+local function next_media(kind)
+  counters[kind] = counters[kind] + 1
+  local media = media_map[kind][counters[kind]]
+  if media == nil then
+    error(
+      "DOCX media map does not contain "
+        .. kind
+        .. " #"
+        .. tostring(counters[kind])
+    )
+  end
+  return media
+end
+
+local function media_image(media)
+  return pandoc.Image(
+    pandoc.Inlines { pandoc.Str(media.alt_text) },
+    media.path,
+    "",
+    {
+      width = media.width,
+      height = media.height
+    }
+  )
+end
+
+local function replace_image(image)
+  local media = next_media("image")
+  image.src = media.path
+  image.caption = pandoc.Inlines { pandoc.Str(media.alt_text) }
+  image.attributes.width = media.width
+  image.attributes.height = media.height
+  return image
+end
+
+local function replace_code_block(block)
+  local kind
+  if contains_class(block.classes, "mermaid") then
+    kind = "mermaid"
+  elseif contains_class(block.classes, "echarts") then
+    kind = "echarts"
+  else
+    return block
+  end
+
+  local media = next_media(kind)
+  local image = media_image(media)
+  if media.caption ~= nil and media.caption ~= "" then
+    return pandoc.Figure(
+      { pandoc.Plain { image } },
+      pandoc.Caption {
+        pandoc.Plain { pandoc.Str(media.caption) }
+      }
+    )
+  end
+  return pandoc.Para { image }
+end
+
+local function validate_counts()
+  for _, kind in ipairs({ "image", "mermaid", "echarts" }) do
+    if counters[kind] ~= #media_map[kind] then
+      error(
+        "DOCX media map contains unused "
+          .. kind
+          .. " resources"
+      )
+    end
+  end
+end
+
+return {
+  Pandoc = function(document)
+    local transformed = document:walk {
+      Image = replace_image,
+      CodeBlock = replace_code_block
+    }
+    validate_counts()
+    return transformed
+  end
+}