Improve callouts and symbol publishing

scripts/pandoc-filters/semantic-blocks.lua

@@ -9,6 +9,58 @@ local env_types = {
   proof = true,
 }
 
+traverse = "topdown"
+
+local callout_labels = {
+  aside = "Aside",
+  caution = "Caution",
+  note = "Note",
+  tip = "Tip",
+  warning = "Warning",
+}
+
+local callout_envs = {
+  aside = "blogasidebox",
+  caution = "blogcautionbox",
+  note = "blognotebox",
+  tip = "blogtipbox",
+  warning = "blogwarningbox",
+}
+
+local callout_markers = {
+  { marker = "📝", key = "note" },
+  { marker = "💡", key = "tip" },
+  { marker = "⚡", key = "tip" },
+  { marker = "⚠️", key = "warning" },
+  { marker = "⚠", key = "warning" },
+}
+
+local chess_macros = {
+  ["♔"] = "\\BlogChessWhite{\\symking}",
+  ["♕"] = "\\BlogChessWhite{\\symqueen}",
+  ["♖"] = "\\BlogChessWhite{\\symrook}",
+  ["♗"] = "\\BlogChessWhite{\\symbishop}",
+  ["♘"] = "\\BlogChessWhite{\\symknight}",
+  ["♙"] = "\\BlogChessWhite{\\sympawn}",
+  ["♚"] = "\\BlogChessBlack{\\symking}",
+  ["♛"] = "\\BlogChessBlack{\\symqueen}",
+  ["♜"] = "\\BlogChessBlack{\\symrook}",
+  ["♝"] = "\\BlogChessBlack{\\symbishop}",
+  ["♞"] = "\\BlogChessBlack{\\symknight}",
+  ["♟"] = "\\BlogChessBlack{\\sympawn}",
+}
+
+local card_macros = {
+  ["♠"] = "\\BlogCardBlack{\\ensuremath{\\spadesuit}}",
+  ["♤"] = "\\BlogCardBlack{\\ensuremath{\\spadesuit}}",
+  ["♣"] = "\\BlogCardBlack{\\ensuremath{\\clubsuit}}",
+  ["♧"] = "\\BlogCardBlack{\\ensuremath{\\clubsuit}}",
+  ["♥"] = "\\BlogCardRed{\\ensuremath{\\heartsuit}}",
+  ["♡"] = "\\BlogCardRed{\\ensuremath{\\heartsuit}}",
+  ["♦"] = "\\BlogCardRed{\\ensuremath{\\diamondsuit}}",
+  ["♢"] = "\\BlogCardRed{\\ensuremath{\\diamondsuit}}",
+}
+
 local function has_class(el, class_name)
   for _, value in ipairs(el.classes) do
     if value == class_name then
@@ -27,6 +79,28 @@ local function first_env_class(el)
   return nil
 end
 
+local function normalize_label(value)
+  value = tostring(value or ""):lower()
+  value = value:gsub("^%s+", ""):gsub("%s+$", "")
+  value = value:gsub("^%[!", ""):gsub("%]$", "")
+  value = value:gsub(":$", "")
+  return value
+end
+
+local function inline_text(inline)
+  if not inline then
+    return ""
+  end
+  return pandoc.utils.stringify(inline)
+end
+
+local function first_inline_text(block)
+  if not block or block.t ~= "Para" or not block.content or #block.content == 0 then
+    return ""
+  end
+  return inline_text(block.content[1])
+end
+
 local function latex_escape(value)
   value = tostring(value or "")
   value = value:gsub("\\", "\\textbackslash{}")
@@ -36,6 +110,154 @@ local function latex_escape(value)
   return value
 end
 
+local function codepoint_at(value, index)
+  return utf8.codepoint(value, index, index)
+end
+
+local function codepoint_len(value, index)
+  local byte = string.byte(value, index)
+  if not byte then
+    return 0
+  elseif byte < 0x80 then
+    return 1
+  elseif byte < 0xE0 then
+    return 2
+  elseif byte < 0xF0 then
+    return 3
+  end
+  return 4
+end
+
+local function char_at(value, index)
+  local len = codepoint_len(value, index)
+  return value:sub(index, index + len - 1), index + len
+end
+
+local function is_variation_selector(cp)
+  return cp == 0xFE0E or cp == 0xFE0F
+end
+
+local function is_skin_tone(cp)
+  return cp >= 0x1F3FB and cp <= 0x1F3FF
+end
+
+local function is_emoji_base(cp)
+  return (cp >= 0x1F000 and cp <= 0x1FAFF)
+    or (cp >= 0x2600 and cp <= 0x27BF)
+    or cp == 0x00A9
+    or cp == 0x00AE
+    or cp == 0x203C
+    or cp == 0x2049
+    or cp == 0x2122
+    or cp == 0x2139
+    or cp == 0x3030
+    or cp == 0x303D
+    or cp == 0x3297
+    or cp == 0x3299
+end
+
+local function emoji_cluster(value, index)
+  local start = index
+  local current, next_index = char_at(value, index)
+  if current == "" then
+    return "", index
+  end
+
+  index = next_index
+
+  while index <= #value do
+    local cp = codepoint_at(value, index)
+    if is_variation_selector(cp) or is_skin_tone(cp) then
+      local _, ni = char_at(value, index)
+      index = ni
+    elseif cp == 0x200D then
+      local _, ni = char_at(value, index)
+      index = ni
+      if index <= #value then
+        local _, after_joined = char_at(value, index)
+        index = after_joined
+      end
+    else
+      break
+    end
+  end
+
+  return value:sub(start, index - 1), index
+end
+
+local function symbol_macro_for_cluster(cluster)
+  local cp = utf8.codepoint(cluster, 1, 1)
+  if not cp then
+    return nil
+  end
+
+  if chess_macros[cluster] then
+    return chess_macros[cluster]
+  end
+
+  if card_macros[cluster] then
+    return card_macros[cluster]
+  end
+
+  if cp >= 0x1F0A0 and cp <= 0x1F0FF then
+    return "\\BlogEmojiText{" .. cluster .. "}"
+  end
+
+  if is_emoji_base(cp) then
+    return "\\BlogEmojiText{" .. cluster .. "}"
+  end
+
+  return nil
+end
+
+local function symbol_str_to_latex(el)
+  if not FORMAT:match("latex") then
+    return nil
+  end
+
+  local text = el.text
+  local pieces = pandoc.List()
+  local plain = {}
+  local changed = false
+  local index = 1
+
+  local function flush_plain()
+    if #plain > 0 then
+      pieces:insert(pandoc.Str(table.concat(plain)))
+      plain = {}
+    end
+  end
+
+  while index <= #text do
+    local cp = codepoint_at(text, index)
+    if cp and (chess_macros[utf8.char(cp)] or card_macros[utf8.char(cp)] or is_emoji_base(cp)) then
+      local cluster, next_index = emoji_cluster(text, index)
+      local macro = symbol_macro_for_cluster(cluster)
+      if macro then
+        flush_plain()
+        pieces:insert(pandoc.RawInline("latex", macro))
+        changed = true
+        index = next_index
+      else
+        local char, ni = char_at(text, index)
+        plain[#plain + 1] = char
+        index = ni
+      end
+    else
+      local char, ni = char_at(text, index)
+      plain[#plain + 1] = char
+      index = ni
+    end
+  end
+
+  flush_plain()
+
+  if changed then
+    return pieces
+  end
+  return nil
+end
+
 local function internal_link_to_cref(el)
   if FORMAT:match("latex") and el.target:sub(1, 1) == "#" then
     local label = el.target:sub(2)
@@ -46,12 +268,24 @@ local function internal_link_to_cref(el)
   return nil
 end
 
+local function latexize_doc(doc)
+  return doc:walk({
+    Link = internal_link_to_cref,
+    Str = symbol_str_to_latex,
+  })
+end
+
 local function markdown_to_latex(markdown)
   local doc = pandoc.read(
     markdown or "",
     "markdown+fenced_code_attributes+tex_math_dollars+tex_math_single_backslash+footnotes+raw_html"
   )
-  doc = doc:walk({ Link = internal_link_to_cref })
+  doc = latexize_doc(doc)
+  return pandoc.write(doc, "latex")
+end
+
+local function blocks_to_latex(blocks)
+  local doc = latexize_doc(pandoc.Pandoc(blocks))
   return pandoc.write(doc, "latex")
 end
 
@@ -88,6 +322,102 @@ local function interactive_to_latex(el)
   )
 end
 
+local function remove_first_inline(block)
+  if not block or block.t ~= "Para" or not block.content or #block.content == 0 then
+    return
+  end
+
+  table.remove(block.content, 1)
+  while #block.content > 0 and block.content[1].t == "Space" do
+    table.remove(block.content, 1)
+  end
+
+  if #block.content > 0 and block.content[1].t == "Str" then
+    block.content[1].text = block.content[1].text:gsub("^:%s*", "")
+    if block.content[1].text == "" then
+      table.remove(block.content, 1)
+    end
+  end
+end
+
+local function strip_text_prefix(block, pattern)
+  if not block or block.t ~= "Para" or not block.content or #block.content == 0 then
+    return
+  end
+
+  local first = block.content[1]
+  if first.t ~= "Str" then
+    return
+  end
+
+  first.text = first.text:gsub(pattern, "")
+  if first.text == "" then
+    remove_first_inline(block)
+  end
+end
+
+local function detect_callout(block)
+  local text = first_inline_text(block)
+  if text == "" then
+    return nil
+  end
+
+  if block.content[1].t == "Strong" then
+    local key = normalize_label(text)
+    if callout_labels[key] then
+      return { key = key, label = callout_labels[key], remover = "strong" }
+    end
+  end
+
+  local bracket = text:match("^%[!([%a-]+)%]")
+  if bracket then
+    local key = normalize_label(bracket)
+    if callout_labels[key] then
+      return { key = key, label = callout_labels[key], remover = "bracket" }
+    end
+  end
+
+  for _, item in ipairs(callout_markers) do
+    if text:sub(1, #item.marker) == item.marker then
+      return { key = item.key, label = callout_labels[item.key], remover = "marker", marker = item.marker }
+    end
+  end
+
+  return nil
+end
+
+local function callout_to_latex(el)
+  if not FORMAT:match("latex") or not el.content or #el.content == 0 then
+    return nil
+  end
+
+  local blocks = pandoc.List(el.content)
+  local callout = detect_callout(blocks[1])
+  if not callout then
+    return nil
+  end
+
+  if callout.remover == "strong" then
+    remove_first_inline(blocks[1])
+  elseif callout.remover == "bracket" then
+    strip_text_prefix(blocks[1], "^%[![%a-]+%]%s*")
+  elseif callout.remover == "marker" then
+    strip_text_prefix(blocks[1], "^" .. callout.marker .. "%s*")
+  end
+
+  if blocks[1] and blocks[1].t == "Para" and #blocks[1].content == 0 then
+    table.remove(blocks, 1)
+  end
+
+  local env = callout_envs[callout.key] or "blognotebox"
+  return pandoc.RawBlock(
+    "latex",
+    "\\begin{" .. env .. "}{}\n"
+      .. blocks_to_latex(blocks)
+      .. "\n\\end{" .. env .. "}"
+  )
+end
+
 local function env_to_latex(el)
   local env_type = el.attributes.type or first_env_class(el)
   if not env_type or not env_types[env_type] then
@@ -126,6 +456,14 @@ function CodeBlock(el)
   return nil
 end
 
+function BlockQuote(el)
+  return callout_to_latex(el)
+end
+
 function Link(el)
   return internal_link_to_cref(el)
 end
+
+function Str(el)
+  return symbol_str_to_latex(el)
+end