Merge pull request #2140 from Omikhleia/feath-mathml-mtext

feat(math): Support MathML mtext and ms elements
sile-typesetter · Oct 29, 2024 · 0cd24ab · 0cd24ab
2 parents d09a58f + 518e497
commit 0cd24ab
Show file tree

Hide file tree

Showing 3 changed files with 28 additions and 5 deletions.
diff --git a/packages/math/base-elements.lua b/packages/math/base-elements.lua
@@ -908,8 +908,8 @@ end
 
 function elements.text:_init (kind, attributes, script, text)
    elements.terminal._init(self)
-   if not (kind == "number" or kind == "identifier" or kind == "operator") then
-      SU.error("Unknown text node kind '" .. kind .. "'; should be one of: number, identifier, operator")
+   if not (kind == "number" or kind == "identifier" or kind == "operator" or kind == "string") then
+      SU.error("Unknown text node kind '" .. kind .. "'; should be one of: number, identifier, operator, string")
    end
    self.kind = kind
    self.script = script

diff --git a/packages/math/texlike.lua b/packages/math/texlike.lua
@@ -42,8 +42,11 @@ local mathGrammar = function (_ENV)
          return ret
       end
    local group = P"{" * V"mathlist" * (P"}" + E("`}` expected"))
+   -- Simple amsmath-like \text command (no embedded math)
+   local textgroup = P"{" * C((1-P"}")^1) * (P"}" + E("`}` expected"))
    local element_no_infix =
       V"def" +
+      V"text" + -- Important: before command
       V"command" +
       group +
       V"argument" +
@@ -115,6 +118,11 @@ local mathGrammar = function (_ENV)
    sub = element_no_infix * _ * P"_" * _ * element_no_infix
    atom = natural + C(utf8code - S"\\{}%^_&") +
       (P"\\{" + P"\\}") / function (s) return string.sub(s, -1) end
+   text = (
+         P"\\text" *
+         Cg(parameters, "options") *
+         textgroup
+      )
    command = (
          P"\\" *
          Cg(ctrl_sequence_name, "command") *
@@ -374,6 +382,8 @@ local function compileToMathML_aux (_, arg_env, tree)
          return compileToMathML_aux(nil, compiledArgs, tree[1])
       end)
       return nil
+   elseif tree.id == "text" then
+      tree.command = "mtext"
    elseif tree.id == "command" and commands[tree.command] then
       local argTypes = commands[tree.command][1]
       local cmdFun = commands[tree.command][2]

diff --git a/packages/math/typesetter.lua b/packages/math/typesetter.lua
@@ -45,7 +45,7 @@ function ConvertMathML (_, content)
       local script = content.options.mathvariant and b.mathVariantToScriptType(content.options.mathvariant)
       local text = content[1]
       if type(text) ~= "string" then
-         SU.error("mi command contains " .. text .. ", which is not text")
+         SU.error("mi command contains content which is not text")
       end
       script = script or (luautf8.len(text) == 1 and b.scriptType.italic or b.scriptType.upright)
       return b.text("identifier", {}, script, text)
@@ -67,15 +67,15 @@ function ConvertMathML (_, content)
          end
       end
       if type(text) ~= "string" then
-         SU.error("mo command contains " .. text .. ", which is not text")
+         SU.error("mo command contains content which is not text")
       end
       return b.text("operator", attributes, script, text)
    elseif content.command == "mn" then
       local script = content.options.mathvariant and b.mathVariantToScriptType(content.options.mathvariant)
          or b.scriptType.upright
       local text = content[1]
       if type(text) ~= "string" then
-         SU.error("mn command contains " .. text .. ", which is not text")
+         SU.error("mn command contains content which is not text")
       end
       if string.sub(text, 1, 1) == "-" then
          text = "−" .. string.sub(text, 2)
@@ -136,6 +136,19 @@ function ConvertMathML (_, content)
       return b.mtr(convertChildren(content))
    elseif content.command == "mtd" then
       return b.stackbox("H", convertChildren(content))
+   elseif content.command == "mtext" or content.command == "ms" then
+      if #content > 1 then
+         SU.error("Wrong number of children in " .. content.command .. ": " .. #content)
+      end
+      local text = content[1] or "" -- empty mtext is allowed, and found in examples...
+      if type(text) ~= "string" then
+         SU.error(content.command .. " command contains content which is not text")
+      end
+      -- MathML Core 3.2.1.1 Layout of <mtext> has some wording about forced line breaks
+      -- and soft wrap opportunities: ignored here.
+      -- There's also some explanations about CSS, italic correction etc. which we ignore too.
+      text = text:gsub("[\n\r]", " ")
+      return b.text("string", {}, b.scriptType.upright, text:gsub("%s+", " "))
    else
       SU.error("Unknown math command " .. content.command)
    end