From bba310e51287a613bda4ae2dbe720bb2d0646eb7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Marcel=20Fabian=20Kr=C3=BCger?= <tex@2krueger.de>
Date: Fri, 30 Apr 2021 19:52:32 +0200
Subject: [PATCH] Better tags

---
 luamml-amsmath.lua | 16 ++----------
 luamml-lr.lua      | 63 ++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 65 insertions(+), 14 deletions(-)
 create mode 100644 luamml-lr.lua

diff --git a/luamml-amsmath.lua b/luamml-amsmath.lua
index 340ede4..b8ac0e3 100644
--- a/luamml-amsmath.lua
+++ b/luamml-amsmath.lua
@@ -4,6 +4,7 @@ local save_result = require'luamml-tex'.save_result
 local store_column = require'luamml-table'.store_column
 local store_tag = require'luamml-table'.store_tag
 local get_table = require'luamml-table'.get_table
+local to_text = require'luamml-lr'
 
 local properties = node.get_properties_table()
 
@@ -42,20 +43,7 @@ token.set_lua('__luamml_amsmath_save_tag:', funcid, 'protected')
 lua.get_functions_table()[funcid] = function()
   local nest = tex.nest.top
   local chars = {}
-  for n, id, sub in node.traverse(nest.head.next) do
-    if id == node.id'glyph' then
-      if sub >= 0x100 then
-        texio.write_nl'WARNING: Already shaped glyph detected in tag. This might lead to wrong output.'
-      end
-      chars[#chars+1] = n.char
-    elseif id == node.id'glue' then
-      chars[#chars+1] = 0x20
-    elseif id == node.id'kern' then
-    else
-      texio.write_nl'WARNING: Unsupported node in tag dropped'
-    end
-  end
-  last_tag = utf8.char(table.unpack(chars))
+  last_tag = to_text(nest.head)
 end
 
 funcid = luatexbase.new_luafunction'__luamml_amsmath_set_tag:'
diff --git a/luamml-lr.lua b/luamml-lr.lua
new file mode 100644
index 0000000..d4921d1
--- /dev/null
+++ b/luamml-lr.lua
@@ -0,0 +1,63 @@
+local properties = node.get_properties_table()
+
+local function to_unicode(head, tail)
+  local result, i = {}, 0
+  local characters, last_fid
+  local iter, state, n = node.traverse(head)
+  while true do
+    local id, sub n, id, sub = iter(state, n)
+    if not n or n == tail then break end
+    local props = properties[n]
+    if props and props.glyph_info then
+      i = i+1
+      result[i] = glyph_info
+    else
+      local char, fid = node.is_glyph(n)
+      if char then
+        if fid ~= last_fid then
+          local fontdir = font.getfont(fid)
+          characters, last_fid = fontdir.characters, fid
+        end
+        local uni = characters[char]
+        local uni = uni and uni.unicode
+        i = i+1
+        if uni then
+          if type(uni) == 'number' then
+            result[i] = utf.char(uni)
+          else
+            result[i] = utf.char(table.unpack(uni))
+          end
+        else
+          if char < 0x110000 then
+            result[i] = utf.char(char)
+          else
+            result[i] = '\u{FFFD}'
+          end
+        end
+      -- elseif node.id'math' == id then
+        -- n = node.end_of_math(n) -- Not sure yet
+      -- elseif node.id'whatsit' == id then
+        -- TODO(?)
+      elseif node.id'glue' == id then
+        if n.width > 1000 then -- FIXME: Coordinate constant with tagpdf
+          i = i+1
+          result[i] = ' '
+        end
+      elseif node.id'hlist' == id then
+        i = i+1
+        result[i] = '\u{FFFD}'
+      elseif node.id'vlist' == id then
+        i = i+1
+        result[i] = '\u{FFFD}'
+      elseif node.id'rule' == id then
+        if n.width ~= 0 then
+          i = i+1
+          result[i] = '\u{FFFD}'
+        end
+      end -- CHECK: Everything else can probably be ignored, otherwise shout at me
+    end
+  end
+  return table.concat(result)
+end
+
+return to_unicode