Support annotations while parsing pdfTeX log

luamml-convert.lua

@@ -505,7 +505,6 @@ function nodes_to_table(head, cur_style)
     local props = properties[n]
     local mathml_core = props and props.mathml_core
     local mathml_table = props and (props.mathml_table or mathml_core)
-    print(mathml_table)
     if mathml_table ~= nil then
       new_node, new_core = mathml_table, mathml_core
     elseif id == noad_t then

luamml-pdf.sty

@@ -0,0 +1,94 @@
+\ProvidesExplPackage {luamml-pdf} {2021-05-31} {0.0.1-alpha}
+  {MathML generation for L̶u̶a̶pdfLaTeX}
+
+\cs_new:Npn \__luamml_pdf_showlists: {
+  \group_begin:
+    \int_set:Nn \l_tmpa_int { \tex_interactionmode:D }
+    \int_set:Nn \tex_interactionmode:D { 0 }
+    \int_set:Nn \tex_showboxdepth:D { \c_max_int }
+    \int_set:Nn \tex_showboxbreadth:D { \c_max_int }
+    \tex_showlists:D
+    \int_set:Nn \tex_interactionmode:D { \l_tmpa_int }
+  \group_end:
+}
+
+\int_new:N \g__luamml_formula_id_int
+\cs_new_protected:Npn \luamml_pdf_write: {
+  \int_gincr:N \g__luamml_formula_id_int
+  \iow_log:x {
+    LUAMML_FORMULA_BEGIN:
+    \int_use:N \g__luamml_formula_id_int
+  }
+  \__luamml_pdf_showlists:
+  \iow_log:x {
+    LUAMML_FORMULA_END
+  }
+}
+
+\cs_new:Npn \luamml_pdf_last_formula: {
+  \int_use:N \g__luamml_formula_id_int
+}
+
+\cs_generate_variant:Nn \tl_to_str:n { e }
+
+% annotate parameters
+%   #1 number of top level nodes to be annotated
+%   #2 annotation
+%   #3 nodes to be annotated
+\int_new:N \g__luamml_annotation_id_int
+\cs_new_protected:Npn \luamml_pdf_annotate:nnn #1#2#3 {
+  \int_gincr:N \g__luamml_annotation_id_int
+  \iow_shipout_x:Nx \c_log_iow {
+    \tl_to_str:e {
+      LUAMML_MARK:
+      \int_use:N \g__luamml_annotation_id_int
+      :
+      count = \int_eval:n {#1},
+      #2
+    }
+    \exp_not:N \iow_newline:
+    LUAMML_MARK_END
+  }
+  #3
+}
+
+% annotate parameters
+%   #1 annotation
+%   #2 nodes to be annotated
+% THIS VERSION IS SIGNIFICANTLY SLOWER
+\cs_new_protected:Npn \luamml_pdf_annotate:nn #1#2 {
+  \int_gincr:N \g__luamml_annotation_id_int
+  \iow_write:Nx \c_log_iow {
+    LUAMML_COUNT_BEGIN:
+    \int_use:N \g__luamml_annotation_id_int
+  }
+  \__luamml_pdf_showlists:
+  \iow_shipout:Nx \c_log_iow {
+    LUAMML_MARK:
+    \int_use:N \g__luamml_annotation_id_int
+    :
+    #1
+  }
+  #2
+  \iow_write:Nx \c_log_iow {
+    LUAMML_COUNT_END:
+    \int_use:N \g__luamml_annotation_id_int
+  }
+  \__luamml_pdf_showlists:
+}
+
+\endinput
+
+\cs_new:Npn \__luamml_patch_package:nn #1 #2 {
+  \@ifpackageloaded {#1} {#2} {
+    \hook_gput_code:nnn {package/after/#1} {luamml} {#2}
+  }
+}
+\cs_new:Npn \__luamml_patch_package:n #1 {
+  \__luamml_patch_package:nn {#1} {
+    \RequirePackage { luamml-patches-#1 }
+  }
+}
+\RequirePackage { luamml-patches-kernel }
+\__luamml_patch_package:n {amsmath}
+\__luamml_patch_package:n {array}

pdfmml-logreader.lua

@@ -2,15 +2,23 @@ local l = lpeg or require'lpeg'
 
 local line = (1-l.P'\n')^0 * '\n'
 
-local list_block = (l.C(l.S'\\._^/ps' * (1-l.P'\n')^0)^-1 * '\n')^0
+local id = l.R'09'/tonumber
+local non_final_list_block = (l.C((1-l.P'\n')^0) * '\n' - '### ')^0
 local math_lists_block = l.Ct('### ' * l.Cg(l.C'display' * ' ', 'display')^-1 * 'math mode entered at line ' * l.Cg(l.R'09'^1 / tonumber, 'line') * '\n'
-                       * list_block)^1
-local generic_list_block = '### ' * line * list_block
-local luamml_block = l.Ct('LUAMML_META_BEGIN\n\n'
- * (math_lists_block + generic_list_block/0)^0
- * (line - 'LUAMML_META_END\n')^0
- * 'LUAMML_META_END\n')
-local log_file = l.Ct((luamml_block + line)^0)
+                       * non_final_list_block)^1
+local generic_list_block = '### ' * (line - 'current page:') * non_final_list_block
+local luamml_block = l.Cg('LUAMML_FORMULA_BEGIN:' * id * l.P'\n'^1 * l.Ct(
+   (math_lists_block + generic_list_block/0)^0
+ * (line - 'LUAMML_FORMULA_END\n')^0
+ * 'LUAMML_FORMULA_END\n') * l.Cc'groups')
+
+local luamml_mark = l.Cg('LUAMML_MARK:' * id * ':' * l.Cs((1 - l.P'\n' + l.Cg('\n' * l.Cc'' - '\nLUAMML_MARK_END\n'))^0) * '\nLUAMML_MARK_END\n' * l.Cc'marks')
+
+local function multi_table_set(t, key, value, table)
+  assert(t[table])[key] = value
+  return t
+end
+local log_file = l.Cf(l.Ct(l.Cg(l.Ct'', 'groups') * l.Cg(l.Ct'', 'marks')) * (luamml_block + luamml_mark + line)^0, multi_table_set)
 
 return function(filename)
   local f

pdfmml-showlists.lua

@@ -1,3 +1,7 @@
+require'pdfmml-emulate-node'
+
+local properties = node.get_properties_table()
+
 local l = lpeg or require'lpeg'
 local hex_digit = l.R('09', 'af')
 local function hex_to_int(s) return tonumber(s, 16) end
@@ -22,12 +26,12 @@ local math_char = l.Ct('\\fam' * l.Cg(l.R'09'^1 / tonumber, 'fam') * ' ' * l.Cg(
 local simple_noad = l.Ct(
     '\\math' * l.Cg(
         'ord' * l.Cc(0)
+      + 'open' * l.Cc(6)
       + 'op\\limits' * l.Cc(2)
       + 'op\\nolimits' * l.Cc(3)
       + 'op' * l.Cc(1)
       + 'bin' * l.Cc(4)
       + 'rel' * l.Cc(5)
-      + 'open' * l.Cc(6)
       + 'close' * l.Cc(7)
       + 'punct' * l.Cc(8)
       + 'inner' * l.Cc(9)
@@ -60,6 +64,8 @@ local fraction_noad = l.Ct('\\fraction, thickness '
 
 local mathchoice_noad = l.Ct('\\mathchoice' * l.Cg(l.Cc'choice', 'id') * -1)
 
+local mark_whatsit = '\\write-{LUAMML_MARK:' * (l.R'09'/tonumber) * ':'
+
 local parse_list
 local function parse_kernel(lines, i, prefix)
   local line = lines[i]
@@ -69,11 +75,14 @@ local function parse_kernel(lines, i, prefix)
   result, i = parse_list(lines, i, prefix)
   return {list = result, id = 'sub_mlist'}, i
 end
-function parse_list(lines, i, prefix)
+function parse_list(lines, i, prefix, marks)
   i = i or 1
   prefix = prefix or ''
   local head, last
+  local mark_environment = {}
+  local current_mark, current_count, current_offset
   while true do
+    local skip
     local line = lines[i]
     if not line or line:sub(1, #prefix) ~= prefix then break end
     local simple = simple_noad:match(line, #prefix+1)
@@ -105,6 +114,13 @@ function parse_list(lines, i, prefix)
             mathchoice.prev, last.next = last, mathchoice
           end
           last = mathchoice
+        else
+          skip = true
+          local mark = mark_whatsit:match(line)
+          if mark then
+            local mark_table = assert(load('return {' .. assert(marks[mark], 'Undefined mark encountered') .. '}', nil, 't', mark_environment))()
+            current_mark, current_count, current_offset = mark_table, mark_table.count or 1, mark_table.offset or 1
+            i = i + 1
           else
             print(line, prefix, i)
             print('unknown noad ' .. line:sub(#prefix+1))
@@ -112,7 +128,18 @@ function parse_list(lines, i, prefix)
           end
         end
       end
+    end
     if not head then head = last end
+    if not skip and current_mark then
+      current_count = current_count - 1
+      current_offset = current_offset - 1
+      if current_offset == 0 then
+        properties[current_mark.nucleus and last.nucleus or last] = {mathml_core = current_mark.core}
+      else
+        properties[last] = {mathml_core = false}
+      end
+      if current_count == 0 then current_mark = nil end
+    end
   end
   return head, i
 end

pdfmml.lua

@@ -17,9 +17,9 @@ end
 local math_lists = assert(parse_log(arg[1]))
 
 local out_stream = arg[2] and arg[2] ~= '-' and assert(io.open(arg[2], 'w')) or io.stdout
-for i, block in ipairs(math_lists) do
+for i, block in ipairs(math_lists.groups) do
   block = block[1]
-  local parsed = parse_showlists(block)
+  local parsed = parse_showlists(block, nil, nil, math_lists.marks)
   local style = block.display and 0 or 2
   out_stream:write(
     to_xml(convert.make_root(convert.process(parsed, style), style))