Pipe table and table caption support (witiko/markdown backport)

bin/lunamark

@@ -347,6 +347,14 @@ environment variable `LUNAMARK_EXTENSIONS` (see ENVIRONMENT below).
     option by the same name: attributes cannot be set on links
     and indirect images (i.e. only direct images support them).
 
+`(-) pipe_tables`
+:   Support "Pipe Table", as with Pandoc's option by the same name,
+    following the syntax introduced in PHP Markdown Extra.
+
+`(-) table_captions`
+:   Enable the Pandoc `table_captions` syntax extension for
+    tables.
+
 # TEMPLATES
 
 By default, lunamark will produce a fragment.  If the
@@ -495,6 +503,8 @@ given in parentheses:
   (-) raw_attribute        Raw pass-through on code elements
   (-) fenced_code_attributes Fenced code block attributes
   (-) link_attributes      Link attributes
+  (-) pipe_tables          PHP Markdown Extra pipe table support
+  (-) table_captions       Table caption syntax extension
 The keyword 'all' may also be used, to set all extensions simultaneously.
 Setting the environment variable LUNAMARK_EXTENSIONS can change the
 defaults.
@@ -561,6 +571,8 @@ local extensions = {  -- defaults
   raw_attribute = false,
   fenced_code_attributes = false,
   link_attributes = false,
+  pipe_tables = false,
+  table_captions = false,
 }
 
 if optarg["0"] then

lunamark/reader/markdown.lua

@@ -85,6 +85,7 @@ parsers.equal                  = P("=")
 parsers.colon                  = P(":")
 parsers.semicolon              = P(";")
 parsers.exclamation            = P("!")
+parsers.pipe                   = P("|")
 parsers.tilde                  = P("~")
 parsers.tab                    = P("\t")
 parsers.newline                = P("\n")
@@ -778,6 +779,13 @@ end
 --         option by the same name: attributes cannot be set on links
 --         and indirect images (i.e. only direct images support them).
 --
+--     `pipe_tables`
+--     :   Support "Pipe Tables", as with Pandoc's option by the same name,
+--         following the syntax introduced in PHP Markdown Extra.
+--
+--     `table_captions`
+--     :   Enable the Pandoc `table_captions` syntax extension for
+--         tables.
 --
 -- *   Returns a converter function that converts a markdown string
 --     using `writer`, returning the parsed document as first result,
@@ -1476,6 +1484,114 @@ function M.new(writer, options)
                        * parsers.optionalspace * parsers.newline
                        / writer.header
 
+  ------------------------------------------------------------------------------
+  -- PipeTable
+  ------------------------------------------------------------------------------
+
+  local function make_pipe_table_rectangular(rows)
+    local num_columns = #rows[2]
+    local rectangular_rows = {}
+    for i = 1, #rows do
+      local row = rows[i]
+      local rectangular_row = {}
+      for j = 1, num_columns do
+        rectangular_row[j] = row[j] or ""
+      end
+      table.insert(rectangular_rows, rectangular_row)
+    end
+    return rectangular_rows
+  end
+
+  local function pipe_table_row(allow_empty_first_column
+                                , nonempty_column
+                                , column_separator
+                                , column)
+    local row_beginning
+    if allow_empty_first_column then
+      row_beginning = -- empty first column
+                      #(parsers.spacechar^4
+                        * column_separator)
+                    * parsers.optionalspace
+                    * column
+                    * parsers.optionalspace
+                    -- non-empty first column
+                    + parsers.nonindentspace
+                    * nonempty_column^-1
+                    * parsers.optionalspace
+    else
+      row_beginning = parsers.nonindentspace
+                    * nonempty_column^-1
+                    * parsers.optionalspace
+    end
+
+    return Ct(row_beginning
+              * (-- single column with no leading pipes
+                #(column_separator
+                  * parsers.optionalspace
+                  * parsers.newline)
+                * column_separator
+                * parsers.optionalspace
+                -- single column with leading pipes or
+                -- more than a single column
+                + (column_separator
+                  * parsers.optionalspace
+                  * column
+                  * parsers.optionalspace)^1
+                * (column_separator
+                  * parsers.optionalspace)^-1))
+  end
+
+  larsers.table_hline_separator = parsers.pipe + parsers.plus
+
+  larsers.table_hline_column = (parsers.dash
+                            - #(parsers.dash
+                                * (parsers.spacechar
+                                  + larsers.table_hline_separator
+                                  + parsers.newline)))^1
+                          * (parsers.colon * Cc("r")
+                            + parsers.dash * Cc("d"))
+                          + parsers.colon
+                          * (parsers.dash
+                            - #(parsers.dash
+                                * (parsers.spacechar
+                                  + larsers.table_hline_separator
+                                  + parsers.newline)))^1
+                          * (parsers.colon * Cc("c")
+                            + parsers.dash * Cc("l"))
+
+  larsers.table_hline = pipe_table_row(false
+                                    , larsers.table_hline_column
+                                    , larsers.table_hline_separator
+                                    , larsers.table_hline_column)
+
+  larsers.table_caption_beginning = parsers.skipblanklines
+                                * parsers.nonindentspace
+                                * (P("Table")^-1 * parsers.colon)
+                                * parsers.optionalspace
+
+  larsers.table_row = pipe_table_row(true
+                                  , (C((parsers.linechar - parsers.pipe)^1)
+                                    / parse_inlines)
+                                  , parsers.pipe
+                                  , (C((parsers.linechar - parsers.pipe)^0)
+                                    / parse_inlines))
+
+  if options.table_captions then
+    larsers.table_caption = #larsers.table_caption_beginning
+                  * larsers.table_caption_beginning
+                  * Ct(parsers.Inline^1) -- N.B. CAVEAT: It was IndentedInline in witiko/markdown
+                  * parsers.newline
+  else
+    larsers.table_caption = parsers.fail
+  end
+
+  larsers.PipeTable = Ct(larsers.table_row * parsers.newline
+                    * larsers.table_hline
+                    * (parsers.newline * larsers.table_row)^0)
+                  / make_pipe_table_rectangular
+                  * larsers.table_caption^-1
+                  / writer.table
+
   ------------------------------------------------------------------------------
   -- Syntax specification
   ------------------------------------------------------------------------------
@@ -1493,6 +1609,7 @@ function M.new(writer, options)
       Blank                 = larsers.Blank,
 
       Block                 = V("Blockquote")
+                            + V("PipeTable")
                             + V("Verbatim")
                             + V("FencedCodeBlock")
                             + V("FencedDiv")
@@ -1518,6 +1635,7 @@ function M.new(writer, options)
       DefinitionList        = larsers.DefinitionList,
       DisplayHtml           = larsers.DisplayHtml,
       Paragraph             = larsers.Paragraph,
+      PipeTable             = larsers.PipeTable,
       Plain                 = larsers.Plain,
 
       Inline                = V("Str")
@@ -1623,6 +1741,10 @@ function M.new(writer, options)
     syntax.TaskList = parsers.fail
   end
 
+  if not options.pipe_tables then
+    syntax.PipeTable = parsers.fail
+  end
+
   if options.alter_syntax and type(options.alter_syntax) == "function" then
     syntax = options.alter_syntax(syntax)
   end

lunamark/writer/generic.lua

@@ -186,7 +186,7 @@ function M.new(options)
   -- `numdelim`, depending on options, may be one of "Default", "OneParen",
   -- "Period".
   -- (Those symbolic names are loosely taken from Pandoc.)
-  function W.orderedlist(items)
+  function W.orderedlist(items, _)
     return util.intersperse(items,W.interblocksep)
   end
 
@@ -341,6 +341,16 @@ function M.new(options)
     return util.intersperse(buffer,W.interblocksep)
   end
 
+  --- A table.
+  -- The two arguments are rows, caption:
+  -- `rows[1]` contains the headers,
+  -- `rows[2]` contains the aligments (r l c d = right, left, centered, default)
+  -- and the other rows follow...
+  -- `caption` is the optional caption,
+  function W.table(_, _)
+    return ""
+  end
+
   --- A cosmo template to be used in producing a standalone document.
   -- `$body` is replaced with the document body, `$title` with the
   -- title, and so on.

lunamark/writer/html.lua

@@ -238,6 +238,49 @@ function M.new(options)
     return {"<dl>", containersep, intersperse(buffer, containersep), containersep, "</dl>"}
   end
 
+  local function tableCellAlign (align)
+    if align == 'l' then
+      return ' style="text-align: left;"'
+    elseif align == 'r' then
+      return ' style="text-align: right;"'
+    elseif align == 'c' then
+      return ' style="text-align: center;"'
+    end
+    return ''
+  end
+
+  function Html.table(rows, caption)
+    -- Mimic Pandoc output for such tables (in terms of carriage returns, class, styles, etc.)
+    local t = {}
+    local aligns = rows[2]
+
+    if caption then
+      t[#t+1] = { "<caption>", caption, "</caption>\n" }
+    end
+
+    local theadrow = { '<tr class="header">\n' }
+    for j, column in ipairs(rows[1]) do
+      local col = { "<th"..tableCellAlign(aligns[j])..">", column, "</th>\n" }
+      theadrow[#theadrow+1] = col
+    end
+    theadrow[#theadrow+1] = "</tr>\n"
+    t[#t+1] = { "<thead>\n", theadrow, "</thead>\n" }
+
+    t[#t+1] = "<tbody>\n"
+    for i = 3, #rows do
+      local tbodyrows = { '<tr class="', (i % 2 == 0) and "even" or "odd",'">\n' }
+      for j, column in ipairs(rows[i]) do
+        local col = { "<td"..tableCellAlign(aligns[j])..">", column, "</td>\n" }
+        tbodyrows[#tbodyrows+1] = col
+      end
+      tbodyrows[#tbodyrows+1]= "</tr>\n"
+      t[#t+1] = tbodyrows
+    end
+    t[#t+1] = "</tbody>\n"
+
+    return { "<table>\n", t, "</table>\n" }
+  end
+
   Html.template = [[
 <html>
 <head>

tests/lunamark/ext_pipe_tables.test

@@ -0,0 +1,38 @@
+lunamark -Xpipe_tables,table_captions
+<<<
+Pipe table
+
+| Right | Left | Default | Center |
+|------:|:-----|---------|:------:|
+|  12   |  12  |    12   |    12  |
+|  123  |  123 |   123   |   123  |
+
+  : Demonstration of a pipe table.
+>>>
+<p>Pipe table</p>
+
+<table>
+<caption>Demonstration of a pipe table.</caption>
+<thead>
+<tr class="header">
+<th style="text-align: right;">Right</th>
+<th style="text-align: left;">Left</th>
+<th>Default</th>
+<th style="text-align: center;">Center</th>
+</tr>
+</thead>
+<tbody>
+<tr class="odd">
+<td style="text-align: right;">12</td>
+<td style="text-align: left;">12</td>
+<td>12</td>
+<td style="text-align: center;">12</td>
+</tr>
+<tr class="even">
+<td style="text-align: right;">123</td>
+<td style="text-align: left;">123</td>
+<td>123</td>
+<td style="text-align: center;">123</td>
+</tr>
+</tbody>
+</table>