Add method normalize to Pandoc objects

This returns a normalized document by merging adjacent spaces in inlines
and by modifying tables.

Author: tarleb

src/Text/Pandoc/Lua/Marshal/Pandoc.hs

@@ -1,4 +1,7 @@
+{-# LANGUAGE FlexibleContexts     #-}
+{-# LANGUAGE LambdaCase           #-}
 {-# LANGUAGE OverloadedStrings    #-}
+{-# LANGUAGE TypeApplications     #-}
 {- |
 Copyright               : © 2021-2024 Albert Krewinkel
 SPDX-License-Identifier : MIT
@@ -25,12 +28,15 @@ import Control.Monad ((<$!>))
 import Data.Aeson (encode)
 import Data.Maybe (fromMaybe)
 import HsLua
+import Text.Pandoc.Definition (Pandoc (..), Meta (..), nullMeta)
 import Text.Pandoc.Lua.Marshal.Block (peekBlocksFuzzy, pushBlocks)
 import Text.Pandoc.Lua.Marshal.Filter
 import Text.Pandoc.Lua.Marshal.MetaValue (peekMetaValue, pushMetaValue)
 import Text.Pandoc.Lua.Marshal.Shared (walkBlocksAndInlines)
 import Text.Pandoc.Lua.Walk (applyStraight)
-import Text.Pandoc.Definition (Pandoc (..), Meta (..), nullMeta)
+import Text.Pandoc.Walk (Walkable (walk))
+import qualified Data.Foldable as Foldable
+import qualified Text.Pandoc.Builder as B
 
 -- | Pushes a 'Pandoc' value as userdata.
 pushPandoc :: LuaError e => Pusher e Pandoc
@@ -74,6 +80,11 @@ typePandoc = deftype "Pandoc"
       <#> parameter peekPandoc "Pandoc" "doc" "self"
       =#> functionResult pushPandoc "Pandoc" "cloned Pandoc document"
 
+  , method $ defun "normalize"
+      ### liftPure normalize
+      <#> udparam typePandoc "self" ""
+      =#> udresult typePandoc "cloned and normalized document"
+
   , method $ defun "walk"
     ### flip applyFully
     <#> parameter peekPandoc "Pandoc" "self" ""
@@ -149,3 +160,17 @@ applyFully filter' doc = case filterWalkingOrder filter' of
   WalkTopdown     -> applyPandocFunction filter' doc
                  >>= applyMetaFunction filter'
                  >>= walkBlocksAndInlines filter'
+
+-- | Normalize a document
+normalize :: (Walkable [B.Inline] a, Walkable B.Block a) => a -> a
+normalize =
+  let normalizeBlock = \case
+        B.Table attr capt specs th tbs tf ->
+          let twidth = length specs
+              th'  = B.normalizeTableHead twidth th
+              tbs' = map (B.normalizeTableBody twidth) tbs
+              tf'  = B.normalizeTableFoot twidth tf
+          in B.Table attr capt specs th' tbs' tf'
+        x -> x
+  in walk normalizeBlock .
+     walk (B.toList . mconcat . map (B.singleton @B.Inline))

test/test-pandoc.lua

@@ -60,6 +60,48 @@ return {
       assert.are_same(Blocks{'different'}, copy.blocks)
     end),
   },
+  group 'normalize' {
+    test('removes repeated whitespace', function ()
+      local doc = Pandoc{
+        Plain{'before', Space(), SoftBreak(), 'after'}
+      }
+      assert.are_equal(
+        Blocks{Plain{'before', SoftBreak(), 'after'}},
+        doc:normalize().blocks
+      )
+    end),
+    test('normalizes table', function ()
+      local attr = Attr('test')
+      local caption = Blocks('Sample caption')
+      local colspecs = {{'AlignDefault', 0.5}, {'AlignLeft', 0.5}}
+      local thead = TableHead({Row{Cell'header cell'}})
+      local tbody = {
+        attr = Attr(),
+        body = List{
+          Row{Cell'body cell 1', Cell'body cell 2'},
+          Row{Cell('hi')}},
+        },
+        head = List{},
+        row_head_columns = 0,
+      }
+      local tfoot = TableFoot()
+      local tbl = Table(caption, colspecs, thead, {tbody}, tfoot)
+      print(tbl.bodies)
+      local expected_body = tbody
+      expected_body.body[2][2]:insert(Cell{})
+      local doc = Pandoc{tbl}
+      assert.are_same(
+        Table(
+          caption,
+          colspecs,
+          TableHead({Row{Cell'header cell', Cell{}}}),
+          {expected_body},
+          tfoot
+        ),
+        doc:normalize().blocks[1]
+      )
+    end)
+  },
   group 'walk' {
     test('uses `Meta` function', function ()
       local meta = {