refactor(languages): Move language specific discretionary handling ou…

…t of generic module
sile-typesetter · Jan 26, 2024 · 95ac070 · 95ac070
1 parent aed271e
commit 95ac070
Show file tree

Hide file tree

Showing 4 changed files with 81 additions and 64 deletions.
diff --git a/core/hyphenator-liang.lua b/core/hyphenator-liang.lua
@@ -95,11 +95,19 @@ SILE.hyphenator = {}
 SILE.hyphenator.languages = {}
 SILE._hyphenators = {}
 
+local function defaultHyphenateSegments (node, segments, _)
+  local hyphen = SILE.shaper:createNnodes(SILE.settings:get("font.hyphenchar"), node.options)
+  return SILE.nodefactory.discretionary({ prebreak = hyphen }), segments
+end
+
 local initHyphenator = function (lang)
   if not SILE._hyphenators[lang] then
     SILE._hyphenators[lang] = { minWord = 5, leftmin = 2, rightmin = 2, trie = {}, exceptions = {}  }
     loadPatterns(SILE._hyphenators[lang], lang)
   end
+  if SILE.hyphenator.languages[lang] and not SILE.hyphenator.languages[lang].hyphenateSegments then
+    SILE.hyphenator.languages[lang].hyphenateSegments = defaultHyphenateSegments
+  end
 end
 
 local hyphenateNode = function (node)
@@ -110,71 +118,25 @@ local hyphenateNode = function (node)
   end
   initHyphenator(node.language)
   local segments = SILE._hyphenate(SILE._hyphenators[node.language], node.text)
+  local hyphen
   if #segments > 1 then
-    local hyphen = SILE.shaper:createNnodes(SILE.settings:get("font.hyphenchar"), node.options)
+    local hyphenateSegments = SILE.hyphenator.languages[node.language].hyphenateSegments
     local newnodes = {}
     for j, segment in ipairs(segments) do
-      local specificDiscretionary
       if segment == "" then
         SU.dump({ j, segments })
         SU.error("No hyphenation segment should ever be empty", true)
       end
-      if node.options.language == "tr" then
-        local nextApostrophe = j < #segments and luautf8.match(segments[j+1], "^['’]")
-        if nextApostrophe then
-          segments[j+1] = luautf8.gsub(segments[j+1], "^['’]", "")
-          local replacement = SILE.shaper:createNnodes(nextApostrophe, node.options)
-          if SILE.settings:get("languages.tr.replaceApostropheAtHyphenation") then
-            -- leading apostrophe (on next segment) cancels when hyphenated
-            specificDiscretionary = SILE.nodefactory.discretionary({ replacement = replacement, prebreak = hyphen })
-          else
-            -- hyphen character substituted for upcomming apostrophe
-            local kesme = SILE.shaper:createNnodes(nextApostrophe, node.options)
-            specificDiscretionary = SILE.nodefactory.discretionary({ replacement = replacement, prebreak = kesme })
-          end
-        end
-      elseif node.options.language == "ca" then
-        -- punt volat (middle dot) cancels when hyphenated
-        -- Catalan typists may use a punt volat or precomposed characters.
-        -- The shaper might behave differently depending on the font, so we need to
-        -- be consistent here with the typist's choice.
-        if luautf8.find(segment, "ŀ$") then -- U+0140
-          segment = luautf8.sub(segment, 1, -2)
-          local ldot = SILE.shaper:createNnodes("ŀ", node.options)
-          local lhyp = SILE.shaper:createNnodes("l" .. SILE.settings:get("font.hyphenchar"), node.options)
-          specificDiscretionary = SILE.nodefactory.discretionary({ replacement = ldot, prebreak = lhyp })
-        elseif luautf8.find(segment, "Ŀ$") then -- U+013F
-          segment = luautf8.sub(segment, 1, -2)
-          local ldot = SILE.shaper:createNnodes("Ŀ", node.options)
-          local lhyp = SILE.shaper:createNnodes("L" .. SILE.settings:get("font.hyphenchar"), node.options)
-          specificDiscretionary = SILE.nodefactory.discretionary({ replacement = ldot, prebreak = lhyp })
-        elseif luautf8.find(segment, "l·$") then -- l + U+00B7
-          segment = luautf8.sub(segment, 1, -3)
-          local ldot = SILE.shaper:createNnodes("l·", node.options)
-          local lhyp = SILE.shaper:createNnodes("l" .. SILE.settings:get("font.hyphenchar"), node.options)
-          specificDiscretionary = SILE.nodefactory.discretionary({ replacement = ldot, prebreak = lhyp })
-        elseif luautf8.find(segment, "L·$") then -- L + U+00B7
-          segment = luautf8.sub(segment, 1, -3)
-          local ldot = SILE.shaper:createNnodes("L·", node.options)
-          local lhyp = SILE.shaper:createNnodes("L" .. SILE.settings:get("font.hyphenchar"), node.options)
-          specificDiscretionary = SILE.nodefactory.discretionary({ replacement = ldot, prebreak = lhyp })
-        end
-      end
-      for _, newNode in ipairs(SILE.shaper:createNnodes(segment, node.options)) do
+      hyphen, segments = hyphenateSegments(node, segments, j)
+      for _, newNode in ipairs(SILE.shaper:createNnodes(segments[j], node.options)) do
         if newNode.is_nnode then
           newNode.parent = node
           table.insert(newnodes, newNode)
         end
       end
       if j < #segments then
-        if specificDiscretionary then
-          specificDiscretionary.parent = node
-          table.insert(newnodes, specificDiscretionary)
-        else
-          local newNode = SILE.nodefactory.discretionary({ prebreak = hyphen })
-          newNode.parent = node
-          table.insert(newnodes, newNode)
-        end
+        hyphen.parent = node
+        table.insert(newnodes, hyphen)
       end
     end
     node.children = newnodes

diff --git a/languages/ca.lua b/languages/ca.lua
@@ -6,6 +6,37 @@ SILE.nodeMakers.ca = pl.class({
 })
 
 SILE.hyphenator.languages["ca"] = {}
+
+SILE.hyphenator.languages["ca"].hyphenateSegments = function (node, segments, j)
+  -- punt volat (middle dot) cancels when hyphenated
+  -- Catalan typists may use a punt volat or precomposed characters.
+  -- The shaper might behave differently depending on the font, so we need to
+  -- be consistent here with the typist's choice.
+  local hyphenChar = SILE.settings:get("font.hyphenchar")
+  local replacement, hyphen
+  if luautf8.find(segments[j], "ŀ$") then -- U+0140
+    segments[j] = luautf8.sub(segments[j], 1, -2)
+    replacement = SILE.shaper:createNnodes("ŀ", node.options)
+    hyphen = SILE.shaper:createNnodes("l" .. hyphenChar, node.options)
+  elseif luautf8.find(segments[j], "Ŀ$") then -- U+013F
+    segments[j] = luautf8.sub(segments[j], 1, -2)
+    replacement = SILE.shaper:createNnodes("Ŀ", node.options)
+    hyphen = SILE.shaper:createNnodes("L" .. hyphenChar, node.options)
+  elseif luautf8.find(segments[j], "l·$") then -- l + U+00B7
+    segments[j] = luautf8.sub(segments[j], 1, -3)
+    replacement = SILE.shaper:createNnodes("l·", node.options)
+    hyphen = SILE.shaper:createNnodes("l" .. hyphenChar, node.options)
+  elseif luautf8.find(segments[j], "L·$") then -- L + U+00B7
+    segments[j] = luautf8.sub(segments[j], 1, -3)
+    replacement = SILE.shaper:createNnodes("L·", node.options)
+    hyphen = SILE.shaper:createNnodes("L" .. hyphenChar, node.options)
+  else
+    hyphen = SILE.shaper:createNnodes(hyphenChar, node.options)
+  end
+  local discretionary = SILE.nodefactory.discretionary({ replacement = replacement, prebreak = hyphen })
+  return discretionary, segments
+end
+
 SILE.hyphenator.languages["ca"].patterns =
    {
 --

diff --git a/languages/tr.lua b/languages/tr.lua
@@ -14,6 +14,25 @@ SILE.nodeMakers.tr = pl.class(SILE.nodeMakers.unicode)
 SILE.nodeMakers.tr.wordTypes = { cm = true, qu = true }
 
 SILE.hyphenator.languages["tr"] = {}
+
+SILE.hyphenator.languages["tr"].hyphenateSegments = function (node, segments, j)
+  local hyphenChar, replacement
+  local maybeNextApostrophe = #segments > j and luautf8.match(segments[j+1], "^['’]")
+  if maybeNextApostrophe then
+    segments[j+1] = luautf8.gsub(segments[j+1], "^['’]", "")
+    if SILE.settings:get("languages.tr.replaceApostropheAtHyphenation") then
+      hyphenChar = SILE.settings:get("font.hyphenchar")
+    else
+      hyphenChar = maybeNextApostrophe
+      replacement = SILE.shaper:createNnodes(maybeNextApostrophe, node.options)
+    end
+  else
+    hyphenChar = SILE.settings:get("font.hyphenchar")
+  end
+  local hyphen = SILE.shaper:createNnodes(hyphenChar, node.options)
+  return SILE.nodefactory.discretionary({ replacement = replacement, prebreak = hyphen }), segments
+end
+
 SILE.hyphenator.languages["tr"].patterns =
    {
 "2a1",

diff --git a/languages/unicode.lua b/languages/unicode.lua
@@ -68,39 +68,44 @@ SILE.nodeMakers.base = pl.class({
       return chardata[cp]
     end,
 
-    isPunctuation = function (self, char)
-      return self.puctuationTypes[self:charData(char).category]
+    isActiveNonBreakingSpace = function (self, char)
+      return self:isNonBreakingSpace(char) and not SILE.settings:get("languages.fixedNbsp")
     end,
 
-    isSpace = function (self, char)
-      return self.spaceTypes[self:charData(char).linebreak]
+    isBreaking = function (self, char)
+      return self.breakingTypes[self:charData(char).linebreak]
     end,
 
     isNonBreakingSpace = function (self, char)
       local c = self:charData(char)
       return c.contextname and c.contextname == "nobreakspace"
     end,
 
-    isActiveNonBreakingSpace = function (self, char)
-      return self:isNonBreakingSpace(char) and not SILE.settings:get("languages.fixedNbsp")
+    isPunctuation = function (self, char)
+      return self.puctuationTypes[self:charData(char).category]
     end,
 
-    isBreaking = function (self, char)
-      return self.breakingTypes[self:charData(char).linebreak]
+    isSpace = function (self, char)
+      return self.spaceTypes[self:charData(char).linebreak]
     end,
+
     isQuote = function (self, char)
       return self.quoteTypes[self:charData(char).linebreak]
-    end
+    end,
+
+    isWord = function (self, char)
+      return self.wordTypes[self:charData(char).linebreak]
+    end,
 
   })
 
 SILE.nodeMakers.unicode = pl.class(SILE.nodeMakers.base)
 
-SILE.nodeMakers.unicode.wordTypes = { cm = true }
-SILE.nodeMakers.unicode.spaceTypes = { sp = true }
 SILE.nodeMakers.unicode.breakingTypes = { ba = true, zw = true }
 SILE.nodeMakers.unicode.puctuationTypes = { po = true }
 SILE.nodeMakers.unicode.quoteTypes = {} -- quote linebreak category is ambiguous depending on the language
+SILE.nodeMakers.unicode.spaceTypes = { sp = true }
+SILE.nodeMakers.unicode.wordTypes = { cm = true }
 
 function SILE.nodeMakers.unicode:dealWith (item)
   local char = item.text
@@ -119,7 +124,7 @@ function SILE.nodeMakers.unicode:dealWith (item)
   elseif self:isQuote(item.text) then
     self:addToken(char, item)
     self:makeToken()
-  elseif self.lasttype and (thistype and thistype ~= self.lasttype and not self.wordTypes[thistype]) then
+  elseif self.lasttype and (thistype and thistype ~= self.lasttype and not self:isWord(thistype)) then
     self:addToken(char, item)
   else
     self:letterspace()