lua-typo (16mar21)

git-svn-id: svn://tug.org/texlive/trunk@58441 c570f23f-e606-0410-a88d-b1316a301751

1 files changed, 149 insertions, 116 deletions

diff --git a/Master/texmf-dist/source/lualatex/lua-typo/lua-typo.dtx b/Master/texmf-dist/source/lualatex/lua-typo/lua-typo.dtx
index d1a4b0068a8..0208839779c 100644
--- a/Master/texmf-dist/source/lualatex/lua-typo/lua-typo.dtx
+++ b/Master/texmf-dist/source/lualatex/lua-typo/lua-typo.dtx
@@ -1,4 +1,4 @@
-% \CheckSum{337}
+% \CheckSum{336}
 %
 % \iffalse meta-comment
 %
@@ -34,6 +34,7 @@
 \postamble
 \endpostamble
 \let\MetaPrefix\DoubleperCent
+\askforoverwritefalse
 \generate{%
    \file{lua-typo.sty}{\from{lua-typo.dtx}{sty}}% {sty,msg,dbg}
    \nopreamble
@@ -618,7 +619,7 @@
 \ProvidesFile{lua-typo.dtx}
 %</dtx>
 %<*dtx|sty>
-                [2021/03/03 v.0.30 Daniel Flipo]
+                [2021/03/14 v.0.32 Daniel Flipo]
 %</dtx|sty>
 %<*sty>
 % \fi
@@ -841,7 +842,7 @@ luatypo = { }
 }
 %    \end{macrocode}
 %
-%    Print the summary of offending pages ---if any-- at the
+%    Print the summary of offending pages ---if any--- at the
 %    (very) end of document unless option |None| has been selected.
 %
 %    \begin{macrocode}
@@ -864,8 +865,8 @@ luatypo = { }
        texio.write_nl('*** lua-typo: No Typo Flaws found.')
     else
        texio.write_nl('*** lua-typo: WARNING *************')
-       texio.write_nl('The following pages need attention: '
-                      .. luatypo.pagelist)
+       texio.write_nl('The following pages need attention: ')
+       texio.write(luatypo.pagelist)
     end
     texio.write_nl('***********************************')
     texio.write_nl(' ')
@@ -955,6 +956,7 @@ local hyphcount = 0
 local parlines = 0
 local pagelines = 0
 local pageno = 0
+local prevno = 0
 local pageflag = false
 
 local char_to_discard = { }
@@ -972,6 +974,8 @@ split_lig[0xFB01] = "fi"
 split_lig[0xFB02] = "fl"
 split_lig[0xFB03] = "ffi"
 split_lig[0xFB04] = "ffl"
+split_lig[0xFB05] = "st"
+split_lig[0xFB06] = "st"
 
 local DISC  = node.id("disc")
 local GLYPH = node.id("glyph")
@@ -982,7 +986,7 @@ local LPAR  = node.id("local_par")
 local MKERN = node.id("margin_kern")
 local PENALTY = node.id("penalty")
 % \end{macrocode}
-%    GLUE subtypes:
+%    Glue subtypes:
 %    \begin{macrocode}
 local USRSKIP  = 0
 local PARSKIP  = 3
@@ -991,7 +995,7 @@ local RGTSKIP  = 9
 local TOPSKIP = 10
 local PARFILL = 15
 % \end{macrocode}
-%    HLIST subtypes:
+%    Hlist subtypes:
 %    \begin{macrocode}
 local LINE    = 1
 local BOX     = 2
@@ -1000,6 +1004,10 @@ local BOX     = 2
 %    \begin{macrocode}
 local USER = 0
 local HYPH = 0x2D
+%    \end{macrocode}
+%    Glyph subtypes:
+%    \begin{macrocode}
+local LIGA = 0x102
 
 local effective_glue = node.effective_glue
 local set_attribute = node.set_attribute
@@ -1018,24 +1026,24 @@ local is_glyph  = node.is_glyph
 %    \begin{macrocode}
 local color_node = function (node, color)
   local attr = oberdiek.luacolor.getattribute()
-  if node.id == DISC then
+  if node and node.id == DISC then
      local pre = node.pre
      local post = node.post
-     local replace = node.replace
+     local repl = node.replace
      if pre then
         set_attribute(pre,attr,color)
-%<dbg>     texio.write_nl('PRE=' .. tostring(pre.char))
+%<dbg>  texio.write_nl('PRE=' .. tostring(pre.char))
      end
      if post then
         set_attribute(post,attr,color)
-%<dbg>     texio.write_nl('POST=' .. tostring(post.char))
+%<dbg>  texio.write_nl('POST=' .. tostring(post.char))
      end
-     if replace then
-        set_attribute(replace,attr,color)
-%<dbg>     texio.write_nl('REPL=' .. tostring(replace.char))
+     if repl then
+        set_attribute(repl,attr,color)
+%<dbg>  texio.write_nl('REPL=' .. tostring(repl.char))
      end
 %<dbg>     texio.write_nl(' ')
-  else
+  elseif node then
      set_attribute(node,attr,color)
   end
 end
@@ -1051,6 +1059,7 @@ local color_hbox = function (head, color)
       color_node(n, color)
   end
 end
+%    \end{macrocode}
 %
 %    This auxillary function colours a whole line. It requires two
 %    arguments: a line’s node and a (named) colour.
@@ -1059,7 +1068,7 @@ end
 local color_line = function (head, color)
   local first = head.head
   for n in traverse(first) do
-      if n.id == HLIST and n.subtype == BOX then
+      if n and n.id == HLIST and n.subtype == BOX then
          color_hbox(n, color)
       else
          color_node(n, color)
@@ -1084,14 +1093,14 @@ end
 local signature = function (node, string, swap)
   local n = node
   local str = string
-  if n.id == GLYPH then
+  if n and n.id == GLYPH then
     local b, id = is_glyph(n)
     if b and not char_to_discard[b] then
 %    \end{macrocode}
 %    Punctuation has to be discarded; the French apostrophe (right quote
 %    U+2019) has a char code ``out of range’’, we replace it with U+0027;
 %    Other glyphs should have char codes less than 0x100 (or 0x180?) or
-%    be ligatures… standard ones (U+FB00 to U+FB04) are converted using
+%    be ligatures… standard ones (U+FB00 to U+FB06) are converted using
 %    table |split_lig|.
 %    \begin{macrocode}
        if b == 0x2019 then b = 0x27 end
@@ -1103,9 +1112,19 @@ local signature = function (node, string, swap)
              c = string.reverse(c)
           end
           str = str .. c
+%    \end{macrocode}
+%    Experimental: store other ligatures as the last two digits of their
+%    decimal code…
+%    \begin{macrocode}
+       elseif n.subtype == LIGA and b > 0xE000 then
+          local c = string.sub(b,-2)
+          if swap then
+             c = string.reverse(c)
+          end
+          str = str .. c
        end
     end
-  elseif n.id == DISC then
+  elseif n and n.id == DISC then
 %    \end{macrocode}
 %    Ligatures are split into |pre| and |post| and both parts are
 %    stored. In case of \emph{ffl, ffi}, the post part is also
@@ -1129,12 +1148,18 @@ local signature = function (node, string, swap)
        end
     end
     if swap then
-       str = str .. c2  .. c1
+       str = str .. c2 .. c1
     else
        str = str .. c1 .. c2
     end
   end
+%    \end{macrocode}
+%    The returned length is the number of \emph{letters}.
+%    \begin{macrocode}
   local len = string.len(str)
+  if string.find(str, "_") then
+     len = len - 1
+  end
   return len, str
 end
 %    \end{macrocode}
@@ -1150,32 +1175,30 @@ end
 local check_last_word = function (old, node, color, flag)
   local match = false
   local new = ""
-  local len = 0
-  if flag then
+  local maxlen = 0
+  if flag and node then
+     local swap = true
 %    \end{macrocode}
-%    Get the last glyph one the line, skipping discretionaries,
-%    margin kerns, etc., unless |flag| cancels the whole process.
+%    Step back to the last glyph or discretionary.
 %    \begin{macrocode}
-     local swap = true
      local lastn = node
-     while lastn and lastn.id ~= GLYPH and lastn.prev do
+     while lastn and lastn.id ~= GLYPH and lastn.id ~= DISC do
        lastn = lastn.prev
      end
 %    \end{macrocode}
 %    A signature is built from the last two words on the current line.
 %    \begin{macrocode}
      local n = lastn
-     repeat
-       len, new = signature (n, new, swap)
+     while n and n.id ~= GLUE do
+       maxlen, new = signature (n, new, swap)
        n = n.prev
-     until not n or n.id == GLUE
+     end
      if n and n.id == GLUE then
         new = new .. "_"
-        len = len + 1
         repeat
           n = n.prev
-          len, new = signature (n, new, swap)
-        until n.id == GLUE or not n.prev
+          maxlen, new = signature (n, new, swap)
+        until not n or n.id == GLUE
      end
      new = string.reverse(new)
 %<dbg>  texio.write_nl('EOLsigold=' .. old)
@@ -1184,24 +1207,24 @@ local check_last_word = function (old, node, color, flag)
      local MinPart = luatypo.MinPart
      MinFull = math.min(MinPart,MinFull)
      local k = MinPart
-     local oldlast = string.gsub (old, '%w+_', '')
-     local newlast = string.gsub (new, '%w+_', '')
-     len = string.len(newlast)
-     if len < MinPart then
+     local oldlast = string.gsub (old, '.*_', '')
+     local newlast = string.gsub (new, '.*_', '')
+     local i = string.find(new, "_")
+     if i and i > maxlen - MinPart + 1 then
         k = MinPart + 1
      end
      local oldsub = string.sub(old,-k)
      local newsub = string.sub(new,-k)
-     local maxlen = string.len(new)
-     if oldsub == newsub then
+     local l = string.len(new)
+     if oldsub == newsub and l >= k then
 %<dbg>  texio.write_nl('EOLnewsub=' .. newsub)
         match = true
-     elseif oldlast == newlast and len >= MinFull then
+     elseif oldlast == newlast and string.len(newlast) >= MinFull then
 %<dbg>  texio.write_nl('EOLnewlast=' .. newlast)
         match = true
         oldsub = oldlast
         newsub = newlast
-        k = len
+        k = string.len(newlast)
      end
      if match then
 %    \end{macrocode}
@@ -1209,7 +1232,7 @@ local check_last_word = function (old, node, color, flag)
 %    \begin{macrocode}
         local osub = oldsub
         local nsub = newsub
-        while osub == nsub and k < maxlen do
+        while osub == nsub and k <= maxlen do
           k = k +1
           osub = string.sub(old,-k)
           nsub = string.sub(new,-k)
@@ -1217,8 +1240,9 @@ local check_last_word = function (old, node, color, flag)
              newsub = nsub
           end
         end
+        newsub = string.gsub(newsub, '^_', '')
 %<dbg>     texio.write_nl('EOLfullmatch=' .. newsub)
-%<msg>     texio.write_nl('***MATCH=' .. newsub ..
+%<msg>     texio.write_nl('***EOLMATCH=' .. newsub ..
 %<msg>                              " on page " .. pageno)
 %<msg>     texio.write_nl(' ')
 %    \end{macrocode}
@@ -1226,18 +1250,16 @@ local check_last_word = function (old, node, color, flag)
 %    \begin{macrocode}
         oldsub = string.reverse(newsub)
         local newsub = ""
-        local k = string.len(oldsub)
         local n = lastn
         repeat
           if n and n.id ~= GLUE then
              color_node(n, color)
-             len, newsub = signature(n, newsub, swap)
+             l, newsub = signature(n, newsub, swap)
           elseif n then
              newsub = newsub .. "_"
-             len = len + 1
           end
           n = n.prev
-        until not n or newsub == oldsub or len >= k
+        until not n or newsub == oldsub or l >= k
      end
   end
   return new, match
@@ -1252,19 +1274,23 @@ local check_first_word = function (old, node, color, flag)
   local match = false
   local swap = false
   local new = ""
-  local len = 0
+  local maxlen = 0
   local start = node
   local n = start
-  while n and n.id ~= GLUE do
-    len, new = signature (n, new, swap)
-    n = n.next
+  while n and n.id ~= GLYPH and n.id ~= DISC do
+     n = n.next
   end
-  n = n.next
-  new = new .. "_"
   while n and n.id ~= GLUE do
-    len, new = signature (n, new, swap)
+    maxlen, new = signature (n, new, swap)
     n = n.next
   end
+  if n and n.id == GLUE then
+     new = new .. "_"
+     repeat
+       n = n.next
+       maxlen, new = signature (n, new, swap)
+     until not n or n.id == GLUE
+  end
 %<dbg>  texio.write_nl('BOLsigold=' .. old)
 %<dbg>  texio.write('   BOLsig=' .. new)
 %    \end{macrocode}
@@ -1277,28 +1303,26 @@ local check_first_word = function (old, node, color, flag)
      local MinPart = luatypo.MinPart
      MinFull = math.min(MinPart,MinFull)
      local k = MinPart
-     local L = MinPart -1
      local oldsub = ""
      local newsub = ""
-     local oldfirst = string.gsub (old, '_%w+', '')
-     local newfirst = string.gsub (new, '_%w+', '')
-     len = string.len(newfirst)
-     if len < MinPart then
+     local oldfirst = string.gsub (old, '_.*', '')
+     local newfirst = string.gsub (new, '_.*', '')
+     local i = string.find(new, "_")
+     if i and i <= MinPart then
         k = MinPart + 1
      end
-     local maxlen = string.len(new)
-     local oldsub = string.sub(old,1,L)
-     local newsub = string.sub(new,1,L)
-     if oldsub == newsub then
+     local oldsub = string.sub(old,1,k)
+     local newsub = string.sub(new,1,k)
+     local l = string.len(newsub)
+     if oldsub == newsub and l >= k then
 %<dbg>  texio.write_nl('BOLnewsub=' .. newsub)
         match = true
-        k = L
-     elseif oldfirst == newfirst  and len >= MinFull then
+     elseif oldfirst == newfirst  and string.len(newfirst) >= MinFull then
 %<dbg>  texio.write_nl('BOLnewfirst=' .. newfirst)
         match = true
         oldsub = oldfirst
         newsub = newfirst
-        k = len
+        k = string.len(newfirst)
      end
      if match then
 %    \end{macrocode}
@@ -1306,7 +1330,7 @@ local check_first_word = function (old, node, color, flag)
 %    \begin{macrocode}
         local osub = oldsub
         local nsub = newsub
-        while osub == nsub and k < maxlen do
+        while osub == nsub and k <= maxlen do
           k = k + 1
           osub = string.sub(old,1,k)
           nsub = string.sub(new,1,k)
@@ -1314,15 +1338,11 @@ local check_first_word = function (old, node, color, flag)
              newsub = nsub
           end
         end
-        if k < MinPart then
-           match =false
-        end
-     end
-     if match then
-%<dbg>     texio.write_nl('BOLfullmatch=' .. newsub)
-%<msg>     texio.write_nl('***MATCH=' .. newsub ..
-%<msg>                              " on page " .. pageno)
-%<msg>     texio.write_nl(' ')
+        newsub = string.gsub(newsub, '_$', '')   --$
+%<dbg>  texio.write_nl('BOLfullmatch=' .. newsub)
+%<msg>  texio.write_nl('***BOLMATCH=' .. newsub ..
+%<msg>                           " on page " .. pageno)
+%<msg>  texio.write_nl(' ')
 %    \end{macrocode}
 %    Lest's colour the matching string.
 %    \begin{macrocode}
@@ -1333,13 +1353,12 @@ local check_first_word = function (old, node, color, flag)
         repeat
           if n and n.id ~= GLUE then
              color_node(n, color)
-             len, newsub = signature(n, newsub, swap)
+             l, newsub = signature(n, newsub, swap)
           elseif n then
              newsub = newsub .. "_"
-             len = len + 1
           end
           n = n.next
-        until not n or newsub == oldsub or len >= k
+        until not n or newsub == oldsub or l >= k
      end
   end
   return new, match
@@ -1370,9 +1389,9 @@ local check_regexpr = function (glyph)
         match = string.find(luatypo.single[lang], string.char(lchar))
         if match then
            color_node(glyph,COLOR)
-%<msg>   texio.write_nl('***MATCH=' .. string.char(lchar) ..
-%<msg>                            " on page " .. pageno)
-%<msg>   texio.write_nl(' ')
+%<msg>     texio.write_nl('***RGXMATCH=' .. string.char(lchar) ..
+%<msg>                              " on page " .. pageno)
+%<msg>     texio.write_nl(' ')
         end
      end
   end
@@ -1392,7 +1411,7 @@ local check_regexpr = function (glyph)
            if match then
               color_node(previous,COLOR)
               color_node(glyph,COLOR)
-%<msg>        texio.write_nl('***MATCH=' .. pattern ..
+%<msg>        texio.write_nl('***RGXMATCH=' .. pattern ..
 %<msg>                                 " on page " .. pageno)
 %<msg>        texio.write_nl(' ')
            end
@@ -1411,7 +1430,7 @@ local check_regexpr = function (glyph)
               if match then
                  color_node(pprev,COLOR)
                  color_node(glyph,COLOR)
-%<msg>           texio.write_nl('***MATCH=' .. pattern ..
+%<msg>           texio.write_nl('***RGXMATCH=' .. pattern ..
 %<msg>                                    " on page " .. pageno)
 %<msg>           texio.write_nl(' ')
               end
@@ -1430,7 +1449,7 @@ end
 %    \begin{macrocode}
 local show_pre_disc = function (disc, color)
   local n = disc
-  while n.id ~= GLUE do
+  while n and n.id ~= GLUE do
     color_node(n, color)
     n = n.prev
   end
@@ -1485,13 +1504,16 @@ luatypo.check_page = function (head)
 %    |hyphcount| hold the number the consecutive hyphenated lines.
 %    \begin{macrocode}
     if head.id == GLUE and head.subtype == TOPSKIP then
-       pageflag = false
-       match1 = false
        pageno = tex.getcount("c@page")
        hyphcount = 0
-       pagelines = 0
-       firstwd = ""
-       lastwd = ""
+       if pageno > prevno then
+          pageflag = false
+          pagelines = 0
+          match1 = false
+          firstwd = ""
+          lastwd = ""
+          prevno = pageno
+       end
     elseif head.id == HLIST and head.subtype == LINE then
 %    \end{macrocode}
 %    The current node is a line, |first| is the line’s first node.
@@ -1507,30 +1529,34 @@ luatypo.check_page = function (head)
 %    \end{macrocode}
 %    Is this line really a text line (one glyph at least)?
 %    \begin{macrocode}
-       local textline = true
-       local n = first
-       while (n and n.id ~= GLYPH) and
-             not (n.id == GLUE and n.subtype == RGTSKIP) do
-         n = n.next
-       end
-       if n.id == GLUE then
-          textline = false
+       local textline = false
+       if first.id == GLYPH then
+          textline = true
+       else
+          local n = first
+          repeat
+            n = n.next
+            if n and n.id == GLYPH then
+               textline = true
+               break
+            end
+          until not n or (n.id == GLUE and n.subtype == RGTSKIP)
        end
 %    \end{macrocode}
 %    Is this line overfull or underfull?
 %    \begin{macrocode}
        if head.glue_set == 1 and head.glue_sign == 2 and
           head.glue_order == 0 and OverfullLines then
-%<msg>  texio.write_nl('***OVERFULL line on page ' .. pageno)
-%<msg>  texio.write_nl(' ')
+%<msg>    texio.write_nl('***OVERFULL line on page ' .. pageno)
+%<msg>    texio.write_nl(' ')
           pageflag = true
           local COLOR = luatypo.colortbl[7]
           color_line (head, COLOR)
        elseif head.glue_set >= Stretch and head.glue_sign == 1 and
               head.glue_order == 0  and UnderfullLines then
-%<msg>  texio.write_nl('***UNDERFULL line on page ' ..
-%<msg>                                    tex.getcount("c@page"))
-%<msg>  texio.write_nl(' ')
+%<msg>    texio.write_nl('***UNDERFULL line on page ' ..
+%<msg>                                      tex.getcount("c@page"))
+%<msg>    texio.write_nl(' ')
           local COLOR = luatypo.colortbl[8]
           pageflag = true
           color_line (head, COLOR)
@@ -1540,7 +1566,7 @@ luatypo.check_page = function (head)
 %    \begin{macrocode}
        if first.id == LPAR then
 %    \end{macrocode}
-%    It starts a paragraph,
+%    It starts a paragraph…
 %    \begin{macrocode}
           hyphcount = 0
           parlines = 1
@@ -1586,7 +1612,7 @@ luatypo.check_page = function (head)
           orphanflag = false
 %    \end{macrocode}
 %    but it is a widow if it is the page’s first line and it does’nt
-%    start a new paragraph. Orphans and widows will be colored later.
+%    start a new paragraph.\\ Orphans and widows will be colored later.
 %    \begin{macrocode}
           if pagelines == 1 and parlines > 1 then
              widowflag = true
@@ -1605,8 +1631,8 @@ luatypo.check_page = function (head)
 %    |llwd| is the line’s length. Is it too short?
 %    \begin{macrocode}
              if llwd < LLminWD then
-%<msg>     texio.write_nl('***Last line too short, page ' .. pageno)
-%<msg>     texio.write_nl(' ')
+%<msg>          texio.write_nl('***Last line too short, page ' .. pageno)
+%<msg>          texio.write_nl(' ')
                 pageflag = true
                 local COLOR  = luatypo.colortbl[6]
                 local attr = oberdiek.luacolor.getattribute()
@@ -1620,8 +1646,8 @@ luatypo.check_page = function (head)
 %    Is this line nearly full? (ending too close to the right margin)
 %    \begin{macrocode}
           if BackParindent and PFskip < BackPI and PFskip > BackFuzz then
-%<msg>     texio.write_nl('***Last line nearly full, page ' .. pageno)
-%<msg>     texio.write_nl(' ')
+%<msg>       texio.write_nl('***Last line nearly full, page ' .. pageno)
+%<msg>       texio.write_nl(' ')
              pageflag = true
              local COLOR  = luatypo.colortbl[12]
              local attr = oberdiek.luacolor.getattribute()
@@ -1657,8 +1683,8 @@ luatypo.check_page = function (head)
              local COLOR = luatypo.colortbl[2]
              local pg = show_pre_disc (pn,COLOR)
              pageflag = true
-%<msg>     texio.write_nl('***HYPH issue page ' .. pageno)
-%<msg>     texio.write_nl(' ')
+%<msg>       texio.write_nl('***HYPH issue page ' .. pageno)
+%<msg>       texio.write_nl(' ')
           end
           if not nextnode and EOPHyphens then
 %    \end{macrocode}
@@ -1758,9 +1784,9 @@ luatypo.check_page = function (head)
        if lwhyphflag and EOPHyphens then
 %<msg>    texio.write_nl('***LAST WORD SPLIT page ' .. pageno)
 %<msg>    texio.write_nl(' ')
+          pageflag = true
           local COLOR = luatypo.colortbl[1]
           local pg = show_pre_disc (pn,COLOR)
-          pageflag = true
        end
 %    \end{macrocode}
 %    Track empty pages: check the number of lines at end of page.
@@ -1776,20 +1802,27 @@ luatypo.check_page = function (head)
 %<msg>    texio.write_nl('***Only ' .. pagelines ..
 %<msg>                          ' lines on page ' .. pageno)
 %<msg>    texio.write_nl(' ')
-          local node = head
-          while node.id ~= HLIST or node.subtype ~= LINE do
-             node = node.prev
+          local n = head
+          while n and (n.id ~= HLIST or n.subtype ~= LINE) do
+             n = n.prev
+          end
+          if n then
+             color_line(n, COLOR)
           end
-          color_line(node, COLOR)
        end
     end
   head = nextnode
   end
 %    \end{macrocode}
 %    Add this page number to the summary if any flaw has been found on it.
+%    Skip duplicates.
 %    \begin{macrocode}
   if pageflag then
-     luatypo.pagelist = luatypo.pagelist .. tostring(pageno) .. ", "
+     local pl = luatypo.pagelist
+     local p = tonumber(string.match(pl, "%s(%d+),%s$"))
+     if not p or pageno > p then
+        luatypo.pagelist = luatypo.pagelist .. tostring(pageno) .. ", "
+     end
   end
   return true
 end