path: root/tex/context/base/mkxl/lang-hup.lmt

if not modules then modules = { } end modules ['lang-hup'] = {
    version   = 1.001,
    comment   = "companion to lang-hup.mkiv",
    author    = "Hans Hagen, PRAGMA-ADE, Hasselt NL",
    copyright = "PRAGMA ADE / ConTeXt Development Team",
    license   = "see context related readme files"
}

local type, next = type, next
local utfchar = utf.char
local concat, sortedhash = table.concat, table.sortedhash
local basename = file.basename

local status        = status
local nodes         = nodes

local is_letter     = characters.is_letter
local is_hyphenator = characters.is_hyphenator

local specialskips  = nodes.specialskipcodes
local nodecodes     = nodes.nodecodes
local disc_code     = nodecodes.disc
local glyph_code    = nodecodes.glyph
local glue_code     = nodecodes.glue
local hlist_code    = nodecodes.hlist
local kern_code     = nodecodes.kern
local par_code      = nodecodes.par
local line_code     = nodes.listcodes.line
local fontkern_code = nodes.kerncodes.fontkern
local nuts          = nodes.nuts
local getlist       = nuts.getlist
local getnext       = nuts.getnext
local getprev       = nuts.getprev
local getid         = nuts.getid
local getsubtype    = nuts.getsubtype
local getreplace    = nuts.getreplace
local getdiscpart   = nuts.getdiscpart
local isnextglyph   = nuts.isnextglyph
local nexthlist     = nuts.traversers.list
local nextglyph     = nuts.traversers.glyph
local traverse      = nuts.traverse

local setcolor      = nodes.tracers.colors.set
local setaction     = nodes.tasks.setaction

local hash          = table.setmetatableindex("number")

local report        = logs.reporter("hyphenated")
local trace_detail  = false

local characters    = fonts.hashes.characters

local word = { }
local w    = 0

----- function collect(word,head)
local function collect(head)
    local last = nil
    while head do
        local nxt, char, id = isnextglyph(head)
        if char then
            local u = characters[id][char].unicode -- we could cache it
            if type(u) == "table" then
                for i=1,#u do
                    local c = u[i]
                    if is_letter[c] or is_hyphenator[c] then
                     -- word[#word+1] = utfchar(c)
                        w = w + 1 ; word[w] = utfchar(c)
                    end
                end
            else
                local c = u or char
                if is_letter[c] or is_hyphenator[c] then
                 -- word[#word+1] = utfchar(c)
                    w = w + 1 ; word[w] = utfchar(c)
                end
            end
            last = head
        elseif id == disc_code then
         -- collect(word,getreplace(head))
            collect(getreplace(head))
        elseif id == kern_code and getsubtype(head,fontkern_code) then
            -- we're ok
        else
            break
        end
        head = nxt
    end
    return last and getdiscpart(last) == 1
end

local function getpostpart(current)
    for n, id, subtype in traverse(current) do
        if id == glue_code then
            if not specialskips[subtype] then
                break
            end
        elseif id == glyph_code or id == disc_code then
            return n
        elseif id ~= par_code then
            break
        end
    end
end

local function findprepart(current)
    for n, id, subtype in traverse(current,true,true) do
        if id == glue_code then
            if not specialskips[subtype] then
                break
            end
        elseif id == glyph_code and getdiscpart(n) == 1 then
            return n
        else
            break
        end
    end
end

local function getprepart(disc)
    local back = disc
    for n, id, subtype in traverse(getprev(disc),true) do
        if id == glyph_code or id == disc_code or (id == kern_code and subtype == fontkern_code) then
            back = n
        else
            return back
        end
    end
    return back
end

function nodes.handlers.showhyphenation(head)
 -- local word = { }
    w = 0
    for current, id, subtype, list in nexthlist, head do
        if list and subtype == line_code then
         -- if #word > 0 then
            if w > 0 then
                local after = getpostpart(list)
             -- local more  = collect(word,after)
                local more  = collect(after)
                if more then
                    goto skip
                else
                 -- local result = concat(word)
                    local result = concat(word,"",1,w)
                    if trace_detail then
                        local r = status.readstate
                        report("around line %s in file %s: %s",r.linenumber or "-",basename(r.filename),result)
                    end
                    hash[result] = hash[result] + 1
                 -- word = { }
                    w = 0
                end
            end
            local last = findprepart(list)
            if last then
                local before = getprepart(last)
             -- collect(word,before)
                collect(before)
            end
          ::skip::
        end
    end
    return head
end

local initialize  initialize = function()
    logs.registerfinalactions(function()
        logs.startfilelogging(report,"hyphenated words")
        if hash and next(hash) then
            local oldname = file.nameonly(tex.jobname) .. "-hyphenated-words-old.txt"
            local newname = file.nameonly(tex.jobname) .. "-hyphenated-words-new.txt"
            local old = string.splitlines(string.strip(io.loaddata(oldname) or "")) or { }
            local hsh = table.tohash(old)
            local new = { }
            for word, count in sortedhash(hash) do
                report("%4i : %s",count,word)
                if not hsh[word] then
                    new[#new+1] = word
                end
            end
            logs.stopfilelogging()
            report("old word list : %a",oldname)
            report("new word list : %a",newname)
            report("to be checked : %a",#new)
            io.savedata(newname,concat(new,"\n"))
        else
            report("nothing hyphenated")
            logs.stopfilelogging()
        end
    end)
    --
    statistics.register("hyphenation",function()
        local n = 0
        local m = 0
        for k, v in sortedhash(hash) do
            n = n + 1
            m = m + v
        end
        return string.format("%i hyphenated words, %i unique words",m,n)
    end)
    initialize = false
end

trackers.register("hyphenation.applied", function(v)
    setaction("finalizers","nodes.handlers.showhyphenation",v)
    if v and initialize then
        initialize()
    end
end)

trackers.register("hyphenation.applied.console", function(v)
    setaction("finalizers","nodes.handlers.showhyphenation",v)
    trace_detail = v
    if v and initialize then
        initialize()
    end
end)

-- local c, f = isglyph(current)
-- local char = chardata[f][c]
-- if char and type(char.unicode) == "table" then -- hackery test

local ligature_code = 0x8000 + nodes.glyphcodes.ligature
local ligature_mode = false

local color_n = { "red",     "green",     "blue"     }
local color_l = { "darkred", "darkgreen", "darkblue" }

function nodes.handlers.visualizehyphenation(head)
    for current, id, subtype, list in nexthlist, head do
        if list and subtype == line_code then
            if ligature_mode then
                for n in nextglyph, list do
                    local d = getdiscpart(n)
                    local s = getsubtype(n)
                    if d > 0 and d < 4 then
                        if s == ligature_code then
                            setcolor(n,color_l[d])
                        else
                            setcolor(n,color_n[d])
                        end
                    elseif s == ligature_code then
                        setcolor(n,"darkgray")
                    end
                end
            else
                for n in nextglyph, list do
                    local d = getdiscpart(n)
                    if d > 0 and d < 4 then
                        setcolor(n,color_n[d])
                    end
                end
            end
        end
    end
    return head
end

trackers.register("hyphenation.applied.visualize", { true, false, "ligatures" }, function(v)
    setaction("finalizers","nodes.handlers.visualizehyphenation",v)
    ligature_mode = v == "ligatures"
end)