if not modules then modules = { } end modules ['lpdf-tag'] = {
version = 1.001,
comment = "companion to lpdf-tag.mkiv",
author = "Hans Hagen, PRAGMA-ADE, Hasselt NL",
copyright = "PRAGMA ADE / ConTeXt Development Team",
license = "see context related readme files"
}
-- instead of taglist we can have a backpointer to the parent but then
-- we also need to adapt the export, it might be a bit slower
--
-- maybe also make specifications a two dimensional table
local next, type, tonumber = next, type, tonumber
local format, match, gmatch, find, gsub = string.format, string.match, string.gmatch, string.find, string.gsub
local concat, sortedhash = table.concat, table.sortedhash
local lpegmatch, P, S, C = lpeg.match, lpeg.P, lpeg.S, lpeg.C
local settings_to_hash = utilities.parsers.settings_to_hash
local formatters = string.formatters
local trace_tags = false
local trace_info = false
local trace_math = false
local trace_blobs = false
local trace_internals = false
local trace_suspects = false
local trace_tree = false
trackers.register("structures.tags", function(v) trace_tags = v end)
trackers.register("structures.tags.info", function(v) trace_info = v end)
trackers.register("structures.tags.math", function(v) trace_math = v attributes.viewerlayers.enable() end) -- somehow has to happen
trackers.register("structures.tags.blobs", function(v) trace_blobs = v end)
trackers.register("structures.tags.internals", function(v) trace_internals = v end)
trackers.register("structures.tags.suspects", function(v) trace_suspects = v end)
trackers.register("structures.tags.showtree", function(v) trace_tree = v end)
local detailedmath = false
local actualtexts = { }
local checklinks = true
directives.register("structures.tags.math.detail", function(v) detailedmath = v end)
local report_tags = logs.reporter("backend","tags")
local pdfbackend = backends.registered.pdf
local nodeinjections = pdfbackend.nodeinjections
local codeinjections = pdfbackend.codeinjections
local enableaction = nodes.tasks.enableaction
local disableaction = nodes.tasks.disableaction
local lpdf = lpdf
local pdfdictionary = lpdf.dictionary
local pdfarray = lpdf.array
local pdfboolean = lpdf.boolean
local pdfconstant = lpdf.constant
local pdfreference = lpdf.reference
local pdfunicode = lpdf.unicode
local pdfstring = lpdf.string
local pdfmakenametree = lpdf.makenametree
local addtocatalog = lpdf.addtocatalog
local addtopageattributes = lpdf.addtopageattributes
local pdfflushobject = lpdf.flushobject
local pdfreserveobject = lpdf.reserveobject
local pdfpagereference = lpdf.pagereference
local version = 1
local texgetcount = tex.getcount
local nodes = nodes
local nodecodes = nodes.nodecodes
local par_code <const> = nodecodes.par
local hlist_code <const> = nodecodes.hlist
local vlist_code <const> = nodecodes.vlist
local glyph_code <const> = nodecodes.glyph
local rule_code <const> = nodecodes.rule
local glue_code <const> = nodecodes.glue
local leaders_code <const> = nodes.gluecodes.leaders
local empty_rule_code <const> = nodes.rulecodes.empty
local a_tagged <const> = attributes.private('tagged')
local a_image <const> = attributes.private('image')
local a_mathblob <const> = attributes.private('mathblob')
local a_taggedpar <const> = attributes.private("taggedpar")
local nuts = nodes.nuts
local nodepool = nuts.pool
local setstate = nodepool.setstate
local register = nodepool.register
local getid = nuts.getid
local getattr = nuts.getattr
local getattrs = nuts.getattrs
local getprev = nuts.getprev
local getnext = nuts.getnext
local getlist = nuts.getlist
local getchar = nuts.getchar
local getwhd = nuts.getwhd
local getleader = nuts.getleader
local getlanguage = nuts.getlanguage
local getruledimensions = nuts.getruledimensions
local tailoflist = nuts.tail
local setlink = nuts.setlink
local setlist = nuts.setlist
local copy_node = nuts.copy
local tosequence = nuts.tosequence
local nextnode = nuts.traversers.node
local nextcontent = nuts.traversers.content
local structure_kids -- delayed
local structure_ref -- delayed
local parent_ref -- delayed
local root -- delayed
local names = { }
local tree = { }
local firstintree = false
local lastintree = false
local elements = { }
local elementsorder = { }
local nofelements = 0
local structurestags = structures.tags
local taglist = structurestags.taglist
local specifications = structurestags.specifications
local usedlabels = structurestags.labels
local properties = structurestags.properties
local overloads = structurestags.overloads
local usewithcare = structurestags.usewithcare
local pushtag = structurestags.push
local poptag = structurestags.pop
local starttag = structurestags.start
local stoptag = structurestags.stop
local usedmapping = { }
local destinations = { }
local references = { }
----- tagsplitter = structurestags.patterns.splitter
-- We used to have a way to embed mathml and bib blobs independent of tagging but
-- that was actually never really used. It used attachments and text notes but is
-- was not really supported by viewers so we never advertized it.
local embeddedtags = false
local embeddedfilelist = pdfarray()
local blobfunctions = { }
local tagtracers = { }
local indirectlocalkids = false
local indirectglobalkids = false
-- When we are generating more pages this will be invalid but there will be an extra
-- run anyway due to storing the final value.
--
-- For some reason links needa struct parent an destinations an object ... smells like
-- application driven stuff.
local pagenumindices = { }
local usedpages = false -- internal -> structure
local linknumoffset = false
local linknumindex = 0
local linknumindices = { }
local linknumentries = table.setmetatableindex(function(t,refatt)
if not linknumoffset then
local tagging = job.variables.collected.tagging
if tagging then
linknumoffset = tagging.basetreesize
else
-- linknumoffset = structures.counters.record("realpage")["last"] + 100
linknumoffset = 1000
end
end
local v = linknumindex + linknumoffset
linknumindex = linknumindex + 1
linknumindices[linknumindex] = false
-- attribute -> parentindex
t[refatt] = v
return v
end)
local function setlinkstructureparent(refatt,objref)
local p = linknumentries[refatt]
-- print("SET",refatt,p,objref)
linknumindices[p] = objref
end
function codeinjections.getlinkstructureparent(refatt)
-- print("GET",refatt, linknumentries[refatt])
return linknumentries[refatt]
end
local referencenumindices = { } -- attribute -> structure
local useddestinations = false -- internal -> structure
function codeinjections.getreferencestructureobject(internal,page)
if not useddestinations then
local tagging = job.variables.collected.tagging
useddestinations = tagging and tagging.destinations or { }
usedpages = tagging and tagging.pages or { }
end
return (internal and useddestinations[internal]) or (page and usedpages[page])
end
--
local function embedsupportedtags(str)
if not embeddedtags then
embeddedtags = { }
end
if str then
for tag in gmatch(str,"([^, ]+)") do
embeddedtags[tag] = true
end
else
for tag in next, blobfunctions do
embeddedtags[tag] = true
end
end
end
directives.register("structures.tags.embed",function(v)
embedsupportedtags(type(v) == "string" and v or nil)
end)
function codeinjections.maptag(original,target,kind)
mapping[original] = { target, kind or "inline" }
end
local namespaces = false
local namespaces = {
mathml = {
url = "http://www.w3.org/1998/Math/MathML",
force = false,
known = true,
map = { },
},
context = {
url = "http://www.contextgarden.net/pdf/context",
map = { },
},
user = {
url = "http://www.contextgarden.net/pdf/user",
map = { },
},
ua1 = {
url = "http://iso.org/pdf/ssn",
force = false,
known = true,
},
ua2 = {
url = "http://iso.org/pdf2/ssn",
force = false,
known = true,
},
}
local missingua = table.setmetatableindex(function(t,k)
local v = "Div"
t[k] = v
return v
end)
statistics.register("pdf tags", function()
local k, v = next(missingua)
if k then
return format("unknown tags mapped to %s: % t",v,table.sortedkeys(missingua))
end
end)
local function concattags(tags)
local l = { } -- table.new
local t = tags.taglist
local n = #t
for i=1,n do
l[i] = t[i] or "ERROR"
end
return concat(l," ",1,n)
end
local function checkoverload(tagname,tagnameused,detail,specification)
local o = overloads[tagnameused]
if o then
local d = detail
local c = o.criterium
if c == "parent" or c == "parents" then
local p = specification.taglist[#specification.taglist-1]
local s = specifications[p]
if c == "parents" then
p = s.parents -- special case
if p then
d = match(p,"%S+")
else
d = s.detail -- just in case
end
else
d = s.detail
end
end
if d then
d = o.mapping[d]
if d then
tagname = d.tag
tagnameused = tagname
detail = nil
end
end
end
return tagname, tagnameused, detail
end
local function finishstructure()
if root and #structure_kids > 0 then
local nums = pdfarray()
local n = 0
if indirectglobalkids then
for i=firstintree,lastintree do
local ti = tree[i]
if ti then
n = n + 1 ; nums[n] = i - 1
n = n + 1 ; nums[n] = pdfreference(pdfflushobject(ti))
else
report_tags("beware: missing page %i in tree", i)
end
end
else
for i=firstintree,lastintree do
local ti = tree[i]
if not ti then
report_tags("beware: missing page %i in tree", i)
elseif #ti > 0 then
n = n + 1 ; nums[n] = i - 1
n = n + 1 ; nums[n] = ti
end
end
end
local usedrolemap = nil
local usednamespaces = nil
local originals = { }
if version == 1 then
for k, v in next, usedmapping do
local k = usedlabels[k] or k
local p = properties[k]
if p then
local pdf = p.pdf
if not pdf then
pdf = missingua[pdf]
end
local r = pdfconstant(pdf)
if usedrolemap then
usedrolemap[k] = r
else
usedrolemap = pdfdictionary { [k] = r }
end
else
report_tags("beware: missing property %s", k)
end
end
else
for k, v in next, usedmapping do
local k = usedlabels[k] or k
local p = properties[k]
if p then
local s = p.namespace
local n = namespaces[s]
if n then
local pdf = p.pdf
local pua = p.pua
local ua = n
if not pdf then
pdf = missingua[pdf]
end
if pua == "ua1" then
ua = namespaces.ua1
ua.force = true
elseif pua == "mathml" then
ua = namespaces.mathml
ua.force = true
else
ua = namespaces.ua2
ua.force = true
end
--
-- Another piece of mid 2025 crap that we now always need to add but we keep
-- the test commented as reference. The more bloat the better I guess.
--
-- if pdf ~= k then
ua = pdfarray { pdfconstant(pdf), ua.ref }
n.map[k] = ua
-- end
--
if not usednamespaces then
usednamespaces = pdfarray()
end
--
local original = p.original
if original then
local o = originals[s]
if not o then
o = pdfdictionary()
originals[s] = o
end
o[k] = pdfdictionary {
S = pdfconstant(original[1]),
T = pdfstring(original[2]),
}
end
--
else
report_tags("beware: missing namespace %s", s)
end
else
report_tags("beware: missing property %s", k)
end
end
if usednamespaces then
for k, v in sortedhash(namespaces) do
local o = originals[k]
if o then
o = pdfreference(pdfflushobject(o))
end
if v.force then
local n = pdfdictionary {
Type = pdfconstant("Namespace"),
NS = pdfunicode(v.url),
LMTX_NameSpace = pdfconstant(k),
LMTX_Originals = o,
}
pdfflushobject(v.num,n)
usednamespaces[#usednamespaces+1] = v.ref
else
local map = v.map
if map and next(map) then
local m = pdfreference(pdfflushobject(pdfdictionary(map)))
local n = pdfdictionary {
Type = pdfconstant("Namespace"),
NS = pdfunicode(v.url),
RoleMapNS = m,
LMTX_NameSpace = pdfconstant(k),
LMTX_Originals = o,
}
pdfflushobject(v.num,n)
usednamespaces[#usednamespaces+1] = v.ref
end
end
end
end
end
if indirectlocalkids then
for i=1,nofelements do
local fulltag = elementsorder[i]
local element = elements[fulltag]
local kids = element.kids
local knum = element.knum
if checklinks then
local tag = element.tag
if tag == "link" then
local ref = element.ref
if ref then
setlinkstructureparent(ref,element.pref)
end
elseif tag == "reference" then
local des = element.des
if des then
referencenumindices[des] = element.dnum
end
end
end
-- if checklinks then
-- for i=1,#kids do
-- local d = kids[i]
-- if type(d) == "table" then
-- local refatt = element.refatt
-- if refatt then
-- local refobj = codeinjections.getrefobj(refatt)
-- if refobj then
-- d.Obj = pdfreference(refobj)
-- end
-- end
-- end
-- end
-- end
pdfflushobject(knum,kids)
end
else
for i=1,nofelements do
local fulltag = elementsorder[i]
local element = elements[fulltag]
local dict = element.dict
if dict then
local dnum = element.dnum
local kids = element.kids
if checklinks then
local tag = element.tag
if tag == "link" then
local ref = element.ref
if ref then
setlinkstructureparent(ref,element.pref)
end
elseif tag == "reference" then
local des = element.des
if des then
referencenumindices[des] = element.dnum
end
elseif tag == "navigationpage" then
pagenumindices[element.pnum] = element.dnum
end
end
-- if checklinks then
-- for i=1,#kids do
-- local d = kids[i]
-- if type(d) == "table" then
-- local refatt = element.refatt
-- if refatt then
-- local refobj = codeinjections.getrefobj(refatt)
-- if refobj then
-- d.Obj = pdfreference(refobj)
-- end
-- end
-- end
-- end
-- end
-- if #kids == 1 then
-- dict.K = kids[1] -- always okay ?
-- end
pdfflushobject(dnum,dict)
end
end
end
--
local getinternalreference = structures.references.getinternalreference
local destinations = { }
for k, v in next, referencenumindices do
destinations[getinternalreference(k)] = v
end
job.variables.tobesaved.tagging = {
basetreesize = lastintree,
destinations = destinations,
pages = pagenumindices,
}
--
table.setmetatableindex(linknumindices)
-- for k, v in next, linknumindices do
-- print("CHECK",k,v)
-- end
for k, v in sortedhash(linknumindices) do
if v then
n = n + 1 ; nums[n] = k
n = n + 1 ; nums[n] = v
end
end
--
local parenttree = pdfdictionary {
Nums = nums
}
local idtree = pdfmakenametree(names)
--
local structuretree = pdfdictionary {
Type = pdfconstant("StructTreeRoot"),
K = pdfreference(pdfflushobject(structure_kids)),
ParentTree = pdfreference(pdfflushobject(parent_ref,parenttree)),
IDTree = idtree,
RoleMap = usedrolemap,
Namespaces = usednamespaces,
-- Experiment with some control (screen readers) but it didn't work out reliable
-- so forget about it.
-- ClassMap = pdfdictionary {
-- Display = pdfdictionary { O = pdfconstant("Layout"), Placement = pdfconstant("Block") },
-- Inline = pdfdictionary { O = pdfconstant("Layout"), Placement = pdfconstant("Inline") },
-- },
}
pdfflushobject(structure_ref,structuretree)
addtocatalog("StructTreeRoot",pdfreference(structure_ref))
--
-- So out of a sudden (sometime 2025) this is now mandate when a document is tagged. Documents
-- that validated before didn't after updating the validator. Well ... but how about those older
-- documents that were found okay before?
--
-- if version == 1 then -- always was optional anyway
local markinfo = pdfdictionary {
Marked = pdfboolean(true) or nil,
-- UserProperties = pdfboolean(true), -- maybe some day
-- Suspects = pdfboolean(true) or nil,
-- AF = #embeddedfilelist > 0 and pdfreference(pdfflushobject(embeddedfilelist)) or nil,
}
addtocatalog("MarkInfo",pdfreference(pdfflushobject(markinfo)))
-- end
end
if trace_tree then
local p = false
local n = tonumber(trace_tree) or true
for i=1,nofelements do
local fulltag = elementsorder[i]
local element = elements[fulltag]
local specification = specifications[fulltag]
local pagenumber = element.pnum
if n == true and i > 1 and p ~= pagenumber then
report_tags("")
end
p = pagenumber
if n == true or n == p then
report_tags("% 5i %3i %s (%s)",i,pagenumber,concattags(specification),(element.des and "destination") or (element.ref and "reference") or "content")
end
end
end
if trace_blobs then
for k, v in sortedhash(actualtexts) do
local shared = v[1] and "-" or "+"
local index = v[2]
local text = v[3]
report_tags("blob %s : %5i : %04X : %s",shared,index,index,text)
end
end
end
lpdf.registerdocumentfinalizer(finishstructure,"document structure")
local index, pageref, pagenum, list = 0, nil, 0, nil
local pdf_mcr = pdfconstant("MCR")
local pdf_struct_element = pdfconstant("StructElem")
local pdf_s = pdfconstant("S")
local pdf_objr = pdfconstant("OBJR")
local c_realpageno <const> = tex.iscount("realpageno")
local function initializepage()
index = 0
pagenum = texgetcount(c_realpageno)
pageref = pdfreference(pdfpagereference(pagenum))
list = pdfarray()
if not namespaced then
for k, v in sortedhash(namespaces) do
v.num = pdfreserveobject()
v.ref = pdfreference(v.num)
end
namespaced = true
end
-- hm, can be later than 1
if not firstintree then
if pagenum > 1 then
report_tags("beware: first page in tree is %i", pagenum)
end
firstintree = pagenum
lastintree = pagenum
end
if pagenum > lastintree then
lastintree = pagenum
else
-- report_tags("beware: page order problem in tree at page %i", pagenum)
end
tree[pagenum] = list -- we can flush after done, todo
end
local function finishpage()
-- flush what can be flushed
addtopageattributes("StructParents",pagenum-1)
end
-- here we can flush and free elements that are finished
local pdf_userproperties = pdfconstant("UserProperties")
-- /O /Table
-- /Headers [ ]
local function makeattribute(t)
if t and next(t) then
local properties = pdfarray()
for k, v in sortedhash(t) do -- easier on comparing pdf
properties[#properties+1] = pdfdictionary {
N = pdfunicode(k),
V = pdfunicode(v),
}
end
return pdfdictionary {
O = pdf_userproperties,
P = properties,
}
end
end
local visualizetags = nil
local visualizespecial = nil
local visualizeblobs = nil
local visualizesuspects = nil
local visualizeinternals = nil
local collectedsuspects = { }
statistics.register("pdf tags", function()
if #collectedsuspects > 0 then
return formatters["suspects: % t"](collectedsuspects)
end
end)
local function checkvisualize()
if not visualizetags then
visualizetags = nodes.visualizers.register("tags")
visualizespecials = nodes.visualizers.register("specials",nil,nil,2.5,true)
visualizeblobs = nodes.visualizers.register("blobs",nil,nil,2.5,true)
visualizesuspects = nodes.visualizers.register("suspects")
visualizeinternals = nodes.visualizers.register("internals")
end
end
local function tagtracer(name,blob)
checkvisualize()
return visualizetags(name), visualizetags()
end
table.setmetatableindex(tagtracers,function(t,k)
t[k] = tagtracer
return tagtracer
end)
local makeelement do
do
local f_tagid = formatters["math-%04X"] -- todo: auto adapt to nofblobs
local f_tagfn = formatters["math-%04X.xml"]
local shared = { }
local bindex = 0
local btags = false
function blobfunctions.math(tagname,specification)
local tagindex = specification.tagindex
local id = f_tagid(tagindex)
local blob = specification.blob
if blob then
local blobindex = mathematics.getblobindex("pdf",blob)
if blobindex then
local index = shared[blobindex]
if not btags then
btags = { }
job.variables.tobesaved.mathblobs = btags
end
if index then
af = index[1]
btags[blob] = index[2]
else
bindex = bindex + 1
local blobname = f_tagid(blobindex)
local blobfile = f_tagfn(blobindex)
local blobdata = mathematics.getmathblob("pdf",blob)
af = codeinjections.embedfile {
force = true,
data = blobdata,
name = blobname,
file = blobfile,
-- hash = hash,
hash = id,
forcereference = true,
-- title = "whatever",
mimetype = "application/mathml+xml",
relation = "Supplement", -- bah
}
af = pdfreference(pdfflushobject(pdfarray { af })) -- maybe also share this
shared[blobindex] = { af, blobindex }
btags[blob] = blobindex
end
actualtext = mathematics.gettextblob("pdf",specification.language or "en",blob)
if actualtext then
if trace_blobs then
actualtexts[blob] = { index and true or false, blobindex, actualtext }
end
actualtext = pdfunicode(actualtext)
end
return id, af, actualtext
end
else
-- af = job.fileobjreferences.collected[id]
-- if af then
-- local r = pdfreference(af)
-- af = pdfarray { r }
-- -- embeddedfilelist[#embeddedfilelist+1] = r
-- end
end
end
local blobdone = { }
function tagtracers.math(name,specification,blob)
checkvisualize()
if blob and not blobdone[blob] then
local bname = btags[blob]
if bname then
bname = "M " .. blob .. " " .. f_tagid(bname)
else
bname = "M " .. blob
end
blobdone[blob] = true
if detailedmath then
return visualizeblobs(bname,name), visualizeblobs()
else
return visualizetags(bname), visualizetags()
end
else
return visualizetags(name), visualizetags()
end
end
function tagtracers.suspect(name)
checkvisualize()
return visualizesuspects("S " .. "mrow"), visualizesuspects()
end
function tagtracers.internallink(internal)
checkvisualize()
return visualizeinternals("L " .. internal), visualizeinternals()
end
function tagtracers.internalreference(internal)
checkvisualize()
return visualizeinternals("R " .. internal), visualizeinternals()
end
end
do
function tagtracers.link(name,specification,blob)
checkvisualize()
return visualizespecials(name), visualizespecials()
end
function tagtracers.reference(name,specification,blob)
checkvisualize()
return visualizespecials(name), visualizespecials()
end
end
do
local f_tagid = formatters["cite-%s"] -- todo: auto adapt to nofblobs
local f_tagfn = formatters["cite-%s.bib"]
local shared = { }
local bindex = 0
local btags = { }
function blobfunctions.cite(tagname,specification)
local detail = specification.detail
if detail then
local dataset, tag = match(detail,"^(.+)::(.+)$")
local index = shared[tag]
local id = f_tagid(tag)
if index then
af = index[1]
btags[tag] = index[2]
else
bindex = bindex + 1
local data = publications.datasets[dataset].luadata[tag] or "no data"
local blobname = f_tagid(tag)
local blobfile = f_tagfn(tag)
local blobdata = publications.savers.bib(false,false,{ [tag] = data }) -- converttoxml(dataset,true,false,true,false,true,true)
af = codeinjections.embedfile {
force = true,
data = gsub(blobdata,"\n+$",""),
name = blobname,
file = blobfile,
-- hash = hash,
hash = id,
forcereference = true,
-- title = "whatever",
mimetype = "application/x-bibtex",
relation = "Supplement", -- bah
}
-- af = pdfarray { af } -- maybe also share this
af = pdfreference(pdfflushobject(pdfarray { af })) -- maybe also share this
shared[tag] = { af, blobname }
btags[tag] = blobname
end
actualtext = publications.meanings[tag]
if actualtext then
actualtext = pdfunicode(actualtext)
end
return id, af, actualtext
end
end
function tagtracers.cite(name,specification)
checkvisualize()
local detail = specification.detail
if detail then
local dataset, tag = match(detail,"^(.-)::(.-)$")
local bname = btags[tag]
if bname then
return visualizetags("C " .. bname), visualizetags()
end
end
return visualizetags(name), visualizetags()
end
end
local lastid = 0
local f_id = formatters["%X"]
local symbols = table.setmetatableindex (
{ -- None
["1"] = "Disc",
["2"] = "Circle",
["3"] = "Square",
["n"] = "Decimal",
["I"] = "UpperRoman",
["i"] = "LowerRoman",
["A"] = "UpperAlpha",
["a"] = "LowerAlpha",
},
function(t,k) return tonumber(k) and "Unordered" or "Ordered" end
)
makeelement = function(fulltag,parent)
local specification = specifications[fulltag]
local tagname = specification and specification.tagname or "ignore"
local tagnameused = tagname
local attributes = nil
-- some catches .. todo
if tagname == "ignore" then
return false
elseif tagname == "mstacker" or tagname == "mstackertop" or tagname == "mstackerbot" or tagname == "mstackermid" then
-- test this in an mp stacker in the math manual, basically any private one
return true
elseif tagname == "mrow" then -- todo: alttext
return false
elseif tagname == "tabulatecell" then
local d = structurestags.gettabulatecell(fulltag)
if d and d.kind == 1 then
tagnameused = "tabulateheadcell"
end
-- elseif tagname == "tablecell" then
elseif tagname == "tablecell" then
-- will become a plugin model
local d = structurestags.gettablecell(fulltag)
if d then
if d.kind == 1 then
tagnameused = "tableheadcell"
end
local rows = d.rows or 1
local cols = d.columns or 1
if rows > 1 or cols > 1 then
attributes = pdfdictionary {
-- The usual inconsistency on short and long keys:
O = pdfconstant("Table"),
RowSpan = rows > 1 and rows or nil,
ColSpan = cols > 1 and cols or nil,
}
end
end
elseif tagname == "itemgroup" then
local d = structurestags.getitemgroup(fulltag)
if d then
local symbol = d.symbol
if symbol then
attributes = pdfdictionary {
ListNumbering = pdfconstant(symbols[symbol] or "None"),
ContinuedList = d.continue and true or nil,
}
end
end
-- elseif tagname == "mtd" then
-- -- only when in detail mode
-- local cols = specification.cols or 1 -- we could set it beforehand ... todo
-- if cols > 1 then
-- attributes = pdfdictionary {
-- O = pdfconstant("Table"),
-- ColSpan = cols,
-- }
-- end
-- elseif tagname == "math" then
-- if specification.mode == "display" then
-- tagnameused = "displaymath"
-- else
-- tagnameused = "inlinemath"
-- end
end
--
local detail = specification.detail
local userdata = specification.userdata
--
if version == 1 then
-- not here
else
-- ugly hack
tagname, tagnameused, detail = checkoverload(tagname,tagnameused,detail,specification)
end
--
usedmapping[tagname] = true
--
-- specification.attribute is unique
--
local af = nil
local id = nil
local actualtext = nil
--
if embeddedtags[tagname] then
local action = blobfunctions[tagname]
if action then
id, af, actualtext = action(tagname,specification)
end
end
-- This can be an option but it bloats the file for little reason.
--
-- if not id then
-- lastid = lastid + 1
-- id = f_id(lastid)
-- end
--
local namespace = nil
if version > 1 then
local p = properties[tagname]
if p then
namespace = namespaces[p.namespace].ref or nil
else
namespace = "user"
properties[tagname] = { namespace = namespace, pdf = "Span", nature = "inline" }
end
end
--
-- local alternate = "Who cares"
local kids = pdfarray()
local tag = usedlabels[tagnameused] or tagnameused
local subtype = pdfconstant(tag)
local pref = parent.pref
local pkids = parent.kids
local element
local dref, dnum
if indirectlocalkids then
local knum = pdfreserveobject()
local dict = pdfdictionary {
-- Type = pdf_struct_element, -- optional, saves bytes
S = subtype,
ID = id,
T = detail and detail or nil,
P = pref,
Pg = pageref,
K = pdfreference(knum),
A = attributes,
Alt = actualtext or nil,
NS = namespace,
-- ActualText = actualtext or nil, -- shared object with Alt?
AF = af or nil,
}
dnum = pdfflushobject(dict)
dref = pdfreference(dnum)
element = {
blob = af and true or false, -- also actualtext
tag = tag,
pnum = pagenum,
pref = dref,
kids = kids,
knum = knum,
dnum = dnum,
ref = tag == "link" and specification.reference or nil,
des = tag == "reference" and specification.destination or nil,
}
else
local dict = pdfdictionary {
-- Type = pdf_struct_element, -- optional, saves bytes
S = subtype,
ID = id,
T = detail and detail or nil,
P = pref,
Pg = pageref,
K = kids,
A = attributes,
Alt = actualtext or nil,
NS = namespace,
-- ActualText = actualtext or nil, -- shared object with Alt?
AF = af or nil,
}
dnum = pdfreserveobject()
dref = pdfreference(dnum)
element = {
blob = af and true or false, -- also actualtext
tag = tag,
pnum = pagenum,
pref = dref,
kids = kids,
dict = dict,
dnum = dnum,
ref = tag == "link" and specification.reference or nil,
des = tag == "reference" and specification.destination or nil,
}
end
if id and names then
names[id] = dref
end
pkids[#pkids+1] = dref
elements[fulltag] = element
nofelements = nofelements + 1
elementsorder[nofelements] = fulltag
return element
end
end
local f_BDC = formatters["/%s <</MCID %s>> BDC"]
local a_destination <const> = attributes.private('destination')
local a_reference <const> = attributes.private('reference')
local references = { }
local function makecontent(start,parent,id,specification,range)
local tag = parent.tag
local kids = parent.kids
local last = index
index = index + 1
if id == "image" then
-- when we are in \startTEXpage .... \stopTEXpage we have not really
-- an element in the end of th elist as all is unstructured
local list = specification.taglist
local data = usewithcare.images[list[#list]]
local d = pdfdictionary {
Type = pdf_mcr,
Pg = pageref,
MCID = last,
}
if data then
local alt = data.alternativetext or ""
parent.dict.Alt = pdfunicode(alt ~= "" and alt or "image")
else
-- not really tagged as image
end
kids[#kids+1] = d
elseif pagenum == parent.pnum then
kids[#kids+1] = last
-- if checklinks and tag == "link" then
-- local ra = range[7]
-- if ra then
-- if not references[ra] then
-- parent.refatt = ra
-- kids[#kids+1] = pdfdictionary {
-- Type = pdf_objr,
-- Obj = pdfreference(0),
-- Pg = pageref,
-- }
-- references[ra] = index
-- end
-- end
-- end
else
local d = pdfdictionary {
Type = pdf_mcr,
Pg = pageref,
MCID = last,
}
-- kids[#kids+1] = pdfreference(pdfflushobject(d))
kids[#kids+1] = d
end
--
list[index] = parent.pref -- page related list
--
return f_BDC(tag,last)
end
local function makeignore(specification,range)
-- inspect(nodes.tonode(range[3]))
return "/Artifact BMC"
end
-- no need to adapt head, as we always operate on lists
local EMCliteral = nil
-- local enabled = true
-- updaters.register("tagging.state.enable", function() enabled = true end)
-- updaters.register("tagging.state.disable", function() enabled = false end)
local tag_ignore_level <const> = 1
local tag_document_level <const> = 2
local tag_image_state <const> = -1
local tag_ignore_state <const> = -2
local tag_link_state <const> = -3
local tag_reference_state <const> = -4
local tag_rule_state <const> = -5
function nodeinjections.addtags(head,ispage)
if tex.systemmodes.export then
return head
elseif not tex.conditionals.c_strc_tags_global then
return head
-- return
end
-- if not enabled then
-- return
-- end
local pdflanguage = languages.pdflanguage
local listtoutf = nodes.listtoutf
if not EMCliteral then
EMCliteral = register(setstate("EMC"))
end
local last = nil
local ranges = { }
local range = nil
local nofranges = 0
if not root then
structure_kids = pdfarray()
structure_ref = pdfreserveobject()
parent_ref = pdfreserveobject()
root = { pref = pdfreference(structure_ref), kids = structure_kids }
names = pdfarray()
end
initializepage()
local mblob = false
local ablob = { }
if ispage then
pushtag() -- tag_document_level
local ac = starttag("navigationpage")
stoptag()
nofranges = nofranges + 1
ranges[nofranges] = { ac, "navigationpage" }
poptag()
end
-- local function check(n,id)
-- local at = getattr(n,a_tagged)
-- if at then
-- local s = taglist[at]
-- if s then
-- print(">>>",nodecodes[id],s.tagname)
-- end
-- end
-- end
-- todo: survive across pages
local lastpar = 0
local lastparat = 0
local function collectranges(head,parent)
for n, id, subtype in nextnode, head do
-- check(n,id)
if id == glyph_code then
-- if getchar(n) ~= 0 then
local at, blob, ap = getattrs(n,a_tagged,a_mathblob,a_taggedpar)
if at == 0 then
at = false
elseif at == 1 then
at = false
elseif at then
--
else
at = false
end
-- if detailedmath then
-- -- skip
-- elseif blob then
-- if blob == mblob then
-- ranges[nofranges] = range
-- last = at
-- else
-- mblob = blob
-- end
-- else
-- mblob = false
-- end
-- if last ~= at then
-- range = { at, "glyph", n, n, parent, blob } -- attr id start stop list
-- nofranges = nofranges + 1
-- ranges[nofranges] = range
-- last = at
-- elseif range then
-- range[4] = n -- stop
-- end
if not blob then
mblob = false
if last ~= at then
-- print("A",lastpar,ap,getattr(n,a_taggedpar),last,at,utf.char(nuts.getchar(n)))
range = { at, "glyph", n, n, parent } -- attr id start stop list
nofranges = nofranges + 1
ranges[nofranges] = range
last = at
lastparat = at
lastpar = ap
elseif range then
-- print("B",lastpar,ap,getattr(n,a_taggedpar),at,lastparat,utf.char(nuts.getchar(n)))
if lastpar ~= ap and at and lastparat == at then
-- local specification = taglist[at]
pushtag(at < tag_document_level and tag_document_level or at)
local ac = starttag("break")
stoptag()
poptag()
range = { ac, "break", n, false }
nofranges = nofranges + 1
ranges[nofranges] = range
lastpar = ap
range = { at, "glyph", n, n, parent } -- , false, false, ap } -- attr id start stop list
nofranges = nofranges + 1
ranges[nofranges] = range
last = at
lastparat = at
else
-- print("C",lastpar,ap,at,lastparat,utf.char(nuts.getchar(n)))
range[4] = n -- stop
end
end
-- inspect(ranges)
-- elseif blob == mblob and at == last then
elseif blob == mblob and (last and last > 0) then
if range then
range[4] = n -- stop
end
last = at
else
mblob = blob
-- if last ~= at then
local a = ablob[blob]
if not a then
a = tag_document_level
if at then
local t = taglist[at].taglist
-- for i=#t,1,-1 do
-- local s = specifications[t[i]]
-- if s.tagname == "math" then
-- a = s.attribute
-- end
-- end
-- we could store the index in specifications but we only need it once
for i=1,#t do
local s = specifications[t[i]]
if s.tagname == "math" then
a = s.attribute
break
end
end
end
ablob[blob] = a
end
range = { a, "math", n, n, parent, blob } -- attr id start stop list
nofranges = nofranges + 1
ranges[nofranges] = range
last = at
-- elseif range then
-- range[4] = n -- stop
-- end
end
-- end
elseif id == hlist_code or id == vlist_code then
local at, img = getattrs(n,a_tagged,a_image)
-- img can be more generic: image, mpgraphic
if img then
range = { at or false, "image", n, n, parent } -- attr id start stop list
nofranges = nofranges + 1
ranges[nofranges] = range
last = tag_image_state
mblob = false
--
-- TODO, plugin
--
local specification = taglist[at] -- todo: img attr
if specification and specification.tagname == "mpgraphic" then
local list = getlist(n)
if list then
collectranges(list,n)
end
end
--
elseif at == 0 then
range = { false, "ignore", n, n, parent } -- attr id start stop list
nofranges = nofranges + 1
ranges[nofranges] = range
last = tag_ignore_state
mblob = false
else -- 1 also process
if at then
local r, d = getattrs(n,a_reference,a_destination)
if r and not references[r] then
-- bah
local b = getattr(n,a_mathblob)
if b then
at = ablob[b]
end
--
pushtag(at < tag_document_level and tag_document_level or at)
local ac = starttag("link", { reference = r })
stoptag()
poptag()
range = { ac, "link", n, false, parent } -- attr id start stop list
nofranges = nofranges + 1
ranges[nofranges] = range
last = tag_link_state
references[r] = true -- ac
end
if d and not destinations[d] then
-- bah
local b = getattr(n,a_mathblob)
if b then
at = ablob[b]
end
--
pushtag(at < tag_document_level and tag_document_level or at)
local ac = starttag("reference", { destination = d })
stoptag()
poptag()
range = { ac, "reference", n, false, parent } -- attr id start stop list
nofranges = nofranges + 1
ranges[nofranges] = range
last = tag_reference_state
mblob = false
destinations[d] = true -- ac
end
end
local list = getlist(n)
if list then
collectranges(list,n)
end
end
-- elseif id == disc_code then
-- -- can't happen
elseif id == glue_code then
if subtype >= leaders_code then
local leader = getleader(n)
if leader then
collectranges(leader,n)
end
end
elseif id == rule_code then
if subtype == empty_rule_code then
-- skip
else
local w, h, d = getruledimensions(n)
if (w ~= 0) and (h + d ~= 0) then
local at, blob = getattrs(n,a_tagged,a_mathblob)
if blob then
at = false
end
-- if blob then
-- else
if not at then
at = false
elseif at == 0 then
at = false
elseif at == 1 then
at = false
end
if last ~= tag_rule_state and last ~= at then
range = { false, "rule", n, n, parent, blob } -- attr id start stop list
nofranges = nofranges + 1
ranges[nofranges] = range
last = at
last = tag_rule_state
mblob = false
elseif range then
range[4] = n -- stop
end
-- end
else
-- print("rule needs tagging",w,h,d,nodes.rulecodes[subtype])
end
end
end
end
end
collectranges(head)
if trace_tags then
report_tags("")
report_tags(ispage and "begin page" or "begin object")
report_tags("")
for i=1,nofranges do
local range = ranges[i]
local attr = range[1]
local id = range[2]
local start = range[3]
local stop = range[4]
-- local blob = range[6]
-- local par = range[8]
local pdf = ""
-- if trace_tags == "pdf" then
-- local specification = taglist[attr]
-- if attr then
-- local tagname = specification.tagname
-- local tagnameused = specification.tagname
-- local detail = specification.detail
-- tagname, tagnameused, detail = checkoverload(tagname,tagname,detail,specification)
-- local p = properties[tagnameused]
-- if p then
-- pdf = p.pdf
-- end
-- if pdf then
-- pdf = " (" .. pdf .. ")"
-- else
-- pdf = ""
-- end
-- end
-- end
local tags = taglist[attr]
if tags then -- not ok ... only first lines
local s = concattags(tags)
if id == "reference" then
report_tags("R %5i %s%s",attr,s,pdf)
elseif id == "link" then
report_tags("L %5i %s%s",attr,s,pdf)
elseif id == "break" then
report_tags("B %5i %s%s",attr,s,pdf)
elseif id == "navigationpage" then
report_tags("P %5i %s : %i%s",attr,s,pagenum or 0,pdf)
else
report_tags("T %5i %s : %s%s",attr,s,listtoutf(start,false,true,stop),pdf)
end
else
report_tags("-------")
end
end
report_tags("")
report_tags(ispage and "end page" or "end object")
report_tags("")
end
local top = nil
local noftop = 0
local blobdone = { }
local function inject(start,stop,list,literal,left,right)
local prev = getprev(start)
if prev then
setlink(prev,literal)
end
if left then
setlink(literal,left,start)
else
setlink(literal,start)
end
if list and not prev then
setlist(list,literal)
end
local finish = copy_node(EMCliteral)
if stop then
-- use insert instead:
local next = getnext(stop)
if next then
setlink(finish,next)
end
if right then
setlink(stop,right,finish)
else
setlink(stop,finish)
end
else
local next = getnext(literal)
if next then
setlink(finish,next)
end
if right then
setlink(literal,right,finish)
else
setlink(literal,finish)
end
end
end
-- not ok (probably when at the start)
-- local function inject(start,stop,list,literal,left,right)
-- setlink(getprev(start) or list or true,literal,left or true,start)
-- setlink(stop,right or true,copy_node(EMCliteral),getnext(stop))
-- end
-- inspect(ranges)
local language = texgetcount("mainlanguagenumber")
for i=1,nofranges do
local range = ranges[i]
local mblob = false
local attr = range[1]
local id = range[2]
local start = range[3]
local stop = range[4]
local list = range[5]
-- local blob = range[6]
-- local par = range[8]
if attr == 0 then -- should be false then
local literal = setstate(makeignore(false,range))
inject(start,stop,list,literal)
elseif attr then
local blob = range[6]
local specification = taglist[attr]
local currentlist = specification.taglist
local noftags = #currentlist
local common = 0
local literal = nil
local ignore = false
if top then
for i=1,noftags >= noftop and noftop or noftags do
if top[i] == currentlist[i] then
common = i
else
break
end
end
end
local prev = common > 0 and elements[currentlist[common]] or root
if blob and not detailedmath then
for j=common+1,noftags do
local tag = currentlist[j]
local prv = elements[tag] or makeelement(tag,prev)
if prv == false then
-- ignore this one
prev = false
ignore = true
break
elseif prv == true then
-- skip this one
else
prev = prv
end
-- this is an ugly hack but ok for now -- elements property
if find(tag,"^math>") then
break
end
end
else
for j=common+1,noftags do
local tag = currentlist[j]
local etg = elements[tag]
local prv = etg or makeelement(tag,prev)
if prv == false then
-- ignore this one
prev = false
ignore = true
break
elseif prv == true then
-- skip this one
else
prev = prv
end
end
end
if prev then
if id == "glyph" then
-- It makes little sense to check if we change a language as that can happen
-- mixed in a running text as well as in nested elements. One should just use
-- proper environments. So, we don't even bother to warn here for the few
-- cases where we can detect it. We delegate that to the \TEX\ end.
local lan = getlanguage(start)
if lan and lan > 0 and lan ~= language then
-- if lan == getlanguage(stop) then
local dict = prev.dict
if dict then
dict.Lang = pdflanguage(lan)
end
-- else -- unreliable
-- report_tags("confusing language %a range: %s",languagenumbers[lan],listtoutf(start,false,true,stop))
-- end
end
end
literal = setstate(makecontent(start,prev,id,specification,range))
elseif ignore then
literal = setstate(makeignore(specification,range))
else
-- maybe ignore too
end
if literal then
local left, right
if trace_info or trace_math then
local name = specification.tagname
if name then
left, right = tagtracers[name](name,specification,trace_math and blob or nil)
end
end
if not left and trace_suspects then
local name = specification.tagname
if name == "mrow" then -- todo
collectedsuspects[#collectedsuspects+1] = formatters["%i:%s"](pagenum,name)
left, right = tagtracers.suspect(name)
end
end
if not left and trace_internals then
if id == "link" then
left, right = tagtracers.internallink(attr or 0)
elseif id == "reference" then
left, right = tagtracers.internalreference(attr or 0)
end
end
inject(start,stop,list,literal,left,right)
end
top = currentlist
noftop = noftags
else
local literal = setstate(makeignore(specification,range))
inject(start,stop,list,literal)
end
end
finishpage()
return head
end
-- variant: more structure but funny collapsing in viewer, only in lua
-- file as commented anyway
-- this belongs elsewhere (export is not pdf related)
local permitted = true
local enabled = false
local shipout = true
directives.register("structures.tags.shipout",function(v) shipout = v end)
function codeinjections.settaggingsupport(option)
if option == false then
if enabled then
disableaction("shipouts","structures.tags.handler")
disableaction("math","noads.handlers.tags")
enabled = false
end
if permitted then
if trace_tags then
report_tags("blocking structure tags")
end
permitted = false
end
end
end
function codeinjections.enabletags()
if permitted and not enabled then
structures.tags.handler = nodeinjections.addtags
if shipout then
enableaction("shipouts","structures.tags.handler")
end
--
-- already done when export is set:
--
enableaction("math","noads.handlers.tags")
--
if not embeddedtags then
embedsupportedtags()
end
--
if trace_tags then
report_tags("enabling structure tags")
end
--
enabled = true
version = lpdf.majorversion()
-- detailedmath = detailedmath or (version < 2) -- always go for it
-- bah
updaters.apply("structures.tagging",version)
structures.references.forceinnermode()
end
end
function codeinjections.discardpages(state)
if state == true then
state = false
else
state = true
end
lpdf.setpagestate(state)
end
-- We now delegate as w need to push and pop anyway:
do
local stack = { }
local level = 0
local count = 0
-- local a_taggedpar <const> = attributes.private("taggedpar")
local unsetvalue <const> = attributes.unsetvalue
local report = logs.reporter("structure","paragraphs")
local trace = false
trackers.register("structure.paragraphs", function(v) trace = v end)
local texsetattribute = tex.setattribute
local texgetattribute = tex.getattribute
-- interfaces.implement {
-- name = "strc_tags_parcounter_check",
-- -- public = true,
-- protected = true,
-- noaligned = true,
-- actions = function()
-- if level == 0 then
-- count = 1
-- level = 1
-- stack[level] = count
-- texsetattribute(a_taggedpar,count,"global")
-- if trace then
-- report("%wforcing, new %s",level,count)
-- end
-- end
-- end
-- }
interfaces.implement {
name = "strc_tags_parcounter_push",
-- public = true,
protected = true,
-- noaligned = true,
actions = function()
local old = texgetattribute(a_taggedpar)
local new = count + 1
level = level + 1
stack[level] = old
texsetattribute(a_taggedpar,new,"global")
if trace then
report("%wpushing, old %s, new %s",level,
old == unsetvalue and "*" or old,
new == unsetvalue and "*" or new
)
end
count = new
end
}
interfaces.implement {
name = "strc_tags_parcounter_pop",
-- public = true,
protected = true,
-- noaligned = true,
actions = function()
local old = texgetattribute(a_taggedpar)
local new = stack[level] or 0
if trace then
report("%wpopping, old %s, new %s",level,
old == unsetvalue and "*" or old,
new == unsetvalue and "*" or new
)
end
if level > 0 then
level = level - 1
else
-- error
end
texsetattribute(a_taggedpar,new,"global")
end
}
end