diff options
Diffstat (limited to 'tex/context/base/strc-tag.lua')
| -rw-r--r-- | tex/context/base/strc-tag.lua | 633 |
1 files changed, 391 insertions, 242 deletions
diff --git a/tex/context/base/strc-tag.lua b/tex/context/base/strc-tag.lua index 7e5c6f993..637d74e8c 100644 --- a/tex/context/base/strc-tag.lua +++ b/tex/context/base/strc-tag.lua @@ -6,189 +6,219 @@ if not modules then modules = { } end modules ['strc-tag'] = { license = "see context related readme files" } --- This is rather experimental code. +-- This is rather experimental code. Tagging happens on the fly and there are two analysers +-- involved: the pdf backend tagger and the exporter. They share data but there are subtle +-- differences. Each tag carries a specification and these can be accessed by attribute (the +-- end of the chain tag) or by so called fullname which is a tagname combined with a number. +local type, next = type, next local insert, remove, unpack, concat = table.insert, table.remove, table.unpack, table.concat -local gsub, find, topattern, format = string.gsub, string.find, string.topattern, string.format -local lpegmatch = lpeg.match +local find, topattern, format = string.find, string.topattern, string.format +local lpegmatch, P, S, C, Cc = lpeg.match, lpeg.P, lpeg.S, lpeg.C, lpeg.Cc local texattribute = tex.attribute local allocate = utilities.storage.allocate local settings_to_hash = utilities.parsers.settings_to_hash +local setmetatableindex = table.setmetatableindex local trace_tags = false trackers.register("structures.tags", function(v) trace_tags = v end) local report_tags = logs.reporter("structure","tags") -local attributes, structures = attributes, structures +local attributes = attributes +local structures = structures +local implement = interfaces.implement local a_tagged = attributes.private('tagged') local unsetvalue = attributes.unsetvalue local codeinjections = backends.codeinjections -local taglist = allocate() -local properties = allocate() -local labels = allocate() -local stack = { } -local chain = { } -local ids = { } -local enabled = false -local tagdata = { } -- used in export -local tagmetadata = { } -- used in export - -local tags = structures.tags -tags.taglist = taglist -- can best be hidden -tags.labels = labels -tags.data = tagdata -tags.metadata = tagmetadata - -local properties = allocate { - - document = { pdf = "Div", nature = "display" }, - - division = { pdf = "Div", nature = "display" }, - paragraph = { pdf = "P", nature = "mixed" }, - p = { pdf = "P", nature = "mixed" }, - construct = { pdf = "Span", nature = "inline" }, - highlight = { pdf = "Span", nature = "inline" }, - - section = { pdf = "Sect", nature = "display" }, - sectiontitle = { pdf = "H", nature = "mixed" }, - sectionnumber = { pdf = "H", nature = "mixed" }, - sectioncontent = { pdf = "Div", nature = "display" }, - - itemgroup = { pdf = "L", nature = "display" }, - item = { pdf = "Li", nature = "display" }, - itemtag = { pdf = "Lbl", nature = "mixed" }, - itemcontent = { pdf = "LBody", nature = "mixed" }, - - description = { pdf = "Div", nature = "display" }, - descriptiontag = { pdf = "Div", nature = "mixed" }, - descriptioncontent = { pdf = "Div", nature = "mixed" }, - descriptionsymbol = { pdf = "Span", nature = "inline" }, -- note reference - - verbatimblock = { pdf = "Code", nature = "display" }, - verbatimlines = { pdf = "Code", nature = "display" }, - verbatimline = { pdf = "Code", nature = "mixed" }, - verbatim = { pdf = "Code", nature = "inline" }, - - lines = { pdf = "Code", nature = "display" }, - line = { pdf = "Code", nature = "mixed" }, - - synonym = { pdf = "Span", nature = "inline" }, - sorting = { pdf = "Span", nature = "inline" }, - - register = { pdf = "Div", nature = "display" }, - registersection = { pdf = "Div", nature = "display" }, - registertag = { pdf = "Span", nature = "mixed" }, - registerentries = { pdf = "Div", nature = "display" }, - registerentry = { pdf = "Span", nature = "mixed" }, - registersee = { pdf = "Span", nature = "mixed" }, - registerpages = { pdf = "Span", nature = "mixed" }, - registerpage = { pdf = "Span", nature = "inline" }, - registerpagerange = { pdf = "Span", nature = "mixed" }, - - table = { pdf = "Table", nature = "display" }, - tablerow = { pdf = "TR", nature = "display" }, - tablecell = { pdf = "TD", nature = "mixed" }, - - tabulate = { pdf = "Table", nature = "display" }, - tabulaterow = { pdf = "TR", nature = "display" }, - tabulatecell = { pdf = "TD", nature = "mixed" }, - - list = { pdf = "TOC", nature = "display" }, - listitem = { pdf = "TOCI", nature = "display" }, - listtag = { pdf = "Lbl", nature = "mixed" }, - listcontent = { pdf = "P", nature = "mixed" }, - listdata = { pdf = "P", nature = "mixed" }, - listpage = { pdf = "Reference", nature = "mixed" }, - - delimitedblock = { pdf = "BlockQuote", nature = "display" }, - delimited = { pdf = "Quote", nature = "inline" }, - subsentence = { pdf = "Span", nature = "inline" }, - - label = { pdf = "Span", nature = "mixed" }, - number = { pdf = "Span", nature = "mixed" }, - - float = { pdf = "Div", nature = "display" }, -- Figure - floatcaption = { pdf = "Caption", nature = "mixed" }, - floatlabel = { pdf = "Span", nature = "inline" }, - floatnumber = { pdf = "Span", nature = "inline" }, - floattext = { pdf = "Span", nature = "mixed" }, - floatcontent = { pdf = "P", nature = "mixed" }, - - image = { pdf = "P", nature = "mixed" }, - mpgraphic = { pdf = "P", nature = "mixed" }, - - formulaset = { pdf = "Div", nature = "display" }, - formula = { pdf = "Div", nature = "display" }, -- Formula - formulacaption = { pdf = "Span", nature = "mixed" }, - formulalabel = { pdf = "Span", nature = "mixed" }, - formulanumber = { pdf = "Span", nature = "mixed" }, - formulacontent = { pdf = "P", nature = "display" }, - subformula = { pdf = "Div", nature = "display" }, - - link = { pdf = "Link", nature = "inline" }, - - margintextblock = { pdf = "Span", nature = "inline" }, - margintext = { pdf = "Span", nature = "inline" }, - - math = { pdf = "Div", nature = "inline" }, -- no display - mn = { pdf = "Span", nature = "mixed" }, - mi = { pdf = "Span", nature = "mixed" }, - mo = { pdf = "Span", nature = "mixed" }, - ms = { pdf = "Span", nature = "mixed" }, - mrow = { pdf = "Span", nature = "display" }, - msubsup = { pdf = "Span", nature = "display" }, - msub = { pdf = "Span", nature = "display" }, - msup = { pdf = "Span", nature = "display" }, - merror = { pdf = "Span", nature = "mixed" }, - munderover = { pdf = "Span", nature = "display" }, - munder = { pdf = "Span", nature = "display" }, - mover = { pdf = "Span", nature = "display" }, - mtext = { pdf = "Span", nature = "mixed" }, - mfrac = { pdf = "Span", nature = "display" }, - mroot = { pdf = "Span", nature = "display" }, - msqrt = { pdf = "Span", nature = "display" }, - mfenced = { pdf = "Span", nature = "display" }, - maction = { pdf = "Span", nature = "display" }, - - mtable = { pdf = "Table", nature = "display" }, -- might change - mtr = { pdf = "TR", nature = "display" }, -- might change - mtd = { pdf = "TD", nature = "display" }, -- might change - - ignore = { pdf = "Span", nature = "mixed" }, - metadata = { pdf = "Div", nature = "display" }, - metavariable = { pdf = "Span", nature = "mixed" }, - - mid = { pdf = "Span", nature = "inline" }, - sub = { pdf = "Span", nature = "inline" }, - sup = { pdf = "Span", nature = "inline" }, - subsup = { pdf = "Span", nature = "inline" }, - - combination = { pdf = "Span", nature = "display" }, - combinationpair = { pdf = "Span", nature = "display" }, - combinationcontent = { pdf = "Span", nature = "mixed" }, - combinationcaption = { pdf = "Span", nature = "mixed" }, +local taglist = allocate() -- access by attribute +local specifications = allocate() -- access by fulltag +local labels = allocate() +local stack = { } +local chain = { } +local ids = { } +local enabled = false +local tagcontext = { } +local tagpatterns = { } +local lasttags = { } +local stacksize = 0 +local metadata = nil -- applied to the next element + +local tags = structures.tags +tags.taglist = taglist -- can best be hidden +tags.labels = labels +tags.patterns = tagpatterns +tags.specifications = specifications + +-- Tags are internally stored as: +-- +-- tag>number tag>number tag>number + +local p_splitter = C((1-S(">"))^1) * P(">") * C(P(1)^1) +tagpatterns.splitter = p_splitter + +local properties = allocate { + + document = { pdf = "Div", nature = "display" }, + + division = { pdf = "Div", nature = "display" }, + paragraph = { pdf = "P", nature = "mixed" }, + p = { pdf = "P", nature = "mixed" }, + construct = { pdf = "Span", nature = "inline" }, + highlight = { pdf = "Span", nature = "inline" }, + + section = { pdf = "Sect", nature = "display" }, + sectiontitle = { pdf = "H", nature = "mixed" }, + sectionnumber = { pdf = "H", nature = "mixed" }, + sectioncontent = { pdf = "Div", nature = "display" }, + + itemgroup = { pdf = "L", nature = "display" }, + item = { pdf = "LI", nature = "display" }, + itemtag = { pdf = "Lbl", nature = "mixed" }, + itemcontent = { pdf = "LBody", nature = "mixed" }, + itemhead = { pdf = "Div", nature = "display" }, + itembody = { pdf = "Div", nature = "display" }, + + description = { pdf = "Div", nature = "display" }, + descriptiontag = { pdf = "Div", nature = "mixed" }, + descriptioncontent = { pdf = "Div", nature = "mixed" }, + descriptionsymbol = { pdf = "Span", nature = "inline" }, -- note reference + + verbatimblock = { pdf = "Code", nature = "display" }, + verbatimlines = { pdf = "Code", nature = "display" }, + verbatimline = { pdf = "Code", nature = "mixed" }, + verbatim = { pdf = "Code", nature = "inline" }, + + lines = { pdf = "Code", nature = "display" }, + line = { pdf = "Code", nature = "mixed" }, + + synonym = { pdf = "Span", nature = "inline" }, + sorting = { pdf = "Span", nature = "inline" }, + + register = { pdf = "Div", nature = "display" }, + registerlocation = { pdf = "Span", nature = "inline" }, + registersection = { pdf = "Div", nature = "display" }, + registertag = { pdf = "Span", nature = "mixed" }, + registerentries = { pdf = "Div", nature = "display" }, + registerentry = { pdf = "Div", nature = "display" }, + registercontent = { pdf = "Span", nature = "mixed" }, + registersee = { pdf = "Span", nature = "mixed" }, + registerpages = { pdf = "Span", nature = "mixed" }, + registerpage = { pdf = "Span", nature = "mixed" }, + registerseparator = { pdf = "Span", nature = "inline" }, + registerpagerange = { pdf = "Span", nature = "mixed" }, + + table = { pdf = "Table", nature = "display" }, + tablerow = { pdf = "TR", nature = "display" }, + tablecell = { pdf = "TD", nature = "mixed" }, + + tabulate = { pdf = "Table", nature = "display" }, + tabulaterow = { pdf = "TR", nature = "display" }, + tabulatecell = { pdf = "TD", nature = "mixed" }, + + list = { pdf = "TOC", nature = "display" }, + listitem = { pdf = "TOCI", nature = "display" }, + listtag = { pdf = "Lbl", nature = "mixed" }, + listcontent = { pdf = "P", nature = "mixed" }, + listdata = { pdf = "P", nature = "mixed" }, + listpage = { pdf = "Reference", nature = "mixed" }, + + delimitedblock = { pdf = "BlockQuote", nature = "display" }, + delimited = { pdf = "Quote", nature = "inline" }, + subsentence = { pdf = "Span", nature = "inline" }, + + label = { pdf = "Span", nature = "mixed" }, + number = { pdf = "Span", nature = "mixed" }, + + float = { pdf = "Div", nature = "display" }, -- Figure + floatcaption = { pdf = "Caption", nature = "mixed" }, + floatlabel = { pdf = "Span", nature = "inline" }, + floatnumber = { pdf = "Span", nature = "inline" }, + floattext = { pdf = "Span", nature = "mixed" }, + floatcontent = { pdf = "P", nature = "mixed" }, + + image = { pdf = "P", nature = "mixed" }, + mpgraphic = { pdf = "P", nature = "mixed" }, + + formulaset = { pdf = "Div", nature = "display" }, + formula = { pdf = "Div", nature = "display" }, -- Formula + formulacaption = { pdf = "Span", nature = "mixed" }, + formulalabel = { pdf = "Span", nature = "mixed" }, + formulanumber = { pdf = "Span", nature = "mixed" }, + formulacontent = { pdf = "P", nature = "display" }, + subformula = { pdf = "Div", nature = "display" }, + + link = { pdf = "Link", nature = "inline" }, + + margintextblock = { pdf = "Span", nature = "inline" }, + margintext = { pdf = "Span", nature = "inline" }, + + math = { pdf = "Div", nature = "inline" }, -- no display + mn = { pdf = "Span", nature = "mixed" }, + mi = { pdf = "Span", nature = "mixed" }, + mo = { pdf = "Span", nature = "mixed" }, + ms = { pdf = "Span", nature = "mixed" }, + mrow = { pdf = "Span", nature = "display" }, + msubsup = { pdf = "Span", nature = "display" }, + msub = { pdf = "Span", nature = "display" }, + msup = { pdf = "Span", nature = "display" }, + merror = { pdf = "Span", nature = "mixed" }, + munderover = { pdf = "Span", nature = "display" }, + munder = { pdf = "Span", nature = "display" }, + mover = { pdf = "Span", nature = "display" }, + mtext = { pdf = "Span", nature = "mixed" }, + mfrac = { pdf = "Span", nature = "display" }, + mroot = { pdf = "Span", nature = "display" }, + msqrt = { pdf = "Span", nature = "display" }, + mfenced = { pdf = "Span", nature = "display" }, + maction = { pdf = "Span", nature = "display" }, + + mstacker = { pdf = "Span", nature = "display" }, -- these are only internally used + mstackertop = { pdf = "Span", nature = "display" }, -- these are only internally used + mstackerbot = { pdf = "Span", nature = "display" }, -- these are only internally used + mstackermid = { pdf = "Span", nature = "display" }, -- these are only internally used + + mtable = { pdf = "Table", nature = "display" }, -- might change + mtr = { pdf = "TR", nature = "display" }, -- might change + mtd = { pdf = "TD", nature = "display" }, -- might change + + ignore = { pdf = "Span", nature = "mixed" }, -- used internally + private = { pdf = "Span", nature = "mixed" }, -- for users (like LS) when they need it + metadata = { pdf = "Div", nature = "display" }, + metavariable = { pdf = "Span", nature = "mixed" }, + + mid = { pdf = "Span", nature = "inline" }, + sub = { pdf = "Span", nature = "inline" }, + sup = { pdf = "Span", nature = "inline" }, + subsup = { pdf = "Span", nature = "inline" }, + + combination = { pdf = "Span", nature = "display" }, + combinationpair = { pdf = "Span", nature = "display" }, + combinationcontent = { pdf = "Span", nature = "mixed" }, + combinationcaption = { pdf = "Span", nature = "mixed" }, } -function tags.detailedtag(tag,detail,attribute) - if not attribute then - attribute = texattribute[a_tagged] - end +tags.properties = properties + +local patterns = setmetatableindex(function(t,tag) + local v = topattern("^" .. tag .. ">") + t[tag] = v + return v +end) + +function tags.locatedtag(tag) + local attribute = texattribute[a_tagged] if attribute >= 0 then - local tl = taglist[attribute] - if tl then - local pattern - if detail and detail ~= "" then - pattern = "^" .. tag .. ":".. detail .. "%-" - else - pattern = "^" .. tag .. "%-" - end - for i=#tl,1,-1 do - local tli = tl[i] - if find(tli,pattern) then - return tli + local specification = taglist[attribute] + if specification then + local taglist = specification.taglist + local pattern = patterns[tag] + for i=#taglist,1,-1 do + local t = taglist[i] + if find(t,pattern) then + return t end end end @@ -198,12 +228,20 @@ function tags.detailedtag(tag,detail,attribute) return false -- handy as bogus index end -tags.properties = properties - -local lasttags = { } -local userdata = { } - -tags.userdata = userdata +function structures.atlocation(str) + local specification = taglist[texattribute[a_tagged]] + if specification then + if list then + local taglist = specification.taglist + local pattern = patterns[str] + for i=#list,1,-1 do + if find(list[i],pattern) then + return true + end + end + end + end +end function tags.setproperty(tag,key,value) local p = properties[tag] @@ -214,15 +252,18 @@ function tags.setproperty(tag,key,value) end end -function tags.registerdata(data) - local fulltag = chain[nstack] - if fulltag then - tagdata[fulltag] = data +function tags.setaspect(key,value) + local tag = chain[stacksize] + if tag then + local p = properties[tag] + if p then + p[key] = value + else + properties[tag] = { [key] = value } + end end end -local metadata - function tags.registermetadata(data) local d = settings_to_hash(data) if metadata then @@ -232,75 +273,92 @@ function tags.registermetadata(data) end end -local nstack = 0 - function tags.start(tag,specification) - local label, detail, user - if specification then - label, detail, user = specification.label, specification.detail, specification.userdata - end if not enabled then codeinjections.enabletags() enabled = true end -- ---~ labels[tag] = label ~= "" and label or tag ---~ local fulltag ---~ if detail and detail ~= "" then ---~ fulltag = tag .. ":" .. detail ---~ else ---~ fulltag = tag ---~ end + labels[tag] = tag -- can go away -- - local fulltag = label ~= "" and label or tag - labels[tag] = fulltag - if detail and detail ~= "" then - fulltag = fulltag .. ":" .. detail - end + local attribute = #taglist + 1 + local tagindex = (ids[tag] or 0) + 1 -- - local t = #taglist + 1 - local n = (ids[fulltag] or 0) + 1 - ids[fulltag] = n - lasttags[tag] = n - local completetag = fulltag .. "-" .. n - nstack = nstack + 1 - chain[nstack] = completetag - stack[nstack] = t - -- a copy as we can add key values for alt and actualtext if needed: - taglist[t] = { unpack(chain,1,nstack) } + local completetag = tag .. ">" .. tagindex -- - if user and user ~= "" then - -- maybe we should merge this into taglist or whatever ... anyway there is room to optimize - -- taglist.userdata = settings_to_hash(user) - userdata[completetag] = settings_to_hash(user) - end - if metadata then - tagmetadata[completetag] = metadata + ids[tag] = tagindex + lasttags[tag] = tagindex + stacksize = stacksize + 1 + -- + chain[stacksize] = completetag + stack[stacksize] = attribute + tagcontext[tag] = completetag + -- + local tagnesting = { unpack(chain,1,stacksize) } -- a copy so we can add actualtext + -- + if specification then + specification.attribute = attribute + specification.tagindex = tagindex + specification.taglist = tagnesting + specification.tagname = tag + if metadata then + specification.metadata = metadata + metadata = nil + end + local userdata = specification.userdata + if user ~= "" and type(userdata) == "string" then + specification.userdata = settings_to_hash(userdata) + end + local detail = specification.detail + if detail == "" then + specification.detail = nil + end + local parents = specification.parents + if parents == "" then + specification.parents = nil + end + else + specification = { + attribute = attribute, + tagindex = tagindex, + taglist = tagnesting, + tagname = tag, + metadata = metadata, + } metadata = nil end - texattribute[a_tagged] = t - return t + -- + taglist[attribute] = specification + specifications[completetag] = specification + -- + texattribute[a_tagged] = attribute + return attribute end -function tags.restart(completetag) - local t = #taglist + 1 - nstack = nstack + 1 - chain[nstack] = completetag - stack[nstack] = t - taglist[t] = { unpack(chain,1,nstack) } - texattribute[a_tagged] = t - return t +function tags.restart(attribute) + stacksize = stacksize + 1 + if type(attribute) == "number" then + local taglist = taglist[attribute].taglist + chain[stacksize] = taglist[#taglist] + else + chain[stacksize] = attribute -- a string + attribute = #taglist + 1 + taglist[attribute] = { taglist = { unpack(chain,1,stacksize) } } + end + stack[stacksize] = attribute + texattribute[a_tagged] = attribute + return attribute end function tags.stop() - if nstack > 0 then - nstack = nstack -1 + if stacksize > 0 then + stacksize = stacksize - 1 end - local t = stack[nstack] + local t = stack[stacksize] if not t then - if trace_tags then - report_tags("ignoring end tag, previous chain: %s",nstack > 0 and concat(chain[nstack],"",1,nstack) or "none") - end + -- if trace_tags then + report_tags("ignoring end tag, previous chain: %s",stacksize > 0 and concat(chain," ",1,stacksize) or "none") + -- end t = unsetvalue end texattribute[a_tagged] = t @@ -308,24 +366,56 @@ function tags.stop() end function tags.getid(tag,detail) - if detail and detail ~= "" then - return ids[tag .. ":" .. detail] or "?" - else - return ids[tag] or "?" - end + return ids[tag] or "?" end function tags.last(tag) return lasttags[tag] -- or false end -function tags.lastinchain() - return chain[nstack] +function tags.lastinchain(tag) + if tag and tag ~= "" then + return tagcontext[tag] + else + return chain[stacksize] + end end -function structures.atlocation(str) - local location = gsub(concat(taglist[texattribute[a_tagged]],"-"),"%-%d+","") - return find(location,topattern(str)) ~= nil +local strip = C((1-S(">"))^1) + +function tags.elementtag() + local fulltag = chain[stacksize] + if fulltag then + return lpegmatch(strip,fulltag) + end +end + +function tags.strip(fulltag) + return lpegmatch(strip,fulltag) +end + +function tags.setuserproperties(tag,list) + if not list or list == "" then + tag, list = chain[stacksize], tag + else + tag = tagcontext[tag] + end + if tag then -- an attribute now + local l = settings_to_hash(list) + local s = specifications[tag] + if s then + local u = s.userdata + if u then + for k, v in next, l do + u[k] = v + end + else + s.userdata = l + end + else + -- error + end + end end function tags.handler(head) -- we need a dummy @@ -334,8 +424,8 @@ end statistics.register("structure elements", function() if enabled then - if nstack > 0 then - return format("%s element chains identified, open chain: %s ",#taglist,concat(chain," => ",1,nstack)) + if stacksize > 0 then + return format("%s element chains identified, open chain: %s ",#taglist,concat(chain," => ",1,stacksize)) else return format("%s element chains identified",#taglist) end @@ -349,6 +439,65 @@ directives.register("backend.addtags", function(v) end end) -commands.starttag = tags.start -commands.stoptag = tags.stop -commands.settagproperty = tags.setproperty +-- interface + +local starttag = tags.start + +implement { + name = "starttag", + actions = starttag, + arguments = { "string" } +} + +implement { + name = "stoptag", + actions = tags.stop, +} + +implement { + name = "starttag_u", + scope = "private", + actions = function(tag,userdata) starttag(tag,{ userdata = userdata }) end, + arguments = { "string", "string" } +} + +implement { + name = "starttag_d", + scope = "private", + actions = function(tag,detail) starttag(tag,{ detail = detail }) end, + arguments = { "string", "string" } +} + +implement { + name = "starttag_c", + scope = "private", + actions = function(tag,detail,parents) starttag(tag,{ detail = detail, parents = parents }) end, + arguments = { "string", "string", "string" } +} + +implement { name = "settagaspect", actions = tags.setaspect, arguments = { "string", "string" } } + +implement { name = "settagproperty", actions = tags.setproperty, arguments = { "string", "string", "string" } } +implement { name = "settagproperty_b", actions = tags.setproperty, arguments = { "string", "'backend'", "string" }, scope = "private" } +implement { name = "settagproperty_n", actions = tags.setproperty, arguments = { "string", "'nature'", "string" }, scope = "private" } + +implement { name = "getelementtag", actions = { tags.elementtag, context } } + +implement { + name = "setelementuserproperties", + scope = "private", + actions = tags.setuserproperties, + arguments = { "string", "string" } +} + +implement { + name = "doifelseinelement", + actions = { structures.atlocation, commands.testcase }, + arguments = "string", +} + +implement { + name = "settaggedmetadata", + actions = structures.tags.registermetadata, + arguments = "string" +} |