Editing Module:Clade/transclude

require('strict')

local DEBUG=false
--DEBUG=true -- comment out or not runtime or debug

local p ={}
local pargs ={}

p.main = function(frame) -- called from template
	pargs = frame:getParent().args
	local output 
	local selectedTree -- subtree extracted from page content
	local modifiedTree -- subtree after pruning and grafting
	
	-- (1) get page
	local page = pargs['page'] or frame.args['page']
	if not page then 
		return p.errorMsg("Target page not provided") 
    end 
	
	-- (2) get content of page (move from _section(), _label, etc)
	local content
	local title = mw.title.new( mw.text.trim(page)) -- , ns)  -- creates object if page doesn't exist (and valid page name)
	                                            --TODO: could use mw.title.makeTitle(), but that needs ns
	if title then  
		if  title.exists then
	  		content = title:getContent()
			if not content then return p.errorMsg("Content of " .. page .. " not loaded.") end
		else
			return p.errorMsg('Page with title "' .. page .. '" not found.') 
    	end
	end
	-- (3) select from content
	
	local section =  pargs['section'] or pargs['section1'] or pargs[1] 
	if section then
		selectedTree = p._section(frame, content, section)
	end
	
	local label =  pargs['label'] or pargs['label1'] or pargs[1] 
	if label then
		selectedTree = p._label(frame, content, label)
	end   
	--TODO does this need to be separate from label?
	local subtree =  pargs['subtree'] or pargs['subtree1'] or pargs[1] 
	if subtree then
		selectedTree = p._label(frame, content, subtree)
	end  
	
    if not selectedTree then -- if none of options retrieve anything
    	p.errorMsg("Nothing retrieved for selection option " .. (label or subtree or section or "none"))
    end

if DEBUG then return selectedTree end      --- returns the code captured without processing
    
	--(4) modify content (excise and replace; prune and graft)
	local exclude = pargs['exclude'] or pargs['exclude1']
	if exclude then
	    if pargs['exclude'] then pargs['exclude1'] = pargs['exclude'] end
		if pargs['replace'] then pargs['replace1'] = pargs['replace'] end
		
		modifiedTree = selectedTree
		
	    local i = 1
	    while pargs['exclude'..i] do
	       local exclude = pargs['exclude'..i]
		   local replace = pargs['replace'..i] or "&nbsp;"  -- must be something
		   modifiedTree = p._xlabel(frame, modifiedTree, exclude, replace)
		   i=i+1
		end
	else
		modifiedTree = selectedTree
	end
	--(5) other options
	----- suppress hidden elements
	if pargs['nohidden'] then
		modifiedTree = modifiedTree:gsub("lade hidden", "lade")
	end
	----- suppress authorities (or anything in small tags)
	if pargs['noauthority'] then
		modifiedTree = modifiedTree:gsub("<small>.-</small>", "")
	end	
	----- suppress images
	if pargs['noimages'] then
		modifiedTree = modifiedTree:gsub("%[%[File:.-%]%]", "")
	end
	----- wrap in outer clade 
	local wrap = pargs['wrap'] 
	if wrap and (label or subtree) then
		local label1 = label or string.lower(subtree)
		local styleString = "" 
		if  pargs['style'] then  styleString = '|style=' .. pargs['style']  end
		
		if wrap ~= "" then label1 = wrap end
		output = "{{clade " .. styleString .. " |label1=" .. p.firstToUpper(label1) .. "|1=" .. modifiedTree .. " }}" -- last space before double brace important
    else
    	output	= modifiedTree
    end
	
	--(6) return final tree
	if output then
		if pargs['raw'] then
			return output
		else
			return frame:preprocess(output)
		end
	end
	
    return p.errorMsg("No valid option for transclusion")
end


--=============================== extract LABELS or SUBTREES=======================================
p.label = function (frame, page, ...)
	local page = frame.args[1] --"User:Jts1882/sandbox/test/Passeriformes"
	local label = frame.args[1] or frame.args['label'] or frame.args['label1']
	local wrap = frame.args['wrap'] 
	
	local output = p._label (frame, page, frame.args[2], frame.args[3], frame.args[4], frame.args[5] )
	if wrap then
		local label1 = string.lower(frame.args[2])
		if wrap ~= "" then label1 = wrap end
		output = "{{clade |label1=" .. p.firstToUpper(label1) .. "|1=" .. output .. "}}"
    end
	return frame:preprocess(output)
end

p._label = function (frame, content, ... )
--	local page = "User:Jts1882/sandbox/test/Passeriformes"
--	local label = frame.args[1] or frame.args['label']
	local args = { ... }
	local output = ""
	
	if not args[1] then return p.errorMsg ("Label name not provided") end
	
    local mode = "label"
	local targetType = "label(%d)"                   -- standard label of form |labelN= (captures N)
	local cladePrefix = "(%d)"                       -- standard node of form |N= (captures N)
	for k,v in pairs(args) do
		local section = mw.text.trim(v)
		if string.upper( section) == section then
			 mode        = "subtree"
			 targetType  = "target(%u)"               -- targets of form targetX (X=uppercase letter)
			 cladePrefix = "subclade(%u)"             -- subclades of form subcladeX (captures X)
	    end
        
        --[=[ the pattern to capture is one of two forms: labelN=Name |N={...} 
                                                         targetX=NAME |subcladeX={...} 
        		          labelN      =  [[        name        ]]        |N           =    {...}
        		   or    targetX      =  [[        name        ]]        |subcladeX   =    {...}
        ]=]
        local pattern = targetType.."=[%s%p]*"..section .. "[%s%p]+.-"..cladePrefix.."=.-(%b{})"
                                                                               -- this .- skips section tags before {{clade ...}}
                                                           -- this .- skips |sublabel and styling following the label (but can return wrong clade when a subtree)

        local index1, index2, selectedTree = string.match( content , pattern )
        -- note index1 and index2 should match (X=X or N=N)

        if selectedTree then 
            --[[ the tree can contain markers for subtrees like {FABIDS} 
                    when the form is |N={FABIDS} we want to substitute the subtree
                    but not when the form is |targetX={FABIDS}
            ]]
            
            local pattern2 = "({%u-})"   -- this captures both |N={FABIDS} and |targetX={FABIDS}
                                         -- we only want to substitute a subtree in the first kind 
                                         -- will exclude second with pattern3 test below
            
            if string.find(selectedTree, pattern2 ) then          -- if a subtree that hasn't been substituted.
            	--local i,j,target = string.find(value, pattern2) -- only one subtree
            	local i=0
            	for bracedMarker in string.gmatch( selectedTree , pattern2 ) do
                    i=i+1

                    -- bracedMarker is either a marker in the tree or part of following
                    --     targetX={bracedMarker} ... |subcladeX=s  then
                    local pattern3 = "target(%u)=[%s]*"..bracedMarker

                    --?? if selectedTree == bracedMarker
			        if not string.find(selectedTree, pattern3 ) then
                     
	            	  	local subtree = p._label (frame, content, bracedMarker) 
		            	if subtree then

		            	 	--[[ method 1: the subtree code is substituted into main tree
		            	 	        this substitutes the subtree within the clade structure before processing
		            	 	        thus there will be a problem with large trees exceeding the expansion depth
		            	 	        however, they can be pruned before processing
		            	 	  ]]
--disable method 1		            	 	selectedTree = string.gsub(selectedTree, bracedMarker, subtree, 1)

		            	 	--[[method 2: add the subtree code before the final double brace
		            	 	    substitute "|targetX={FABIDS} |subcladeX=subtree" before last double brace of selectedTree
		            	 	    use capture in pattern3 to find X
		            	 	  ]]
		            	 	local i,j,X = string.find(content, pattern3)
		            	 	  
		            	 	if selectedTree == bracedMarker then
		            	 	    selectedTree = subtree
		            	 	else 
		            	 	   	  selectedTree = selectedTree:sub(1,-3)  -- trim final double brace
		            	 	          .. "\n|target" .. X .. "=" .. bracedMarker 
		            	 	          .. "\n|subclade" .. X .. "=" .. subtree .. ""
		            	 	          .. "\n }}"
		            	 	end 
		            	end
	            	end --substitution of subtree
            	end
            end

        	output = output .. selectedTree
        else
        	output = output .. p.errorMsg ("Failed to capture subclade with " .. mode .. " " ..section)
        end

	end
	
    if output ~= "" then 
		return output -- preprocess moved to entry function
	else 
		return '<span class="error">Section for label not found</span>' 
    end
end	

--================================== exclude LABEL ================================================

p.xlabel = function (frame, page, ...)
	local page = frame.args[1] --"User:Jts1882/sandbox/test/Passeriformes"
	local label = frame.args[1] or frame.args['label'] or frame.args['label1']
	                      -- page , target tree,  subtrees to exclude ...
	
	
	
	--                       page,   include clade, multple clades to exclude |
	return p._xlabel (frame, page, frame.args[2], frame.args[3], frame.args[4], frame.args[5])



end
p._xlabel = function (frame, targetTree, exclude, replace)

	
	local fullOutput =  targetTree
	--local fullOutput =  p._section(frame, page, target) 

	local output=targetTree -- return unmodified tree if nothing happens
	local section = exclude
	
	local targetType = "label%d"
	local cladePrefix = "%d"
	if string.upper( section) == section then
		 targetType = "target%u"               -- by convention subtrees must be uppercase
		 cladePrefix = "subclade%u"
    end

	--                label      = [[        name       ]]                 |n=   {...}
    local pattern = "("..targetType.."=[%s%p]*"..section .. "[%s%p]*.-"..cladePrefix.."=.-)(%b{})"
                                                                                -- ^^ this .- skips section tags before clade
                                                           -- ^^this .- skips |sublabel and styling following the label (but can return wrong clade when a subtree)
  
    local value = string.match( fullOutput , pattern ) 
    if value then
    	local trimmedTree, matches = string.gsub(fullOutput, pattern, "%1"..replace)--replaces pattern with capture %1
        return trimmedTree
    else
    	local message = ""
    	if string.upper(section) == section then 
    		message = "; subtree may have been substituted, try label"
        end
    	output = output .. p.warningMsg ("Failed to capture subclade for exclusion with label "..section..message)
    end



    if output ~= "" then 
		return  output .. '<span class="error">Nothing pruned</span>' 
		--return frame:preprocess(fullOutput)
	else 
		return '<span class="error">Section for label not found</span>' -- shouldn't get here 
    end
end	

--======================================== SECTION ==================================
p.section = function (frame)
	-------------------------target page  ---- sections
	return frame:preprocess(p._section(frame, mw.text.trim(frame.args[1]),frame.args[2],frame.args[3],frame.args[4],frame.args[5]))
end
p._section = function (frame,content,...)
	local args = { ... }
	local output = ""

	for k,v in pairs(args) do
		local section = mw.text.trim(v)
		--[[ note: using the non-greedy - in (.-) to allow capture of several sections 
		    this allows internal clade structures to be closed without capturing sisters clades
		    e.g. see section Tyranni in User:Jts1882/sandbox/test/Passeriformes
		]]
		local pattern = "<section begin="..section.."[ ]*/>(.-)<section end="..section.."[ ]*/>"

		for value in string.gmatch( content , pattern ) do
		    if value then 
				if frame.args.wrap or frame:getParent().args.wrap then
					local label1 = frame.args.wrap or frame:getParent().args.wrap 
					if label1 == "" then label1 = section end
					value =  "{{clade |label1=" .. label1 .. "|1=" .. value .. "}}"
			    end
				output = output .. value
			end

		end
	end
    if pargs['norefs'] or pargs['noref'] then                                                -- strip out references
	   --output =   mw.text.killMarkers( output ) 
	   if output:find("<ref")  then                       
			output = output:gsub('<ref[%w%p%s]-%/>', "") 
			output = output:gsub("<ref.-<%/ref>", "")                                      -- %C works, %w%p%s%c doesn't
	   end
	end
    
    if output ~= "" then 
		--return  frame:preprocess(output)
		return output -- leave preprocessing for entry function
	else 
		return '<span class="error">Section not found</span>' 
    end

end 

p.xsection = function (frame)
	local page = frame.args[1] --"User:Jts1882/sandbox/test/Passeriformes"
	local label = frame.args[1] or frame.args['label'] or frame.args['label1']
	                                       -- page , target tree,  sections to exclude ...	
	return frame:preprocess(p._xsection(frame, page ,frame.args[2],frame.args[3],frame.args[4],frame.args[5]))
end

p._xsection = function (frame,page, target, ...)
	local args = { ... }
	local output = ""
	local title = mw.title.new( page) -- , ns)  -- creates object if page doesn't exist (and valid page name)
	                                            --TODO: could use mw.title.makeTitle(), but that needs ns
	                                            
	
	
	                                            
	if title and title.exists then 
		local content = title:getContent()
		local fullOutput =  p._section(frame, page, target) 
	    output=fullOutput
	    
		
		for k,v in pairs(args) do
			local section = mw.text.trim(v)
			--[[ note: using the non-greedy - in (.-) to allow capture of several sections 
			    this allows internal clade structures to be closed without capturing sisters clades
			    e.g. see section Tyranni in User:Jts1882/sandbox/test/Passeriformes
			]]
			local pattern = "(<section begin="..section.."[ ]*/>)(.-)(<section end="..section.."[ ]*/>)"

            local value = string.match( fullOutput , pattern ) 

            
            if value then
            	local trimmedTree, matches = string.gsub(fullOutput, pattern, "replacement string")--replaces pattern with capture %1
 
            	output = output .. trimmedTree
            	output = output .. "<pre>" .. trimmedTree .. "</pre>"
                fullOutput = trimmedTree
            else
            	output = output .. p.errorMsg ("Failed to capture subclade with label "..section)
            end

		end
		

    else
    	return  '<span class="error">No page title found</span>'
	end
    
    if output ~= "" then 
		--return  frame:preprocess(output)
		return output -- leave preprocessing for entry function
	else 
		return '<span class="error">Section not found</span>' 
    end

end 

function p.firstToUpper(str)
    return (str:gsub("^%l", string.upper))
end
p.errorMsg = function (message)
	return '<span class="error">' .. message .. '</span>' 
end	
p.warningMsg = function (message)
	return '<span class="warning" style="color:#ac6600;font-size:larger;">' .. message .. '</span>'
end	
return p