Module:Cite taxon/utilities

 localp={}

 p.format_species_list=function(frame,taxa)

 locallist=frame.args[1]orframe:getParent().args[1]ortaxa
 localexpand=frame.args['expand']orframe:getParent().args['expand']-- want to subst template code
 localcompare=frame.args['compare']orframe:getParent().args['compare']
 localoption=frame.args['option']orframe:getParent().args['option']
 localmode=frame.args['mode']orframe:getParent().args['mode']
 localnolink=frame.args['nolink']orframe:getParent().args['nolink']

 list=mw.text.trim(list)

 localnames=mw.text.split(list,"\n")-- could use plain option
 --local genus, species, authority
 localoutput={}-- table of species names for output

 locali=1
 whilenames[i]do

 localname,match,skip
 localsep=" "-- space between genus and species
 localinfrasep=""
 localitalics="''"
 localgenus,species,subspecies,authority
 --local subspecies = ""

 names[i]=mw.text.trim(names[i])

 -- custom preprocess options
 ifnames[i]==""thenskip=trueend-- ignore blank lines
 ifcompareandnotnames[i]:find("^"..compare)then-- only consider lines beginning with compare text (e.g. =genus) 
 skip=true
 end
 ifoptionandstring.lower(option)=="worms"then-- process WoRMS list
 ifmode=="genus"then
 names[i]=names[i]:gsub("^Genus ","")-- delete leading "Genus"
 else
 names[i]=names[i]:gsub("^Species ","")-- delete leading "Species"
 end
 ifnames[i]:find("accepted as")then-- ignore invalid species, i.e. those "accepted as" something else
 skip=true
 end
 end
 ifoptionandstring.lower(option)=="gbif"then-- process GBIF list (copies with Species on alternate lines)
 ifnames[i]=="Species"ornames[i]=="Unranked"then
 skip=true
 end
 end
 ifoptionandstring.lower(option)=="tpl"then-- process tpl list 
 ifnames[i]:find("Accepted")then
 names[i]=names[i]:gsub("^(.+)	Accepted.+","%1")-- only include up to Accepted
 names[i]=mw.text.trim(names[i])-- needed as separate line as gsub returns str, i
 else
 skip=true-- skip if not accepted
 end
 end

 ifoptionandstring.lower(option)=="algaebase"then-- process algaebase
 ifnames[i]:find("C$")then
 names[i]=names[i]:gsub("C$","")-- only include C
 names[i]=mw.text.trim(names[i])-- needed as separate line as gsub returns str, i
 else
 skip=true-- skip if not accepted
 end
 end
 -- make list
 -- TODO try "(%S+)([%s×ばつ]+)(%S+) (.*)" 
 ifmode=="genus"ormode=="taxon"then-- assume form taxon authority
 forg,ainstring.gmatch(names[i],"(%S+) (.*)")do-- match: genus ×ばつ species authority
 genus=g
 species=""
 sep=""
 ifa~=""thenauthority=aend-- no authority after space
 match=true-- we have a match 
 ifmode=="taxon"thenitalics=""end
 end

 else-- match species list (various forms w/wo authority, hybrid)

 ifnotmatchthen
 --for g, s, f, ss, a in string.gmatch(names[i], "(%S+) (%S+)( subsp%. )(%S+) (.*)" ) do -- match: genus species subsp. subspecies authority
 forg,s,f,ss,ainstring.gmatch(names[i],"(%S+) (%S+)( [svfb][ubsparomiv]*%. )(%S+) (.*)")do-- match: genus species subsp. subspecies authority
 genus=g
 species=s
 subspecies=ss
 infrasep=f--" subsp. "
 sep=" "
 ifa~=""thenauthority=aend-- no authority after space
 match=true-- we have a match 
 end
 end
 ifnotmatchthen
 --for g, s, f, ss in string.gmatch(names[i], "(%S+) (%S+)( subsp%. )(%S+)" ) do -- match: genus species subsp. subspecies
 forg,s,f,ssinstring.gmatch(names[i],"(%S+) (%S+)( [svfb][ubsparomiv]*%. )(%S+)")do-- match: genus species subsp. subspecies
 genus=g
 species=s
 subspecies=ss
 infrasep=f-- " subsp. "
 sep=" "
 --if a ~= "" then authority = a end -- no authority after space
 match=true-- we have a match 
 end
 end
 ifnotmatchthen
 forg,s,ainstring.gmatch(names[i],"(%S+) ×ばつ (%S+) (.*)")do-- match: genus ×ばつ species authority
 genus=g
 species=s
 sep=" ×ばつ "
 ifa~=""thenauthority=aend-- no authority after space
 match=true-- we have a match 
 end
 end

 ifnotmatchthen
 forg,sinstring.gmatch(names[i],"(%S+) ×ばつ (%S+).*")do-- match: genus ×ばつ species
 genus=g
 species=s
 sep=" ×ばつ "
 ifa~=""thenauthority=aend
 match=true
 end
 end
 ifnotmatchthen
 forg,s,ainstring.gmatch(names[i],"(%S+) (%S+) (.*)")do-- match: genus species authority
 genus=g
 species=s
 ifa~=""thenauthority=aend-- no authority after space
 match=true-- we have a match for genus, species, authority
 end
 end
 ifnotmatchthen
 forg,sinstring.gmatch(names[i],"(%S+) (%S+).*")do-- match: genus species
 genus=g
 species=s
 match=true
 end
 end
 end

 ifmatchandnotskipthen
 localspecies_name=genus..sep..species
 ifsubspeciesthen
 name="''"..species_name.."''"..infrasep.."''"..subspecies.."''"
 ifnotnolinkthen
 name="[["..species_name..infrasep..subspecies.."|"..name.."]]"-- wikilinked name with redirect
 end
 else
 name=species_name
 ifnotnolinkthenname="[["..name.."]]"end-- add wikilink
 name=italics..name..italics-- add italics 
 end

 ifauthorityandexpandthen
 name=name.." "..frame:expandTemplate{title='small',args={authority}}--expand template
 elseifauthoritythen
 name=name.." {{small|"..authority.."}}"-- don't expand template
 end

 end
 ifnamethen
 table.insert(output,name)
 elseifnotskipthen
 table.insert(output,'<span class="error">unsupported format: expects "genus species authority"</span>')
 end


 i=i+1
 end

 return"*"..table.concat(output,"\n*")

 end

 p.format_taxon_list=function(frame)

 locallist=frame.args[1]orframe:getParent().args[1]
 localexpand=frame.args['expand']orframe:getParent().args['expand']-- want to subst template code

 list=mw.text.trim(list)

 localnames=mw.text.split(list,"\n",plain)
 --local genus, species, authority
 localoutput={}-- table of species names for output
 localpatterns={"(%S+)( ×ばつ )(%S+) (.*)",-- genus ×ばつ species authority
 "(%S+)( ×ばつ )(%S+).*",-- genus ×ばつ species
 "(%S+)( )(%S+) (.*)",-- genus species authority
 "(%S+)( )(%S+).*",-- genus species 
 -- taxon or genus authority
 }
 locali=1
 whilenames[i]do
 localname,match
 --local sep = " " -- space between genus and species
 localgenus,separator,species,authority
 -- TODO try "(%S+)([%s×ばつ]+)(%S+) (.*)" 
 fork,vinpairs(patterns)do
 forg,sep,s,ainstring.gmatch(names[i],v)do-- match: genus ×ばつ species authority
 genus=g
 species=s
 separator=sep
 ifaanda~=""thenauthority=aend-- no authority after space
 match=true-- we have a match 
 end
 ifmatchthenbreakend
 end

 ifmatchthen
 name="''[["..genus..separator..species.."]]''"
 ifauthorityandexpandthen
 name=name..frame:expandTemplate{title='small',args={authority}}--expand template
 elseifauthoritythen
 name=name.."{{small|"..authority.."}}"-- don't expand template
 end

 end
 ifnamethen
 table.insert(output,name)
 else
 table.insert(output,'<span class="error">unsupported format: expects "genus species authority"</span>')
 end


 i=i+1
 end

 return"*"..table.concat(output,"\n*")

 end
 returnp