imported>Jts1882: add algaebase preprocess

2021-11-24T14:23:42Z

add algaebase preprocess

New page

local p = {}

p.format_species_list = function(frame, taxa)

local list = frame.args[1] or frame:getParent().args[1] or taxa
local expand = frame.args['expand'] or frame:getParent().args['expand'] -- want to subst template code
local compare = frame.args['compare'] or frame:getParent().args['compare']
local option = frame.args['option'] or frame:getParent().args['option']
local mode = frame.args['mode'] or frame:getParent().args['mode']
local nolink = frame.args['nolink'] or frame:getParent().args['nolink']

list = mw.text.trim( list )

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

local i = 1
while names[i] do

local name, match, skip
local sep = " " -- space between genus and species
local infrasep = ""
local italics = "''"
local genus, species, subspecies, authority
--local subspecies = ""

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

-- custom preprocess options
if names[i] == "" then skip = true end -- ignore blank lines
if compare and not names[i]:find("^" .. compare) then -- only consider lines beginning with compare text (e.g. =genus)
skip = true
end
if option and string.lower(option) == "worms" then -- process WoRMS list
names[i] = names[i]:gsub("^Species ", "") -- delete leading "Species"
if names[i]:find("accepted as") then -- ignore invalid species, i.e. those "accepted as" something else
skip = true
end
end
if option and string.lower(option) == "gbif" then -- process GBIF list (copies with Species on alternate lines)
if names[i] == "Species" or names[i] == "Unranked" then
skip = true
end
end
if option and string.lower(option) == "tpl" then -- process tpl list
if names[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

if option and string.lower(option) == "algaebase" then -- process algaebase
if names[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×]+)(%S+) (.*)"
if mode == "genus" or mode == "taxon" then -- assume form taxon authority
for g, a in string.gmatch(names[i], "(%S+) (.*)" ) do -- match: genus × species authority
genus = g
species = ""
sep = ""
if a ~= "" then authority = a end -- no authority after space
match = true -- we have a match
if mode == "taxon" then italics = "" end
end

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

if not match then
--for g, s, f, ss, a in string.gmatch(names[i], "(%S+) (%S+)( subsp%. )(%S+) (.*)" ) do -- match: genus species subsp. subspecies authority
for g, s, f, ss, a in string.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 = " "
if a ~= "" then authority = a end -- no authority after space
match = true -- we have a match
end
end
if not match then
--for g, s, f, ss in string.gmatch(names[i], "(%S+) (%S+)( subsp%. )(%S+)" ) do -- match: genus species subsp. subspecies
for g, s, f, ss in string.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
if not match then
for g, s, a in string.gmatch(names[i], "(%S+) × (%S+) (.*)" ) do -- match: genus × species authority
genus = g
species = s
sep = " × "
if a ~= "" then authority = a end -- no authority after space
match = true -- we have a match
end
end

if not match then
for g, s in string.gmatch(names[i], "(%S+) × (%S+).*" ) do -- match: genus × species
genus = g
species = s
sep = " × "
if a ~= "" then authority = a end
match = true
end
end
if not match then
for g, s, a in string.gmatch(names[i], "(%S+) (%S+) (.*)" ) do -- match: genus species authority
genus = g
species = s
if a ~= "" then authority = a end -- no authority after space
match = true -- we have a match for genus, species, authority
end
end
if not match then
for g, s in string.gmatch(names[i], "(%S+) (%S+).*" ) do -- match: genus species
genus = g
species = s
match = true
end
end
end

if match and not skip then
local species_name = genus .. sep .. species
if subspecies then
name = "''" .. species_name .. "''" .. infrasep .. "''" .. subspecies .."''"
if not nolink then
name = "[[" .. species_name .. infrasep .. subspecies .. "|" .. name .. "]]" -- wikilinked name with redirect
end
else
name = species_name
if not nolink then name = "[[" .. name .. "]]" end -- add wikilink
name = italics .. name .. italics -- add italics
end

if authority and expand then
name = name.. " " .. frame:expandTemplate{ title = 'small', args = { authority } } --expand template
elseif authority then
name = name .. " {{small|" .. authority .."}}" -- don't expand template
end

end
if name then
table.insert ( output , name )
elseif not skip then
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)

local list = frame.args[1] or frame:getParent().args[1]
local expand = frame.args['expand'] or frame:getParent().args['expand'] -- want to subst template code

list = mw.text.trim( list )

local names = mw.text.split( list, "\n", plain )
--local genus, species, authority
local output = { } -- table of species names for output
local patterns = { "(%S+)( × )(%S+) (.*)", -- genus × species authority
"(%S+)( × )(%S+).*", -- genus × species
"(%S+)( )(%S+) (.*)", -- genus species authority
"(%S+)( )(%S+).*", -- genus species
-- taxon or genus authority
}
local i = 1
while names[i] do
local name, match
--local sep = " " -- space between genus and species
local genus, separator, species, authority
-- TODO try "(%S+)([%s×]+)(%S+) (.*)"
for k, v in pairs (patterns) do
for g, sep, s, a in string.gmatch(names[i], v ) do -- match: genus × species authority
genus = g
species = s
separator = sep
if a and a ~= "" then authority = a end -- no authority after space
match = true -- we have a match
end
if match then break end
end

if match then
name = "''[[" .. genus .. separator .. species .. "]]''"
if authority and expand then
name = name .. frame:expandTemplate{ title = 'small', args = { authority } } --expand template
elseif authority then
name = name .. "{{small|" .. authority .."}}" -- don't expand template
end

end
if name then
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
return p

Module:FishRef/utilities - Revision history

imported>Jts1882: add algaebase preprocess