Module:Sandbox/genewiki/geneboxdev
--This code is originally from https://en.wikipedia.org/wiki/Module:Sandbox/RexxS/AA and copied here (9/17/2015) for modifications and reusablity
-- Testing for arbitrary access
-- Intended to be:
-- Use : {{#invoke:Wikidata|getValueFromID|
-- E.g.: {{#invoke:Wikidata|getValueFromID|Q151973|P26|FETCH_WIKIDATA}} - to fetch value of 'spouse' (P26) from 'Richard Burton' (Q151973)
-- While in sandbox:
-- Use : {{#invoke:Sandbox/RexxS/AA|getValueFromID|
-- E.g.: {{#invoke:Sandbox/RexxS/AA|getValueFromID|Q151973|P26|FETCH_WIKIDATA}} - to fetch value of 'spouse' (P26) from 'Richard Burton' (Q151973)
-- Added Extra parameters for chosing rank (preferred, normal, truth, depreciated and random value (just first in list)
-- E.g.: {{#invoke:Sandbox/RexxS/AA|getValueFromID|Q151973|P26|FETCH_WIKIDATA|RANK_PREFERRED|RANDOM}}
local p = {}
-- This is used to get a value, or a comma separated list of them if multiple values exist
p.getValueFromID = function(frame)
local itemID = mw.text.trim(frame.args[1] or "")
local propertyID = mw.text.trim(frame.args[2] or "")
local input_parm = mw.text.trim(frame.args[3] or "")
local input_rank = mw.text.trim(frame.args[4] or "")
local random = mw.text.trim(frame.args[5] or "")
if input_parm == "FETCH_WIKIDATA" then
local entity = mw.wikibase.getEntity(itemID)
local claims = entity.claims[propertyID]
if claims then
-- if wiki-linked value output as link if possible
if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid" ) then
local out = {}
for k, v in pairs(claims) do
local sitelink = mw.wikibase.sitelink("Q" .. v.mainsnak.datavalue.value["numeric-id"])
local label = mw.wikibase.label("Q" .. v.mainsnak.datavalue.value["numeric-id"])
if label == nil then label = "Q" .. v.mainsnak.datavalue.value["numeric-id"] end
if sitelink then
out[#out + 1] = "" .. label .. ""
else
--out[#out + 1] = "" .. label .. "[*]"
out[#out + 1] = "" .. label .. ""
end
end
return table.concat(out, ", ")
else
local results
if input_rank == "" then
results = entity:formatPropertyValues(propertyID, mw.wikibase.entity.claimRanks).value
else
results = entity:formatPropertyValues(propertyID, mw.wikibase.entity.claimRanks.input_rank).value
end
if random == "" then
return results
else
local results_split = mw.text.split(results, ",")
return results_split[1]
end
end
else
return ""
end
else
return input_parm
end
end
-- A function to return the QID of a property value, rather than its text label
-- May be useful for constructing chains of calls to get properties of properties, etc.
-- It returns the QID of only the first property value if more than one
-- Use like this: {{#invoke:Sandbox/RexxS/AA|getQIDFromID|Q151973|P26|FETCH_WIKIDATA}}
-- That will fetch the QID of the first value for the spouse (P26) of Richard Burton (Q151973)
-- Returns an empty string if the value doesn't exist or has no QID.
p.getQIDFromID = function(frame)
local itemID = mw.text.trim(frame.args[1] or "")
local propertyID = mw.text.trim(frame.args[2] or "")
local input_parm = mw.text.trim(frame.args[3] or "")
local entity = mw.wikibase.getEntity(itemID)
local claims = entity.claims[propertyID]
if claims then
-- if wiki-linked value return the QID of the first value of the property
if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid") then
return "Q" .. claims[1].mainsnak.datavalue.value["numeric-id"]
else
return ""
end
else
return ""
end
end
p.getRawValueFromID = function(frame)
local itemID = mw.text.trim(frame.args[1] or "")
local propertyID = mw.text.trim(frame.args[2] or "")
local input_parm = mw.text.trim(frame.args[3] or "")
if input_parm == "FETCH_WIKIDATA" then
local entity = mw.wikibase.getEntity(itemID)
local claims = entity.claims[propertyID]
if claims then
local result = entity:formatPropertyValues(propertyID, mw.wikibase.entity.claimRanks).value
-- if number type: remove thousand separators
if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "quantity") then
result = mw.ustring.gsub(result, "(%d),(%d)", "%1%2")
end
return result
else
return ""
end
else
return input_parm
end
end
--This could be expanded into more levels taking a list in. But the deeper one abstracts the more likely one level won't return a value and error checking could be tedious.
--{{#invoke:Sandbox/RexxS/AA|getQIDFromID|Q151973|P26|P35|FETCH_WIKIDATA}}
p.getQIDFromID_two_levels = function(frame)
local parent_itemID = mw.text.trim(frame.args[1] or "") ---first Qid
local parent_propertyID = mw.text.trim(frame.args[2] or "")
local child_propertyID = mw.text.trim(frame.args[3] or "")
local input_parm = mw.text.trim(frame.args[4] or "")
local entity_parent = mw.wikibase.getEntity(parent_itemID)
local claims_parent = entity_parent.claims[parent_propertyID]
if claims_parent then
-- if wiki-linked value return the QID of the first value of the property
if (claims_parent[1] and claims_parent[1].mainsnak.snaktype == "value" and claims_parent[1].mainsnak.datavalue.type == "wikibase-entityid") then
local child_itemID = "Q" .. claims_parent[1].mainsnak.datavalue.value["numeric-id"]
local entity_child = mw.wikibase.getEntity(child_itemID)
local claims_child = entity_child.claims[child_propertyID]
--wiki-linked value of the second QID
if (claims_child[1] and claims_child[1].mainsnak.snaktype == "value" and claims_child[1].mainsnak.datavalue.type == "wikibase-entityid") then
return "Q"..claims_child[1].mainsnak.datavalue.value["numeric-id"]
else
return ""
end
else
return ""
end
else
return ""
end
end
p.getQualifierID = function(frame)
local propertyID = mw.text.trim(frame.args[1] or "")
local qualifierID = mw.text.trim(frame.args[2] or "")
local input_parm = mw.text.trim(frame.args[3] or "")
local itemID = mw.text.trim(frame.args[4] or "") --if direct link from wikidata item
if input_parm == "FETCH_WIKIDATA" then
local entity = ""
if itemID ~= "" then
entity = mw.wikibase.getEntityObject(itemID)
else
entity = mw.wikibase.getEntityObject()
end
if entity.claims[propertyID] ~= nil then
local out = {}
for k, v in pairs(entity.claims[propertyID]) do
for k2, v2 in pairs(v.qualifiers[qualifierID]) do
if v2.snaktype == 'value' then
out[#out + 1] = "Q" .. v2.datavalue.value["numeric-id"]
end
end
end
return table.concat(out, ", ")
else
return ""
end
else
return input_parm
end
end
p.getAliasFromGenomeAssembly_hs = function(frame)
-- will contain the numeric value for the requested coordinate
local output = ""
local sep = " "
-- can only be P644 (genomic start) or P645 (genomic end) for this to work
-- should probably try to catch that. Might also increase legibility to use specific variable names when possible
local propertyID = mw.text.trim(frame.args[1] or "")
-- this can really only be P659 right now. I'm not sure of the value of including it as a parameter as other values will likely break this function
local qualifierID = mw.text.trim(frame.args[2] or "")
-- Why do we include this here? What should happen if FETCH_WIKIDATA is not included?
local input_parm = mw.text.trim(frame.args[3] or "")
-- this can needs to be fed to the function either by a call to {{#invoke:Wikidata|pageId}} or by setting it directly (e.g. if the function was applied on a page other than the targeted gene)
--alert if this id is not a valid thing in wikidata, a Lua error will occur that says
--The ID entered is unknown to the system. Please use a valid entity ID.
local itemID = mw.text.trim(frame.args[4] or "")
-- will track the different builds pulled from the qualifiers
local newest_build = "0"
-- starts the process
local entity = mw.wikibase.getEntityObject(itemID)
local claims
--gets a table of claims on the (genomic start or end) property Q19847637
if entity and entity.claims then
claims = entity.claims[propertyID]
end
--will return nothing if no claims are found
if claims then
--checking to be sure claims is populated, not sure it its needed
if (claims[1] ) then
--useful for debugging
--local out = {}
--pulls the genome location from the claim
for k, v in pairs(claims) do
local quals = v.qualifiers.P659
--if there are any
if quals then
for qk, qv in pairs(quals) do
local qual_obj_id = "Q"..qv.datavalue.value["numeric-id"]
--get to the entity targeted by the qualifier property. Genome builds are Items in wikidata
local qual_obj = mw.wikibase.getEntityObject(qual_obj_id)
local alias = ""
--this uses the aliases to pull out version numbers
--seems like there ought to be a better way to do this, but likely would need to change the data added by the bot
if qual_obj["aliases"] ~= nil then
local test = qual_obj["aliases"]["en"]
for key, value in ipairs(test) do
if string.match(value['value'], '^hg') then
alias = value['value']
local build_no = alias:gsub("hg","")
--report only the most location associated with the most recent build
--if there is more than one location per build, just give one back as that is not our problem right now.
if build_no > newest_build then
newest_build = build_no
end
end
end
end
end
--in case there are no qualifiers, but there is a location, might as well return it
else output = location
end
end
return "hg"..newest_build
else
return ""
end
else
return ""
--debug
--"no claims for "..itemID.." prop "..propertyID
end
end
--in future could just combine this with getChromosomeLoc once all in once code
p.getAliasFromGenomeAssembly_mm = function(frame)
-- will contain the numeric value for the requested coordinate
local output = ""
local sep = " "
-- can only be P644 (genomic start) or P645 (genomic end) for this to work
-- should probably try to catch that. Might also increase legibility to use specific variable names when possible
local propertyID = mw.text.trim(frame.args[1] or "")
-- this can really only be P659 right now. I'm not sure of the value of including it as a parameter as other values will likely break this function
local qualifierID = mw.text.trim(frame.args[2] or "")
-- Why do we include this here? What should happen if FETCH_WIKIDATA is not included?
local input_parm = mw.text.trim(frame.args[3] or "")
-- this can needs to be fed to the function either by a call to {{#invoke:Wikidata|pageId}} or by setting it directly (e.g. if the function was applied on a page other than the targeted gene)
--alert if this id is not a valid thing in wikidata, a Lua error will occur that says
--The ID entered is unknown to the system. Please use a valid entity ID.
local itemID = mw.text.trim(frame.args[4] or "")
-- will track the different builds pulled from the qualifiers
local newest_build = "0"
-- starts the process
local mouse_propertyID = "P684"
--get the QID for the mouse gene
local mouse_itemID
--use itemID (QID) to get the mouse QID to return start or end location (propertyID)
local entity_gene = mw.wikibase.getEntity(itemID)
local claims_gene = entity_gene.claims[mouse_propertyID]
if claims_gene then
local claims_mouse
if (claims_gene[1] and claims_gene[1].mainsnak.snaktype == "value" and claims_gene[1].mainsnak.datavalue.type == "wikibase-entityid") then
mouse_itemID = "Q" .. claims_gene[1].mainsnak.datavalue.value["numeric-id"]
local entity_mouse = mw.wikibase.getEntity(mouse_itemID)
claims_mouse = entity_mouse.claims[propertyID]
end --will return nothing if no claims are found
if claims_mouse then
--checking to be sure claims is populated, not sure it its needed
if (claims_mouse[1] ) then
--useful for debugging
--local out = {}
--pulls the genome location from the claim
for k, v in pairs(claims_mouse) do
local quals = v.qualifiers.P659
--if there are any
if quals then
for qk, qv in pairs(quals) do
local qual_obj_id = "Q"..qv.datavalue.value["numeric-id"]
--get to the entity targeted by the qualifier property. Genome builds are Items in wikidata
local qual_obj = mw.wikibase.getEntityObject(qual_obj_id)
local alias = ""
--this uses the aliases to pull out version numbers
--seems like there ought to be a better way to do this, but likely would need to change the data added by the bot
if qual_obj["aliases"] ~= nil then
local test = qual_obj["aliases"]["en"]
for key, value in ipairs(test) do
if string.match(value['value'], '^mm') then
alias = value['value']
local build_no = alias:gsub("mm","")
--report only the most location associated with the most recent build
--if there is more than one location per build, just give one back as that is not our problem right now.
if build_no > newest_build then
newest_build = build_no
end
end
end
end
end
--in case there are no qualifiers, but there is a location, might as well return it
else
end
end
return "mm"..newest_build
else
return ""
end
else
return ""
--debug
--"no claims for "..itemID.." prop "..propertyID
end
end
end
---getChromosomeLoc
---input propertyID ie(Genomic start) P644
--- qualifierID (ie GenLoc Assembly) P659
--- input_parm (ie FETCH_WIKIDATA)
--- {{#invoke:Sandbox/genewiki/geneboxdev|getChromosomeLoc|P644|P659|FETCH_WIKIDATA}}
---output preferred chromosome location start value in this case it would be 49893092
--for debug window -- Q14865053
--frame = mw.getCurrentFrame()
--frame.args = {"P644","P659","FETCH_WIKIDATA","Q14865053"}
--print(p.getChromosomeLoc(frame))
p.getChromosomeLoc = function(frame)
-- will contain the numeric value for the requested coordinate
local output = ""
local sep = " "
-- can only be P644 (genomic start) or P645 (genomic end) for this to work
-- should probably try to catch that. Might also increase legibility to use specific variable names when possible
local propertyID = mw.text.trim(frame.args[1] or "")
-- this can really only be P659 right now. I'm not sure of the value of including it as a parameter as other values will likely break this function
local qualifierID = mw.text.trim(frame.args[2] or "")
-- Why do we include this here? What should happen if FETCH_WIKIDATA is not included?
local input_parm = mw.text.trim(frame.args[3] or "")
-- this can needs to be fed to the function either by a call to {{#invoke:Wikidata|pageId}} or by setting it directly (e.g. if the function was applied on a page other than the targeted gene)
--alert if this id is not a valid thing in wikidata, a Lua error will occur that says
--The ID entered is unknown to the system. Please use a valid entity ID.
local itemID = mw.text.trim(frame.args[4] or "")
-- will track the different builds pulled from the qualifiers
local newest_build = "0"
-- starts the process
local entity = mw.wikibase.getEntityObject(itemID)
local claims
--gets a table of claims on the (genomic start or end) property Q19847637
if entity and entity.claims then
claims = entity.claims[propertyID]
end
--will return nothing if no claims are found
if claims then
--checking to be sure claims is populated, not sure it its needed
if (claims[1] ) then
--useful for debugging
--local out = {}
--pulls the genome location from the claim
for k, v in pairs(claims) do
local location = v.mainsnak.datavalue.value
--debugging
--out[#out + 1] = k.." location:" .. location.. " || "
--gets the qualifiers linked to the current claim
local quals = v.qualifiers.P659
--if there are any
if quals then
for qk, qv in pairs(quals) do
local qual_obj_id = "Q"..qv.datavalue.value["numeric-id"]
--get to the entity targeted by the qualifier property. Genome builds are Items in wikidata
local qual_obj = mw.wikibase.getEntityObject(qual_obj_id)
local alias = ""
--this uses the aliases to pull out version numbers
--seems like there ought to be a better way to do this, but likely would need to change the data added by the bot
if qual_obj["aliases"] ~= nil then
local test = qual_obj["aliases"]["en"]
for key, value in ipairs(test) do
if string.match(value['value'], '^hg') then
alias = value['value']
local build_no = alias:gsub("hg","")
--report only the most location associated with the most recent build
--if there is more than one location per build, just give one back as that is not our problem right now.
if build_no > newest_build then
output = location
newest_build = build_no
end
end
end
end
end
--in case there are no qualifiers, but there is a location, might as well return it
else output = location
end
end
return output
else
return ""
end
else
return ""
--debug
--"no claims for "..itemID.." prop "..propertyID
end
end
p.getChromosomeLoc_mm = function(frame)
-- will contain the numeric value for the requested coordinate
local output = ""
local sep = " "
-- can only be P644 (genomic start) or P645 (genomic end) for this to work
-- should probably try to catch that. Might also increase legibility to use specific variable names when possible
local propertyID = mw.text.trim(frame.args[1] or "")
-- this can really only be P659 right now. I'm not sure of the value of including it as a parameter as other values will likely break this function
local qualifierID = mw.text.trim(frame.args[2] or "")
-- Why do we include this here? What should happen if FETCH_WIKIDATA is not included?
local input_parm = mw.text.trim(frame.args[3] or "")
-- this can needs to be fed to the function either by a call to {{#invoke:Wikidata|pageId}} or by setting it directly (e.g. if the function was applied on a page other than the targeted gene)
--alert if this id is not a valid thing in wikidata, a Lua error will occur that says
--The ID entered is unknown to the system. Please use a valid entity ID.
local itemID = mw.text.trim(frame.args[4] or "")
-- will track the different builds pulled from the qualifiers
local newest_build = "0"
-- starts the process
local mouse_propertyID = "P684"
--get the QID for the mouse gene
local mouse_itemID
--use itemID (QID) to get the mouse QID to return start or end location (propertyID)
local entity_gene = mw.wikibase.getEntity(itemID)
local claims_gene = entity_gene.claims[mouse_propertyID]
if claims_gene then
local claims_mouse
if (claims_gene[1] and claims_gene[1].mainsnak.snaktype == "value" and claims_gene[1].mainsnak.datavalue.type == "wikibase-entityid") then
mouse_itemID = "Q" .. claims_gene[1].mainsnak.datavalue.value["numeric-id"]
local entity_mouse = mw.wikibase.getEntity(mouse_itemID)
claims_mouse = entity_mouse.claims[propertyID]
end --will return nothing if no claims are found
if claims_mouse then
--checking to be sure claims is populated, not sure it its needed
if (claims_mouse[1] ) then
--useful for debugging
--local out = {}
--pulls the genome location from the claim
for k, v in pairs(claims_mouse) do
local location = v.mainsnak.datavalue.value
--debugging
--out[#out + 1] = k.." location:" .. location.. " || "
--gets the qualifiers linked to the current claim
local quals = v.qualifiers.P659
--if there are any
if quals then
for qk, qv in pairs(quals) do
local qual_obj_id = "Q"..qv.datavalue.value["numeric-id"]
--get to the entity targeted by the qualifier property. Genome builds are Items in wikidata
local qual_obj = mw.wikibase.getEntityObject(qual_obj_id)
local alias = ""
--this uses the aliases to pull out version numbers
--seems like there ought to be a better way to do this, but likely would need to change the data added by the bot
if qual_obj["aliases"] ~= nil then
local test = qual_obj["aliases"]["en"]
for key, value in ipairs(test) do
if string.match(value['value'], '^mm') then
alias = value['value']
local build_no = alias:gsub("mm","")
--report only the most location associated with the most recent build
--if there is more than one location per build, just give one back as that is not our problem right now.
if build_no > newest_build then
output = location
newest_build = build_no
end
end
end
end
end
--in case there are no qualifiers, but there is a location, might as well return it
else output = location
end
end
return output
else
return ""
end
else
return ""
--debug
--"no claims for "..itemID.." prop "..propertyID
end
end
end
--eg:{{#invoke:Sandbox/genewiki/geneboxdev|getChromosomeLoc|P644|FETCH_WIKIDATA}}
--This function is used to generate a list of aliases
--To Do: exclude gene symbol that is already displayed
function p.get_aliases(frame)
entity = mw.wikibase.getEntityObject(frame.args['from'])
a = ''
if entity['aliases'] ~= nil then
test = entity['aliases']['en']
for key, value in ipairs(test) do
a = a .. ', ' .. value['value']
end
return a
end
end
p.trimChromosome = function(frame)
local string_to_trim = mw.text.trim(frame.args[1] or "")
local out = '';
if string.find(string_to_trim, 'chromosome.') then
out = string_to_trim:gsub("chromosome ", "")
end
if string.find(string_to_trim, 'mouse.chromosome.') then
out = string_to_trim:gsub("mouse chromosome ", "")
end
return out
end
--EG: {{#invoke:Sandbox/genewiki/geneboxdev|getGO|Q14864805|P680|P686|FETCH_WIKIDATA}}
p.getGO = function(frame)
local itemID_parent = mw.text.trim(frame.args[1] or "")
local propertyID_parent = mw.text.trim(frame.args[2] or "") -- ie molecular, cellular, function
local propertyID_child = mw.text.trim(frame.args[3] or "") -- Gene Ontology ID
local input_parm = mw.text.trim(frame.args[4] or "")
if input_parm == "FETCH_WIKIDATA" then
local entity = mw.wikibase.getEntity(itemID_parent)
local claims = entity.claims[propertyID_parent]
if claims then
if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid") then
local out = {}
for k, v in pairs(claims) do
local itemID_child = "Q" .. v.mainsnak.datavalue.value["numeric-id"]
local entity = mw.wikibase.getEntityObject(itemID_child)
local claims
local result_GOID = ''
if entity then claims = entity.claims[propertyID_child] end
if claims then
result_GOID = entity:formatPropertyValues(propertyID_child, mw.wikibase.entity.claimRanks).value
else
return ""
end
local sitelink = "http://amigo.geneontology.org/amigo/term/GO:"
local label = mw.wikibase.label("Q" .. v.mainsnak.datavalue.value["numeric-id"])
if label == nil then label = "Q" .. v.mainsnak.datavalue.value["numeric-id"] end
if sitelink then
---out[#out + 1] = "*[" .. sitelink .. "|" .. label .."]\n"
out[#out + 1] = "*[" .. sitelink .. result_GOID .. " " .. label .."]\n"
else
out[#out + 1] = "*" .. label .. "\n"
end
end
return table.concat(out, "")
else
return entity:formatPropertyValues(propertyID, mw.wikibase.entity.claimRanks).value
end
else
return ""
end
else
return input_parm
end
end
p.getPDB = function(frame)
local itemID = mw.text.trim(frame.args[1] or "")
local propertyID = mw.text.trim(frame.args[2] or "")
local input_parm = mw.text.trim(frame.args[3] or "")
if input_parm == "FETCH_WIKIDATA" then
local entity = mw.wikibase.getEntity(itemID)
local claims = entity.claims[propertyID]
local sitelink = "http://www.rcsb.org/pdb/explore/explore.do?pdbId="
if claims then
if (claims[1] and claims[1].mainsnak.snaktype == "value") then
local out = {}
for k, v in pairs(claims) do
--don't think we want the Q here, right ?
--local label = mw.wikibase.label("Q" .. v.mainsnak.datavalue.value)
--if label == nil then label = "Q" .. v.mainsnak.datavalue.value end
local label = mw.wikibase.label(v.mainsnak.datavalue.value)
if label == nil then label = v.mainsnak.datavalue.value end
if sitelink then
out[#out + 1] = "[" .. sitelink .. label .. " " ..label .. "]"
else
out[#out + 1] = "" .. label .. ""
end
end
return table.concat(out, ", ")
else
return entity:formatPropertyValues(propertyID, mw.wikibase.entity.claimRanks).value
end
else
return ""
end
else
return input_parm
end
end
function p.ViewSomething(frame)
local itemID = mw.text.trim(frame.args[1] or "")
local data = mw.wikibase.getEntityObject(itemID)
if not data then
return nil
end
local f = frame.args[1] and frame or frame:getParent()
local i = 1
while true do
local index = f.args[i]
if not index then
if type(data) == "table" then
return mw.text.jsonEncode(data, mw.text.JSON_PRESERVE_KEYS + mw.text.JSON_PRETTY)
else
return tostring(data)
end
end
data = data[index] or data[tonumber(index)]
if not data then
return
end
i = i + 1
end
end
return p