require('strict');
localdata=mw.loadData('Module:Footnotes/anchor id list/data');
localwhitelist=mw.loadData('Module:Footnotes/whitelist');
localLang_obj=mw.language.getContentLanguage();-- used by template_list_add() to uppercase first letter of template name TODO: better way to do that?
localredirects_date={
['date']=true,
['datetomos']=true,
['formatdate']=true,
['isotodmymdy']=true,
['isotomos']=true,
}
localredirects_patent={-- special case cs1-like templates because uses different parameters for name and date in anchor ID
['Cite patent']=true,
['Citeref patent']=true,
['Ref patent']=true,
}
localredirects_sfnref={
['sfnref']=true,
['harvid']=true,
}
localaliases_author={-- these use pseudo-patterns in the same way as cs1|2; '#' represents 1 or more enumerator digits
'last#',
'author#',
'surname#',
'author-last#',
'author#-last',
'subject#',
'host#',
}
localaliases_contributor={
'contributor#',
'contributor-last#',
'contributor#-last',
'contributor-surname#',
'contributor#-surname',
}
localaliases_editor={
'editor#',
'editor-last#',
'editor#-last',
'editor-surname#',
'editor#-surname',
}
localaliases_harvc_author={
'last#',
'author#',
}
localaliases_inventor={-- cite patent
'inventor#',
'inventor-last#',
'inventor#-last',
'inventor-surname#',
'inventor#-surname',
'invent#',
'invent-#',
}
localalias_patterns_date={-- normal lua patterns for most cs1|2-like templates
'|%s*year%s*=%s*',
'|%s*date%s*=%s*',
'|%s*publication%-?date%s*=%s*',
'|%s*air%-?date%s*=%s*',
}
localalias_patterns_harvc_date={-- normal lua patterns for harvc template
'|%s*anchor%-year%s*=%s*',
'|%s*year%s*=%s*',
}
localalias_patterns_patent_date={-- normal lua patterns for cite patent templates
'|%s*issue%-date%s*=%s*',
'|%s*gdate%s*=%s*',
'|%s*publication%-date%s*=%s*',
'|%s*pubdate%s*=%s*',
}
localpatterns_date={-- normal lua patterns
--	'(%d%d%d%d–%d%d%d%d%l?)$',													-- YYYY–YYYY four-digit year range at end (Season YYYY–YYYY); with or without dab
'(%d%d%d%d)%D+(%d%d%d%d%l?)$',-- any range with four-digit years; with or without dab; not two captures
'^(%d%d%d%d–%d%d%l?)$',-- YYYY–YY two-digit year range; with or without dab
'^(c%. %d%d%d%d?%l?)$',-- three- or four-digit circa year; with or without dab
'(%d%d%d%d?%l?)$',-- three- or four-digit year at end of date (dmy or mdy); with or without dab
'^(%d%d%d%d?%l?)',-- three- or four-digit year at beginning of date (ymd or YYYY); with or without dab
'^(n%.d%.%l?)$',-- 'no date' with dots; with or without dab
'^(nd%l?)$',-- 'no date' without dots; with or without dab
}
localpatterns_tags={
'<nowiki>.-</nowiki>',
'<!%-%-.-%-%->',
'<pre>.-</pre>',
'<syntaxhighlight.->.-</syntaxhighlight>',
'<source.->.-</source>',-- deprecated alias of syntaxhighlight tag
}
localtemplate_skip={-- templates to be skipped for whatever reason; mostly because they resemble cs1-like templates
['Citation-attribution']=true,
}
localglobal_article_content=nil
localglobal_anchor_id_list=nil-- exported tables
localglobal_template_list=nil
localglobal_article_whitelist=nil
--[[--------------------------< A R T I C L E _ C O N T E N T _ G E T >----------------------------------------
get article content, remove certain html-like tags and their content so that this code doesn't include any citation
templates inside the tags as valid tagets; they are not.
]]
localfunctionarticle_content_get()
ifglobal_article_contentthenreturnglobal_article_contentend
localarticle_content=mw.title.getCurrentTitle():getContent()or'';-- get the content of the article or ''; new pages edited w/ve do not have 'content' until saved; ve does not preview; phab:T221625
for_,taginipairs(patterns_tags)do
article_content=article_content:gsub(tag,'');-- remove certain html-like tags and their content
end
global_article_content=article_content
returnarticle_content
end
--[[--------------------------< S F N R E F _ G E T >----------------------------------------------------------
make an anchor id from the contents of {{sfnref}} or {{harvid}}. this function assumes that {{sfnref}} and {{harvid}}
are correctly formed.
]]
localfunctionsfnref_get(template)
template=template:gsub('{{%s*(.-)%s*}}','%1');-- strip bounding template markup and trim
localparts=mw.text.split(template,'%s*|%s*');-- split at the pipe and remove extraneous space characters
localanchor_id={};
ifredirects_sfnref[parts[1]:lower()]then
anchor_id[1]='CITEREF';
else
returnnil;-- not an sfnref or harvid template
end
locali=2;-- indexer into parts{} table
localj=2;-- indexer into anchor_id{} table which already has 'CITEREF' at [1]
whileparts[i]and7>jdo-- loop through what should be just positional parameters for names and year (2-6 four names and a date)
ifnotparts[i]:find('=')then-- look for equal sign (named paraneter in a template that doesn't support named parameters)
anchor_id[j]=parts[i];-- positional parameters are saved
j=j+1;-- bump the anchor_id{} indexer
end
i=i+1;-- bump the parts{} indexer
end
returntable.concat(anchor_id,'');
end
--[[--------------------------< D A T E _ G E T >--------------------------------------------------------------
extract year from one of |year=, |date=, |publicationdate=, or |publication-date in that order. Does not error
check (that is left to the cs1|2 templates to do)
also gets date from |<date alias>={{date|...}}
]]
localfunctiondate_get(template,aliases)
localrvalue;
for_,patterninipairs(aliases)do-- spin through the date alias patterns
rvalue=tostring(template):match(pattern);-- is this |<date alias>= used (tostring() because something makes match() think template is a table)
ifrvaluethen
rvalue=tostring(template):match(pattern..'(%b{})');-- is rvalue a template?
ifrvaluethen
rvalue=rvalue:gsub('{{%s*(.-)%s*}}','%1');-- strip bounding template markup and trim
localparts=mw.text.split(rvalue,'%s*|%s*');-- split at the pipe and remove extraneous space characters
ifredirects_date[parts[1]:lower()]then-- if parts[1] names {{date}} or redirect
rvalue=parts[2];-- assume that date template is properly formed, first positional parameter is the date
else
return'';-- |date= holds some other template than {{date}} or redirect
end
else
rvalue=template:match(pattern..'([^|}]+)');
ifrvaluethen-- if rvalue is something
rvalue=mw.text.trim(rvalue);-- trim it
end
ifnotrvalueor''==rvaluethen-- if rvale was nothing or trimed to nothing
rvalue=nil;-- ensure that it is unset so we can try the next parameter in the list
end
end
ifrvaluethen
for_,patterninipairs(patterns_date)do-- spin through the recognized date formats
--					date = rvalue:match (pattern);								-- attempt to extract year portion according to the pattern
localdate,date2=rvalue:match(pattern);-- attempt to extract year portion according to the pattern; <date2> gets second year in any range
ifdatethen
ifdate2then-- when a second year
date=table.concat({date,'–',date2});-- build a date range
end
returndate;-- matched so return;
end
end
break;-- found a date but it was malformed so abandon
end
end
end
return'';-- no date param or date param doesn't hold a recognized date; empty string for concatenation
end
--[[--------------------------< V N A M E S _ G E T >----------------------------------------------------------
extract names from |vauthors= or |veditors=; there is no |vcontributors= parameter.
splits the v parameter value at the comma; correctly handles accept-as-witten markup when used to wrap a comma-
separated names (corporate)
]]
localfunctionvnames_get(params,vparam)
localvnames={};-- first four author or editor names go here
localsplit={};-- temp table to assist in decoding accept-as-witten-markup
ifparams[vparam]then-- test for |vauthors= or |veditor=
split=mw.text.split(params[vparam],'%s*,%s*');-- this will separate portions of ((Black, Brown, White, an Co.))
locali=1;-- an indexer
whilesplit[i]do
ifsplit[i]:match('^%(%(.*[^%)][^%)]$')then-- first segment of comma-separated accept-as-witten; this segment has the opening doubled parens
localname=split[i];
i=i+1;-- bump indexer to next segment
whilesplit[i]do
name=name..', '..split[i];-- concatenate with previous segments
ifsplit[i]:match('^.*%)%)$')then-- if this table member has the closing doubled parens
break;-- and done reassembling so
end
i=i+1;-- bump indexer
end
table.insert(vnames,name);-- and add accept-as-witten name to the vnames table
else
table.insert(vnames,split[i]);-- and add name to the vnames table
end
i=i+1;-- bump indexer
if5==ithenbreak;end-- limit to four names
end
fori,vnameinipairs(vnames)do
ifnotvname:match('%(%(.-%)%)')then-- without accept-this-value-as-written markup
vnames[i]=vname:gsub('(.-)%s+%u+$','%1');-- extract and save surname(s)
end
end
fori,vnameinipairs(vnames)do-- repeat, this time for accept-this-value-as-written markup
vnames[i]=vname:gsub('%(%((.-)%)%)','%1');-- remove markup if present and save the whole name
end
end
return0~=#vnamesandtable.concat(vnames)ornil-- return a concatenation of the vnames; nil else
end
--[[--------------------------< N A M E S _ G E T >------------------------------------------------------------
cs1|2 makes anchor id from contributor, author, or editor name-lists in that order
get the names from the cs1|2 template; if there are no contributor names, try author names, then try editor names.
returns concatenated names in enumeration order when successful; nil else
empty name (nameholding parameter n is present without value) and missing name (nameholding parameter n is not
present) are included as empty string with all other names
]]
localfunctionnames_get(params,aliases_list)
localnames={};-- first four author or editor names go here
localenum_alias;-- alias with '#' replaced with a digit
forenum=1,4do-- four names only
fori,aliasinipairs(aliases_list)do
ifnotnames[enum]then-- hanven't found a previous alias with this [enum]? see if we can find this alias with this enum
enum_alias=alias:gsub('#',enum);-- replace '#' to make 'lastn'
if1==enumthen-- because |last= and last1= are exact aliases
ifparams[enum_alias]then-- test |last1= first
names[enum]=params[enum_alias];-- found so save the value assigned to |last1=
break;-- next enum
else
enum_alias=alias:gsub('#','');-- replace '#' to make 'last'
ifparams[enum_alias]then
names[enum]=params[enum_alias];-- found so save the value assigned to |last=
break;-- next enum
end
end
else-- here for enum 2, 3, 4
ifparams[enum_alias]then
names[enum]=params[enum_alias];-- found so save the value assigned to |lastn=
break;-- next enum
end
end
end
end
end
forenum=1,4do-- spin through the names table and
localname=names[enum];
ifnotnamethen-- when nameholding parameter n is not present (nil)
name='';-- convert to empty string for concatenation
end
name=name:gsub('%(%((.-)%)%)','%1');-- remove accept-as-written markup if present
names[enum]=name;-- save the modified name
end
localname_str=table.concat(names);-- concatenate the names
return''~=name_strandname_strornil;-- return the concatenation if not empty string; nil else
end
--[[--------------------------< T E M P L A T E _ S T R I P >--------------------------------------------------
removes the citation or havrc template's {{ and }} markup then removes, in whole, any templates found inside the
citation or harvc template.
Templates are not allowed in parameters that are made part of COinS metadata; yet, they will appear. cs1|2 does
not see the template markup but instead sees the result of the template as html. cs1|2 strips the html which
leaves the displayed value for the anchor id. We can't do that here so, because templates aren't allowed in
parameters, we simply discard any templates found in the cs1|2 template.
this may leave a |lastn= parameter empty which will be treated as if it were really empty as cs1|2 do (three authors,
|last2= empty -> CITEREFLast1Last3YYYY (the harv and sfn render: 'Last1, & Last3 YYYY' with CITEREFLast1Last3YYYY).
]]
localfunctiontemplate_strip(template)
template=template:gsub('^{{%s*',''):gsub('%s*}}$','',1);-- remove outer {{ and }} (cs1|2 template delimiters with trailing/leading whitespace)
template=template:gsub('%b{}','');-- remove any templates from the cs1|2 template
returntemplate;
end
--[[--------------------------< E S C A P E _ L U A _ M A G I C _ C H A R S >----------------------------------
Returns a string where all of lua's magic characters have been escaped. This is important because functions like
string.gsub() treat their pattern and replace strings as patterns, not literal strings.
]]
localfunctionescape_lua_magic_chars(argument)
argument=argument:gsub("%%","%%%%");-- replace % with %%
argument=argument:gsub("([%^%$%(%)%.%[%]%*%+%-%?])","%%%1");-- replace all other lua magic pattern characters
returnargument;
end
--[=[-------------------------< W I K I L I N K _ S T R I P >--------------------------------------------------
Wikilink markup does not belong in an anchor id and can / does confuse the code that parses apart citation and
harvc templates so here we remove any wiki markup:
	[[link|label]] -> label
	[[link]] -> link
]=]
localfunctionwikilink_strip(template)
forwikilinkintemplate:gmatch('%[%b[]%]')do-- get a wikilink
template=template:gsub('%[%b[]%]','__57r1P__',1);-- install a marker
ifwikilink:match('%[%[.-|(.-)%]%]')then
wikilink=wikilink:match('%[%[.-|(.-)%]%]');-- extract label from complex [[link|label]] wikilink
else
wikilink=wikilink:match('%[%[(.-)%]%]');-- extract link from simple [[link]] wikilinks
end
wikilink=escape_lua_magic_chars(wikilink);-- in case there are lua magic characters in wikilink
template=template:gsub('__57r1P__',wikilink,1);-- replace the marker with the appropriate text
end
returntemplate;
end
--[[--------------------------< T E M P L A T E _ N A M E _ G E T >--------------------------------------------
return the citation or harvc template's name; convert to lower case and trim leading and trailing whitespace;
when the template is a sandbox the subpage portion of the template name is omitted from the returned template name
	{{Cite book/new |...}} returns cite book
]]
localfunctiontemplate_name_get(template)
localtemplate_name=template:match('^{{%s*([^/|}]+)');-- get template name; ignore subpages ~/new, ~/sandbox; parser functions
ifnottemplate_nameortemplate_name:match('^#')then-- parser functions, magic words don't count as templates
returnnil;-- could not get template name from (possibly corrupt) template; extraneous opening { mid template can cause this;
end;
template_name=template_name:gsub('%s*$','');-- trim trailing whitespace; leading whitespace already removed
returnLang_obj:ucfirst(template_name);-- first character in template name must be uppercase (same as canonical template name) TODO: better way to do this?
end
--[[--------------------------< T E M P L A T E _ P A R A M S _ G E T >----------------------------------------
parse apart a template's parameters and store in the params table where key is the parameter's name and value is
the parameter's value; empty parameters are not saved
]]
localfunctiontemplate_params_get(template,params_t)
template=wikilink_strip(template);-- because piped wikilinks confuse code that builds params_t{} and because wikilinks not allowed in an anchor id
-- strip templates after getting |ref= value because |ref={{sfnref}} and |ref={{harvid}} are allowed
template=template_strip(template);-- because template markup can confuse code that builds params_t{} and because templates in name parameters are not allowed
localtemp_t=mw.text.split(template,'%s*|%s*');--split on the pipe
for_,paraminipairs(temp_t)do
ifparam:find('=',1,true)then-- a named parameter?
localk,v=param:match('%s*([^=]-)%s*=%s*([^|}]+)');
ifvthen-- there must be a value
if''~=vandnotv:match('^%s$')then-- skip when value is empty string or only whitespace
params_t[k]=mw.text.trim(v);-- add trimmed value else
end
end
end
end
end
--[[--------------------------< C I T E R E F _ M A K E _ H A R V C >------------------------------------------
makes anchor_id from {{harvc}} or redirects
]]
localfunctionanchor_id_make_harvc(template)
localdate=date_get(template,alias_patterns_harvc_date);-- get date; done here because might be in {{date}}; return date if valid; empty string else
localanchor_id;
localparams={};-- table of harvc parameters
localid;-- custom anchor id for this {{harvc}} template
id=template:match('|%s*id%s*=%s*(%b{})');-- in case |id={{sfnref}}; done here because templates will be stripped
template_params_get(template,params);-- build a table of template parameters and their values; this strips wikilinks and templates
ifidthen-- when set is {{sfnref}} or {{harvid}} template
returnsfnref_get(id);-- returns content of {{sfnref}} or {{harvid}}; nil else
end
ifparams.idthen-- custom anchor for this {{harvc}} template (text)
returnparams.id;-- |id= value as written
end
anchor_id=names_get(params,aliases_harvc_author);-- get the harvc contributor names
ifanchor_idthen-- if names were gotten
return'CITEREF'..anchor_id..date;
end
returnnil;-- no names; no anchor_id
end
--[[--------------------------< A N C H O R _ I D _ M A K E _ W R A P P E R >----------------------------------
for wrapper templates
inspect externally visible |ref= to decide what to do:
	|ref=										- empty or missing: get names and date from whitelist defaults; override defaults from externally visible template parameters
	|ref=harv									- same as empty or missing
	|ref={{SfnRef|name|name|name|name|year}}	- assemble an anchor id from {{sfnref}} positional parameters
	|ref={{Harvid|name|name|name|name|year}}	- assemble an anchor id from {{harvid}} positional parameters
	|ref=none									- skip; do nothing because an anchor id intentionally suppressed; TODO: keep with a type code of '0'?
	|ref=<text>									- save param value because may match an anchor id override value in {{harv}} template |ref= parameter or {{harvc}} |id= parameter
]]
localfunctionanchor_id_make_wrapper(template)
localref;-- content of |ref=
localtemplate_name;-- name of the template
localanchor_id;-- the assembled anchor id from this template
localdate;
localname_default;
localdate_default;
localvol;
localparams={};-- table of template parameters
template_name=template_name_get(template);-- get first char uppercase trimmed template name; ignore subpages ~/new, ~/sandbox
ifnottemplate_nameortemplate_skip[template_name]then
returnnil;-- could not extract template name from (possibly corrupted) template (extraneous opening { in the template will cause this)
end
date=date_get(template,alias_patterns_date);-- get date; done here because might be in {{date}}
--	if '' == date then
--		date = whitelist.wrapper_templates[template_name][2] or '';				-- no externally visible date so get default date
--	end
ref=template:match('|%s*ref%s*=%s*(%b{})');-- first look for |ref={{sfnref}} or |ref={{harvid}} because we will strip templates from the wrapper template
ifnotrefthen
iftemplate:match('|%s*ref%s*=([^|}]+)')then-- |ref={{template}} not found; if there is a |ref= param with an assigned value
ref=template:match('|%s*ref%s*=([^|}]+)');-- get the value; whitespace is a 'value'
ifrefthen-- nil when |ref=|... or when |ref=}} (no spaces between assignment operator and pipe or closing brace)
ref=mw.text.trim(ref);-- something, could be just whitespace, so trim leading / trailing whitespace
if''==refthen-- trimming a string of whitespace makes an empty string
ref=nil;-- make empty ref same as missing ref
end
end
end
end
template_params_get(template,params);-- build a table of template parameters and their values
localwrap_data=whitelist.wrapper_templates[template_name]
ifwrap_data[1]then-- is this wrapper a simple-default wrapper?
name_default=wrap_data[1];-- get the default names
date_default=wrap_data[2];-- get the default date
else
vol=params['volume']or'default';
localfascicle=params['fascicle']-- some templates use "fascicle" to mean "subvolume"
iffasciclethen
localsubvol=vol..'/'..fascicle-- if fascicle is used, subvolume = "vol/fascicle"
ifwrap_data[subvol]then-- if subvolume exists, use it, otherwise fall back to volume
vol=subvol
end
end
ifnotwrap_data[vol]then-- make sure this volume exists
vol='default';-- doesn't exist, use default volume
end
name_default=wrap_data[vol][1];-- get the default names
date_default=wrap_data[vol][2];-- get the default date
end
if'harv'==refornotrefthen-- |ref=harv specified or |ref= missing or empty
anchor_id=names_get(params,aliases_contributor)or-- get contributor, author, or editor names
names_get(params,aliases_author)or
vnames_get(params,'vauthors')or-- |vauthors=
names_get(params,aliases_editor)or
vnames_get(params,'veditors')or-- |veditors=
name_default;-- default names from whitelist
--			whitelist.wrapper_templates[template_name][1];						-- default names from whitelist
if''==datethen-- if date not provided in the template
date=date_default;-- use the default date from whitelist
end
ifanchor_idthen-- if names were gotten
anchor_id='CITEREF'..anchor_id..date;
end
elseifref:match('%b{}')then-- ref holds a template
anchor_id=sfnref_get(ref);-- returns content of {{sfnref}} or {{harvid}}; nil else
elseif'none'==refthen-- |ref=none
returnnil;-- anchor id expicitly suppressed
else
anchor_id=ref;-- |ref=<text> may match an anchor id override value in {{harv}} template |ref= parameter
end
returnanchor_id;-- anchor_id text; nil else
end
--[[--------------------------< A N C H O R _ I D _ M A K E _ C S 1 2 >----------------------------------------
for cs1|2 template and cs1-like templates
inspect |ref= to decide what to do:
	|ref=										- empty or missing: get names and date from template parameters; all cs1|2 create CITEREF anchor IDs
	|ref=harv									- get names and date from template parameters
	|ref={{SfnRef|name|name|name|name|year}}	- assemble an anchor id from {{sfnref}} positional parameters
	|ref={{Harvid|name|name|name|name|year}}	- assemble an anchor id from {{harvid}} positional parameters
	|ref=none									- skip; do nothing because an anchor id intentionally suppressed; TODO: keep with a type code of '0'?
	|ref=<text>									- save param value because may match an anchor id override value in {{harv}} template |ref= parameter or {{harvc}} |id= parameter
]]
localfunctionanchor_id_make_cs12(template)
localref;-- content of |ref=
localtemplate_name;-- name of the template
localanchor_id;-- the assembled anchor id from this template
localdate;
localparams={};-- table of template parameters
template_name=template_name_get(template);-- get first char uppercase trimmed template name; ignore subpages ~/new, ~/sandbox
ifnottemplate_nameortemplate_skip[template_name]then
returnnil;-- could not extract template name from (possibly corrupted) template (extraneous opening { in the template will cause this)
end
ifredirects_patent[template_name]then
date=date_get(template,alias_patterns_patent_date);-- get date; done here because might be in {{date}} 
else
date=date_get(template,alias_patterns_date);
end
ref=template:match('|%s*ref%s*=%s*(%b{})');-- first look for |ref={{sfnref}} or |ref={{harvid}} because we will strip templates from the cs1|2 template
ifnotrefthen
iftemplate:match('|%s*ref%s*=([^|}]+)')then-- |ref={{template}} not found; if there is a |ref= param with an assigned value
ref=template:match('|%s*ref%s*=([^|}]+)');-- get the value; whitespace is a 'value'
ifrefthen-- nil when |ref=|... or when |ref=}} (no spaces between assignment operator and pipe or closing brace)
ref=mw.text.trim(ref);-- something, could be just whitespace, so trim leading / trailing whitespace
if''==refthen-- trimming a string of whitespace makes an empty string
ref=nil;-- make empty ref same as missing ref
end
end
end
end
template_params_get(template,params);-- build a table of template parameters and their values
if'harv'==refornotrefthen-- |ref=harv specified or |ref= missing or empty
ifredirects_patent[template_name]then-- if this is a cite patent template
anchor_id=names_get(params,aliases_inventor);-- inventor names only
else-- cs1|2 template
anchor_id=names_get(params,aliases_contributor)or-- get contributor, author, or editor names
names_get(params,aliases_author)or
vnames_get(params,'vauthors')or-- |vauthors=
names_get(params,aliases_editor)or
vnames_get(params,'veditors');-- |veditors=
end
ifanchor_idthen-- if names were gotten
anchor_id='CITEREF'..anchor_id..date;
end
elseifref:match('%b{}')then-- ref holds a template
anchor_id=sfnref_get(ref);-- returns content of {{sfnref}} or {{harvid}}; nil else
elseif'none'==refandnotredirects_patent[template_name]then-- |ref=none; not supported by cite patent
returnnil;-- anchor id expicitly suppressed
else
anchor_id=ref;-- |ref=<text> may match an anchor id override value in {{harv}} template |ref= parameter
end
returnanchor_id;-- anchor_id text; nil else
end
--[[--------------------------< L I S T _ A D D >--------------------------------------------------------------
adds an <item> to <list> table; for anchor IDs, the boolean <encode> argument must be set true; no return value
]]
localfunctionlist_add(item,list,encode)
ifitemthen-- if there was an item
ifencodethen-- for anchor IDs ...
item=mw.uri.anchorEncode(item);-- encode to remove wikimarkup, convert spaces to underscores etc
end
ifnotlist[item]then-- if not already saved
list[item]=1;-- save it 
else-- here when this item already saved
list[item]=list[item]+1;-- to indicate that there are multiple items
end
end
end
--[[--------------------------< A N C H O R _ I D _ M A K E _ A N C H O R >------------------------------------
make anchor IDs from {{anchor}}; there may be more than one because {{anchor}} is not limited to the number of
anchors it may hold.
]]
localfunctionanchor_id_make_anchor(template,anchor_id_list)
template=template:gsub('^{{[^|]+|',''):gsub('}}$','',1);-- remove outer {{ and }} and template name
template=wikilink_strip(template);-- strip any wikilink markup (there shouldn't be any but just in case)
localparams={};
localanchor_id;
forparamintemplate:gmatch('%b{}')do-- loop through the template; remove and save templates (presumed to be sfnref or harvid)
table.insert(params,param);-- save it
template=template:gsub('%b{}','',1);-- remove it from source template
end
for_,tinipairs(params)do-- spin through the templates in params
anchor_id=sfnref_get(t);-- attempt to decode {{sfnref}} and {{harvid}}
ifanchor_idthen-- nil when not {{sfnref}} or {{harvid}}
list_add(anchor_id,anchor_id_list,true);-- add anchor ID to the list
end
end
template=template:gsub('|%s*|','|');-- when pipe follows pipe with or without white space, remove extraneous pipe
template=template:gsub('^|',''):gsub('|$','');-- remove extraneous leading and trailing pipes
params=mw.text.split(template,'%s*|%s*');-- split at the pipe and remove extraneous space characters
for_,tinipairs(params)do-- spin through the anchor IDs
anchor_id=mw.text.trim(t);-- trim white space
if''~=anchor_idthen-- should always have something
list_add(anchor_id,anchor_id_list,true);-- add anchor ID to the list
end
end
end
--[[--------------------------< T E M P L A T E _ L I S T _ A D D >--------------------------------------------
makes a list of templates use in the article.
]]
localfunctiontemplate_list_add(template,template_list)
localtemplate=template:match('{{%s*(.-)[|}]');-- keep the case of the template - this is different from template_name_get()
iftemplateandnottemplate:match('^#')then-- found a template or magic word; ignore magic words
template=mw.text.trim(template);-- trim whitespace
template=Lang_obj:ucfirst(template);-- first character in template name must be uppercase (same as canonical template name) TODO: better way to do this?
list_add(template,template_list);-- add to list with (unused) tally
end
end
--[[--------------------------< A N C H O R _ I D _ L I S T _ M A K E >----------------------------------------
makes a list of anchor ids from cs1|2, cs1|2-like, vcite xxx, harvc, anchor, wikicite templates
Because cs1|2 wrapper templates can, and often do, hide the author and date parameters inside the wrapper,
these parameters are not available in the article's wikisource so {{harv}}, {{sfn}}, and {{harvc}} templates that
link correctly to those wrapper templates will incorrectly show error messages. Use |ignore-err=yes in the {{harv}},
{{sfn}}, and {{harvc}} templates to supress the error message.
creates a list of templates used in the article for use with the whitelist
creates a list of article-local whitelisted anchor IDs from {{sfn whitelist}}
]]
localfunctionanchor_id_list_make()
localanchor_id_list={}
localtemplate_list={}
localarticle_whitelist={}
localarticle_content=article_content_get();-- attempt to get this article's content
ifarticle_content==''then-- when there is no article content
return'';-- no point in continuing
end
localtemplate;-- place to hold the template that we found
localtemplate_name;
localanchor_id;-- place to hold an anchor id as it is extracted / decoded
localfind_pattern='%f[{]{{[^{]';
localtstart,tend=article_content:find(find_pattern);-- find the first template; do not find template variables: {{{template var|}}} 
whiletstartdo
template=article_content:match('%b{}',tstart);-- get the whole template
ifnottemplatethen
break;-- template is nil for some reason (last template missing closing }} for example) so declare ourselves done
end
template_name=template_name_get(template);-- get first char uppercase trimmed template name; ignore subpages ~/new, ~/sandbox
template_list_add(template,template_list);-- add this template's name to the list
ifdata.known_templates_cs12[template_name]then
anchor_id=anchor_id_make_cs12(template);-- extract an anchor id from this template
list_add(anchor_id,anchor_id_list,true)
elseifdata.known_templates_vcite[template_name]then
localref=template:match('|%s*ref%s*=%s*(%b{})');-- first look for |ref={{sfnref}} or |ref={{harvid}} because we will strip templates from the vcite template
ifrefthen-- |ref={{template}}
anchor_id=sfnref_get(ref);-- returns content of {{sfnref}} or {{harvid}}; nil else
list_add(anchor_id,anchor_id_list,true);
else
localparams={};
template_params_get(template,params);-- build a table of template parameters and their values
anchor_id=params['ref'];-- when both set, vcite uses value from |ref=
ifnotanchor_idandparams['harvid']then
anchor_id='CITEREF'..params['harvid'];-- in vcite, |harvid= auto-adds 'CITEREF' prefix to the value in |harvid=
end
list_add(anchor_id,anchor_id_list,true);
end
elseifdata.known_templates_harvc[template_name]then
anchor_id=anchor_id_make_harvc(template);-- extract an anchor id from this template
list_add(anchor_id,anchor_id_list,true);
elseifdata.known_templates_wikicite[template_name]then
localref=template:match('|%s*ref%s*=%s*(%b{})');-- first look for |ref={{sfnref}} or |ref={{harvid}}
ifrefthen
anchor_id=sfnref_get(ref);
elseiftemplate:match('|%s*ref%s*=([^|}]+)')then
anchor_id=template:match('|%s*ref%s*=([^|}]+)');-- plain-text
elseiftemplate:match('|%s*id%s*=%s*(%b{})')then
ref=template:match('|%s*id%s*=%s*(%b{})');
anchor_id='Reference-'..sfnref_get(ref);
elseiftemplate:match('|%s*id%s*=([^|}]+)')then
anchor_id='Reference-'..template:match('|%s*id%s*=([^|}]+)');-- plain-text
else
anchor_id=nil;-- no matches, ensure that anchor_id has no value
end
ifanchor_idthen
list_add(anchor_id,anchor_id_list,true);
end
elseifdata.known_templates_anchor[template_name]then
anchor_id_make_anchor(template,anchor_id_list);-- extract anchor ids from this template if any
elseifdata.known_templates_sfn_whitelist[template_name]then
template=template:gsub('^{{[^|]+|',''):gsub('}}$','',1);-- remove outer {{ and }} and template name
template=mw.text.trim(template,'%s|');-- trim leading trailing white space and pipes
template=mw.text.split(template,'%s*|%s*');-- make a table of the template's parameters
for_,anchor_idinipairs(template)do-- spin through this template's parameter
if''~=anchor_idandnotarticle_whitelist[anchor_id]then
anchor_id=mw.uri.anchorEncode(anchor_id)
article_whitelist[anchor_id]=1;-- add to the whitelist
end
end
elseiftemplate_nameandwhitelist.wrapper_templates[template_name]then
anchor_id=anchor_id_make_wrapper(template);-- extract an anchor id from this template if possible
list_add(anchor_id,anchor_id_list,true);
elseiftemplate_nameandtemplate_name:match('^Cit[ea]')then-- not known, not known wrapper; last gasp, try as cs1-like
anchor_id=anchor_id_make_cs12(template);-- extract an anchor id from this template if possible
list_add(anchor_id,anchor_id_list,true);
end
tstart,tend=article_content:find(find_pattern,tend);-- search for another template; begin at end of last search
end
mw.logObject(anchor_id_list,'anchor_id_list');
mw.logObject(template_list,'template_list');
mw.logObject(article_whitelist,'article_whitelist');
global_anchor_id_list=anchor_id_list
global_template_list=template_list
global_article_whitelist=article_whitelist
end
--[[--------------------------< C I T E R E F _ P A T T E R N S _ M A K E >--------------------------------------------
Scans template_list to look for wrapper templates that generate citerefs that require Lua patterns.
This scan is only done once per page load, to save time
]]
localfunctionciteref_patterns_make()
ifnotglobal_template_listthenreturnend
localciteref_patterns={}
localtemplate_patterns=whitelist.wrapper_template_patterns
for_,pinipairs(template_patterns)do
for_,tinipairs(p[1])do-- loop through list of template wrappers
ifglobal_template_list[t]then-- if wrapper is found in article, record corresponding patterns
for_,patinipairs(p[2])do
table.insert(citeref_patterns,pat)
end
break
end
end
end
mw.logObject(citeref_patterns,'citeref_patterns')
returnciteref_patterns
end
--[[--------------------------< E X P O R T E D _ T A B L E S >------------------------------------------------
]]
-- First create global_anchor_id_list, global_template_list, global_article_whitelist
anchor_id_list_make()
-- Then stuff them (and derived tables) into return table
return{
anchor_id_list=global_anchor_id_listor{},-- table of anchor ids available in this article
article_whitelist=global_article_whitelistor{},-- table of anchor ids with false-positive error message to be suppressed
template_list=global_template_listor{},-- table of templates used in this article
citeref_patterns=citeref_patterns_make()or{}-- table of Lua patterns to search for citeref from wrappers
}