Editing Module:Citation/CS1/COinS
The edit can be undone. Please check the comparison below to verify that this is what you want to do, and then publish the changes below to finish undoing the edit.
Latest revision | Your text | ||
Line 1: | Line 1: | ||
--[[ | |||
History of changes since last sync: 2015-12-12 | |||
2016-01-10: bug fix; article title (aka chapter) missing from cite conference book metadata; | |||
]] | |||
local coins = {}; | |||
--[[--------------------------< F O R W A R D D E C L A R A T I O N S >-------------------------------------- | --[[--------------------------< F O R W A R D D E C L A R A T I O N S >-------------------------------------- | ||
]] | ]] | ||
local is_set, in_array, remove_wiki_link | local is_set, in_array, remove_wiki_link; -- functions in Module:Citation/CS1/Utilities | ||
local cfg; -- table of configuration tables that are defined in Module:Citation/CS1/Configuration | local cfg; -- table of configuration tables that are defined in Module:Citation/CS1/Configuration | ||
--[[--------------------------< S T R I P _ A P O S T R O P H E _ M A R K U P >-------------------------------- | |||
Strip wiki italic and bold markup from argument so that it doesn't contaminate COinS metadata. | |||
This function strips common patterns of apostrophe markup. We presume that editors who have taken the time to | |||
markup a title have, as a result, provided valid markup. When they don't, some single apostrophes are left behind. | |||
]] | |||
local function strip_apostrophe_markup (argument) | |||
if not is_set (argument) then return argument; end | |||
if argument:find ( "''", 1, true ) == nil then -- Is there at least one double apostrophe? If not, exit. | |||
return argument; | |||
end | |||
while true do | |||
if argument:find ( "'''''", 1, true ) then -- bold italic (5) | |||
argument=argument:gsub("%'%'%'%'%'", ""); -- remove all instances of it | |||
elseif argument:find ( "''''", 1, true ) then -- italic start and end without content (4) | |||
argument=argument:gsub("%'%'%'%'", ""); | |||
elseif argument:find ( "'''", 1, true ) then -- bold (3) | |||
argument=argument:gsub("%'%'%'", ""); | |||
elseif argument:find ( "''", 1, true ) then -- italic (2) | |||
argument=argument:gsub("%'%'", ""); | |||
else | |||
break; | |||
end | |||
end | |||
return argument; -- done | |||
end | |||
Line 21: | Line 62: | ||
title = strip_apostrophe_markup (title); -- strip any apostrophe markup | title = strip_apostrophe_markup (title); -- strip any apostrophe markup | ||
else | else | ||
title = ''; -- if not set, make sure title is an empty string | title=''; -- if not set, make sure title is an empty string | ||
end | end | ||
if is_set (script) then | if is_set (script) then | ||
Line 27: | Line 68: | ||
script = strip_apostrophe_markup (script); -- strip any apostrophe markup | script = strip_apostrophe_markup (script); -- strip any apostrophe markup | ||
else | else | ||
script = ''; | script=''; -- if not set, make sure script is an empty string | ||
end | end | ||
if is_set (title) and is_set (script) then | if is_set (title) and is_set (script) then | ||
Line 38: | Line 79: | ||
--[[--------------------------< E S C A P E _ L U A _ M A G I C _ C H A R S >---------------------------------- | --[[--------------------------< 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 | 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. | string.gsub() treat their pattern and replace strings as patterns, not literal strings. | ||
]] | ]] | ||
Line 44: | Line 85: | ||
local function escape_lua_magic_chars (argument) | local function escape_lua_magic_chars (argument) | ||
argument = argument:gsub("%%", "%%%%"); -- replace % with %% | argument = argument:gsub("%%", "%%%%"); -- replace % with %% | ||
argument = argument:gsub("([%^%$%(%)%.%[%]%*%+%-%?])", "%%%1"); -- replace all other | argument = argument:gsub("([%^%$%(%)%.%[%]%*%+%-%?])", "%%%1"); -- replace all other lua magic pattern characters | ||
return argument; | return argument; | ||
end | end | ||
Line 60: | Line 101: | ||
while true do | while true do | ||
pattern = pages:match("%[(%w*:?//[^ ]+%s+)[%w%d].*%]"); -- pattern is the opening bracket, the | pattern = pages:match("%[(%w*:?//[^ ]+%s+)[%w%d].*%]"); -- pattern is the opening bracket, the url and following space(s): "[url " | ||
if nil == pattern then break; end -- no more | if nil == pattern then break; end -- no more urls | ||
pattern = escape_lua_magic_chars (pattern); -- pattern is not a literal string; escape | pattern = escape_lua_magic_chars (pattern); -- pattern is not a literal string; escape lua's magic pattern characters | ||
pages = pages:gsub(pattern, ""); -- remove as many instances of pattern as possible | pages = pages:gsub(pattern, ""); -- remove as many instances of pattern as possible | ||
end | end | ||
pages = pages:gsub("[%[%]]", ""); -- remove the brackets | pages = pages:gsub("[%[%]]", ""); -- remove the brackets | ||
pages = pages:gsub("–", "-" ); | pages = pages:gsub("–", "-" ); -- replace endashes with hyphens | ||
pages = pages:gsub("&%w+;", "-" ); | pages = pages:gsub("&%w+;", "-" ); -- and replace html entities (– etc.) with hyphens; do we need to replace numerical entities like   and the like? | ||
return pages; | return pages; | ||
end | end | ||
Line 80: | Line 121: | ||
MathML with SVG or PNG fallback | MathML with SVG or PNG fallback | ||
All three are heavy with | All three are heavy with html and css which doesn't belong in the metadata. | ||
Without this function, the metadata saved in the raw wikitext contained the rendering determined by the settings | Without this function, the metadata saved in the raw wikitext contained the rendering determined by the settings | ||
Line 87: | Line 128: | ||
This function gets the rendered form of an equation according to the editor's preference before the page is saved. It | This function gets the rendered form of an equation according to the editor's preference before the page is saved. It | ||
then searches the rendering for the text equivalent of the rendered equation and replaces the rendering with that so | then searches the rendering for the text equivalent of the rendered equation and replaces the rendering with that so | ||
that the page is saved without extraneous | that the page is saved without extraneous html/css markup and with a reasonably readable text form of the equation. | ||
When a replacement is made, this function returns true and the value with replacement; otherwise false and the | When a replacement is made, this function returns true and the value with replacement; otherwise false and the intital | ||
value. To replace multipe equations it is | value. To replace multipe equations it is necesary to call this function from within a loop. | ||
]=] | ]=] | ||
local function coins_replace_math_stripmarker (value) | local function coins_replace_math_stripmarker (value) | ||
local stripmarker = | local stripmarker = '\127UNIQ%-%-math%-[%a%d]+%-QINU\127'; -- math stripmarker pattern | ||
local rendering = value:match (stripmarker); -- is there a math stripmarker | local rendering = value:match (stripmarker); -- is there a math stripmarker | ||
Line 120: | Line 161: | ||
--[[--------------------------< C O I N S _ C L E A N U P >---------------------------------------------------- | --[[--------------------------< C O I N S _ C L E A N U P >---------------------------------------------------- | ||
Cleanup parameter values for the metadata by removing or replacing invisible characters and certain | Cleanup parameter values for the metadata by removing or replacing invisible characters and certain html entities. | ||
2015-12-10: there is a bug in mw.text.unstripNoWiki (). It | 2015-12-10: there is a bug in mw.text.unstripNoWiki (). It replaced math stripmarkers with the appropriate content | ||
when it shouldn't. See https://phabricator.wikimedia.org/T121085 and Wikipedia_talk:Lua#stripmarkers_and_mw.text.unstripNoWiki.28.29 | when it shouldn't. See https://phabricator.wikimedia.org/T121085 and Wikipedia_talk:Lua#stripmarkers_and_mw.text.unstripNoWiki.28.29 | ||
Line 132: | Line 173: | ||
local function coins_cleanup (value) | local function coins_cleanup (value) | ||
local replaced = true; -- default state to get the do loop running | local replaced = true; -- default state to get the do loop running | ||
while replaced do -- loop until all math stripmarkers replaced | while replaced do -- loop until all math stripmarkers replaced | ||
replaced, value = coins_replace_math_stripmarker (value); -- replace math stripmarker with text representation of the equation | replaced, value = coins_replace_math_stripmarker (value); -- replace math stripmarker with text representation of the equation | ||
end | end | ||
value = value:gsub ( | value = value:gsub ('\127UNIQ%-%-math%-[%a%d]+%-QINU\127', "MATH RENDER ERROR"); -- one or more couldn't be replaced; insert vague error message | ||
value = mw.text.unstripNoWiki (value); -- replace nowiki stripmarkers with their content | value = mw.text.unstripNoWiki (value); -- replace nowiki stripmarkers with their content | ||
value = value:gsub ('<span class="nowrap" style="padding%-left:0%.1em;">' | value = value:gsub ('<span class="nowrap" style="padding%-left:0%.1em;">'s</span>', "'s"); -- replace {{'s}} template with simple apostrophe-s | ||
value = value:gsub ('‍\226\128\138\039\226\128\139', "'"); -- replace {{'}} with simple apostrophe | |||
value = value:gsub ('\226\128\138\039\226\128\139', "'"); -- replace {{'}} with simple apostrophe (as of 2015-12-11) | |||
value = value:gsub (' ', ' '); -- replace entity with plain space | value = value:gsub (' ', ' '); -- replace entity with plain space | ||
value = value:gsub ('\226\128\138', ' '); -- replace hair space with plain space | value = value:gsub ('\226\128\138', ' '); -- replace hair space with plain space | ||
value = value:gsub ('‍', ''); -- remove ‍ entities | |||
value = value:gsub ('[\226\128\141\226\128\139]', '') -- remove zero-width joiner, zero-width space | |||
value = value:gsub ('[\194\173\009\010\013]', ' '); -- replace soft hyphen, horizontal tab, line feed, carriage return with plain space | |||
value = value:gsub ('[\009\010\013 ] | |||
return value; | return value; | ||
end | end | ||
Line 180: | Line 221: | ||
}); | }); | ||
if in_array (class, {'arxiv | if in_array (class, {'arxiv', 'journal', 'news'}) or (in_array (class, {'conference', 'interview', 'map', 'press release', 'web'}) and is_set(data.Periodical)) or | ||
('citation' == class and is_set(data.Periodical) and not is_set (data.Encyclopedia)) then | ('citation' == class and is_set(data.Periodical) and not is_set (data.Encyclopedia)) then | ||
OCinSoutput.rft_val_fmt = "info:ofi/fmt:kev:mtx:journal"; -- journal metadata identifier | OCinSoutput.rft_val_fmt = "info:ofi/fmt:kev:mtx:journal"; -- journal metadata identifier | ||
if | if 'arxiv' == class then -- set genre according to the type of citation template we are rendering | ||
OCinSoutput["rft.genre"] = "preprint"; -- cite arxiv | OCinSoutput["rft.genre"] = "preprint"; -- cite arxiv | ||
elseif 'conference' == class then | elseif 'conference' == class then | ||
OCinSoutput["rft.genre"] = "conference"; -- cite conference (when Periodical set) | OCinSoutput["rft.genre"] = "conference"; -- cite conference (when Periodical set) | ||
Line 197: | Line 238: | ||
-- these used only for periodicals | -- these used only for periodicals | ||
OCinSoutput["rft.ssn"] = data.Season; -- keywords: winter, spring, summer, fall | OCinSoutput["rft.ssn"] = data.Season; -- keywords: winter, spring, summer, fall | ||
OCinSoutput["rft.chron"] = data.Chron; -- free-form date components | OCinSoutput["rft.chron"] = data.Chron; -- free-form date components | ||
OCinSoutput["rft.volume"] = data.Volume; -- does not apply to books | OCinSoutput["rft.volume"] = data.Volume; -- does not apply to books | ||
Line 221: | Line 261: | ||
end | end | ||
end | end | ||
else -- {'audio-visual', 'AV-media-notes', 'DVD-notes', 'episode', 'interview', 'mailinglist', 'map', 'newsgroup', 'podcast', 'press release', 'serial', 'sign', 'speech', 'web'} | else --{'audio-visual', 'AV-media-notes', 'DVD-notes', 'episode', 'interview', 'mailinglist', 'map', 'newsgroup', 'podcast', 'press release', 'serial', 'sign', 'speech', 'web'} | ||
OCinSoutput["rft.genre"] = "unknown"; | OCinSoutput["rft.genre"] = "unknown"; | ||
end | end | ||
Line 237: | Line 277: | ||
OCinSoutput['rft.inst'] = data.PublisherName; -- book and dissertation | OCinSoutput['rft.inst'] = data.PublisherName; -- book and dissertation | ||
end | end | ||
-- and now common parameters (as much as possible) | -- and now common parameters (as much as possible) | ||
OCinSoutput["rft.date"] = data.Date; -- book, journal, dissertation | OCinSoutput["rft.date"] = data.Date; -- book, journal, dissertation | ||
for k, v in pairs( data.ID_list ) do -- what to do about these? For now assume that they are common to all? | for k, v in pairs( data.ID_list ) do -- what to do about these? For now assume that they are common to all? | ||
-- if k == 'ISBN' then v = clean_isbn( v ) end | |||
if k == 'ISBN' then v = v:gsub( "[^-0-9X]", "" ); end | if k == 'ISBN' then v = v:gsub( "[^-0-9X]", "" ); end | ||
local id = cfg.id_handlers[k].COinS; | local id = cfg.id_handlers[k].COinS; | ||
if string.sub( id or "", 1, 4 ) == 'info' then -- for ids that are in the info:registry | if string.sub( id or "", 1, 4 ) == 'info' then -- for ids that are in the info:registry | ||
OCinSoutput["rft_id"] = table.concat{ id, "/", v }; | OCinSoutput["rft_id"] = table.concat{ id, "/", v }; | ||
elseif string.sub (id or "", 1, 3 ) == 'rft' then -- for isbn, issn, eissn, etc | elseif string.sub (id or "", 1, 3 ) == 'rft' then -- for isbn, issn, eissn, etc that have defined COinS keywords | ||
OCinSoutput[ id ] = v; | OCinSoutput[ id ] = v; | ||
elseif id then -- when cfg.id_handlers[k].COinS is not nil | |||
OCinSoutput["rft_id"] = table.concat{ cfg.id_handlers[k].prefix, v }; -- others; provide a url | |||
elseif id then -- when cfg.id_handlers[k].COinS is not nil | |||
OCinSoutput["rft_id"] = table.concat{ cfg.id_handlers[k].prefix, v | |||
end | end | ||
end | end | ||
--[[ | |||
for k, v in pairs( data.ID_list ) do -- what to do about these? For now assume that they are common to all? | |||
local id, value = cfg.id_handlers[k].COinS; | |||
if k == 'ISBN' then value = clean_isbn( v ); else value = v; end | |||
if string.sub( id or "", 1, 4 ) == 'info' then | |||
OCinSoutput["rft_id"] = table.concat{ id, "/", v }; | |||
else | |||
OCinSoutput[ id ] = value; | |||
end | |||
end | |||
]] | |||
local last, first; | local last, first; | ||
for k, v in ipairs( data.Authors ) do | for k, v in ipairs( data.Authors ) do | ||
last, first = coins_cleanup (v.last), coins_cleanup (v.first or ''); -- replace any nowiki | last, first = coins_cleanup (v.last), coins_cleanup (v.first or ''); -- replace any nowiki strip markers, non-printing or invisible characers | ||
if k == 1 then -- for the first author name only | if k == 1 then -- for the first author name only | ||
if is_set(last) and is_set(first) then -- set these COinS values if |first= and |last= specify the first author name | if is_set(last) and is_set(first) then -- set these COinS values if |first= and |last= specify the first author name | ||
OCinSoutput["rft.aulast"] = last; -- book, journal, dissertation | OCinSoutput["rft.aulast"] = last; -- book, journal, dissertation | ||
OCinSoutput["rft.aufirst"] = first; -- book, journal, dissertation | OCinSoutput["rft.aufirst"] = first; -- book, journal, dissertation | ||
Line 271: | Line 320: | ||
OCinSoutput["rft.au"] = last; -- book, journal, dissertation | OCinSoutput["rft.au"] = last; -- book, journal, dissertation | ||
end | end | ||
end | end | ||
end | end | ||
Line 277: | Line 325: | ||
OCinSoutput.rft_id = data.URL; | OCinSoutput.rft_id = data.URL; | ||
OCinSoutput.rfr_id = table.concat{ "info:sid/", mw.site.server:match( "[^/]*$" ), ":", data.RawPage }; | OCinSoutput.rfr_id = table.concat{ "info:sid/", mw.site.server:match( "[^/]*$" ), ":", data.RawPage }; | ||
OCinSoutput = setmetatable( OCinSoutput, nil ); | |||
-- sort with version string always first, and combine. | -- sort with version string always first, and combine. | ||
table.sort( OCinSoutput ); | |||
table.insert( OCinSoutput, 1, "ctx_ver=" .. ctx_ver ); -- such as "Z39.88-2004" | table.insert( OCinSoutput, 1, "ctx_ver=" .. ctx_ver ); -- such as "Z39.88-2004" | ||
return table.concat(OCinSoutput, "&"); | return table.concat(OCinSoutput, "&"); | ||
end | end | ||
Line 302: | Line 343: | ||
cfg = cfg_table_ptr; | cfg = cfg_table_ptr; | ||
is_set = utilities_page_ptr.is_set; -- import functions from | is_set = utilities_page_ptr.is_set; -- import functions from select Module:Citation/CS1/Utilities module | ||
in_array = utilities_page_ptr.in_array; | in_array = utilities_page_ptr.in_array; | ||
remove_wiki_link = utilities_page_ptr.remove_wiki_link; | remove_wiki_link = utilities_page_ptr.remove_wiki_link; | ||
end | end | ||
return { | return { |