Module:USN fleet totals/data

require('strict');

local patterns_tags = { ' .- ',	'<!%-%-.-%-%->',	' .- ',	' ',	' ',													-- deprecated alias of syntaxhighlight tag }

local Article_content; local grand_total = 0; local planned_total = 0;

local wikitables_t = { ['Commissioned_t'] = {},													-- k/v table where k is ship type and v is the number of that type ['Non-commissioned_t'] = {}, ['Support_t'] = {}, ['Ready Reserve Force ships_t'] = {}, ['Reserve fleet_t'] = {}, ['Under construction_t'] = {}, ['On order_t'] = {}, ['Retirements_t'] = {},														-- this one is filled separately from the others; }

local wikitable_names_t = {'Commissioned_t', 'Non-commissioned_t', 'Support_t', 'Ready Reserve Force ships_t', 'Reserve fleet_t', 'Under construction_t', 'On order_t', 'Retirements_t'}; local fleet_wikitables_t = {'Commissioned_t', 'Non-commissioned_t', 'Support_t', 'Ready Reserve Force ships_t', 'Reserve fleet_t'};	-- only these for n_of_m local headers_t = {																-- headers for the fleet totals section 'Commissioned (USS)', 'Non-commissioned (USNS)', 'Support (MV, RV – \'\' or no prefix \'\')', 'Ready Reserve Force ships (MV, SS, GTS)', 'Reserve Fleet ships (USS, USNS)', 'Under construction', 'On order', 'Expected to retire', }

local totals_t = {																-- table to hold total number of ships in these categories ['Commissioned_t'] = 0, ['Non-commissioned_t'] = 0, ['Support_t'] = 0, ['Ready Reserve Force ships_t'] = 0, ['Reserve fleet_t'] = 0, ['Under construction_t'] = 0, ['On order_t'] = 0, ['Retirements_t'] = 0, }

local n_of_m_t = {}																-- table of total counts (m in 'n of m') local retirements_year_max = 0;

--[[--< 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.

]]

local function article_content_get if not Article_content then Article_content = mw.title.new ('List of current ships of the United States Navy'):getContent; for _, tag in ipairs (patterns_tags) do			Article_content = Article_content:gsub (tag, '');					-- remove certain html-like tags and their content end while Article_content:match ('([\r\n]+|%-) *[\r\n]+|%-') do				-- are there multiple row markers without intervening column data? Article_content = Article_content:gsub ('([\r\n]+|%-) *[\r\n]+|%-', '%1');	-- remove duplicate row markers end end end

--[[--< C O L U M N _ I N D E X _ G E T >--

get the labels from the tops of the current wikitable; we need indexes for '!Type' and '!Note' and return the associated column number. This function assumes that label markup for each table is one-line-per-label; label markup all-on-one-line (! !! !! ...) is not supported.

is a {| ... |} delimited wikitable

]]

local function column_index_get (wikitable) local i = 0; local type_col, note_col; for label in wikitable:gmatch ('[\r\n](![^\r\n]+)') do						-- spin through each header row i = i + 1;																-- bump the indexer if label:match ('! *Type') then											-- if this header row is for 'Type' type_col = i;														-- save the indexer as column number elseif label:match ('! *Note') then										-- if this header row is for 'Note' note_col = i;														-- save the indexer as column number end end return type_col, note_col; end

--[[--< S H I P _ T Y P E S _ G E T >--

find first row markup (|-) and fetch the column numbers for Type and Note. Then find each row markup, count lines until Type column data are located, extract the contents (ship type) and add entry to the appropriate wikitable in wikitables_t; ship type already present, bump the count.

Continue to locate the Note column. If Note column holds a (commissioned ships) or  (non=commissioned ships) template, add ship type to ; ship type already present, bump the count.

]]

local function ship_types_get (wikitable, wikitable_index) local find_pattern = '|%-'; local type_label_index, note_label_index; local tstart, tend = wikitable:find (find_pattern);							-- find the table row marker in ; should be column headers local i=0;

if tstart then																-- if we found row marker (|-) for headers type_label_index, note_label_index = column_index_get (wikitable);		-- try to find index of type label (!Type) and note label (!Note) tstart, tend = wikitable:find (find_pattern, tend);						-- look for the next row else return nil;																-- TODO: error message? end

while tstart do		local i = 0; local column_text;														-- the text that is the wikitable column markup + content (a row of text that is from:  ) local rstart = tend;													-- row start and end return values from string.find local rend; local pattern = '[\r\n]+|([^\r\n]*)';									-- newline, pipe, everything before the next newline local ship_type;														-- the ship type; we need this for the Note column

while rstart do															-- nil if not found rstart, rend, column_text = wikitable:find (pattern, rstart);		-- get the row's individual column contents

if rstart then i = i + 1;														-- bump the column counter

if type_label_index == i then									-- is this the column that has the ship type entry (Type)? ship_type = column_text;									-- save a copy of the Type column text for use with Note for the expected-to-be-retired section of this function

if wikitables_t[wikitable_names_t[wikitable_index]][ship_type] then		-- if we found a ship type that we already know about: wikitables_t[wikitable_names_t[wikitable_index]][ship_type] = wikitables_t[wikitable_names_t[wikitable_index]][ship_type] + 1;	-- bump the count else wikitables_t[wikitable_names_t[wikitable_index]][ship_type] = 1;	-- add ship type to the table with a count of 1 else end totals_t[wikitable_names_t[wikitable_index]] = totals_t[wikitable_names_t[wikitable_index]] + 1;	-- tally

elseif note_label_index == i then								-- here we are looking for and  templates if column_text:match ('{{%s*[Dd]ecommission%s*|%s*scheduled') or column_text:match ('{{%s*[Dd]ecommission%s*|%s*proposed') or						column_text:match ('{{%s*[Ee]nd of service%s*|%s*scheduled') or column_text:match ('{{%s*[Ee]nd of service%s*|%s*proposed') then

local year = column_text:match ('{{%s*[Dd]ecommission%s*|%s*scheduled%s*|%s*(%d%d%d%d)') or column_text:match ('{{%s*[Dd]ecommission%s*|%s*proposed%s*|%s*(%d%d%d%d)') or						column_text:match ('{{%s*[Ee]nd of service%s*|%s*scheduled%s*|%s*(%d%d%d%d)') or column_text:match ('{{%s*[Ee]nd of service%s*|%s*proposed%s*|%s*(%d%d%d%d)');

local decom_ship_type; if year then decom_ship_type = table.concat ({ship_type, ' (', year, ')'}); if tonumber (year) > retirements_year_max then retirements_year_max = tonumber (year); end else decom_ship_type = ship_type; end

if wikitables_t['Retirements_t'][decom_ship_type] then	-- if we found a ship type that we already know about: wikitables_t['Retirements_t'][decom_ship_type] = wikitables_t['Retirements_t'][decom_ship_type] + 1;	-- bump the count else wikitables_t['Retirements_t'][decom_ship_type] = 1;	-- add ship type to the table with a count of 1 else end

totals_t['Retirements_t'] = totals_t['Retirements_t'] + 1;	-- tally end end rstart = rend;														-- not yet on the correct column; reset the starting index to the end of the last find end end

tstart, tend = wikitable:find (find_pattern, tend);							-- search for another row marker (|-) in ; begin at end of last row marker search end end

--[[--< R E N D E R _ O U T P U T >

extract data from various tables and append to 

]]

local function render_output (index, output_string) local out_t = {};															-- here we compose this section of  local temp_t = {};															-- a sequence of ship types and counts taken from wikitables_t; separate table for sorting local function sort (a, b)		a = a:gsub ('%[%[(.+)%]%]', '%1');										-- remove outer wikilink markup b = b:gsub ('%[%[(.+)%]%]', '%1'); a = a:gsub ('%* ', '');													-- remove unordered list markup b = b:gsub ('%* ', ''); a = a:gsub ('^[^|]+|', '');												-- remove all but the display text from complex wikilinks b = b:gsub ('^[^|]+|', ''); a = a:gsub (' *– *%d+', '');											-- remove the count b = b:gsub (' *– *%d+', ''); return a < b;	end table.insert (out_t, table.concat ({'', headers_t[index], ' – ', totals_t[wikitable_names_t[index]]}));		-- header table.insert (out_t, '{{Div col|colwidth=22em}}');							-- start of columnar data for ship_type, count in pairs (wikitables_t[wikitable_names_t[index]]) do	-- make an unordered, unsorted, list local ship_type_total = n_of_m_t[ship_type] and table.concat ({' (of ', n_of_m_t[ship_type], ')'}) or ''; if wikitable_names_t[index] == fleet_wikitables_t[index] then table.insert (temp_t, table.concat ({'* ', ship_type, ' – ', count, ship_type_total}));	-- make and add an unordered list item else table.insert (temp_t, table.concat ({'* ', ship_type, ' – ', count}));	-- make and add an unordered list item end end table.sort (temp_t, sort);													-- ascending sort; TODO: is there a better way to get a sorted list of ship types? table.insert (out_t, table.concat (temp_t, '\n'));							-- make a big string and add it to out_t table.insert (out_t, '{{div col end}}');									-- end of columnar data table.insert (out_t, ' \n');											-- horizontal rule output_string = table.concat ({output_string, table.concat (out_t, '\n')});	-- make a big string and add it to output_string return output_string;														-- and done end

--[[--< R E N D E R _ T O T A L S >

render a wikitable list of totals; exclude Under construction, on-order, and expected-to-be-retired ships

This is a presentation table so that the numbers are right justified, each above and below the preceding number

]]

local function render_totals (output_string) local out_t = {'Totals'}; local tally = 0;

table.insert (out_t, '{| role="presentation" style="margin-left:1.5em"');	-- start a wikitable indented 1.5em for i, wikitable_name in ipairs (wikitable_names_t) do						-- spin through the list of table names if 6 > i then															-- not under construction of on order local wikitable = wikitable_name:gsub ('_t', '');					-- remove suffix table.insert (out_t, '|-');											-- add row markup table.insert (out_t, table.concat ({'| ', wikitable, ': || style="text-align:right" | ', totals_t[wikitable_names_t[i]]}));	-- make a row tally = tally + totals_t[wikitable_names_t[i]];						-- and bump the grand total else break; end end grand_total = tally;														-- this will be rounded to the nearest integer evenly divisible by 5 table.insert (out_t, '|-');														-- add row markup table.insert (out_t, table.concat ({'| Grand total: || style="text-align:right" | ', tally, ''}));	-- add the grand total row table.insert (out_t, '|}');													-- close the wikitable

output_string = table.concat ({output_string, table.concat (out_t, '\n')});	-- make a big string and add it to output_string return output_string;														-- and done end

--[[--< N _ O F _ M _ G E T >--

For ship types in commissioned, non-commissioned, support, ready researve, and reserve tables, count the number of same-type ships so that we can render '(of M)' annotation for those ship types.

First gather a list of ship types that are listed in more than one table. Compare commissioned to non-commissioned; commissioned to support; etc. Then non-commissioned to support; non-commissioned to ready researve; etc. Continue until we compare ready researve to reserve. Do not compare reserve to itself.

Second, scroll through the list of duplicates and accumulate tallies of each ship type.

]]

local function n_of_m_get for i, wikitable_name in ipairs (fleet_wikitables_t) do						-- for each wikitable if #fleet_wikitables_t == i then										-- when we get to the last wikitable, don't compare it to itself break; end for ship_type, _ in pairs (wikitables_t[wikitable_name]) do for j=i+1, #fleet_wikitables_t do									-- index 1 looks in indexes 2, 3, ... if wikitables_t[fleet_wikitables_t[j]][ship_type] then n_of_m_t[ship_type] = 0;									-- this ship type appears in more than one wikitable end end end end

for ship_type, _ in pairs (n_of_m_t) do										-- for each ship type in n_of_m table for _, wikitable_name in ipairs (fleet_wikitables_t) do					-- for each wikitable if wikitables_t[wikitable_name][ship_type] then						-- if the ship type is found in this wikitable n_of_m_t[ship_type] = n_of_m_t[ship_type] + wikitables_t[wikitable_name][ship_type];	-- add to the tally end end end end

--[[--< U S N _ S H I P _ C O U N T E R >--

tables in 'List of current ships of the United States Navy' are (in this order): Commissioned Non-commissioned Support Ready Reserve Force ships Reserve fleet Under construction On order

count the ship types in the Type columns of these tables, and then render pretty sorted lists of ship types with their counts.

]]

local function fleet_totals article_content_get ;														-- attempt to get this article's content

local wikitable_index = 0; local find_pattern = '{|'; local tstart, tend = Article_content:find (find_pattern);					-- find the first wikitable

while tstart do		local wikitable = Article_content:match ('%b{}', tstart);				-- get the whole wikitable

if not wikitable then break;																-- wikitable is nil for some reason (missing closing |} for example) so declare ourselves done end

wikitable_index = wikitable_index + 1; ship_types_get (wikitable, wikitable_index);

tstart, tend = Article_content:find (find_pattern, tend);				-- search for another template; begin at end of last wikitable search end

n_of_m_get ;

local output_string = ''; for i, _ in ipairs (wikitable_names_t) do		output_string = render_output (i, output_string);						-- make a pretty unordered list of ship types and counts for each wikitable end

output_string = render_totals (output_string);								-- make a prety unordered list of 'classification' total numbers

local frame = mw.getCurrentFrame;											-- do this to get access to frame:preprocess return frame:preprocess (output_string); end

----< E X P O R T S >

return { fleet_totals_str = fleet_totals, grand_total = grand_total, planned_total = totals_t['Under construction_t'] + totals_t['On order_t'], retirements_total = totals_t['Retirements_t'], retirements_year_max = retirements_year_max,								-- for {{USN fleet totals|retire-year}}

decommission_t = {															-- for (commissioned ships) ['scheduled'] = 'Scheduled to be decommissioned', ['proposed'] = 'Proposed to be decommissioned', },	end_of_service_t = {														-- for (non-commissioned ships) ['scheduled'] = 'Scheduled end of service', ['proposed'] = 'Proposed end of service', },	}