Merge "Add text module"

Scribunto.php

@@ -112,6 +112,7 @@ $wgAutoloadClasses['Scribunto_LuaUstringLibrary'] = $dir.'engines/LuaCommon/Ustr
 $wgAutoloadClasses['Scribunto_LuaLanguageLibrary'] = $dir.'engines/LuaCommon/LanguageLibrary.php';
 $wgAutoloadClasses['Scribunto_LuaMessageLibrary'] = $dir.'engines/LuaCommon/MessageLibrary.php';
 $wgAutoloadClasses['Scribunto_LuaTitleLibrary'] = $dir.'engines/LuaCommon/TitleLibrary.php';
+$wgAutoloadClasses['Scribunto_LuaTextLibrary'] = $dir.'engines/LuaCommon/TextLibrary.php';
 
 /***** Configuration *****/
 

common/Hooks.php

@@ -315,6 +315,7 @@ WIKI;
 			'engines/LuaCommon/UstringLibraryTest.php',
 			'engines/LuaCommon/MessageLibraryTest.php',
 			'engines/LuaCommon/TitleLibraryTest.php',
+			'engines/LuaCommon/TextLibraryTest.php',
 		);
 		foreach ( $tests as $test ) {
 			$files[] = dirname( __FILE__ ) .'/../tests/' . $test;

engines/LuaCommon/LuaCommon.php

@@ -12,6 +12,7 @@ abstract class Scribunto_LuaEngine extends ScribuntoEngineBase {
 		'mw.language' => 'Scribunto_LuaLanguageLibrary',
 		'mw.message' => 'Scribunto_LuaMessageLibrary',
 		'mw.title' => 'Scribunto_LuaTitleLibrary',
+		'mw.text' => 'Scribunto_LuaTextLibrary',
 	);
 
 	/**

engines/LuaCommon/TextLibrary.php

@@ -0,0 +1,31 @@
+<?php
+
+class Scribunto_LuaTextLibrary extends Scribunto_LuaLibraryBase {
+	function register() {
+		$lib = array(
+			'unstrip' => array( $this, 'textUnstrip' ),
+			'getEntityTable' => array( $this, 'getEntityTable' ),
+		);
+		$this->getEngine()->registerInterface( 'mw.text.lua', $lib, array(
+			'comma' => wfMessage( 'comma-separator' )->inContentLanguage()->text(),
+			'and' => wfMessage( 'and' )->inContentLanguage()->text() .
+				wfMessage( 'word-separator' )->inContentLanguage()->text(),
+			'ellipsis' => wfMessage( 'ellipsis' )->inContentLanguage()->text(),
+		) );
+	}
+
+	function textUnstrip( $s ) {
+		$this->checkType( 'unstrip', 1, $s, 'string' );
+		return array( $this->getParser()->mStripState->unstripBoth( $s ) );
+	}
+
+	function getEntityTable() {
+		$flags = ENT_QUOTES;
+		// PHP 5.3 compat
+		if ( defined( "ENT_HTML5" ) ) {
+			$flags |= constant( "ENT_HTML5" );
+		}
+		$table = array_flip( get_html_translation_table( HTML_ENTITIES, $flags, "UTF-8" ) );
+		return array( $table );
+	}
+}

engines/LuaCommon/lualib/libraryUtil.lua

@@ -21,6 +21,18 @@ function libraryUtil.checkTypeForIndex( index, value, expectType )
 	end
 end
 
+function libraryUtil.checkTypeForNamedArg( name, argName, arg, expectType, nilOk )
+	if arg == nil and nilOk then
+		return
+	end
+	if type( arg ) ~= expectType then
+		local msg = string.format( "bad named argument %s to '%s' (%s expected, got %s)",
+			argName, name, expectType, type( arg )
+		)
+		error( msg, 3 )
+	end
+end
+
 function libraryUtil.makeCheckSelfFunction( libraryName, varName, selfObj, selfObjDesc )
 	return function ( self, method )
 		if self ~= selfObj then

engines/LuaCommon/lualib/mw.text.lua

@@ -0,0 +1,257 @@
+local mwtext = {}
+local php
+local options
+
+local util = require 'libraryUtil'
+local checkType = util.checkType
+local checkTypeForNamedArg = util.checkTypeForNamedArg
+
+function mwtext.setupInterface( opts )
+	-- Boilerplate
+	mwtext.setupInterface = nil
+	php = mw_interface
+	mw_interface = nil
+	options = opts
+
+	-- Register this library in the "mw" global
+	mw = mw or {}
+	mw.text = mwtext
+
+	package.loaded['mw.text'] = mwtext
+end
+
+function mwtext.trim( s, charset )
+	charset = charset or '\t\r\n\f '
+	s = mw.ustring.gsub( s, '^[' .. charset .. ']*(.-)[' .. charset .. ']*$', '%1' )
+	return s
+end
+
+local htmlencode_map = {
+	['>'] = '>',
+	['<'] = '&lt;',
+	['&'] = '&amp;',
+	['"'] = '&quot;',
+	["'"] = '&#039;',
+	['\194\160'] = '&#nbsp;',
+}
+local htmldecode_map = {}
+for k, v in pairs( htmlencode_map ) do
+	htmldecode_map[v] = k
+end
+local decode_named_entities = nil
+
+function mwtext.encode( s, charset )
+	charset = charset or '<>&"\'\194\160'
+	s = mw.ustring.gsub( s, '[' .. charset .. ']', function ( m )
+		if not htmlencode_map[m] then
+			local e = string.format( '&#%d;', mw.ustring.codepoint( m ) )
+			htmlencode_map[m] = e
+			htmldecode_map[e] = m
+		end
+		return htmlencode_map[m]
+	end )
+	return s
+end
+
+function mwtext.decode( s, decodeNamedEntities )
+	local dec
+	if decodeNamedEntities then
+		if decode_named_entities == nil then
+			decode_named_entities = php.getEntityTable()
+			setmetatable( decode_named_entities, { __index = htmldecode_map } )
+		end
+		dec = decode_named_entities
+	else
+		dec = htmldecode_map
+	end
+	-- string.gsub is safe here, because only ASCII chars are in the pattern
+	s = string.gsub( s, '(&(#?x?)([a-zA-Z0-9]+);)', function ( m, flg, name )
+		if not dec[m] then
+			local n = nil
+			if flg == '#' then
+				n = tonumber( name, 10 )
+			elseif flg == '#x' then
+				n = tonumber( name, 16 )
+			end
+			if n and n <= 0x10ffff then
+				n = mw.ustring.char( n )
+				if n then
+					htmldecode_map[m] = n
+					htmlencode_map[n] = m
+				end
+			end
+		end
+		return dec[m]
+	end )
+	return s
+end
+
+local nowikiRepl1 = {
+	['"'] = '&#34;',
+	['&'] = '&#38;',
+	["'"] = '&#39;',
+	['<'] = '&#60;',
+	['='] = '&#61;',
+	['>'] = '&#62;',
+	['['] = '&#91;',
+	[']'] = '&#93;',
+	['{'] = '&#123;',
+	['|'] = '&#124;',
+	['}'] = '&#125;',
+}
+
+local nowikiRepl2 = {
+	["\n#"] = "\n&#35;",
+	["\n*"] = "\n&#42;",
+	["\n:"] = "\n&#58;",
+	["\n;"] = "\n&#59;",
+}
+
+function mwtext.nowiki( s )
+	-- string.gsub is safe here, because we're only caring about ASCII chars
+	s = string.gsub( s, '["&\'<=>%[%]{|}]', nowikiRepl1 )
+	s = string.sub( string.gsub( '\n' .. s, '\n[#*:;]', nowikiRepl2 ), 2 )
+	s = string.gsub( s, '://', '&#58;//' )
+	s = string.gsub( s, 'ISBN ', 'ISBN&#32;' )
+	s = string.gsub( s, 'RFC ', 'RFC&#32;' )
+
+	return s
+end
+
+function mwtext.tag( name, attrs, content )
+	local named = false
+	if type( name ) == 'table' then
+		named = true
+		name, attrs, content = name.name, name.attrs, name.content
+		checkTypeForNamedArg( 'tag', 'name', name, 'string' )
+		checkTypeForNamedArg( 'tag', 'attrs', attrs, 'table', true )
+	else
+		checkType( 'tag', 1, name, 'string' )
+		checkType( 'tag', 2, attrs, 'table', true )
+	end
+
+	local ret = { '<' .. name }
+	for k, v in pairs( attrs or {} ) do
+		if type( k ) ~= 'string' then
+			error( "bad named argument attrs to 'tag' (keys must be strings, found " .. type( k ) .. ")", 2 )
+		end
+		if string.match( k, '[\t\r\n\f /<>"\'=]' ) then
+			error( "bad named argument attrs to 'tag' (invalid key '" .. k .. "')", 2 )
+		end
+		local tp = type( v )
+		if tp == 'boolean' then
+			if v then
+				ret[#ret+1] = ' ' .. k
+			end
+		elseif tp == 'string' or tp == 'number' then
+			ret[#ret+1] = string.format( ' %s="%s"', k, mwtext.encode( tostring( v ) ) )
+		else
+			error( "bad named argument attrs to 'tag' (value for key '" .. k .. "' may not be " .. tp .. ")", 2 )
+		end
+	end
+
+	local tp = type( content )
+	if content == nil then
+		ret[#ret+1] = '>'
+	elseif content == false then
+		ret[#ret+1] = ' />'
+	elseif tp == 'string' or tp == 'number' then
+		ret[#ret+1] = '>'
+		ret[#ret+1] = content
+		ret[#ret+1] = '</' .. name .. '>'
+	else
+		if named then
+			checkTypeForNamedArg( 'tag', 'content', content, 'string, number, nil, or false' )
+		else
+			checkType( 'tag', 3, content, 'string, number, nil, or false' )
+		end
+	end
+
+	return table.concat( ret )
+end
+
+function mwtext.unstrip( s )
+	return php.unstrip( s )
+end
+
+function mwtext.split( text, pattern, plain )
+	local ret = {}
+	for m in mwtext.gsplit( text, pattern, plain ) do
+		ret[#ret+1] = m
+	end
+	return ret
+end
+
+function mwtext.gsplit( text, pattern, plain )
+	local s, l = 1, mw.ustring.len( text )
+	return function ()
+		if s then
+			local e, n = mw.ustring.find( text, pattern, s, plain )
+			local ret
+			if not e then
+				ret = mw.ustring.sub( text, s )
+				s = nil
+			elseif n < e then
+				-- Empty separator!
+				ret = mw.ustring.sub( text, s, e )
+				if e < l then
+					s = e + 1
+				else
+					s = nil
+				end
+			else
+				ret = e > s and mw.ustring.sub( text, s, e - 1 ) or ''
+				s = n + 1
+			end
+			return ret
+		end
+	end, nil, nil
+end
+
+function mwtext.listToText( list, separator, conjunction )
+	separator = separator or options.comma
+	conjunction = conjunction or options['and']
+	local n = #list
+
+	local ret
+	if n > 1 then
+		local tmp = list[n-1]
+		list[n-1] = list[n-1] .. conjunction .. list[n]
+		ret = table.concat( list, separator, 1, n - 1 )
+		list[n-1] = tmp
+	else
+		ret = tostring( list[1] or '' )
+	end
+
+	return ret
+end
+
+function mwtext.truncate( text, length, ellipsis, adjustLength )
+	local l = mw.ustring.len( text )
+	if l <= math.abs( length ) then
+		return text
+	end
+
+	ellipsis = ellipsis or options.ellipsis
+	local elen = 0
+	if adjustLength then
+		elen = mw.ustring.len( ellipsis )
+	end
+
+	local ret
+	if math.abs( length ) <= elen then
+		ret = ellipsis
+	elseif length > 0 then
+		ret = mw.ustring.sub( text, 1, length - elen ) .. ellipsis
+	else
+		ret = ellipsis .. mw.ustring.sub( text, length + elen )
+	end
+
+	if mw.ustring.len( ret ) < l then
+		return ret
+	else
+		return text
+	end
+end
+
+return mwtext

tests/engines/LuaCommon/TextLibraryTest.php

@@ -0,0 +1,23 @@
+<?php
+
+class Scribunto_LuaTextLibraryTests extends Scribunto_LuaEngineTestBase {
+	protected static $moduleName = 'TextLibraryTests';
+
+	function setUp() {
+		parent::setUp();
+
+		// For unstrip test
+		$interpreter = $this->getEngine()->getInterpreter();
+		$interpreter->callFunction(
+			$interpreter->loadString( 'mw.text.stripTest = ...', 'fortest' ),
+			$this->getEngine()->getParser()->insertStripItem( 'ok' )
+		);
+	}
+
+
+	function getTestModules() {
+		return parent::getTestModules() + array(
+			'TextLibraryTests' => __DIR__ . '/TextLibraryTests.lua',
+		);
+	}
+}

tests/engines/LuaCommon/TextLibraryTests.lua

@@ -0,0 +1,239 @@
+local testframework = require 'Module:TestFramework'
+
+-- Force the argument list to be ordered
+local tagattrs = { absent = false, present = true, key = 'value', n = 42 }
+setmetatable( tagattrs, { __pairs = function ( t )
+	local keys = { 'absent', 'present', 'key', 'n' }
+	local i = 0
+	return function()
+		i = i + 1
+		if i <= #keys then
+			return keys[i], t[keys[i]]
+		end
+	end
+end } )
+
+-- Tests
+local tests = {
+	{ name = 'trim',
+	  func = mw.text.trim, args = { '  foo bar  ' },
+	  expect = { 'foo bar' }
+	},
+	{ name = 'trim right',
+	  func = mw.text.trim, args = { 'foo bar  ' },
+	  expect = { 'foo bar' }
+	},
+	{ name = 'trim left',
+	  func = mw.text.trim, args = { '  foo bar' },
+	  expect = { 'foo bar' }
+	},
+	{ name = 'trim none',
+	  func = mw.text.trim, args = { 'foo bar' },
+	  expect = { 'foo bar' }
+	},
+	{ name = 'trim charset',
+	  func = mw.text.trim, args = { 'xxx foo bar xxx', 'x' },
+	  expect = { ' foo bar ' }
+	},
+
+	{ name = 'encode',
+	  func = mw.text.encode, args = { '<b>foo "bar"</b> & \'baz\'' },
+	  expect = { '&lt;b&gt;foo &quot;bar&quot;&lt;/b&gt; &amp; &#039;baz&#039;' }
+	},
+	{ name = 'encode charset',
+	  func = mw.text.encode, args = { '<b>foo "bar"</b> & \'baz\'', 'aeiou' },
+	  expect = { '<b>f&#111;&#111; "b&#97;r"</b> & \'b&#97;z\'' }
+	},
+
+	{ name = 'decode',
+	  func = mw.text.decode,
+	  args = { '&lt;&gt;&amp;&quot; &#102;&#111;&#x6f; &#x0066;&#00111;&#x6F; &hearts; &amp;quot;' },
+	  expect = { '<>&" foo foo &hearts; &quot;' }
+	},
+	{ name = 'decode named',
+	  func = mw.text.decode,
+	  args = { '&lt;&gt;&amp;&quot; &#102;&#111;&#x6f; &#x0066;&#00111;&#x6F; &hearts; &amp;quot;', true },
+	  expect = { '<>&" foo foo ♥ &quot;' }
+	},
+
+	{ name = 'nowiki',
+	  func = mw.text.nowiki,
+	  args = { '*"&\'<=>[]{|}#*:;\n*\n#\n:\n;\nhttp://example.com:80/\nRFC 123, ISBN 456' },
+	  expect = {
+		  '&#42;&#34;&#38;&#39;&#60;&#61;&#62;&#91;&#93;&#123;&#124;&#125;#*:;' ..
+		  '\n&#42;\n&#35;\n&#58;\n&#59;\nhttp&#58;//example.com:80/' ..
+		  '\nRFC&#32;123, ISBN&#32;456'
+	  }
+	},
+
+	{ name = 'tag, simple',
+	  func = mw.text.tag,
+	  args = { { name = 'b' } },
+	  expect = { '<b>' }
+	},
+	{ name = 'tag, simple with content',
+	  func = mw.text.tag,
+	  args = { { name = 'b', content = 'foo' } },
+	  expect = { '<b>foo</b>' }
+	},
+	{ name = 'tag, simple self-closing',
+	  func = mw.text.tag,
+	  args = { { name = 'br', content = false } },
+	  expect = { '<br />' }
+	},
+	{ name = 'tag, args',
+	  func = mw.text.tag,
+	  args = { { name = 'b', attrs = tagattrs } },
+	  expect = { '<b present key="value" n="42">' }
+	},
+	{ name = 'tag, args with content',
+	  func = mw.text.tag,
+	  args = { { name = 'b', attrs = tagattrs, content = 'foo' } },
+	  expect = { '<b present key="value" n="42">foo</b>' }
+	},
+	{ name = 'tag, args self-closing',
+	  func = mw.text.tag,
+	  args = { { name = 'br', attrs = tagattrs, content = false } },
+	  expect = { '<br present key="value" n="42" />' }
+	},
+	{ name = 'tag, args, positional params',
+	  func = mw.text.tag,
+	  args = { 'b', tagattrs },
+	  expect = { '<b present key="value" n="42">' }
+	},
+	{ name = 'tag, args with content, positional params',
+	  func = mw.text.tag,
+	  args = { 'b', tagattrs, 'foo' },
+	  expect = { '<b present key="value" n="42">foo</b>' }
+	},
+
+	{ name = 'unstrip',
+	  func = mw.text.unstrip, args = { mw.text.stripTest },
+	  expect = { 'ok' }
+	},
+
+	{ name = 'split, simple',
+	  func = mw.text.split, args = { 'a,b,c,d', ',' },
+	  expect = { { 'a', 'b', 'c', 'd' } }
+	},
+	{ name = 'split, no separator',
+	  func = mw.text.split, args = { 'xxx', ',' },
+	  expect = { { 'xxx' } }
+	},
+	{ name = 'split, empty string',
+	  func = mw.text.split, args = { '', ',' },
+	  expect = { { '' } }
+	},
+	{ name = 'split, with empty items',
+	  func = mw.text.split, args = { ',,', ',' },
+	  expect = { { '', '', '' } }
+	},
+	{ name = 'split, with empty items (1)',
+	  func = mw.text.split, args = { 'x,,', ',' },
+	  expect = { { 'x', '', '' } }
+	},
+	{ name = 'split, with empty items (2)',
+	  func = mw.text.split, args = { ',x,', ',' },
+	  expect = { { '', 'x', '' } }
+	},
+	{ name = 'split, with empty items (3)',
+	  func = mw.text.split, args = { ',,x', ',' },
+	  expect = { { '', '', 'x' } }
+	},
+	{ name = 'split, with empty items (4)',
+	  func = mw.text.split, args = { ',x,x', ',' },
+	  expect = { { '', 'x', 'x' } }
+	},
+	{ name = 'split, with empty items (5)',
+	  func = mw.text.split, args = { 'x,,x', ',' },
+	  expect = { { 'x', '', 'x' } }
+	},
+	{ name = 'split, with empty items (7)',
+	  func = mw.text.split, args = { 'x,x,', ',' },
+	  expect = { { 'x', 'x', '' } }
+	},
+	{ name = 'split, with empty pattern',
+	  func = mw.text.split, args = { 'xxx', '' },
+	  expect = { { 'x', 'x', 'x' } }
+	},
+	{ name = 'split, with empty pattern (2)',
+	  func = mw.text.split, args = { 'xxx', ',?' },
+	  expect = { { 'x', 'x', 'x' } }
+	},
+
+	{ name = 'listToText (0)',
+	  func = mw.text.listToText, args = { {} },
+	  expect = { '' }
+	},
+	{ name = 'listToText (1)',
+	  func = mw.text.listToText, args = { { 1 } },
+	  expect = { '1' }
+	},
+	{ name = 'listToText (2)',
+	  func = mw.text.listToText, args = { { 1, 2 } },
+	  expect = { '1 and 2' }
+	},
+	{ name = 'listToText (3)',
+	  func = mw.text.listToText, args = { { 1, 2, 3 } },
+	  expect = { '1, 2 and 3' }
+	},
+	{ name = 'listToText (4)',
+	  func = mw.text.listToText, args = { { 1, 2, 3, 4 } },
+	  expect = { '1, 2, 3 and 4' }
+	},
+	{ name = 'listToText, alternate separator',
+	  func = mw.text.listToText, args = { { 1, 2, 3, 4 }, '; ' },
+	  expect = { '1; 2; 3 and 4' }
+	},
+	{ name = 'listToText, alternate conjunction',
+	  func = mw.text.listToText, args = { { 1, 2, 3, 4 }, nil, ' or ' },
+	  expect = { '1, 2, 3 or 4' }
+	},
+
+	{ name = 'truncate, no truncation',
+	  func = mw.text.truncate, args = { 'foobarbaz', 9 },
+	  expect = { 'foobarbaz' }
+	},
+	{ name = 'truncate, no truncation (2)',
+	  func = mw.text.truncate, args = { 'foobarbaz', -9 },
+	  expect = { 'foobarbaz' }
+	},
+	{ name = 'truncate, tail truncation',
+	  func = mw.text.truncate, args = { 'foobarbaz', 3 },
+	  expect = { 'foo...' }
+	},
+	{ name = 'truncate, head truncation',
+	  func = mw.text.truncate, args = { 'foobarbaz', -3 },
+	  expect = { '...baz' }
+	},
+	{ name = 'truncate, avoid silly truncation',
+	  func = mw.text.truncate, args = { 'foobarbaz', 8 },
+	  expect = { 'foobarbaz' }
+	},
+	{ name = 'truncate, avoid silly truncation (2)',
+	  func = mw.text.truncate, args = { 'foobarbaz', 6 },
+	  expect = { 'foobarbaz' }
+	},
+	{ name = 'truncate, alternate ellipsis',
+	  func = mw.text.truncate, args = { 'foobarbaz', 3, '!' },
+	  expect = { 'foo!' }
+	},
+	{ name = 'truncate, with adjusted length',
+	  func = mw.text.truncate, args = { 'foobarbaz', 6, nil, true },
+	  expect = { 'foo...' }
+	},
+	{ name = 'truncate, with adjusted length (2)',
+	  func = mw.text.truncate, args = { 'foobarbaz', -6, nil, true },
+	  expect = { '...baz' }
+	},
+	{ name = 'truncate, ridiculously short',
+	  func = mw.text.truncate, args = { 'foobarbaz', 1, nil, true },
+	  expect = { '...' }
+	},
+	{ name = 'truncate, ridiculously short (2)',
+	  func = mw.text.truncate, args = { 'foobarbaz', -1, nil, true },
+	  expect = { '...' }
+	},
+}
+
+return testframework.getTestProvider( tests )