r90928 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r90927‎ \| r90928 \| r90929 >
Date:	01:09, 28 June 2011
Author:	aaron
Status:	ok
Tags:
Comment:	* Added truncate() & truncateHTML() tests * Some fixes/changes to truncateHTML() based on tests Something like "<span>hello</span>" ends up as "<span>...</span>" instead of just "..." for relevant cases) If we get something like "<span></span" in, just return it back instead of "" * Renamed $dispLength -> $dispLen in truncateHTML()
Modified paths:	/trunk/phase3/languages/Language.php (modified) (history) /trunk/phase3/tests/phpunit/languages/LanguageTest.php (modified) (history)

Diff [purge]

Index: trunk/phase3/tests/phpunit/languages/LanguageTest.php
—	—	@@ -112,7 +112,109 @@
113	113	);
114	114	}
115	115
	116	+ function testTruncate() {
	117	+ $this->assertEquals(
	118	+ "XXX",
	119	+ $this->lang->truncate( "1234567890", 0, 'XXX' ),
	120	+ 'truncate prefix, len 0, small ellipsis'
	121	+ );
	122	+
	123	+ $this->assertEquals(
	124	+ "12345XXX",
	125	+ $this->lang->truncate( "1234567890", 8, 'XXX' ),
	126	+ 'truncate prefix, small ellipsis'
	127	+ );
	128	+
	129	+ $this->assertEquals(
	130	+ "123456789",
	131	+ $this->lang->truncate( "123456789", 5, 'XXXXXXXXXXXXXXX' ),
	132	+ 'truncate prefix, large ellipsis'
	133	+ );
	134	+
	135	+ $this->assertEquals(
	136	+ "XXX67890",
	137	+ $this->lang->truncate( "1234567890", -8, 'XXX' ),
	138	+ 'truncate suffix, small ellipsis'
	139	+ );
	140	+
	141	+ $this->assertEquals(
	142	+ "123456789",
	143	+ $this->lang->truncate( "123456789", -5, 'XXXXXXXXXXXXXXX' ),
	144	+ 'truncate suffix, large ellipsis'
	145	+ );
	146	+ }
	147	+
116	148	/**
	149	+ * @dataProvider provideHTMLTruncateData()
	150	+ */
	151	+ function testTruncateHTML( $len, $ellipsis, $input, $expected ) {
	152	+ // Actual HTML...
	153	+ $this->assertEquals(
	154	+ $expected,
	155	+ $this->lang->truncateHTML( $input, $len, $ellipsis )
	156	+ );
	157	+ }
	158	+
	159	+ /**
	160	+ * Array format is ($len, $ellipsis, $input, $expected)
	161	+ */
	162	+ function provideHTMLTruncateData() {
	163	+ return array(
	164	+ array( 0, 'XXX', "1234567890", "XXX" ),
	165	+ array( 8, 'XXX', "1234567890", "12345XXX" ),
	166	+ array( 5, 'XXXXXXXXXXXXXXX', '1234567890', "1234567890" ),
	167	+ array( 2, '***',
	168	+ '<p><span style="font-weight:bold;"></span></p>',
	169	+ '<p><span style="font-weight:bold;"></span></p>',
	170	+ ),
	171	+ array( 2, '***',
	172	+ '<p><span style="font-weight:bold;">123456789</span></p>',
	173	+ '<p><span style="font-weight:bold;">***</span></p>',
	174	+ ),
	175	+ array( 2, '***',
	176	+ '<p><span style="font-weight:bold;"> 23456789</span></p>',
	177	+ '<p><span style="font-weight:bold;">***</span></p>',
	178	+ ),
	179	+ array( 3, '***',
	180	+ '<p><span style="font-weight:bold;">123456789</span></p>',
	181	+ '<p><span style="font-weight:bold;">***</span></p>',
	182	+ ),
	183	+ array( 4, '***',
	184	+ '<p><span style="font-weight:bold;">123456789</span></p>',
	185	+ '<p><span style="font-weight:bold;">1***</span></p>',
	186	+ ),
	187	+ array( 5, '***',
	188	+ '<tt><span style="font-weight:bold;">123456789</span></tt>',
	189	+ '<tt><span style="font-weight:bold;">12***</span></tt>',
	190	+ ),
	191	+ array( 6, '***',
	192	+ '<p><a href="www.mediawiki.org">123456789</a></p>',
	193	+ '<p><a href="www.mediawiki.org">123***</a></p>',
	194	+ ),
	195	+ array( 6, '***',
	196	+ '<p><a href="www.mediawiki.org">12 456789</a></p>',
	197	+ '<p><a href="www.mediawiki.org">12 ***</a></p>',
	198	+ ),
	199	+ array( 7, '***',
	200	+ '<small><span style="font-weight:bold;">123<p id="#moo">456</p>789</span></small>',
	201	+ '<small><span style="font-weight:bold;">123<p id="#moo">4***</p></span></small>',
	202	+ ),
	203	+ array( 8, '***',
	204	+ '<div><span style="font-weight:bold;">123<span>4</span>56789</span></div>',
	205	+ '<div><span style="font-weight:bold;">123<span>4</span>5***</span></div>',
	206	+ ),
	207	+ array( 9, '***',
	208	+ '<p><table style="font-weight:bold;"><tr><td>123456789</td></tr></table></p>',
	209	+ '<p><table style="font-weight:bold;"><tr><td>123456789</td></tr></table></p>',
	210	+ ),
	211	+ array( 10, '***',
	212	+ '<p><font style="font-weight:bold;">123456789</font></p>',
	213	+ '<p><font style="font-weight:bold;">123456789</font></p>',
	214	+ ),
	215	+ );
	216	+ }
	217	+
	218	+ /**
117	219	* Test Language::isValidBuiltInCode()
118	220	* @dataProvider provideLanguageCodes
119	221	*/
Index: trunk/phase3/languages/Language.php
—	—	@@ -2770,31 +2770,33 @@
2771	2771	return $text; // string short enough even with HTML (short-circuit)
2772	2772	}
2773	2773
2774		~~- $displayLen = 0; // innerHTML legth so far~~
	2774	+ $dispLen = 0; // innerHTML legth so far
2775	2775	$testingEllipsis = false; // checking if ellipses will make string longer/equal?
2776	2776	$tagType = 0; // 0-open, 1-close
2777	2777	$bracketState = 0; // 1-tag start, 2-tag name, 0-neither
2778	2778	$entityState = 0; // 0-not entity, 1-entity
2779		~~- $tag = $ret = $pRet = ''; // accumulated tag name, accumulated result string~~
	2779	+ $tag = $ret = ''; // accumulated tag name, accumulated result string
2780	2780	$openTags = array(); // open tag stack
2781		~~- $pOpenTags = array();~~
	2781	+ $maybeState = null; // possible truncation state
2782	2782
2783	2783	$textLen = strlen( $text );
2784	2784	$neLength = max( 0, $length - strlen( $ellipsis ) ); // non-ellipsis len if truncated
2785	2785	for ( $pos = 0; true; ++$pos ) {
2786	2786	# Consider truncation once the display length has reached the maximim.
	2787	+ # We check if $dispLen > 0 to grab tags for the $neLength = 0 case.
2787	2788	# Check that we're not in the middle of a bracket/entity...
2788		~~- if ( $displayLen >= $neLength && $bracketState == 0 && $entityState == 0 ) {~~
	2789	+ if ( $dispLen && $dispLen >= $neLength && $bracketState == 0 && !$entityState ) {
2789	2790	if ( !$testingEllipsis ) {
2790	2791	$testingEllipsis = true;
2791	2792	# Save where we are; we will truncate here unless there turn out to
2792	2793	# be so few remaining characters that truncation is not necessary.
2793		~~- $pOpenTags = $openTags; // save state~~
2794		~~- $pRet = $ret; // save state~~
2795		~~- } elseif ( $displayLen > $length && $displayLen > strlen( $ellipsis ) ) {~~
	2794	+ if ( !$maybeState ) { // already saved? ($neLength = 0 case)
	2795	+ $maybeState = array( $ret, $openTags ); // save state
	2796	+ }
	2797	+ } elseif ( $dispLen > $length && $dispLen > strlen( $ellipsis ) ) {
2796	2798	# String in fact does need truncation, the truncation point was OK.
2797		~~- $openTags = $pOpenTags; // reload state~~
2798		~~- $ret = $this->removeBadCharLast( $pRet ); // reload state, multi-byte char fix~~
	2799	+ list( $ret, $openTags ) = $maybeState; // reload state
	2800	+ $ret = $this->removeBadCharLast( $ret ); // multi-byte char fix
2799	2801	$ret .= $ellipsis; // add ellipsis
2800	2802	break;
2801	2803	}
—	—	@@ -2832,25 +2834,27 @@
2833	2835	if ( $entityState ) {
2834	2836	if ( $ch == ';' ) {
2835	2837	$entityState = 0;
2836		~~- $displayLen++; // entity is one displayed char~~
	2838	+ $dispLen++; // entity is one displayed char
2837	2839	}
2838	2840	} else {
	2841	+ if ( $neLength == 0 && !$maybeState ) {
	2842	+ // Save state without $ch. We want to hit the first
	2843	+ // display char (to get tags) but not use it if truncating.
	2844	+ $maybeState = array( substr( $ret, 0, -1 ), $openTags );
	2845	+ }
2839	2846	if ( $ch == '&' ) {
2840	2847	$entityState = 1; // entity found, (e.g. " ")
2841	2848	} else {
2842		~~- $displayLen++; // this char is displayed~~
	2849	+ $dispLen++; // this char is displayed
2843	2850	// Add the next $max display text chars after this in one swoop...
2844		~~- $max = ( $testingEllipsis ? $length : $neLength ) - $displayLen;~~
	2851	+ $max = ( $testingEllipsis ? $length : $neLength ) - $dispLen;
2845	2852	$skipped = $this->truncate_skip( $ret, $text, "<>&", $pos + 1, $max );
2846		~~- $displayLen += $skipped;~~
	2853	+ $dispLen += $skipped;
2847	2854	$pos += $skipped;
2848	2855	}
2849	2856	}
2850	2857	}
2851	2858	}
2852		~~- if ( $displayLen == 0 ) {~~
2853		~~- return ''; // no text shown, nothing to format~~
2854		~~- }~~
2855	2859	// Close the last tag if left unclosed by bad HTML
2856	2860	$this->truncate_endBracket( $tag, $text[$textLen - 1], $tagType, $openTags );
2857	2861	while ( count( $openTags ) > 0 ) {

Sign-offs

User	Flag	Date
Werdna	inspected	18:26, 15 July 2011

Past revisions this follows-up on

Revision	Commit summary	Author	Date
r84660	(follow-up 79778) Make $wgLang->truncate function consider the length of the ...	bawolff	02:54, 24 March 2011
r84693	* Follow-up for r84660:...	aaron	18:39, 24 March 2011

Status & tagging log

12:42, 8 August 2011 😂 (talk | contribs) changed the status of r90928 [removed: new added: ok]