Add linker tests, fix firstOnly behavior.

- Fix: The firstOnly option finally also works if a page contains a link to a given other page that was not currently added by the extension, i.e. that existed prior to an edit or that was manually added.

Closes #12.

includes/Linker.php

@@ -86,6 +86,14 @@ class Linker {
 				continue;
 			}
 
+			// Dealing with existing links if the firstOnly option is set:
+			// A link to the current page should only be recognized if it appears in
+			// clear text, i.e. we do not count piped links as existing links.
+			// (Similarly, by design, redirections should not be counted as existing links.)
+			if ( $limit == 1 && preg_match( '/[[' . $target->getCaseSensitiveLinkValueRegex() . ']]/' , $text ) ) {
+				continue;
+			}
+
 			// Split the page content by non-linkable sections.
 			// Credits to inhan @ StackOverflow for suggesting preg_split.
 			// See http://stackoverflow.com/questions/10672286
@@ -161,7 +169,7 @@ class Linker {
 			// we need to ignore the first letter of the page titles, as
 			// it does not matter for linking.
 			if ( strcmp( substr( $this->targetTitleText, 1 ), substr( $matches[ 0 ], 1) ) == 0 ) {
-				// Case-sensitive match: no need to bulid piped link.
+				// Case-sensitive match: no need to build piped link.
 				return '[[' . $matches[ 0 ]  . ']]';
 			} else  {
 				// Case-insensitive match: build piped link.
@@ -171,7 +179,7 @@ class Linker {
 			// If $wgCapitalLinks is false, we can use the simple variant
 			// of the callback function.
 			if ( strcmp( $this->targetTitleText, $matches[ 0 ] ) == 0 ) {
-				// Case-sensitive match: no need to bulid piped link.
+				// Case-sensitive match: no need to build piped link.
 				return '[[' . $matches[ 0 ] . ']]';
 			} else  {
 				// Case-insensitive match: build piped link.

includes/Target.php

@@ -60,6 +60,8 @@ class Target {
 	 */
 	private $config;
 
+	private $caseSensitiveLinkValueRegex;
+
 	/**
 	 * Constructs a new Target object
 	 *
@@ -103,20 +105,7 @@ class Target {
 	 * @return String regular expression for this title.
 	 */
 	public function getCaseSensitiveRegex() {
-		$regexSafeTitle = $this->getRegexSafeTitle();
+		return $this->buildRegex( $this->getCaseSensitiveLinkValueRegex() );
-
-		// Depending on the $config->capitalLinks setting,
-		// the title has to be searched for either in a strictly case-sensitive
-		// way, or in a 'fuzzy' way where the first letter of the title may
-		// be either case.
-		//
-		if ( $this->config->capitalLinks && ( $regexSafeTitle[0] != '\\' )) {
-			$searchTerm = '((?i)' . $regexSafeTitle[0] . '(?-i)' . substr($regexSafeTitle, 1) . ')';
-		}	else {
-			$searchTerm = '(' . $regexSafeTitle . ')';
-		}
-
-		return $this->buildRegex( $searchTerm );
 	}
 
 	/**
@@ -138,6 +127,27 @@ class Target {
 		return '/(?<![\:\.\@\/\?\&])' . $this->wordStart . $searchTerm . $this->wordEnd . '/S';
 	}
 
+	/**
+	 * Gets the (cached) regex for the link value.
+	 *
+	 * Depending on the $config->capitalLinks setting, the title has to be
+	 * searched for either in a strictly case-sensitive way, or in a 'fuzzy' way
+	 * where the first letter of the title may be either case.
+	 *
+	 * @return String regular expression pattern for the link value.
+	 */
+	public function getCaseSensitiveLinkValueRegex() {
+		if ( $this->caseSensitiveLinkValueRegex === null ) {
+			$regexSafeTitle = $this->getRegexSafeTitle();
+			if ( $this->config->capitalLinks && ( $regexSafeTitle[0] != '\\' )) {
+				$this->caseSensitiveLinkValueRegex = '((?i)' . $regexSafeTitle[0] . '(?-i)' . substr($regexSafeTitle, 1) . ')';
+			}	else {
+				$this->caseSensitiveLinkValueRegex = '(' . $regexSafeTitle . ')';
+			}
+		}
+		return $this->caseSensitiveLinkValueRegex;
+	}
+
 	/**
 	 * Returns the \Content of the target page.
 	 *

tests/phpunit/LinkerTest.php

@@ -1,5 +1,13 @@
 <?php
 /**
+ * Unit tests for the LinkTitles\Linker class.
+ *
+ * The test class is prefixed with 'LinkTitles' to avoid a naming collision
+ * with a class that exists in the MediaWiki core.
+ *
+ * Ideally the test classes should be namespaced, but when you do that, they
+ * will no longer be automatically discovered.
+ *
  * @group bovender
  * @group Database
  */
@@ -17,12 +25,13 @@ class LinkTitlesLinkerTest extends LinkTitles\TestCase {
   public function testLinkContentSmartMode( $capitalLinks, $smartMode, $input, $expectedOutput) {
     $this->setMwGlobals( 'wgCapitalLinks', $capitalLinks );
     $config = new LinkTitles\Config();
+    $config->firstOnly = false;
     $config->smartMode = $smartMode;
     $linker = new LinkTitles\Linker( $config );
     $this->assertSame( $expectedOutput, $linker->linkContent( $this->title, $input ));
   }
 
-  public static function provideLinkContentSmartModeData() {
+  public function provideLinkContentSmartModeData() {
     return [
       [
         true, // wgCapitalLinks
@@ -86,4 +95,39 @@ class LinkTitlesLinkerTest extends LinkTitles\TestCase {
       ],
     ];
   }
+
+  /**
+   * @dataProvider provideLinkContentFirstOnlyData
+   */
+  public function testLinkContentFirstOnly( $firstOnly, $input, $expectedOutput ) {
+    $config = new LinkTitles\Config();
+    $config->firstOnly = $firstOnly;
+    $linker = new LinkTitles\Linker( $config );
+    $this->assertSame( $expectedOutput, $linker->linkContent( $this->title, $input ));
+  }
+
+  public function provideLinkContentFirstOnlyData() {
+    return [
+      [
+        false, // firstOnly
+        'With firstOnly = false, link target is a link target multiple times',
+        'With firstOnly = false, [[link target]] is a [[link target]] multiple times'
+      ],
+      [
+        false, // firstOnly
+        'With firstOnly = false, [[link target]] is a link target multiple times',
+        'With firstOnly = false, [[link target]] is a [[link target]] multiple times'
+      ],
+      [
+        true, // firstOnly
+        'With firstOnly = true, link target is a link target only once',
+        'With firstOnly = true, [[link target]] is a link target only once'
+      ],
+      [
+        true, // firstOnly
+        'With firstOnly = true, [[link target]] is a link target only once',
+        'With firstOnly = true, [[link target]] is a link target only once'
+      ],
+    ];
+  }
 }