fix for URLs preceded with text are not linked correctly gregjacobs#159

src/matcher/url-matcher.ts

@@ -201,6 +201,25 @@ export class UrlMatcher extends Matcher {
 				}
 			}
 
+			// The autolinker accepts many characters in a url's scheme (like `fake://test.com`).
+			// However, in cases where a URL is missing whitespace before an obvious link,
+			// (for example: `nowhitespacehttp://www.test.com`), we only want the match to start
+			// at the http:// part. We will check if the match contains a common scheme and then 
+			// shift the match to start from there. 		
+			const foundCommonScheme = [ 'http://', 'https://' ].find(
+				(commonScheme) => !!schemeUrlMatch && schemeUrlMatch.indexOf( commonScheme ) !== -1
+			);
+			if ( foundCommonScheme  ) {
+				// If we found an overmatched URL, we want to find the index
+				// of where the match should start and shift the match to
+				// start from the beginning of the common scheme
+				const indexOfSchemeStart = matchStr.indexOf( foundCommonScheme );
+
+				matchStr = matchStr.substr( indexOfSchemeStart );
+				schemeUrlMatch = schemeUrlMatch.substr( indexOfSchemeStart );
+				offset = offset + indexOfSchemeStart;
+			}
+
 			let urlMatchType: UrlMatchTypeOptions = schemeUrlMatch ? 'scheme' : ( wwwUrlMatch ? 'www' : 'tld' ),
 			    protocolUrlMatch = !!schemeUrlMatch;
 

tests/autolinker-url.spec.ts

@@ -304,6 +304,13 @@ describe( "Autolinker Url Matching -", () => {
 				expect( result5 ).toBe( 'do not link first char: .<a href="aa://example">aa://example</a>' );
 			} );
 
+			it( "should autolink protocol starting at http:// or http:// if URL is preceded with text", function() {
+				let result1 = autolinker.link( 'link this: xxxhttp://example.com' );
+				expect( result1 ).toBe( 'link this: xxx<a href="http://example.com">example.com</a>' );
+
+				let result2 = autolinker.link( 'link this: abchttps://www.example.com' );
+				expect( result2 ).toBe( 'link this: abc<a href="https://www.example.com">example.com</a>' );
+			} );
 
 			it( "should NOT autolink possible URLs with the 'javascript:' URI scheme", function() {
 				let result = autolinker.link( "do not link javascript:window.alert('hi') please" );

tests/matcher/url-matcher.spec.ts

@@ -180,6 +180,18 @@ describe( "Autolinker.matcher.Url", function() {
 				expect( Date.now() - start ).toBeLessThan( 100 );
 			} );
 
+			it( 'if scheme contains a common protocol, should only begin match from start of common protocol', function() {
+				let matches1 = matcher.parseMatches( 'text with a typohttp://www.example.com mid-sentence' );
+
+				expect( matches1.length ).toBe( 1 );
+				MatchChecker.expectUrlMatch( matches1[ 0 ], 'http://www.example.com', 16 );
+
+				let matches2 = matcher.parseMatches( 'text with a typohttps://www.example.com mid-sentence' );
+
+				expect( matches2.length ).toBe( 1 );
+				MatchChecker.expectUrlMatch( matches2[ 0 ], 'https://www.example.com', 16 );
+			} );
+
 		} );
 
 	} );