Fix issue with unquoted anchor href's that contain an '=' sign (as part of the query string) causing the HTML tag to not properly parse

Fixes #253

Author: gregjacobs

package.json

@@ -1,6 +1,6 @@
 {
   "name": "autolinker",
-  "version": "3.0.0",
+  "version": "3.0.1",
   "description": "Utility to automatically link the URLs, email addresses, phone numbers, hashtags, and mentions (Twitter, Instagram) in a given block of text/HTML",
   "main": "./dist/commonjs/index.js",
   "typings": "./dist/commonjs/index.d.ts",

src/htmlParser/parse-html.ts

@@ -81,6 +81,7 @@ export function parseHtml( html: string, { onOpenTag, onCloseTag, onText, onComm
 		var char = html.charAt( charIdx );
 
 		// For debugging: search for other "For debugging" lines
+		// ALSO: Temporarily remove the 'const' keyword on the State enum
 		// table.push( 
 		// 	[ charIdx, char, State[ state ], currentDataIdx, currentTag.idx, currentTag.idx === -1 ? '' : currentTag.type ] 
 		// );
@@ -113,6 +114,7 @@ export function parseHtml( html: string, { onOpenTag, onCloseTag, onText, onComm
 		}
 
 		// For debugging: search for other "For debugging" lines
+		// ALSO: Temporarily remove the 'const' keyword on the State enum
 		// table.push( 
 		// 	[ charIdx, char, State[ state ], currentDataIdx, currentTag.idx, currentTag.idx === -1 ? '' : currentTag.type ] 
 		// );
@@ -125,7 +127,7 @@ export function parseHtml( html: string, { onOpenTag, onCloseTag, onText, onComm
 	}
 
 	// For debugging: search for other "For debugging" lines
-	//console.log( '\n' + table.toString() );
+	// console.log( '\n' + table.toString() );
 
 
 
@@ -356,12 +358,6 @@ export function parseHtml( html: string, { onOpenTag, onCloseTag, onText, onComm
 			// start of another tag (ignore the previous, incomplete one)
 			startNewTag();
 
-		} else if( quoteRe.test( char ) || /[=`]/.test( char ) ) {
-			// "Parse error" characters that, according to the spec, should be
-			// appended to the attribute value, but we'll treat these characters
-			// as not forming a real HTML tag
-			resetToDataState();
-
 		} else {
 			// Any other character, treat it as part of the attribute value
 		}

tests/autolinker.spec.ts

@@ -141,6 +141,15 @@ describe( "Autolinker", function() {
 			} );
 
 
+			it( "when unquoted anchor href's exist, should NOT automatically link the text inside", function() {
+				let result = autolinker.link( '<p>Joe went to <a href=http://www.yahoo.com>yahoo</a></p>' );
+				expect( result ).toBe( '<p>Joe went to <a href=http://www.yahoo.com>yahoo</a></p>' );
+
+				let result2 = autolinker.link( '<p>Joe went to <a href=http://www.yahoo.com?query=1>yahoo</a></p>' );
+				expect( result2 ).toBe( '<p>Joe went to <a href=http://www.yahoo.com?query=1>yahoo</a></p>' );
+			} );
+
+
 			it( "should NOT automatically link URLs within self-closing tags", function() {
 				let result = autolinker.link( 'Just a flower image <img src="https://farm9.staticflickr.com/8378/8578790632_83c6471f3f_b.jpg" />' );
 				expect( result ).toBe( 'Just a flower image <img src="https://farm9.staticflickr.com/8378/8578790632_83c6471f3f_b.jpg" />' );

tests/htmlParser/parse-html.spec.ts

@@ -136,6 +136,28 @@ describe( "Autolinker.htmlParser.HtmlParser", () => {
 		} );
 
 
+		it( `should handle html tags with unqouted attributes`, () => {
+			let nodes = parseHtmlAndCapture( `<div id=hi class=some-class align=center>Test</div>` );
+			
+			expect( nodes ).toEqual( [
+				{ type: 'openTag', tagName: 'div', offset: 0 },
+				{ type: 'text', text: 'Test', offset: 41 },
+				{ type: 'closeTag', tagName: 'div', offset: 45 }
+			] );
+		} );
+
+
+		it( `should handle anchor tags with unqouted href's and query strings (Issue #263)`, () => {
+			let nodes = parseHtmlAndCapture( `<a href=https://www.facebook.com/hashtag/muntcentrum?epa=HASHTAG>#Muntcentrum</a>` );
+			
+			expect( nodes ).toEqual( [
+				{ type: 'openTag', tagName: 'a', offset: 0 },
+				{ type: 'text', text: '#Muntcentrum', offset: 65 },
+				{ type: 'closeTag', tagName: 'a', offset: 77 }
+			] );
+		} );
+
+
 		it( `when there are two << characters, the first one should be ignored
 			 if the second one forms a tag`,
 		() => {