only count email addresses as one word

Author: antislice

Diff for: README.md

@@ -6,7 +6,7 @@ JavaScript port of [TextStatistics.php](https://github.com/DaveChild/Text-Statis
 I've done what I think is a reasonably faithful port. Documentation incoming!
 I removed a lot of the original comments during the port, but seeing as the API remained largely the same, I'll add them in shortly.
 
-Same goes for a test suite - I'll get something working in node in a bit. :)
+The beginning of a test suite in [Mocha](https://mochajs.org/) is here, covering cleaning the text and some cases of word and sentence counting.
 
 ## Installation
 

Diff for: index.js

@@ -19,6 +19,7 @@
 			.replace(/[,:;()\/&+]|\-\-/g, " ")				// Replace commas, hyphens etc (count them as spaces)
 			.replace(/[\.!?]/g, ".")					// Unify terminators
 			.replace(/^\s+/, "")						// Strip leading whitespace
+			.replace(/[\.]?(\w+)[\.]?(\w+)@(\w+)[\.](\w+)[\.]?/g, "$1$2@$3$4")	// strip periods in email addresses (so they remain counted as one word)
 			.replace(/[ ]*(\n|\r\n|\r)[ ]*/g, ".")	// Replace new lines with periods
 			.replace(/([\.])[\.]+/g, ".")			// Check for duplicated terminators
 			.replace(/[ ]*([\.])/g, ". ")				// Pad sentence terminators
@@ -28,7 +29,6 @@
 		if(text.slice(-1) != '.') {
 			text += "."; // Add final terminator, just in case it's missing.
 		}
-		
 		return text;
 	}
 

Diff for: test/testCleanText.js

@@ -102,6 +102,33 @@ describe('TextStatistics', function() {
       });
     });
 
+    context('stripping periods from email addresses', function() {
+      it('should replace a single period', function() {
+        var ts = TextStatistics('[email protected]');
+        assert.equal(ts.text, 'textstatistics@examplecom.');
+      });
+
+      it('should replace a single period in the first part', function() {
+        var ts = TextStatistics('[email protected]');
+        assert.equal(ts.text, 'textstatistics@examplecom.');
+      });
+
+      it('should replace two periods in the first part', function() {
+        var ts = TextStatistics('[email protected]');
+        assert.equal(ts.text, 'textstatistics@examplecom.');
+      });
+
+      it('should replace periods with a subdomain', function() {
+        var ts = TextStatistics('[email protected]');
+        assert.equal(ts.text, 'textstatistics@testexamplecom.');
+      });
+
+      it('should replace periods with a subdomain and before the @', function() {
+        var ts = TextStatistics('[email protected]');
+        assert.equal(ts.text, 'textstatistics@testexamplecom.');
+      });
+    });
+
     context('replacing non-terminator punctuation', function() {
       it('should replace commas with spaces', function() {
         var ts = TextStatistics('Hello, hi, friend.');