feat: support for redacting URLs (#8)

Author: JaredBett

lib/firstNames.json

@@ -329,7 +329,6 @@
   "barbar",
   "barbara",
   "barbera",
-  "barbie",
   "barbra",
   "barrie",
   "basilia",
@@ -869,7 +868,6 @@
   "dayna",
   "daysi",
   "deadra",
-  "dean",
   "deana",
   "deandra",
   "deandrea",
@@ -994,7 +992,6 @@
   "donella",
   "donetta",
   "donette",
-  "dong",
   "donita",
   "donna",
   "donnetta",
@@ -1275,7 +1272,6 @@
   "florentina",
   "floretta",
   "floria",
-  "florida",
   "florinda",
   "florine",
   "florrie",
@@ -1295,7 +1291,6 @@
   "francisca",
   "francisco",
   "francoise",
-  "frank",
   "frankie",
   "fransisca",
   "fred",
@@ -1557,7 +1552,6 @@
   "jacelyn",
   "jacinda",
   "jacinta",
-  "jack",
   "jackeline",
   "jackelyn",
   "jacki",
@@ -2423,7 +2417,6 @@
   "lynne",
   "lynnette",
   "lynsey",
-  "ma 50",
   "mabel",
   "mabelle",
   "mable",
@@ -2818,7 +2811,6 @@
   "myrtis",
   "myrtle",
   "myung",
-  "nada",
   "nadene",
   "nadia",
   "nadine",
@@ -2844,6 +2836,7 @@
   "natalya",
   "natasha",
   "natashia",
+  "nate",
   "nathalie",
   "natisha",
   "natividad",
@@ -2872,7 +2865,6 @@
   "nettie",
   "neva",
   "nevada",
-  "nga",
   "ngan",
   "ngoc",
   "nguyet",
@@ -2924,7 +2916,6 @@
   "norma",
   "norman",
   "nubia",
-  "numbers",
   "nydia",
   "nyla",
   "obdulia",
@@ -2946,6 +2937,7 @@
   "olivia",
   "ollie",
   "olympia",
+  "omar",
   "oneida",
   "onie",
   "onita",
@@ -3069,6 +3061,7 @@
   "raylene",
   "raymond",
   "raymonde",
+  "raymund",
   "rayna",
   "rea",
   "reagan",
@@ -3457,7 +3450,6 @@
   "sixta",
   "skye",
   "slyvia",
-  "so 10",
   "socorro",
   "sofia",
   "soila",
@@ -3504,13 +3496,11 @@
   "stevie",
   "suanne",
   "sudie",
-  "sue",
   "sueann",
   "suellen",
   "suk",
   "sulema",
   "sumiko",
-  "sunday",
   "sunni",
   "susan",
   "susana",
@@ -3804,7 +3794,6 @@
   "venice",
   "venita",
   "vennie",
-  "venus",
   "veola",
   "vera",
   "verda",
@@ -3837,7 +3826,6 @@
   "vicki",
   "vickie",
   "vicky",
-  "victor",
   "victoria",
   "victorina",
   "vida",

lib/patterns.js

@@ -20,5 +20,6 @@ module.exports = {
   company: /([A-Z&][\w,]* )+(I[Nn][Cc](orporated)?|C[Oo](rp(oration)?)?|LLP|llc|LLC|plc|gmbh)\.?(\b|$)/g,
   salutation: /(^|\n(\s+)?)(dear|hi|hey|hello|greetings) ([^,:;\s]+(,? )?){1,5}[,;\n]/gi,
   valediction: /([Tt]hank(s| you)( for [^!,.]+| again)?|[Cc]heers|[Ss]incerely|[Rr]egards|[Rr]espectfully|[Bb]est|[Bb]est regards|[Yy]ours truly)\s*[!,.]?\s*([A-Z&]([\w&]+)?\.?( )?)+[^a-z]*$/g,
-  digits: /\d+/g
+  digits: /\d+/g,
+  url: /([^\s:/?#]+):\/\/([^/?#\s]*)([^?#\s]*)(\?([^#\s]*))?(#([^\s]*))?\b/
 };

test/index.test.js

@@ -167,4 +167,18 @@ defineTest('index.js', function (Redactor) {
     redactor.redact('codeA: 123, codeB: 678').should.equal('codeA: DIGITS, codeB: DIGITS');
   });
 
+  it('should replace URLs', function () {
+    redactor.redact('My homepage is http://example.com').should.equal('My homepage is URL');
+    redactor.redact('ip http://127.0.01/example.html test').should.equal('ip URL test');
+    redactor.redact('custom protocol myapp://example.com').should.equal('custom protocol URL');
+    redactor.redact('Reset password url is https://example.com/reset/password/12345').should.equal('Reset password url is URL');
+    redactor.redact('complex http://user@pass:example.com:8080/reset/password/12345?foo=bar&hi=there#/app works?').should.equal('complex URL works?');
+    redactor.redact('before http://www.example.com after').should.equal('before URL after');
+    redactor.redact('before http://www.example.com:123 after').should.equal('before URL after');
+    redactor.redact('before http://www.example.com/foo after').should.equal('before URL after');
+    redactor.redact('before http://www.example.com/foo/bar after').should.equal('before URL after');
+    redactor.redact('before http://www.example.com/foo/bar?foo=bar after').should.equal('before URL after');
+    redactor.redact('before http://www.example.com/foo/bar?foo=bar#/foo/bar after').should.equal('before URL after');
+  });
+
 });