started to implemented the regexes from the current LibreJS into the contact finder

1 files changed, 193 insertions, 29 deletions

diff --git a/contact_finder.js b/contact_finder.js
index 759b3b0..69b9254 100644
--- a/contact_finder.js
+++ b/contact_finder.js
@@ -7,6 +7,100 @@
 */
 // Now, the contact finder will load on every page and you can test it where ever you want.
 
+
+//*********************************************************************************************
+//Regexes taken from "contact_regex.js" in the current LibreJS
+//Copyright (C) 2011, 2012, 2014 Loic J. Duros
+//Copyright (C) 2014, 2015 Nik Nyby
+
+// email address regexp
+var reEmail = /^mailto\:(admin|feedback|webmaster|info|contact|support|comments|team|help)\@[a-z0-9.\-]+\.[a-z]{2,4}$/i;
+
+var reAnyEmail = /^mailto\:.*?\@[a-z0-9\.\-]+\.[a-z]{2,4}$/i;
+
+// twitter address regexp
+var reTwitter = /twitter\.com\/(\!?#\/)?[a-z0-9]*/i;
+
+// identi.ca address regexp
+var reIdentiCa = /identi\.ca\/(?!notice\/)[a-z0-9]*/i;
+
+/**
+ * contactSearchStrings
+ * Contains arrays of strings classified by language
+ * and by degree of certainty.
+ */
+var contactStr = {
+    'da': {
+        'certain': [
+            '^[\\s]*Kontakt os[\\s]*$',
+            '^[\\s]*Email Os[\\s]*$',
+            '^[\\s]*Kontakt[\\s]*$'
+        ],
+        'probable': ['^[\\s]Kontakt', '^[\\s]*Email'],
+        'uncertain': [
+            '^[\\s]*Om Us',
+            '^[\\s]*Om',
+            'Hvem vi er'
+        ]
+    },
+    'en': {
+        'certain': [
+            '^[\\s]*Contact Us[\\s]*$',
+            '^[\\s]*Email Us[\\s]*$',
+            '^[\\s]*Contact[\\s]*$',
+            '^[\\s]*Feedback[\\s]*$',
+            '^[\\s]*Web.?site Feedback[\\s]*$'
+        ],
+        'probable': ['^[\\s]Contact', '^[\\s]*Email'],
+        'uncertain': [
+            '^[\\s]*About Us',
+            '^[\\s]*About',
+            'Who we are',
+            'Who I am',
+            'Company Info',
+            'Customer Service'
+        ]
+    },
+    'es': {
+        'certain': [
+            '^[\\s]*contáctenos[\\s]*$',
+            '^[\\s]*Email[\\s]*$'
+        ],
+        'probable': ['^[\\s]contáctenos', '^[\\s]*Email'],
+        'uncertain': [
+            'Acerca de nosotros'
+        ]
+    },
+    'fr': {
+        'certain': [
+            '^[\\s]*Contactez nous[\\s]*$',
+            '^[\\s]*(Nous )?contacter[\\s]*$',
+            '^[\\s]*Email[\\s]*$',
+            '^[\\s]*Contact[\\s]*$',
+            '^[\\s]*Commentaires[\\s]*$'
+        ],
+        'probable': ['^[\\s]Contact', '^[\\s]*Email'],
+        'uncertain': [
+            '^[\\s]*(A|À) propos',
+            'Qui nous sommes',
+            'Qui suis(-| )?je',
+            'Info',
+            'Service Client(e|è)le'
+        ]
+    }
+};
+
+var usaPhoneNumber = new RegExp(/(?:\+ ?1 ?)?\(?[2-9]{1}[0-9]{2}\)?(?:\-|\.| )?[0-9]{3}(?:\-|\.| )[0-9]{4}(?:[^0-9])/mg);
+
+//*********************************************************************************************
+
+
+/**
+*
+*	Creates a transparent floating button from a name string and a callback
+*
+*
+*/
 var button_i = 0;
 if(document.getElementById("abc123_main_div") !== null){
 	document.getElementById("abc123_main_div").remove();
@@ -22,55 +116,125 @@ function new_debug_button(name_text,callback){
 	button_i = button_i + 1;
 }
 /**
+*	returns input with all elements not of type string removed
+*/
+function remove_not_str(a){
+	var new_a = [];	
+	for(var i in a){
+		if(typeof(a[i]) == "string"){
+			new_a.push(a[i])
+		}	
+	}
+	return new_a;
+}
+/**
+*	Tests all links on the page for regexes under a certain certainty level.
+*
+*	Will return either the first regex match from the selected certainty level or all regexes that
+*	match on that certainty level.
+*
+*	certainty_lvl can be "certain" > "probable" > "uncertain" 
+*/
+function attempt(certainty_lvl, first=true){
+	// There needs to be some kind of max so that people can't troll by for example leaving a comment with a bunch of emails
+	// to cause LibreJS users to slow down.
+	var fail_flag = true;
+	var flag;
+	var matches = [];
+	var result = [];	
+	var str_under_test = "";	
+	for(var i in document.links){
+		if( typeof(document.links[i].innerText) != "string" || typeof(document.links[i].href) != "string"){
+			continue;
+		}
+		str_under_test = document.links[i].innerText + " " + document.links[i].href;
+		flag = true;
+		for(var j in contactStr){
+			for(var k in contactStr[j][certainty_lvl]){
+				if(flag){
+					result = [];				
+					result = str_under_test.match(new RegExp(contactStr[j][certainty_lvl][k],"g"));
+					result = remove_not_str(result);
+					if(result !== undefined && typeof(result[0]) == "string" ){				
+						if(first){
+							return {"fail":false,"result":document.links[i]};
+						} else{
+							console.log(document.links[i].href + " matched " + contactStr[j][certainty_lvl][k]);
+							matches.push(document.links[i]);
+							fail_flag = false;
+							flag = false;
+						}
+					}
+				}
+			}
+		}
+	}
+	console.log(matches);						
+	return {"fail":fail_flag,"result":matches};
+}
+
+/**
 *	"LibreJS detects contact pages, email addresses that are likely to be owned by the 
 *	maintainer of the site, Twitter and identi.ca links, and phone numbers."
 */
 function find_contacts(){
-	var all = document.documentElement.innerHTML;
-	var emails = [];	
-	emails.push(all.match(/\S+@\S+\.\S+\b/g));
-	// 1.555.123.4567
-	//+1.555.123.4567
-	var phone_num = []; 
-	phone_num.push(all.match(/(\d{1,3}\.)?(\d\d\d)\.(\d\d\d)\.(\d\d\d\d)/g));
-	// 1-555-123-4567
-	//+1-555-123-4567
-	phone_num.push(all.match(/(\+?\d)?([\-|\.])(\d\d\d)\2(\d\d\d)\2(\d\d\d\d)/g));
-	// +15554567890
-	phone_num.push(all.match(/\+?\d{10,15}\b/g));
-	// twitter handles 
-	var twitter = [];
-	twitter.push(all.match(/@\w{3,15}\b/g));
-	// twitter links
+	var all = document.documentElement.innerText;
+	var phone_num = [];
 	var twitlinks = [];
-	twitlinks.push(all.match(/twitter\.com\/\w{3,15}\b/g));
-	// identi.ca link	
-	// 25 is my guess at the max username length (I don't actually know)
 	var identi = [];
-	identi.push(all.match(/identi\.ca\/\w{3,25}\b/g));
-	// Attempt to find contact pages
-	var contact_pages = [];	
-	var links = document.getElementsByTagName("a");
-	for(i in links){
-		if(links[i].href !== undefined && links[i].href.indexOf("contact") != -1){
-			contact_pages.push(links[i]);
+	var contact_pages = [];
+	console.log("certain:");
+	var res = attempt("certain");
+	var flag = true;	
+	if(res["fail"] == false){
+		console.log("certain contact found:" + res["result"]);
+		res = res["result"];
+		flag = false;	
+	}
+	if(flag){
+		console.log("probable:");
+		res = attempt("probable");
+		if(res["fail"] == false){
+			console.log("probable contact found:" + res["result"]);
+			res = res["result"];
+			flag = false;	
+		}
+	}
+	if(flag){
+		console.log("uncertain:");
+		res = attempt("uncertain");
+		console.log(res);
+		if(res["fail"] == false){
+			console.log("uncertain contact found:" + res["result"]);
+			res = res["result"];
+			flag = false;		
 		}
 	}
+	if(flag){
+		console.log("No contact found");
+	}
+	console.log("final result:");
+	console.log(res);
+	/*
 	console.log("********************************************************");
 	console.log("%c RESULTS: ","color: #dd0000;");
 	console.log("%c " + phone_num.length + "%c phone numbers","color: red;","color: purple;");
-	console.log("%c " + twitter.length + "%c twitter handles","color: red;","color: purple;");
 	console.log("%c " + twitlinks.length + "%c twitter links","color: red;","color: purple;");
 	console.log("%c " + identi.length + "%c identi.ca links","color: red;","color: purple;");
 	console.log("%c " + contact_pages.length + "%c possible contact pages","color: red;","color: purple;");
 	console.log("********************************************************");
+	*/
 
 
 
-
+}
+// need to have this so the handler doesn't take too long
+function handler(){
+	find_contacts();
+	return 0;
 }
 
-new_debug_button("Complain to website",find_contacts);
+new_debug_button("Complain to website",handler);
 new_debug_button("Remove these buttons",function(){
 	if(document.getElementById("abc123_main_div") !== null){
 		document.getElementById("abc123_main_div").remove();