fix gmail replies

lib/emailreplyparser.js

@@ -142,15 +142,36 @@ Email.prototype = {
 
 		// Check for multi-line reply headers. Some clients break up
 		// the "On DATE, NAME <EMAIL> wrote:" line into multiple lines.
-		var patt = /^(On\s(\n|.)*wrote:)$/m;
-		var doubleOnPatt = /^(On\s(\n|.)*(^(> )?On\s)((\n|.)*)wrote:)$/m;
+		var englishPatt = /^(On\s(\n|.)*wrote:)$/mi;
+		var englishDoubleOnPatt = /^(On\s(\n|.)*(^(> )?On\s)((\n|.)*)wrote:)$/mi;
 
-		if(patt.test(text) && !doubleOnPatt.test(text)) {
-			var reply_header = (patt.exec(text))[0];
-			// Remove all new lines from the reply header.
+		if(englishPatt.test(text) && !englishDoubleOnPatt.test(text)) {
+			var reply_header = (englishPatt.exec(text))[0];
 			text = text.replace(reply_header, reply_header.replace(/\n/g, ' '));
 		}
 
+		var frenchPatt = /^(Le\s(\n|.)*a écrit:)$/mi;
+		var frenchDoubleOnPatt = /^(Le\s(\n|.)*(\s)((\n|.)*)a écrit :)$/mi;
+
+		if(frenchPatt.test(text) && !frenchDoubleOnPatt.test(text)) {
+			var reply_header = (frenchPatt.exec(text))[0];
+			text = text.replace(reply_header, '');
+		} else if (frenchDoubleOnPatt.test(text)) {
+			var reply_header = (frenchDoubleOnPatt.exec(text))[0];
+			text = text.replace(reply_header, '');
+		}
+
+		// Check gmail pattern
+		// DATE NAME <EMAIL>:
+		var GmailPatt = /^(\n.*?\s.*?\<[\n]?.*?\>\:)$/m;
+		if(GmailPatt.test(text)) {
+			var reply_header = (GmailPatt.exec(text))[0];
+			text = text.replace(reply_header, ' ');
+		}
+
+		text = this.remove_headers(text);
+
+
 		// The text is reversed initially due to the way we check for hidden
 		// fragments.
 		text = text.reverse();
@@ -234,6 +255,15 @@ Email.prototype = {
 		return (new RegExp('^:etorw.*nO$')).test(line);
 	},
 
+	remove_headers: function(text) {
+		var pattern = /^From:\s.*$/mi;
+		if(pattern.test(text)) {
+			var reply_header = (pattern.exec(text))[0];
+			text = text.substring(0, text.indexOf(reply_header));
+		}
+		return text;
+	},
+
 	// Builds the fragment string and reverses it, after all lines have been
 	// added.  It also checks to see if this Fragment is hidden.  The hidden
 	// Fragment check reads from the bottom to the top.
@@ -316,4 +346,4 @@ Fragment.prototype = {
 
 module.exports.EmailReplyParser = EmailReplyParser;
 
-//console.log(EmailReplyParser.read("I get proper rendering as well.\n\nSent from a magnificent torch of pixels\n\nOn Dec 16, 2011, at 12:47 PM, Corey Donohoe\n<reply@reply.github.com>\nwrote:\n\n> Was this caching related or fixed already?  I get proper rendering here.\n>\n> ![](https://img.skitch.com/20111216-m9munqjsy112yqap5cjee5wr6c.jpg)\n>\n> ---\n> Reply to this email directly or view it on GitHub:\n> https://github.com/github/github/issues/2278#issuecomment-3182418\n"));
+//console.log(EmailReplyParser.read("I get proper rendering as well.\n\nSent from a magnificent torch of pixels\n\nOn Dec 16, 2011, at 12:47 PM, Corey Donohoe\n<reply@reply.github.com>\nwrote:\n\n> Was this caching related or fixed already?  I get proper rendering here.\n>\n> ![](https://img.skitch.com/20111216-m9munqjsy112yqap5cjee5wr6c.jpg)\n>\n> ---\n> Reply to this email directly or view it on GitHub:\n> https://github.com/github/github/issues/2278#issuecomment-3182418\n"));

package.json

@@ -8,6 +8,16 @@
     "url": "https://github.com/mowens/emailreplyparser.git"
   },
   "main": "./lib/emailreplyparser",
-  "engines": { "node": ">= 0.4.0" },
-  "keywords": ["email", "parser", "emailreplyparser", "email_reply_parser"]
+  "engines": {
+    "node": ">= 0.4.0"
+  },
+  "keywords": [
+    "email",
+    "parser",
+    "emailreplyparser",
+    "email_reply_parser"
+  ],
+  "devDependencies": {
+    "underscore": "^1.8.3"
+  }
 }

test/email_reply_parser_test.js

@@ -5,13 +5,13 @@ var _  = require('underscore');
 var EmailReplyParser = require('../lib/emailreplyparser').EmailReplyParser;
 
 function get_email(name) {
-	var data = fs.readFileSync(__dirname + '/emails/' + name + '.txt', 'ascii');
+	var data = fs.readFileSync(__dirname + '/emails/' + name + '.txt', 'utf-8');
 
 	return EmailReplyParser.read(data);
 }
 
 function get_raw_email(name) {
-  	return fs.readFileSync(__dirname + '/emails/' + name + '.txt', 'ascii');
+  	return fs.readFileSync(__dirname + '/emails/' + name + '.txt', 'utf-8');
 }
 
 exports.test_reads_simple_body = function(test){
@@ -192,3 +192,22 @@ exports.test_correctly_reads_top_post_when_line_starts_with_On = function(test){
     test.ok((new RegExp('^_')).test(reply.fragments[4].to_s()));
     test.done();
 }
+
+exports.test_parse_out_send_from_multiword_gmail = function(test){
+    body = get_raw_email('test_parse_out_send_from_multiword_gmail');
+    test.equal("Awesome! I haven't had another problem with it.", EmailReplyParser.parse_reply(body));
+    test.done();
+}
+
+
+exports.test_parse_out_send_from_hotmail = function(test){
+    body = get_raw_email('email_hotmail');
+    test.equal("I replied", EmailReplyParser.parse_reply(body));
+    test.done();
+}
+
+exports.test_parse_out_send_from_french = function(test){
+    body = get_raw_email('email_french');
+    test.equal("On a 60% de test sur toute l'Infra", EmailReplyParser.parse_reply(body));
+    test.done();
+}

test/emails/email_french.txt

@@ -0,0 +1,4 @@
+On a 60% de test sur toute l'Infra
+
+Le 14 oct. 2016 7:54 PM, "www.acme.com" <
+session_7d0qe51e-4c4-4q24-c90f-8cbd5bf0d93f+in@mail.test.im> a écrit :

test/emails/email_hotmail.txt

@@ -0,0 +1,8 @@
+I replied
+
+From: examples@email.goalengine.com
+To: reply@reply.github.com
+Subject: GitHub | You received a reply
+Date: Sat, 30 Apr 2016 17:00:42 +0000
+
+Message

test/emails/test_parse_out_send_from_multiword_gmail.txt

@@ -0,0 +1,5 @@
+Awesome! I haven't had another problem with it.
+
+2016-04-30 14:17 GMT+02:00 defunkt<reply@reply.github.com>:
+
+> Loader seems to be working well.