Improve URL recognition (sacarasc)

2 files changed, 39 insertions, 0 deletions

diff --git a/src/common/url.c b/src/common/url.c
index 9fa2d75c..94a03401 100644
--- a/src/common/url.c
+++ b/src/common/url.c
@@ -166,6 +166,29 @@ url_check_word (char *word, int len)
 		{ D(".html") },
 		{ D(".info") },
 		{ D(".name") },
+		/* Some extra common suffixes.
+		foo.blah/baz.php etc should work now, rather than
+		needing  http:// at the beginning. */
+		{ D(".php") },
+		{ D(".htm") },
+		{ D(".aero") },
+		{ D(".asia") },
+		{ D(".biz") },
+		{ D(".cat") },
+		{ D(".coop") },
+		{ D(".int") },
+		{ D(".jobs") },
+		{ D(".mobi") },
+		{ D(".museum") },
+		{ D(".name") },
+		{ D(".pro") },
+		{ D(".tel") },
+		{ D(".travel") },
+		{ D(".xxx") },
+		{ D(".asp") },
+		{ D(".aspx") },
+		{ D(".shtml") },
+		{ D(".xml") },
 	};
 #undef D
 	const char *at, *dot;
diff --git a/src/common/util.c b/src/common/util.c
index fb4cd23e..27f0aa40 100644
--- a/src/common/util.c
+++ b/src/common/util.c
@@ -953,6 +953,7 @@ static const domain_t domain[] =
 		{"AC", N_("Ascension Island") },
 		{"AD", N_("Andorra") },
 		{"AE", N_("United Arab Emirates") },
+		{"AERO", N_("Aviation-Related Fields") },
 		{"AF", N_("Afghanistan") },
 		{"AG", N_("Antigua and Barbuda") },
 		{"AI", N_("Anguilla") },
@@ -964,6 +965,7 @@ static const domain_t domain[] =
 		{"AR", N_("Argentina") },
 		{"ARPA", N_("Reverse DNS") },
 		{"AS", N_("American Samoa") },
+		{"ASIA", N_("Asia-Pacific Region") },
 		{"AT", N_("Austria") },
 		{"ATO", N_("Nato Fiel") },
 		{"AU", N_("Australia") },
@@ -991,6 +993,7 @@ static const domain_t domain[] =
 		{"BY", N_("Belarus") },
 		{"BZ", N_("Belize") },
 		{"CA", N_("Canada") },
+		{"CAT", N_("Catalan") },
 		{"CC", N_("Cocos Islands") },
 		{"CD", N_("Democratic Republic of Congo") },
 		{"CF", N_("Central African Republic") },
@@ -1003,6 +1006,7 @@ static const domain_t domain[] =
 		{"CN", N_("China") },
 		{"CO", N_("Colombia") },
 		{"COM", N_("Internic Commercial") },
+		{"COOP", N_("Cooperatives") },
 		{"CR", N_("Costa Rica") },
 		{"CS", N_("Serbia and Montenegro") },
 		{"CU", N_("Cuba") },
@@ -1010,6 +1014,7 @@ static const domain_t domain[] =
 		{"CX", N_("Christmas Island") },
 		{"CY", N_("Cyprus") },
 		{"CZ", N_("Czech Republic") },
+		{"DD", N_("East Germany") },
 		{"DE", N_("Germany") },
 		{"DJ", N_("Djibouti") },
 		{"DK", N_("Denmark") },
@@ -1072,6 +1077,7 @@ static const domain_t domain[] =
 		{"JE", N_("Jersey") },
 		{"JM", N_("Jamaica") },
 		{"JO", N_("Jordan") },
+		{"JOBS", N_("Company Jobs") },
 		{"JP", N_("Japan") },
 		{"KE", N_("Kenya") },
 		{"KG", N_("Kyrgyzstan") },
@@ -1098,6 +1104,7 @@ static const domain_t domain[] =
 		{"MA", N_("Morocco") },
 		{"MC", N_("Monaco") },
 		{"MD", N_("Moldova") },
+		{"ME", N_("Montenegro") },
 		{"MED", N_("United States Medical") },
 		{"MG", N_("Madagascar") },
 		{"MH", N_("Marshall Islands") },
@@ -1107,18 +1114,21 @@ static const domain_t domain[] =
 		{"MM", N_("Myanmar") },
 		{"MN", N_("Mongolia") },
 		{"MO", N_("Macau") },
+		{"MOBI", N_("Mobile Devices") },
 		{"MP", N_("Northern Mariana Islands") },
 		{"MQ", N_("Martinique") },
 		{"MR", N_("Mauritania") },
 		{"MS", N_("Montserrat") },
 		{"MT", N_("Malta") },
 		{"MU", N_("Mauritius") },
+		{"MUSEUM", N_("Museums") },
 		{"MV", N_("Maldives") },
 		{"MW", N_("Malawi") },
 		{"MX", N_("Mexico") },
 		{"MY", N_("Malaysia") },
 		{"MZ", N_("Mozambique") },
 		{"NA", N_("Namibia") },
+		{"NAME", N_("Individual's Names") },
 		{"NC", N_("New Caledonia") },
 		{"NE", N_("Niger") },
 		{"NET", N_("Internic Network") },
@@ -1143,6 +1153,7 @@ static const domain_t domain[] =
 		{"PM", N_("St. Pierre and Miquelon") },
 		{"PN", N_("Pitcairn") },
 		{"PR", N_("Puerto Rico") },
+		{"PRO", N_("Professions") },
 		{"PS", N_("Palestinian Territory") },
 		{"PT", N_("Portugal") },
 		{"PW", N_("Palau") },
@@ -1151,6 +1162,7 @@ static const domain_t domain[] =
 		{"RE", N_("Reunion") },
 		{"RO", N_("Romania") },
 		{"RPA", N_("Old School ARPAnet") },
+		{"RS", N_("Serbia") },
 		{"RU", N_("Russian Federation") },
 		{"RW", N_("Rwanda") },
 		{"SA", N_("Saudi Arabia") },
@@ -1168,6 +1180,7 @@ static const domain_t domain[] =
 		{"SN", N_("Senegal") },
 		{"SO", N_("Somalia") },
 		{"SR", N_("Suriname") },
+		{"SS", N_("South Sudan") },
 		{"ST", N_("Sao Tome and Principe") },
 		{"SU", N_("Former USSR") },
 		{"SV", N_("El Salvador") },
@@ -1175,6 +1188,7 @@ static const domain_t domain[] =
 		{"SZ", N_("Swaziland") },
 		{"TC", N_("Turks and Caicos Islands") },
 		{"TD", N_("Chad") },
+		{"TEL", N_("Internet Communication Services") },
 		{"TF", N_("French Southern Territories") },
 		{"TG", N_("Togo") },
 		{"TH", N_("Thailand") },
@@ -1186,6 +1200,7 @@ static const domain_t domain[] =
 		{"TO", N_("Tonga") },
 		{"TP", N_("East Timor") },
 		{"TR", N_("Turkey") },
+		{"TRAVEL", N_("Travel and Tourism") },
 		{"TT", N_("Trinidad and Tobago") },
 		{"TV", N_("Tuvalu") },
 		{"TW", N_("Taiwan") },
@@ -1205,6 +1220,7 @@ static const domain_t domain[] =
 		{"VU", N_("Vanuatu") },
 		{"WF", N_("Wallis and Futuna Islands") },
 		{"WS", N_("Samoa") },
+		{"XXX", N_("Adult Entertainment") },
 		{"YE", N_("Yemen") },
 		{"YT", N_("Mayotte") },
 		{"YU", N_("Yugoslavia") },