Merge pull request #703 from hippietrail/countries_and_capitals

feat: capitalize multi-word country and capital city names

harper-core/dictionary.dict

@@ -684,7 +684,8 @@ Astoria/2M
 Astrakhan/21M
 AstroTurf/1M
 Asturias/2M
-Asuncion/M
+Asuncion/2M
+Asunción/2M
 Aswan/2M
 Atacama/2M
 Atahualpa/M
@@ -1273,6 +1274,7 @@ Boer/12SM
 Boethius/2M
 Bogart/2M
 Bogota/2M
+Bogotá/2M
 Bohemia/21M
 Bohemian/152SM
 Bohr/2M
@@ -2045,6 +2047,7 @@ Chiquita/M
 Chirico/2M
 Chisholm/2M
 Chisinau/2M
+Chișinău/2M
 Chittagong/2M
 Chivas/M
 Chloe/2M
@@ -2432,6 +2435,7 @@ Cunard/2M
 Cunningham/2M
 Cupid/2M
 Curacao/2M
+Curaçao/2M
 Curie/2M
 Curitiba/2M
 Currier/2M
@@ -5849,7 +5853,8 @@ Lollobrigida/2M
 Lombard/125M
 Lombardi/2M
 Lombardy/2M
-Lome/M
+Lome/2M
+Lomé/2M
 Lompoc/2M
 Lon/2M
 London/2MRZ
@@ -6144,6 +6149,7 @@ Maldives/29M
 Maldivian/15MS
 Maldonado/2M
 Male/2M
+Malé/2M
 Mali/21M
 Malian/15SM
 Malibu/2M
@@ -7269,7 +7275,8 @@ Nosferatu/1M
 Nostradamus/21M
 Nottingham/2M
 Nouakchott/2M
-Noumea/M
+Noumea/2M
+Nouméa/2M
 Nov/2M
 Nova/21M
 Novartis/M
@@ -8321,6 +8328,7 @@ Revlon/2M
 Rex/2M
 Reyes/2M
 Reykjavik/2M
+Reykjavík/2M
 Reyna/M
 Reynaldo/M
 Reynolds/2M
@@ -9836,6 +9844,7 @@ Torrance/2M
 Torrens/2M
 Torres/2M
 Torricelli/2M
+Tórshavn/2M
 Tortola/2M
 Tortuga/2M
 Torvalds/2M
@@ -9947,6 +9956,7 @@ Turkestan/2M
 Turkey/215M
 Turkic/52MS
 Turkish/25M
+Türkiye/2M
 Turkmenistan/2M
 Turlock/2M
 Turner/21M
@@ -10639,6 +10649,7 @@ Yank/1SM
 Yankee/14SM
 Yaobang/M
 Yaounde/2M
+Yaoundé/2M
 Yaqui/12M
 Yaren/2
 Yaroslavl/2M

harper-core/src/linting/matcher.rs

@@ -177,6 +177,77 @@ impl Matcher {
             "World","War","Ii" => "World War II"
         });
 
+        // countries and capitals with special casing or punctuation
+        triggers.extend(pt! {
+           "andorra","la","vella" => "Andorra la Vella",
+           "Andorra","la","vella" => "Andorra la Vella",
+           "Andorra","La","Vella" => "Andorra la Vella",
+           "guinea","bissau" => "Guinea-Bissau",
+           "Guinea","bissau" => "Guinea-Bissau",
+           "Guinea","Bissau" => "Guinea-Bissau",
+           "ndjamena" => "N'Djamena",
+           "Ndjamena" => "N'Djamena",
+           "n'djamena" => "N'Djamena",
+           "N'djamena" => "N'Djamena",
+           "port","au","prince" => "Port-au-Prince",
+           "Port","au","prince" => "Port-au-Prince",
+           "Port","Au","Prince" => "Port-au-Prince",
+           // port-au-prince won't work here because the left side has hyphens
+           // Port-au-prince ditto
+           // Port-Au-Prince ditto
+           "porto","novo" => "Porto-Novo",
+           "Porto","novo" => "Porto-Novo",
+           "Porto","Novo" => "Porto-Novo",
+           "st","georges" => "St. George's",
+           // "st.","georges" => "St. George's",
+           "st","george's" => "St. George's",
+           // "st.","george's" => "St. George's",
+           "St","georges" => "St. George's",
+           // "St.","georges" => "St. George's",
+           "St","george's" => "St. George's",
+           // "St.","george's" => "St. George's",
+           "St","Georges" => "St. George's",
+           // "St.","Georges" => "St. George's",
+           "St","George's" => "St. George's"
+        });
+
+        // countries and capitals with accents and diacritics
+        triggers.extend(pt! {
+            "asuncion" => "Asunción",
+            "Asuncion" => "Asunción",
+            "chisinau" => "Chișinău",
+            "Chisinau" => "Chișinău",
+            "bogota" => "Bogotá",
+            "Bogota" => "Bogotá",
+            "curacao" => "Curaçao",
+            "curacao" => "Curaçao",
+            "lome" => "Lomé",
+            "Lome" => "Lomé",
+            "male" => "Malé",
+            "Male" => "Malé",
+            "noumea" => "Nouméa",
+            "Noumea" => "Nouméa",
+            "nukualofa" => "Nukuʻalofa",
+            "Nukualofa" => "Nukuʻalofa",
+            "nuku'alofa" => "Nukuʻalofa",
+            "Nuku'alofa" => "Nukuʻalofa",
+            "reykjavik" => "Reykjavík",
+            "Reykjavik" => "Reykjavík",
+            "san","jose" => "San José",
+            "San","jose" => "San José",
+            "sao","tome" => "São Tomé",
+            "Sao","Tome" => "São Tomé",
+            "sao","tome","and","principe" => "São Tomé and Príncipe",
+            "Sao","Tome","and","Principe" => "São Tomé and Príncipe",
+            "Sao","Tome","And","Principe" => "São Tomé and Príncipe",
+            "torshavn" => "Tórshavn",
+            "Torshavn" => "Tórshavn",
+            "turkiye" => "Türkiye",
+            "Turkiye" => "Türkiye",
+            "yaounde" => "Yaoundé",
+            "Yaounde" => "Yaoundé"
+        });
+
         triggers.push(Rule {
             pattern: vec![pt!("L"), pt!(Period), pt!("L"), pt!(Period), pt!("M")],
             replace_with: vecword!("large language model"),