protect against private use region codes in the regions field of a langtag (#1382)

Author: hahn-kev

SIL.WritingSystems.Tests/LanguageLookupTests.cs

@@ -232,7 +232,7 @@ public void SuggestLanguages_ExactCountryMatch_SortsAboveFuzzyNames()
 			{
 				new AllTagEntry { name = "Crimean Tatar", tag="crh-Cyrl-UA", region = "UA", regions = new List<string> {"BG", "KG", "US"}},
 				new AllTagEntry { name = "English", tag = "en", region = "US"},
-				new AllTagEntry {name = "French", tag = "fr", region = "FR", regions = new List<string> {"FR", "US"}} 
+				new AllTagEntry {name = "French", tag = "fr", region = "FR", regions = new List<string> {"FR", "US"}}
 			};
 			var lookup = new LanguageLookup(entries);
 			var languages = lookup.SuggestLanguages("United States");
@@ -314,7 +314,7 @@ public void SuggestLanguages_LanguageIsOromo_HasNoPejorativeLanguageNames()
 			languages = lookup.SuggestLanguages("hae").ToArray();
 			// “Kwottu” (pej.), “Qottu” (pej.), “Quottu” (pej.), “Qwottu” (pej.)
 			Assert.False(languages.Any(l => l.Names.Contains("ottu")));
-			// “Qotu Oromo” (pej.), 
+			// “Qotu Oromo” (pej.),
 			Assert.False(languages.Any(l => l.Names.Contains("Qotu")));
 		}
 
@@ -439,6 +439,59 @@ public void SuggestLanguages_SearchingForSign_ReturnsSignLanguages()
 			}
 		}
 
+		[Test]
+		public void EnsureTagsWithPrivateUseRegions_WorksAsExpected()
+		{
+			var entries = new List<AllTagEntry>
+			{
+				new AllTagEntry
+				{
+					full = "sq-Todr-AL",
+					iana = new List<string>() { "Albanian" },
+					iso639_3 = "sqi",
+					name = "Albanian",
+					names = new List<string>()
+					{
+						"Albanian, Tosk",
+						"Arnaut",
+						"Arvanitika",
+						"Camerija",
+						"Shkip",
+						"Shqip",
+						"Skchip",
+						"Tosk",
+						"Zhgabe"
+					},
+
+					region = "AL",
+					regionName = "Albania",
+					regions = new List<string>()
+					{
+						"AT",
+						"AU",
+						"BE",
+						"DE",
+						"GR",
+						"HR",
+						"IT",
+						"MK",
+						"MN",
+						"RS",
+						"SE",
+						"TR",
+						"UA",
+						"US",
+						"XK"
+					},
+					sldr = false,
+					tag = "sq-Todr",
+					tags = new List<string>() { "als-Todr", "als-Todr-AL" },
+				}
+			};
+
+			Assert.DoesNotThrow(() => new LanguageLookup(entries, false));
+		}
+
 		[Test]
 		public void EnsureDefaultTags_WorksAsExpected()
 		{

SIL.WritingSystems/LanguageLookup.cs

@@ -152,14 +152,7 @@ private bool AddLanguage(string code, string threelettercode, string full = null
 			}
 			else if (StandardSubtags.IsValidIso3166RegionCode(region))
 			{
-				if (StandardSubtags.IsPrivateUseRegionCode(region))
-				{
-					primaryCountry = region == "XK" ? "Kosovo" : "Unknown private use";
-				}
-				else
-				{
-					primaryCountry = StandardSubtags.RegisteredRegions[region].Name; // convert to full region name
-				}
+				primaryCountry = RegionToCountry(region);
 			}
 			else
 			{
@@ -180,7 +173,7 @@ private bool AddLanguage(string code, string threelettercode, string full = null
 				{
 					if (!country.Contains('?') && country != "")
 					{
-						language.Countries.Add(StandardSubtags.RegisteredRegions[country].Name);
+						language.Countries.Add(RegionToCountry(country));
 					}
 				}
 			}
@@ -258,6 +251,24 @@ private bool AddLanguage(string code, string threelettercode, string full = null
 			return true;
 		}
 
+		private static string RegionToCountry(string region)
+		{
+			string country;
+			if (StandardSubtags.IsPrivateUseRegionCode(region))
+			{
+				country = region == "XK" ? "Kosovo" : "Unknown private use";
+			}
+			else if (StandardSubtags.RegisteredRegions.TryGet(region, out var regionTag))
+			{
+				country = regionTag.Name; // convert to full region name
+			}
+			else
+			{
+				country = "Invalid region " + region;
+			}
+			return country;
+		}
+
 
 		/// <summary>
 		///  For testing; used to detect if we need more special cases where LanguageDataIndex()