Map -u-co- to provider variant

unicode-org · Mar 31, 2022 · c90040c · c90040c
1 parent cb9b12a
commit c90040c
Showing 1 changed file with 78 additions and 7 deletions.
diff --git a/experimental/collator/src/lib.rs b/experimental/collator/src/lib.rs
@@ -70,6 +70,7 @@ pub mod error;
 pub mod provider;
 
 extern crate alloc;
+use crate::alloc::string::ToString;
 use crate::provider::CollationDataV1Marker;
 use crate::provider::CollationDiacriticsV1Marker;
 use crate::provider::CollationJamoV1Marker;
@@ -2217,14 +2218,63 @@ impl<'data> Collator {
             > + ?Sized,
     {
         let locale = locale.into();
+
+        // TODO: Alias mapping here.
+
         let langid: icu_locid::LanguageIdentifier = locale.clone().into();
 
+        // TODO: Is there a compile-time macro for this?
+        let key = "co".parse::<icu_locid::extensions::unicode::Key>().unwrap();
+
+        // In the provider, the variant called "standard" in CLDR is represented
+        // as `variant: None`. It is the default for all but two languages:
+        // sv and zh. Since there are only two special cases, hard-coding them
+        // here for now instead of making the defaulting fancy and data driven.
+        // The Swedish naming seems ad hoc from
+        // https://unicode-org.atlassian.net/browse/CLDR-679 .
+        let variant = if let Some(extension) = locale.get_unicode_extension(&key) {
+            // Surely there should be a non-allocating way to get a
+            // &str out of a single-component extension.
+            let extension_string = extension.to_string();
+            match extension_string.as_str() {
+                "trad" => Some("traditional".into()),
+                "phonebk" => Some("phonebook".into()),
+                "dict" => Some("dictionary".into()),
+                "gb2312" => Some("gb2312han".into()),
+                "standard" => None,
+                _ => {
+                    // XXX Test that the provider ignores non-matching
+                    // variants.
+
+                    // `Locale` should enforce that it's impossible
+                    // to get these values that are longer than
+                    // 8 ASCII characters:
+                    debug_assert_ne!(extension_string, "traditional");
+                    debug_assert_ne!(extension_string, "phonebook");
+                    debug_assert_ne!(extension_string, "dictionary");
+                    debug_assert_ne!(extension_string, "gb2312han");
+                    Some(extension_string.into())
+                }
+            }
+        } else {
+            if &langid.language == &"zh" {
+                Some("pinyin".into())
+            } else if &langid.language == &"sv" {
+                Some("reformed".into())
+            } else {
+                None
+            }
+        };
+
+        // TODO: Handle POSIX, which is part of the language instead
+        // of the -u-co- extension.
+
         let metadata_payload: DataPayload<crate::provider::CollationMetadataV1Marker> =
             data_provider
                 .load_resource(&DataRequest {
                     options: ResourceOptions {
                         langid: Some(langid.clone()),
-                        variant: None,
+                        variant: variant.clone(),
                     },
                     metadata: Default::default(),
                 })?
@@ -2239,7 +2289,7 @@ impl<'data> Collator {
                         .load_resource(&DataRequest {
                             options: ResourceOptions {
                                 langid: Some(langid.clone()),
-                                variant: None,
+                                variant: variant,
                             },
                             metadata: Default::default(),
                         })?
@@ -2267,6 +2317,8 @@ impl<'data> Collator {
                     } else {
                         Some(LanguageIdentifier::und())
                     },
+                    // XXX: Check if this should go inside the condition
+                    // above.
                     variant: None,
                 },
                 metadata: Default::default(),
@@ -2280,11 +2332,13 @@ impl<'data> Collator {
         let jamo: DataPayload<CollationJamoV1Marker> = data_provider
             .load_resource(&DataRequest {
                 options: ResourceOptions {
-                    langid: if metadata.tailored_jamo() {
-                        Some(langid.clone())
-                    } else {
-                        Some(LanguageIdentifier::und())
-                    },
+                    // TODO: load other jamo tables.
+                    // langid: if metadata.tailored_jamo() {
+                    //     Some(langid.clone())
+                    // } else {
+                    //     Some(LanguageIdentifier::und())
+                    // },
+                    langid: Some(LanguageIdentifier::und()),
                     variant: None,
                 },
                 metadata: Default::default(),
@@ -3580,6 +3634,23 @@ mod tests {
         }
     }
 
+    // TODO: This test should eventually test fallback
+    // TODO: Test Swedish and Chinese, also, since they have unusual
+    // variant defaults. (But are currently not part of the test data.)
+    #[test]
+    fn test_region_fallback() {
+        // There's no explicit fi-FI data.
+        let locale: Locale = "fi-u-co-standard".parse().unwrap();
+
+        // let locale: Locale = langid!("fi-FI").into();
+
+        let data_provider = icu_testdata::get_provider();
+
+        let collator: Collator =
+            Collator::try_new(locale, &data_provider, CollatorOptions::new()).unwrap();
+        assert_eq!(collator.compare("ä", "z"), Ordering::Greater);
+    }
+
     // TODO: frcoll requires support for fr-CA
 
     // TODO: Write a test for Bangla