bump

unicode-org · May 2, 2023 · 4c9283b · 4c9283b
1 parent 1a24d05
commit 4c9283b
Show file tree

Hide file tree

Showing 27 changed files with 16,843 additions and 14,994 deletions.
diff --git a/components/segmenter/tests/testdata/provider/segmenter/dictionary/w_auto@1/ja.postcard b/components/segmenter/tests/testdata/provider/segmenter/dictionary/w_auto@1/ja.postcard
diff --git a/components/segmenter/tests/testdata/provider/segmenter/dictionary/wl_ext@1/lo.postcard b/components/segmenter/tests/testdata/provider/segmenter/dictionary/wl_ext@1/lo.postcard
diff --git a/components/segmenter/tests/testdata/provider/segmenter/dictionary/wl_ext@1/th.postcard b/components/segmenter/tests/testdata/provider/segmenter/dictionary/wl_ext@1/th.postcard
diff --git a/provider/datagen/data/segmenter/dictionary/burmesedict.toml b/provider/datagen/data/segmenter/dictionary/burmesedict.toml
diff --git a/provider/datagen/data/segmenter/dictionary/cjdict.toml b/provider/datagen/data/segmenter/dictionary/cjdict.toml
diff --git a/provider/datagen/data/segmenter/dictionary/khmerdict.toml b/provider/datagen/data/segmenter/dictionary/khmerdict.toml
diff --git a/provider/datagen/data/segmenter/dictionary/laodict.toml b/provider/datagen/data/segmenter/dictionary/laodict.toml
diff --git a/provider/datagen/data/segmenter/dictionary/thaidict.toml b/provider/datagen/data/segmenter/dictionary/thaidict.toml
diff --git a/provider/datagen/src/source.rs b/provider/datagen/src/source.rs
@@ -49,7 +49,7 @@ impl SourceData {
     pub const LATEST_TESTED_CLDR_TAG: &'static str = "43.0.0";
 
     /// The latest ICU export tag that has been verified to work with this version of `icu_datagen`.
-    pub const LATEST_TESTED_ICUEXPORT_TAG: &'static str = "release-73-1";
+    pub const LATEST_TESTED_ICUEXPORT_TAG: &'static str = "icu4x/2023-04-27/73.x";
 
     /// The latest `SourceData` that has been verified to work with this version of `icu_datagen`.
     ///

diff --git a/provider/datagen/src/transform/segmenter/mod.rs b/provider/datagen/src/transform/segmenter/mod.rs
@@ -773,9 +773,14 @@ impl crate::DatagenProvider {
 
         let toml_data: &SegmenterDictionaryData = self
             .source
-            .builtin()
-            .read_and_parse_toml(filename)
-            .expect("built-in file present and readable");
+            .icuexport()
+            .and_then(|e| e.read_and_parse_toml(filename))
+            .or_else(|e| {
+                self.source
+                    .builtin()
+                    .read_and_parse_toml(filename)
+                    .map_err(|_| e)
+            })?;
 
         Ok(UCharDictionaryBreakDataV1 {
             trie_data: ZeroVec::alloc_from_slice(&toml_data.trie_data),

diff --git a/provider/repodata/data/icuexport/segmenter/dictionary/burmesedict.toml b/provider/repodata/data/icuexport/segmenter/dictionary/burmesedict.toml
diff --git a/provider/repodata/data/icuexport/segmenter/dictionary/cjdict.toml b/provider/repodata/data/icuexport/segmenter/dictionary/cjdict.toml
diff --git a/provider/repodata/data/icuexport/segmenter/dictionary/khmerdict.toml b/provider/repodata/data/icuexport/segmenter/dictionary/khmerdict.toml
diff --git a/provider/repodata/data/icuexport/segmenter/dictionary/laodict.toml b/provider/repodata/data/icuexport/segmenter/dictionary/laodict.toml
diff --git a/provider/repodata/data/icuexport/segmenter/dictionary/thaidict.toml b/provider/repodata/data/icuexport/segmenter/dictionary/thaidict.toml
diff --git a/provider/repodata/data/json/fingerprints.csv b/provider/repodata/data/json/fingerprints.csv