HPLT seg_langs hotfix for Croatian

ZJaume · ZJaume · commit 9c89be6cf75e · 2026-02-20T16:26:11.000Z
diff --git a/pipeline/data/hplt.py b/pipeline/data/hplt.py
@@ -191,7 +191,8 @@ def _run_download(self):
             read_lines(shuffled_shard_urls, on_enter_location=self.stats.count_shards_visited)
         )
         # Fix for Bosnian which has "hbs" in the seg_langs
-        if self.hplt_locale == "bos_Latn":
+        # TODO remove this in v4
+        if self.hplt_locale in ("bos_Latn", "hrv_Latn"):
             self.hplt_locale = "hbs_Latn"
             logger.warn(f"Changed HPLT locale to {self.hplt_locale} for bos_Latn")
 

Original file line number	Diff line number	Diff line change
`@@ -191,7 +191,8 @@ def _run_download(self):`
`191`	`191`	`read_lines(shuffled_shard_urls, on_enter_location=self.stats.count_shards_visited)`
`192`	`192`	`)`
`193`	`193`	`# Fix for Bosnian which has "hbs" in the seg_langs`
`194`		`- if self.hplt_locale == "bos_Latn":`
	`194`	`+ # TODO remove this in v4`
	`195`	`+ if self.hplt_locale in ("bos_Latn", "hrv_Latn"):`
`195`	`196`	`self.hplt_locale = "hbs_Latn"`
`196`	`197`	`logger.warn(f"Changed HPLT locale to {self.hplt_locale} for bos_Latn")`
`197`	`198`