Disable language detection in corpus building process

backend/local_rag.py

@@ -769,6 +769,7 @@ def _run_prepare_pipeline(slug: str, on_progress=None, **opts):
             out=paths["corpus"],
             on_progress=build_progress,
             emit="per-file",
+            lang_detect=False,
         )
         _mark_pipeline_stage(slug, "build", corpus_signature)
         states["has_corpus"] = True
@@ -1087,6 +1088,7 @@ async def build_library(slug: str):
             root=stage_dir(slug),
             out=_collect_library_paths(slug)["corpus"],
             emit="per-file",
+            lang_detect=False,
             stage_signature=payload.get("corpus_signature"),
         )
     return {"job_id": job_id}

backend/rag/corpus_builder.py

@@ -267,6 +267,11 @@ try:
 except Exception:
     _ld_detect = None
 
+_LANGID_RUNTIME_SAFE = (
+    langid is not None and
+    not (sys.platform == "darwin" and sys.version_info >= (3, 13))
+)
+
 # Progress
 try:
     from tqdm import tqdm
@@ -501,11 +506,11 @@ def detect_language(text: str) -> Optional[str]:
     else:
         sample = text
     try:
-        if langid is not None:
-            lang, _ = langid.classify(sample)
-            return lang
         if _ld_detect is not None:
             return _ld_detect(sample)
+        if _LANGID_RUNTIME_SAFE:
+            lang, _ = langid.classify(sample)
+            return lang
     except Exception:
         pass
     return None