Clarify LLM paragraph fallback behavior

engine_distribute.py

@@ -76,6 +76,11 @@ def run_distribution(in_queue, args):
                 llm_paragraph_msg = f"[LLM-PARAGRAPH]: {speaker_prefix}{paragraph_text}"
                 print(llm_paragraph_msg, flush=True)
                 log_debug(llm_paragraph_msg)
+
+            if payload.get("paragraph_fallback") and paragraph_text:
+                paragraph_msg = f"[PARAGRAPH]: {speaker_prefix}{paragraph_text}"
+                print(paragraph_msg, flush=True)
+                log_debug(paragraph_msg)
             
             for lang in ["es", "fr", "ar", "en"]:
                 if payload.get(lang):

engine_llm.py

@@ -29,6 +29,23 @@ def run_llm_processor(in_queue, out_queue, args):
     is_openai = args.post_correct_model.startswith("gpt-")
     llm_state = {"line_warned": False, "paragraph_warned": False}
 
+    def check_ollama_health():
+        try:
+            response = requests.get("http://127.0.0.1:11434/api/tags", timeout=2.5)
+            response.raise_for_status()
+            data = response.json()
+            models = data.get("models", [])
+            available = {model.get("name") for model in models if model.get("name")}
+            if args.post_correct_model not in available and f"{args.post_correct_model}:latest" not in available:
+                print(f"[LLM] Configured Ollama model '{args.post_correct_model}' is not installed.")
+            return True
+        except Exception as exc:
+            print(f"[LLM] Ollama health check failed ({exc}). Local LLM features may fall back.")
+            return False
+
+    if not is_openai:
+        check_ollama_health()
+
     def normalize_english_caption(text):
         normalized = " ".join((text or "").split()).strip()
         if normalized and normalized[0].isalpha():
@@ -86,7 +103,8 @@ def run_llm_processor(in_queue, out_queue, args):
             return call_ollama(prompt, temperature)
         except Exception as exc:
             if not llm_state[warn_key]:
-                print(f"[LLM] {warn_key.replace('_', ' ').title()} unavailable ({exc}). Falling back to deterministic text.")
+                label = "Paragraph LLM" if warn_key == "paragraph_warned" else "Line LLM"
+                print(f"[LLM] {label} unavailable ({exc}). Falling back to deterministic text.")
                 llm_state[warn_key] = True
             return ""
 
@@ -182,6 +200,7 @@ def run_llm_processor(in_queue, out_queue, args):
             elif len(pending_buffer) >= 1 and (time.time() - last_llm_call_time) > 15: should_refine = True
 
             structured_paragraph = None
+            paragraph_from_llm = False
             if args.llm_paragraph and should_refine:
                 new_batch = " ".join(pending_buffer)
                 result = call_llm_rolling_refine(new_batch, active_context)
@@ -199,6 +218,7 @@ def run_llm_processor(in_queue, out_queue, args):
                         # No break, just update the context
                         active_context = result
                         structured_paragraph = result
+                    paragraph_from_llm = True
                 else:
                     structured_paragraph = new_batch
                     active_context = new_batch
@@ -212,7 +232,7 @@ def run_llm_processor(in_queue, out_queue, args):
                 "paragraph": structured_paragraph,
                 "en_bridge": bridge_text,
                 "used_llm_line": bool(getattr(args, "post_correct_llm", False)),
-                "used_llm_paragraph": bool(args.llm_paragraph and structured_paragraph),
+                "used_llm_paragraph": paragraph_from_llm,
                 "detected_lang": item.get("detected_lang"),
                 "speaker": item.get("speaker"),
                 "ts": item.get("ts", time.time())

engine_translate.py

@@ -77,6 +77,7 @@ def run_translation(in_queue, out_queue, args):
                 "en_bridge": item.get("en_bridge"),
                 "used_llm_line": item.get("used_llm_line", False),
                 "used_llm_paragraph": item.get("used_llm_paragraph", False),
+                "paragraph_fallback": bool(paragraph_text) and not item.get("used_llm_paragraph", False),
                 "speaker": item.get("speaker"),
                 "ts": item.get("ts", time.time())
             }