Merge branch 'main' of http://truenas.local:30008/david/personal_development

video_transcription/ai_transcriber_v2/translator.py

@@ -1,5 +1,7 @@
 import os
 import sys
+import time
+from datetime import datetime
 from google import genai
 from google.genai import types
 from tenacity import retry, stop_after_attempt, wait_exponential, retry_if_exception_type
@@ -15,36 +17,58 @@ from utils import LANGUAGE_MAP
 def translate_via_ollama(source_srt_content, target_language="English", model="llama3"):
     """
     Translates SRT content using a local Ollama model (Line-by-Line for progress).
+    Includes retries and debug logging.
     """
+    debug_log_path = "ollama_debug.log"
+    
     try:
         subs = pysubs2.SSAFile.from_string(source_srt_content)
         
         # Using tqdm for progress bar
-        for line in tqdm(subs, desc="  Ollama Progress", unit="line"):
+        # dynamic_ncols=True helps it resize properly. leave=True ensures it stays after completion.
+        for line in tqdm(subs, desc="  Ollama Progress", unit="line", dynamic_ncols=True, leave=True):
             text = line.text.strip()
             
             # Skip empty, numeric-only, or extremely short non-word text
             if not text or text.isdigit() or len(text) < 2:
                 continue
 
-            if text:
             prompt = (
                 f"Translate this subtitle text to {target_language}. Output ONLY the translation.\n"
                 f"Text: {text}"
             )
+            
+            # Retry loop for stability
+            max_retries = 3
+            for attempt in range(max_retries):
                 try:
                     response = ollama.chat(model=model, messages=[{'role': 'user', 'content': prompt}])
+                    
+                    # Check for "model not found" or other soft errors in response if API wraps them
+                    # Usually ollama library raises ResponseError for 404
+                    
                     translated_text = response['message']['content'].strip()
                     if translated_text:
                         line.text = translated_text
+                    break # Success, exit retry loop
+                    
                 except Exception as e:
-                   # Silent fail on line, logs would be too spammy in progress bar
+                    # Log error details
+                    with open(debug_log_path, "a") as log:
+                        log.write(f"[{datetime.now()}] Error on line '{text}': {str(e)}\n")
+                    
+                    if attempt < max_retries - 1:
+                        time.sleep(2) # Wait before retry
+                    else:
+                        # If all retries fail, keep original text or empty?
+                        # Keeping original might be safer than silence, or just skip.
+                        # For now, we skip updating 'line.text' so it stays as source language (better than corruption)
                         pass
 
         return subs.to_string(format_="srt")
 
     except Exception as e:
-        print(f"  [Local LLM] Error: {e}")
+        print(f"  [Local LLM] Critical Error: {e}")
         return None
 
 def translate_fallback_mymemory(source_srt_content, target_language="en"):