Improve summarization model output quality

.env.example

@@ -45,3 +45,14 @@ STASHCAST_USER_TOKEN=608AF9E5-E989-4729-9C05-7FFB6EA86FE4
 # Optional: Slug generation settings
 # STASHCAST_SLUG_MAX_WORDS=6
 # STASHCAST_SLUG_MAX_CHARS=40
+
+# Optional: Summarization settings
+# Number of sentences in summary (set to 0 to disable summarization)
+# STASHCAST_SUMMARY_SENTENCES=8
+
+# Summarization backend: 'extractive' (default, fast, CPU-only) or 'ollama' (better quality, requires Ollama)
+# STASHCAST_SUMMARIZER=extractive
+
+# Ollama settings (only used when STASHCAST_SUMMARIZER=ollama)
+# STASHCAST_OLLAMA_HOST=http://localhost:11434
+# STASHCAST_OLLAMA_MODEL=qwen2.5:1.5b

media/context_processors.py

@@ -2,10 +2,13 @@
 
 from django.conf import settings
 
+from media.service.ollama import get_summarizer_status
+
 
 def stashcast_settings(request):
     """Make StashCast settings available to all templates."""
     return {
         'user_token': settings.STASHCAST_USER_TOKEN,
         'require_user_token_for_feeds': settings.REQUIRE_USER_TOKEN_FOR_FEEDS,
+        'summarizer_status': get_summarizer_status(),
     }

media/management/commands/check_ollama.py

@@ -0,0 +1,80 @@
+"""
+Django management command to check Ollama configuration and status.
+
+Usage:
+    ./manage.py check_ollama
+"""
+
+from django.conf import settings
+from django.core.management.base import BaseCommand
+
+from media.service.ollama import get_ollama_status, get_summarizer_status
+
+
+class Command(BaseCommand):
+    help = 'Check Ollama configuration and availability for summarization'
+
+    def handle(self, *args, **options):
+        self.stdout.write('\n=== Summarization Configuration ===\n')
+
+        # Show current settings
+        summarizer = settings.STASHCAST_SUMMARIZER
+        sentences = settings.STASHCAST_SUMMARY_SENTENCES
+        ollama_host = settings.STASHCAST_OLLAMA_HOST
+        ollama_model = settings.STASHCAST_OLLAMA_MODEL
+
+        self.stdout.write(f'STASHCAST_SUMMARIZER: {summarizer}')
+        self.stdout.write(f'STASHCAST_SUMMARY_SENTENCES: {sentences}')
+        self.stdout.write(f'STASHCAST_OLLAMA_HOST: {ollama_host}')
+        self.stdout.write(f'STASHCAST_OLLAMA_MODEL: {ollama_model}')
+
+        self.stdout.write('\n=== Status ===\n')
+
+        # Get overall status
+        status = get_summarizer_status()
+        self.stdout.write(f"Mode: {status['mode']}")
+        self.stdout.write(f"Status: {status['status']}")
+        self.stdout.write(f"Message: {status['message']}")
+
+        if summarizer == 'ollama':
+            self.stdout.write('\n=== Ollama Details ===\n')
+
+            ollama_status = get_ollama_status()
+
+            if ollama_status.available:
+                self.stdout.write(self.style.SUCCESS('Ollama service: Running'))
+            else:
+                self.stdout.write(self.style.ERROR('Ollama service: Not reachable'))
+                self.stdout.write(f'  Error: {ollama_status.error}')
+                self.stdout.write('\n  To start Ollama, run: ollama serve')
+
+            if ollama_status.model_loaded:
+                self.stdout.write(self.style.SUCCESS(f'Model {ollama_model}: Available'))
+            elif ollama_status.available:
+                self.stdout.write(self.style.WARNING(f'Model {ollama_model}: Not found'))
+                self.stdout.write(f'\n  To pull the model, run: ollama pull {ollama_model}')
+
+            if ollama_status.ready:
+                self.stdout.write(
+                    self.style.SUCCESS('\nOllama is ready for summarization!')
+                )
+            else:
+                self.stdout.write(
+                    self.style.ERROR('\nOllama is NOT ready for summarization.')
+                )
+
+        elif summarizer == 'extractive':
+            self.stdout.write(
+                self.style.SUCCESS(
+                    '\nExtractive summarizer (LexRank) is ready. No external service required.'
+                )
+            )
+
+        if sentences <= 0:
+            self.stdout.write(
+                self.style.WARNING(
+                    '\nNote: Summarization is DISABLED (STASHCAST_SUMMARY_SENTENCES=0)'
+                )
+            )
+
+        self.stdout.write('')

media/service/ollama.py

@@ -0,0 +1,180 @@
+"""Ollama service for LLM-based summarization."""
+
+import urllib.request
+import urllib.error
+import json
+from dataclasses import dataclass
+from typing import Optional
+
+from django.conf import settings
+
+
+@dataclass
+class OllamaStatus:
+    """Status of Ollama service and model availability."""
+
+    available: bool
+    model_loaded: bool
+    error: Optional[str] = None
+
+    @property
+    def ready(self) -> bool:
+        """Return True if Ollama is available and model is loaded."""
+        return self.available and self.model_loaded
+
+
+def get_ollama_status() -> OllamaStatus:
+    """
+    Check if Ollama is running and the configured model is available.
+
+    Returns:
+        OllamaStatus with availability information
+    """
+    host = settings.STASHCAST_OLLAMA_HOST
+    model = settings.STASHCAST_OLLAMA_MODEL
+
+    try:
+        # Check if Ollama is running by listing models
+        url = f"{host}/api/tags"
+        req = urllib.request.Request(url, method='GET')
+        req.add_header('Content-Type', 'application/json')
+
+        with urllib.request.urlopen(req, timeout=5) as response:
+            data = json.loads(response.read().decode('utf-8'))
+
+        # Check if the configured model is available
+        available_models = [m.get('name', '') for m in data.get('models', [])]
+
+        # Model names in Ollama can be with or without :latest tag
+        model_base = model.split(':')[0]
+        model_found = any(
+            m == model or m.startswith(f"{model_base}:") or m == f"{model}:latest"
+            for m in available_models
+        )
+
+        if model_found:
+            return OllamaStatus(available=True, model_loaded=True)
+        else:
+            return OllamaStatus(
+                available=True,
+                model_loaded=False,
+                error=f"Model '{model}' not found. Run: ollama pull {model}",
+            )
+
+    except urllib.error.URLError as e:
+        return OllamaStatus(
+            available=False,
+            model_loaded=False,
+            error=f"Ollama not reachable at {host}: {e.reason}",
+        )
+    except Exception as e:
+        return OllamaStatus(
+            available=False, model_loaded=False, error=f"Error checking Ollama: {e}"
+        )
+
+
+def generate_summary_ollama(text: str, max_sentences: int = 8) -> Optional[str]:
+    """
+    Generate a summary using Ollama.
+
+    Args:
+        text: The full text to summarize (e.g., from subtitles)
+        max_sentences: Target number of sentences for the summary
+
+    Returns:
+        The generated summary, or None if generation failed
+    """
+    host = settings.STASHCAST_OLLAMA_HOST
+    model = settings.STASHCAST_OLLAMA_MODEL
+
+    # Construct the prompt
+    prompt = f"""Summarize the following transcript in approximately {max_sentences} sentences.
+Focus on the main topics, key points, and any important conclusions discussed.
+Write in a clear, informative style suitable for a podcast description.
+Do not include phrases like "This transcript discusses" or "The speaker talks about".
+Just provide the summary directly.
+
+Transcript:
+{text}
+
+Summary:"""
+
+    try:
+        url = f"{host}/api/generate"
+        payload = json.dumps({
+            'model': model,
+            'prompt': prompt,
+            'stream': False,
+            'options': {
+                'temperature': 0.3,  # Lower temperature for more focused summaries
+                'num_predict': 500,  # Limit output length
+            },
+        }).encode('utf-8')
+
+        req = urllib.request.Request(url, data=payload, method='POST')
+        req.add_header('Content-Type', 'application/json')
+
+        with urllib.request.urlopen(req, timeout=120) as response:
+            data = json.loads(response.read().decode('utf-8'))
+
+        summary = data.get('response', '').strip()
+        return summary if summary else None
+
+    except Exception as e:
+        # Log the error but don't raise - summarization failure shouldn't block anything
+        print(f"Ollama summarization failed: {e}")
+        return None
+
+
+def get_summarizer_status() -> dict:
+    """
+    Get the current summarizer configuration and status for display.
+
+    Returns:
+        Dict with 'mode', 'status', and 'message' keys
+    """
+    summarizer = settings.STASHCAST_SUMMARIZER
+    sentences = settings.STASHCAST_SUMMARY_SENTENCES
+
+    if sentences <= 0:
+        return {
+            'mode': 'disabled',
+            'status': 'disabled',
+            'message': 'Summarization disabled (STASHCAST_SUMMARY_SENTENCES=0)',
+        }
+
+    if summarizer == 'extractive':
+        return {
+            'mode': 'extractive',
+            'status': 'ready',
+            'message': f'Extractive (LexRank, {sentences} sentences)',
+        }
+
+    if summarizer == 'ollama':
+        ollama_status = get_ollama_status()
+        model = settings.STASHCAST_OLLAMA_MODEL
+
+        if ollama_status.ready:
+            return {
+                'mode': 'ollama',
+                'status': 'ready',
+                'message': f'Ollama ({model})',
+            }
+        elif ollama_status.available:
+            return {
+                'mode': 'ollama',
+                'status': 'model_missing',
+                'message': f'Ollama: model not found ({model})',
+            }
+        else:
+            return {
+                'mode': 'ollama',
+                'status': 'unavailable',
+                'message': 'Ollama: service not running',
+            }
+
+    return {
+        'mode': 'unknown',
+        'status': 'error',
+        'message': f"Unknown summarizer: {summarizer}",
+    }