Function parse_programs_for_source

apps/epg/tasks.py:2062–2418

Parse programs for all MAPPED channels from an EPG source in a single pass. This is an optimized version that: 1. Only processes EPG entries that are actually mapped to channels 2. Parses the XML file ONCE instead of once per channel 3. Skips programmes for unmapped channels en

(epg_source, tvg_id=None)

def parse_programs_for_source(epg_source, tvg_id=None):
    """
    Parse programs for all MAPPED channels from an EPG source in a single pass.

    This is an optimized version that:
    1. Only processes EPG entries that are actually mapped to channels
    2. Parses the XML file ONCE instead of once per channel
    3. Skips programmes for unmapped channels entirely during parsing

    This dramatically improves performance when an EPG source has many channels
    but only a fraction are mapped.
    """
    # Send initial programs parsing notification
    send_epg_update(epg_source.id, "parsing_programs", 0)
    should_log_memory = False
    process = None
    initial_memory = 0
    source_file = None

    # Add memory tracking only in trace mode or higher
    try:
        # Get current log level as a number
        current_log_level = logger.getEffectiveLevel()

        # Only track memory usage when log level is TRACE or more verbose
        should_log_memory = current_log_level <= 5 or settings.DEBUG  # Assuming TRACE is level 5 or lower

        if should_log_memory:
            process = psutil.Process()
            initial_memory = process.memory_info().rss / 1024 / 1024
            logger.info(f"[parse_programs_for_source] Initial memory usage: {initial_memory:.2f} MB")
    except ImportError:
        logger.warning("psutil not available for memory tracking")
        process = None
        should_log_memory = False

    try:
        # Only get EPG entries that are actually mapped to channels
        mapped_epg_ids = set(
            Channel.objects.filter(
                epg_data__epg_source=epg_source,
                epg_data__isnull=False
            ).values_list('epg_data_id', flat=True)
        )

        if not mapped_epg_ids:
            total_epg_count = EPGData.objects.filter(epg_source=epg_source).count()
            logger.info(f"No channels mapped to any EPG entries from source: {epg_source.name} "
                       f"(source has {total_epg_count} EPG entries, 0 mapped)")
            # Update status - this is not an error, just no mapped entries
            epg_source.status = 'success'
            epg_source.last_message = f"No channels mapped to this EPG source ({total_epg_count} entries available)"
            epg_source.save(update_fields=['status', 'last_message'])
            send_epg_update(epg_source.id, "parsing_programs", 100, status="success")
            return True

        # Get the mapped EPG entries with their tvg_ids
        mapped_epgs = EPGData.objects.filter(id__in=mapped_epg_ids).values('id', 'tvg_id')