Method select_screened_validation_batch

scripts/source_inventory.py:184–199

(
        self,
        existing_urls: Optional[set] = None,
        target_gap: int = 0,
        screened_sources: Optional[List[Dict]] = None,
    )

        return self.refresh_candidate_pool(merged, save=save)

    def select_screened_validation_batch(
        self,
        existing_urls: Optional[set] = None,
        target_gap: int = 0,
        screened_sources: Optional[List[Dict]] = None,
    ) -> List[Dict]:
        existing_urls = existing_urls or set()
        screened = screened_sources or self.load_screened_sources()

        batch_size = max(self.screened_validation_batch, target_gap * self.validation_oversample_factor)
        available = [
            deepcopy(source)
            for source in screened
            if str(source.get("bookSourceUrl", "")).strip() not in existing_urls
        ]
        return self._sort_sources(available)[:batch_size]

    def _apply_existing_bonus(self, current_sources: List[Dict], candidate_sources: List[Dict]) -> List[Dict]:
        current_urls = {str(item.get("bookSourceUrl", "")).strip() for item in current_sources}