anya/structs/story.py

import logging
from collections import OrderedDict

from services.dates import date_part
from structs.headline import Headline

logger = logging.getLogger(__name__)


class Story:
    """A cluster of headlines, drawn from one or more sources, that all report
    the same underlying news story."""

    def __init__(self, headlines: list[Headline]):
        self.headlines = list(headlines)
        logger.debug("Created Story with %d headline(s)", len(self.headlines))

    @property
    def sources(self) -> list[str]:
        """Distinct outlet domains for this story, in first-seen order."""
        seen = OrderedDict()
        for h in self.headlines:
            if h.domain:
                seen.setdefault(h.domain, None)
        return list(seen.keys())

    @property
    def source_count(self) -> int:
        return len(self.sources)

    @property
    def representative(self) -> str:
        """The most descriptive headline text for this story (longest, first on tie)."""
        return max(self.headlines, key=lambda h: len(h.display_text)).display_text

    @property
    def latest_date(self):
        """The most recent publication date among this story's headlines, or None."""
        dates = [date_part(h.published_at) for h in self.headlines if h.published_at is not None]
        return max(dates) if dates else None