curated_feed.export
===================

.. py:module:: curated_feed.export

.. autoapi-nested-parse::

   Render a corpus file as compact Markdown for pasting into a chat session.

   A collection day can run to several hundred items, so the output is deliberately
   terse: two lines per item, numbered for reference and carrying the record `id` so
   that picks can be traced back to the corpus.

   The numbering is the same one `policy.py` puts in the prompt, and comes from the
   same renderer. A chat session and a pipeline run therefore refer to an article by
   the same number, which is what lets a response saved from one drive the other.



Functions
---------

.. autoapisummary::

   curated_feed.export.render_markdown
   curated_feed.export.render
   curated_feed.export.build_parser
   curated_feed.export.resolve_corpus_file
   curated_feed.export.main


Module Contents
---------------

.. py:function:: render_markdown(candidates: collections.abc.Sequence[curated_feed.models.Candidate], *, title: str, max_items: int | None = None) -> str

   Numbered Markdown under a heading, one item per two lines.


.. py:function:: render(candidates: collections.abc.Sequence[curated_feed.models.Candidate], *, config: curated_feed.config.Config, title: str, max_items: int | None = None, as_prompt: bool = False, prompt_path: pathlib.Path | None = None) -> str

   The listing on its own, or the whole prompt the pipeline would send.

   The prompt form is what a chat session wants: the same policy documents and
   the same numbering the pipeline uses, so a response saved from that session
   can be fed straight back with `curate-score --provider file`.


.. py:function:: build_parser() -> argparse.ArgumentParser

.. py:function:: resolve_corpus_file(args: argparse.Namespace, config: curated_feed.config.Config) -> pathlib.Path

.. py:function:: main(argv: collections.abc.Sequence[str] | None = None) -> int