Class FetchAndFormatTests

tests/flows/test_paper.py:97–177

            cfg,
            _summary_provider=provider,
        ).build(LocalFilePath(path=_FIXTURE))

        self.assertEqual(len(provider.calls), 1)
        source_seen, chunk_set_seen, _ = provider.calls[0]
        self.assertIs(source_seen, result.source_revision)
        self.assertIs(chunk_set_seen, result.chunk_set)
        self.assertEqual(len(result.source_revision.parsed.pages), 4)
        self.assertTrue(result.chunk_set.chunks)
        self.assertTrue(result.global_summary.summary)
        self.assertGreaterEqual(len(result.global_summary.citations), 3)
        self.assertGreaterEqual(
            len(
                {
                    citation.page_number
                    for citation in result.global_summary.citations
                }
            ),
            2,
        )
        self.assertTrue(
            all(
                chunk.source_revision_id == result.source_revision.id
                and chunk.source_spans
                for chunk in result.chunk_set.chunks
            )
        )

    async def test_exact_arxiv_source_facts_are_code_owned(self) -> None:
        raw = RawPaper(
            bytes=_FIXTURE.read_bytes(),
            content_type="application/pdf",
            source_url="https://arxiv.org/pdf/1706.03762v7.pdf",
            resolved_url="https://arxiv.org/pdf/1706.03762v7.pdf",
            fetched_at=_WHEN,
            arxiv_id="1706.03762v7",
            title="Attention Is All You Need",
            authors=("Ashish Vaswani",),
            published_at=_WHEN,
            updated_at=_WHEN,
        )
        with patch(
            "quantmind.flows.paper.fetch_arxiv",
            new=AsyncMock(return_value=raw),
        ):
            result = await PaperFlow(
                PaperSemanticCfg(chunk_size=256, chunk_overlap=32),
                _summary_provider=_FakeSummaryProvider(),
            ).build(ArxivIdentifier(id="1706.03762v7"))

        self.assertEqual(result.source_revision.arxiv_id, "1706.03762v7")
        self.assertEqual(result.source_revision.source.kind, "arxiv")
        self.assertEqual(
            result.source_revision.title, "Attention Is All You Need"
        )
        self.assertEqual(
            result.source_revision.source.content_hash,
            result.source_revision.parsed.source_hash,
        )