from app.models.chat import SourceRef
from app.pipeline.nodes.generate import _dedup_sources, _reindex_citations_and_sources


def test_reindex_citations_compacts_in_first_mention_order() -> None:
    sources = [
        SourceRef(title="A", url="https://x/a", section="s", source_type="project"),
        SourceRef(title="B", url="https://x/b", section="s", source_type="project"),
        SourceRef(title="C", url="https://x/c", section="s", source_type="project"),
        SourceRef(title="D", url="https://x/d", section="s", source_type="project"),
    ]
    answer = "Uses D [4], then B [2], then D again [4]."

    new_answer, new_sources = _reindex_citations_and_sources(answer, sources)

    assert "[1]" in new_answer
    assert "[2]" in new_answer
    assert "[4]" not in new_answer
    assert [s.title for s in new_sources] == ["D", "B"]


def test_reindex_citations_merges_same_document_indices() -> None:
    # Two different old indices point to the same document URL.
    sources = [
        SourceRef(title="Resume", url="https://darshanchheda.com/resume", section="Experience", source_type="cv"),
        SourceRef(title="Resume", url="https://darshanchheda.com/resume", section="Skills", source_type="cv"),
        SourceRef(title="TextOps", url="https://darshanchheda.com/projects/textops", section="Overview", source_type="project"),
    ]
    answer = "He has QA experience [1][2] and built TextOps [3]."

    new_answer, new_sources = _reindex_citations_and_sources(answer, sources)

    assert "[1][1]" not in new_answer
    assert "[1]" in new_answer
    assert "[2]" in new_answer
    assert [s.url for s in new_sources] == [
        "https://darshanchheda.com/resume",
        "https://darshanchheda.com/projects/textops",
    ]


def test_reindex_citations_drops_out_of_range_markers() -> None:
    sources = [SourceRef(title="Only", url="https://x/only", section="s", source_type="project")]
    answer = "Valid [1], invalid [9]."

    new_answer, new_sources = _reindex_citations_and_sources(answer, sources)

    assert "[1]" in new_answer
    assert "[9]" not in new_answer
    assert len(new_sources) == 1


def test_reindex_citations_merges_url_variants_same_document() -> None:
    sources = [
        SourceRef(
            title="Resume",
            url="https://darshanchheda.com/resume/",
            section="Experience",
            source_type="cv",
        ),
        SourceRef(
            title="Resume",
            url="https://www.darshanchheda.com/resume?ref=nav#top",
            section="Skills",
            source_type="cv",
        ),
    ]
    answer = "Resume evidence appears in both chunks [1][2]."

    new_answer, new_sources = _reindex_citations_and_sources(answer, sources)

    assert new_answer.count("[1]") == 1
    assert "[2]" not in new_answer
    assert len(new_sources) == 1
    assert new_sources[0].title == "Resume"


def test_dedup_sources_merges_url_variants() -> None:
    sources = [
        SourceRef(title="Resume", url="darshanchheda.com/resume", section="", source_type="cv"),
        SourceRef(title="Resume", url="https://darshanchheda.com/resume/", section="", source_type="cv"),
        SourceRef(title="Project", url="https://darshanchheda.com/projects/textops", section="", source_type="project"),
    ]

    deduped = _dedup_sources(sources)

    assert [s.title for s in deduped] == ["Resume", "Project"]