build-small-hackathon/hackathon-advisor

1from pathlib import Path2 3from tests.helpers import load_test_index4import json5 6from hackathon_advisor.data import (7    Project,8    ProjectIndex,9    public_project_summary,10    public_project_title,11)12 13 14def test_project_index_searches_snapshot() -> None:15    index = load_test_index()16 17    hits = index.search("lullaby children audio", limit=3)18 19    assert hits20    assert hits[0].project.id.startswith("build-small-hackathon/")21    assert hits[0].page_number >= 122    assert index.index_algorithm == "llama-cpp-embedding-v1"23 24 25def test_project_index_whitespace() -> None:26    index = load_test_index()27 28    items = index.find_whitespace(limit=3)29 30    assert len(items) == 331    assert all(item.label for item in items)32 33 34def test_public_project_cards_hide_generic_submission_copy() -> None:35    assert public_project_title("My Build Small Hackathon") == "Untitled project"36    assert public_project_summary("This is my submission for the build-small-hackathon") == ""37    assert public_project_summary("Todo") == ""38    assert public_project_summary("Local-first personal knowledge agent") == "Local-first personal knowledge agent"39 40    project = Project(41        id="build-small-hackathon/my-build-small-hackathon",42        title="My Build Small Hackathon",43        summary="This is my submission for the build-small-hackathon",44        tags=(),45        models=(),46        datasets=(),47        likes=0,48        sdk="gradio",49        license="",50        created_at="",51        last_modified="",52        host="",53        url="https://example.test",54    )55 56    public = project.to_public_dict()57 58    assert public["title"] == "Untitled project"59    assert public["summary"] == ""60 61 62def test_searchable_text_includes_main_app_file_signals() -> None:63    project = Project(64        id="build-small-hackathon/idea-canvas",65        title="Idea Canvas",66        summary="",67        tags=("gradio",),68        models=(),69        datasets=(),70        likes=0,71        sdk="gradio",72        license="",73        created_at="",74        last_modified="",75        host="",76        url="https://example.test",77        app_file="app.py",78        app_file_embedding_text="score_idea\ngr.Textbox\nProject idea",79    )80 81    searchable = project.searchable_text82 83    assert "main app file: app.py" in searchable84    assert "score_idea" in searchable85    assert "Project idea" in searchable86 87 88def test_public_project_tags_exclude_hosting_metadata() -> None:89    project = Project.from_dict(90        {91            "id": "build-small-hackathon/idea-canvas",92            "title": "Idea Canvas",93            "summary": "",94            "tags": ["gradio", "region:us", "local-first", "region:eu", "gradio"],95            "models": [],96            "datasets": [],97            "url": "https://example.test",98        }99    )100 101    assert project.tags == ("gradio", "region:us", "local-first", "region:eu", "gradio")102    assert project.to_public_dict()["tags"] == ["gradio", "local-first"]103 104 105def test_searchable_text_excludes_refresh_readme_body_for_stable_reuse() -> None:106    project = Project(107        id="build-small-hackathon/long-readme",108        title="Long README",109        summary="",110        tags=(),111        models=(),112        datasets=(),113        likes=0,114        sdk="gradio",115        license="",116        created_at="",117        last_modified="",118        host="",119        url="https://example.test",120        readme_body="a" * 2500 + "middle should not be embedded" + "b" * 2500,121    )122 123    searchable = project.searchable_text124 125    assert "readme:" not in searchable126    assert "middle should not be embedded" not in searchable127 128 129def test_project_index_rejects_mismatched_snapshot(tmp_path: Path) -> None:130    payload = json.loads(Path("data/project_index.json").read_text(encoding="utf-8"))131    payload["snapshot_generated_at"] = "2000-01-01T00:00:00+00:00"132    bad_index = tmp_path / "project_index.json"133    bad_index.write_text(json.dumps(payload), encoding="utf-8")134 135    try:136        ProjectIndex.from_files(Path("data/projects.json"), bad_index)137    except ValueError as error:138        assert "different snapshot timestamp" in str(error)139    else:140        raise AssertionError("mismatched index should be rejected")141 142 143def test_project_index_retains_validated_payload() -> None:144    payload = json.loads(Path("data/project_index.json").read_text(encoding="utf-8"))145    index = ProjectIndex.from_files(Path("data/projects.json"), Path("data/project_index.json"))146 147    assert index.index_payload["snapshot_digest"] == payload["snapshot_digest"]148    assert len(index.index_payload["documents"]) == len(index.projects)149