build-small-hackathon/hackathon-advisor

1from __future__ import annotations2 3from types import SimpleNamespace4 5import pytest6 7from scripts import crawl_hf_spaces8 9 10def test_readme_frontmatter_extracts_app_file() -> None:11    frontmatter = crawl_hf_spaces.readme_frontmatter(12        """---13title: Tiny Demo14app_file: "src/app.py" # main entrypoint15tags:16  - gradio17---18# Tiny Demo19"""20    )21 22    assert frontmatter["app_file"] == "src/app.py"23 24 25def test_validate_app_file_rejects_untrusted_paths() -> None:26    with pytest.raises(RuntimeError, match="invalid app_file path"):27        crawl_hf_spaces.validate_app_file("../app.py", space_id="build-small-hackathon/demo")28 29 30def test_project_from_space_downloads_frontmatter_app_file(monkeypatch) -> None:31    downloads = {32        ("build-small-hackathon/demo", "README.md"): "---\napp_file: app.py\n---\n# Demo\nREADME body evidence.\n",33        ("build-small-hackathon/demo", "app.py"): "import gradio as gr\ngr.Textbox(label='Idea')\n",34    }35 36    def fake_download(repo_id: str, filename: str) -> str:37        return downloads[(repo_id, filename)]38 39    monkeypatch.setattr(crawl_hf_spaces, "download_repo_text", fake_download)40    space = SimpleNamespace(41        id="build-small-hackathon/demo",42        card_data={"title": "Demo", "short_description": "Advisor demo", "sdk": "gradio"},43        siblings=[44            SimpleNamespace(rfilename="README.md"),45            SimpleNamespace(rfilename="app.py"),46        ],47        tags=["gradio", "region:us"],48        models=[],49        datasets=[],50        likes=3,51        created_at=None,52        last_modified=None,53        host="https://example.test",54        private=False,55    )56 57    project = crawl_hf_spaces.project_from_space(space)58 59    assert project["app_file"] == "app.py"60    assert project["readme_body"] == "# Demo\nREADME body evidence."61    assert project["app_file_source"] == "import gradio as gr\ngr.Textbox(label='Idea')\n"62    assert "gr.Textbox" in project["app_file_embedding_text"]63    assert "Idea" in project["app_file_embedding_text"]64    assert project["tags"] == ["gradio"]65 66 67def test_project_from_space_tolerates_stale_frontmatter_app_file(monkeypatch) -> None:68    downloads = {69        ("build-small-hackathon/demo", "README.md"): "---\napp_file: app.py\n---\n",70    }71 72    def fake_download(repo_id: str, filename: str) -> str:73        if (repo_id, filename) in downloads:74            return downloads[(repo_id, filename)]75        raise crawl_hf_spaces.EntryNotFoundError("missing file")76 77    monkeypatch.setattr(crawl_hf_spaces, "download_repo_text", fake_download)78    space = SimpleNamespace(79        id="build-small-hackathon/demo",80        card_data={"title": "Demo", "short_description": "Advisor demo", "sdk": "gradio"},81        siblings=[82            SimpleNamespace(rfilename="README.md"),83            SimpleNamespace(rfilename="app.py"),84        ],85        tags=["gradio"],86        models=[],87        datasets=[],88        likes=3,89        created_at=None,90        last_modified=None,91        host="https://example.test",92        private=False,93    )94 95    project = crawl_hf_spaces.project_from_space(space)96 97    assert project["app_file"] == "app.py"98    assert project["readme_body"] == ""99    assert project["app_file_source"] == ""100    assert project["app_file_embedding_text"] == ""101 102 103def test_download_repo_text_uses_bounded_raw_request(monkeypatch) -> None:104    calls: list[tuple[str, tuple[int, int]]] = []105 106    class Response:107        status_code = 200108        encoding = ""109        text = "hello"110 111        def raise_for_status(self) -> None:112            return None113 114    def fake_get(url: str, *, timeout: tuple[int, int]):115        calls.append((url, timeout))116        return Response()117 118    monkeypatch.setattr(crawl_hf_spaces.requests, "get", fake_get)119 120    text = crawl_hf_spaces.download_repo_text("build-small-hackathon/demo space", "src/app file.py")121 122    assert text == "hello"123    assert calls == [124        (125            "https://huggingface.co/spaces/build-small-hackathon/demo%20space/resolve/main/src/app%20file.py",126            crawl_hf_spaces.DOWNLOAD_TIMEOUT_SECONDS,127        )128    ]129 130 131def test_crawl_workers_rejects_non_integer_env(monkeypatch) -> None:132    monkeypatch.setenv("ADVISOR_CRAWL_WORKERS", "many")133 134    with pytest.raises(RuntimeError, match="ADVISOR_CRAWL_WORKERS"):135        crawl_hf_spaces.crawl_workers()136