sphinx_thebe/tests/test_build.py

from bs4 import BeautifulSoup
from pathlib import Path
from subprocess import run
from shutil import copy, copytree, rmtree
import pytest


path_tests = Path(__file__).parent.resolve()
path_docs = path_tests.joinpath("..", "docs")


@pytest.fixture(scope="session")
def sphinx_build(tmpdir_factory):
    class SphinxBuild:
        path_tmp = Path(tmpdir_factory.mktemp("build"))
        path_tmp_docs = path_tmp.joinpath("docs")
        path_build = path_tmp_docs.joinpath("_build")
        path_html = path_build.joinpath("html")
        path_pg_index = path_html.joinpath("index.html")
        path_pg_config = path_html.joinpath("configure.html")
        path_pg_ntbk = path_html.joinpath("examples/notebooks.html")
        # Test using our documentation but the Alabaster theme
        # this avoids version pinning clashes with the Book Theme
        cmd_base = ["sphinx-build", ".", "_build/html", "-D", "html_theme=alabaster", "-a"]

        def copy(self, path=None):
            """Copy the specified book to our tests folder for building."""
            if path is None:
                path = path_docs
                path_changelog = path / "../CHANGELOG.md"
            if not self.path_tmp_docs.exists():
                copytree(path, self.path_tmp_docs)
                # Copy since it's loaded with an `include` directive
                copy(path_changelog, self.path_tmp)

        def build(self, cmd=None):
            """Build the test book"""
            cmd = [] if cmd is None else cmd
            output = run(self.cmd_base + cmd, cwd=self.path_tmp_docs, check=True, capture_output=True)
            self.output = output

        def clean(self):
            """Clean the _build folder so files don't clash with new tests."""
            rmtree(self.path_build)

    return SphinxBuild()


def test_sphinx_thebe(file_regression, sphinx_build):
    """Test building with thebe."""
    sphinx_build.copy()

    # Basic build with defaults
    sphinx_build.build()

    # No build warnings that were raised by sphinx-thebe
    errors = list(sphinx_build.output.stderr.decode().split("\n"))
    assert all(["[sphinx-thebe]" not in ii for ii in errors])

    # Testing index for base config
    soup_ix = BeautifulSoup(Path(sphinx_build.path_pg_index).read_text(), "html.parser")
    config = soup_ix("script", {"type": "text/x-thebe-config"})
    assert len(config) == 1
    config = config[0]
    file_regression.check(config.prettify(), basename="config_index", extension=".html")

    # Testing the configure page which has a custom kernel
    soup_conf = BeautifulSoup(
        Path(sphinx_build.path_pg_config).read_text(), "html.parser"
    )
    config = soup_conf("script", {"type": "text/x-thebe-config"})
    assert len(config) == 1
    config = config[0]
    file_regression.check(
        config.prettify(), basename="config_custom", extension=".html"
    )

    # Test launch buttons for custom text and structure
    launch_buttons = soup_conf.select(".thebe-launch-button")
    lb_text = "\n\n".join([ii.prettify() for ii in launch_buttons])
    file_regression.check(lb_text, basename="launch_buttons", extension=".html")

    # Check for MyST-NB cell structure to make sure it stays the same
    # If this breaks, we'll need to update our default cell selectors
    soup_nb = BeautifulSoup(
        Path(sphinx_build.path_pg_ntbk).read_text(), "html.parser"
    )
    cell = soup_nb.select(".cell")[0]
    
    # Remove the *content* of input/output since we only care about the DOM structure
    cell.select(".cell_input pre")[0].clear()
    cell.select(".cell_output")[0].clear()
    file_regression.check(
        cell.prettify(), basename="myst-nb__cell", extension=".html"
    )


def test_lazy_load(file_regression, sphinx_build):
    """Test building with thebe."""
    sphinx_build.copy()
    url = "https://unpkg.com/thebe@0.8.2/lib/index.js"  # URL to search for

    # Thebe JS should not be loaded by default (is loaded lazily)
    sphinx_build.build()
    soup_ix = BeautifulSoup(Path(sphinx_build.path_pg_index).read_text(), "html.parser")
    sources = [ii.attrs.get("src") for ii in soup_ix.select("script")]
    thebe_source = [ii for ii in sources if ii == url]
    assert len(thebe_source) == 0

    # always_load=True should force this script to load on all pages
    sphinx_build.build(cmd=["-D", "thebe_config.always_load=true"])
    soup_ix = BeautifulSoup(Path(sphinx_build.path_pg_index).read_text(), "html.parser")
    sources = [ii.attrs.get("src") for ii in soup_ix.select("script")]
    thebe_source = [ii for ii in sources if ii == url]
    assert len(thebe_source) == 1