republisher/tests/test_config.py

from os import path as os_path
from pathlib import Path

import pytest

from repub.config import (
    FeedConfig,
    RepublisherConfig,
    build_base_settings,
    build_feed_settings,
    load_config,
)


def test_load_config_resolves_relative_out_dir_and_merges_imported_feeds(
    tmp_path: Path,
) -> None:
    manifest_path = tmp_path / "imports" / "manifest.toml"
    manifest_path.parent.mkdir(parents=True)
    manifest_path.write_text(
        """
[[feeds]]
name = "Info Martí "
slug = "info-marti"
url = "file:///srv/pygea/info-marti/rss.xml"
""".strip()
        + "\n",
        encoding="utf-8",
    )

    config_path = tmp_path / "configs" / "repub.toml"
    config_path.parent.mkdir(parents=True)
    manifest_ref = os_path.relpath(manifest_path, start=config_path.parent)
    config_path.write_text(
        f"""
out_dir = "../mirror"
feed_config_files = ["{manifest_ref}"]

[[feeds]]
name = "Guardian Project Podcast"
slug = "gp-pod"
url = "https://guardianproject.info/podcast/podcast.xml"
""".strip()
        + "\n",
        encoding="utf-8",
    )

    config = load_config(config_path)

    assert config.out_dir == (tmp_path / "mirror").resolve()
    assert config.feeds == (
        FeedConfig(
            name="Guardian Project Podcast",
            slug="gp-pod",
            url="https://guardianproject.info/podcast/podcast.xml",
        ),
        FeedConfig(
            name="Info Martí ",
            slug="info-marti",
            url="file:///srv/pygea/info-marti/rss.xml",
        ),
    )


def test_load_config_preserves_absolute_out_dir(tmp_path: Path) -> None:
    absolute_out_dir = (tmp_path / "absolute-out").resolve()
    config_path = tmp_path / "repub.toml"
    config_path.write_text(
        f"""
out_dir = "{absolute_out_dir}"

[[feeds]]
name = "NASA Breaking News"
slug = "nasa"
url = "https://www.nasa.gov/rss/dyn/breaking_news.rss"
""".strip()
        + "\n",
        encoding="utf-8",
    )

    config = load_config(config_path)

    assert config.out_dir == absolute_out_dir


def test_load_config_rejects_duplicate_imported_slugs(tmp_path: Path) -> None:
    manifest_path = tmp_path / "manifest.toml"
    manifest_path.write_text(
        """
[[feeds]]
name = "Imported Feed"
slug = "shared-slug"
url = "file:///srv/pygea/shared-slug/rss.xml"
""".strip()
        + "\n",
        encoding="utf-8",
    )

    config_path = tmp_path / "repub.toml"
    config_path.write_text(
        f"""
out_dir = "out"
feed_config_files = ["{manifest_path.name}"]

[[feeds]]
name = "Local Feed"
slug = "shared-slug"
url = "https://example.com/feed.xml"
""".strip()
        + "\n",
        encoding="utf-8",
    )

    with pytest.raises(ValueError, match="Feed slug"):
        load_config(config_path)


def test_build_feed_settings_derives_output_paths_from_feed_slug(
    tmp_path: Path,
) -> None:
    out_dir = (tmp_path / "mirror").resolve()
    config = RepublisherConfig(
        config_path=tmp_path / "repub.toml",
        out_dir=out_dir,
        feeds=(
            FeedConfig(
                name="Info Martí ",
                slug="info-marti",
                url="file:///srv/pygea/info-marti/rss.xml",
            ),
        ),
        scrapy_settings={"LOG_LEVEL": "DEBUG"},
    )

    base_settings = build_base_settings(config)
    feed_settings = build_feed_settings(
        base_settings, out_dir=out_dir, feed_slug="info-marti"
    )

    assert base_settings["LOG_LEVEL"] == "DEBUG"
    assert feed_settings["REPUBLISHER_OUT_DIR"] == str(out_dir)
    assert feed_settings["LOG_FILE"] == str(out_dir / "logs" / "info-marti.log")
    assert feed_settings["HTTPCACHE_DIR"] == str(out_dir / "httpcache")
    assert feed_settings["IMAGES_STORE"] == str(out_dir / "info-marti" / "images")
    assert feed_settings["AUDIO_STORE"] == str(out_dir / "info-marti" / "audio")
    assert feed_settings["VIDEO_STORE"] == str(out_dir / "info-marti" / "video")
    assert feed_settings["FILES_STORE"] == str(out_dir / "info-marti" / "files")
    assert feed_settings["FEEDS"] == {
        str(out_dir / "info-marti.rss"): {
            "format": "rss",
            "postprocessing": [],
            "feed_name": "info-marti",
        }
    }


def test_build_feed_settings_uses_runtime_media_dir_overrides(tmp_path: Path) -> None:
    out_dir = (tmp_path / "mirror").resolve()
    config = RepublisherConfig(
        config_path=tmp_path / "repub.toml",
        out_dir=out_dir,
        feeds=(
            FeedConfig(
                name="Guardian Project Podcast",
                slug="gp-pod",
                url="https://guardianproject.info/podcast/podcast.xml",
            ),
        ),
        scrapy_settings={
            "REPUBLISHER_VIDEO_DIR": "videos-custom",
            "REPUBLISHER_AUDIO_DIR": "audio-custom",
        },
    )

    base_settings = build_base_settings(config)
    feed_settings = build_feed_settings(
        base_settings,
        out_dir=out_dir,
        feed_slug="gp-pod",
    )

    assert feed_settings["REPUBLISHER_VIDEO_DIR"] == "videos-custom"
    assert feed_settings["REPUBLISHER_AUDIO_DIR"] == "audio-custom"
    assert feed_settings["VIDEO_STORE"] == str(out_dir / "gp-pod" / "videos-custom")
    assert feed_settings["AUDIO_STORE"] == str(out_dir / "gp-pod" / "audio-custom")