fix: publish feeds atomically

tests/test_config.py

@@ -154,7 +154,7 @@ def test_build_feed_settings_derives_output_paths_from_feed_slug(
         out_dir / "feeds" / "info-marti" / "files"
     )
     assert feed_settings["FEEDS"] == {
-        str(out_dir / "feeds" / "info-marti" / "feed.rss"): {
+        str(out_dir / "feeds" / "info-marti" / ".feed.rss.next"): {
             "format": "rss",
             "postprocessing": [],
             "feed_name": "info-marti",

tests/test_job_runner.py

@@ -2,8 +2,11 @@ from pathlib import Path
 
 import pytest
 
-from repub.config import FeedConfig
-from repub.job_runner import _build_crawl_settings
+from repub import job_runner as job_runner_module
+from repub.config import FeedConfig, feed_output_path, staged_feed_output_path
+from repub.job_runner import JobSourceConfig, _build_crawl_settings
+
+VALID_FEED = '<rss version="2.0"><channel><title>new</title></channel></rss>\n'
 
 
 def test_build_crawl_settings_passes_feed_url_to_spider(tmp_path: Path) -> None:
@@ -35,3 +38,128 @@ def test_build_crawl_settings_requires_non_empty_feed_url(
             stats_path=tmp_path / "stats.jsonl",
             feed_url="",
         )
+
+
+def test_main_publishes_staged_feed_after_successful_crawl(
+    monkeypatch: pytest.MonkeyPatch, tmp_path: Path
+) -> None:
+    out_dir = tmp_path / "out"
+    public_path = feed_output_path(out_dir=out_dir, feed_slug="demo")
+    staged_path = staged_feed_output_path(out_dir=out_dir, feed_slug="demo")
+    public_path.parent.mkdir(parents=True)
+    public_path.write_text("<rss>old</rss>\n", encoding="utf-8")
+    staged_path.write_text(VALID_FEED, encoding="utf-8")
+
+    _patch_worker_dependencies(monkeypatch, exit_code=0)
+
+    exit_code = job_runner_module.main(
+        [
+            "--job-id",
+            "1",
+            "--execution-id",
+            "2",
+            "--db-path",
+            str(tmp_path / "republisher.db"),
+            "--out-dir",
+            str(out_dir),
+            "--stats-path",
+            str(tmp_path / "stats.jsonl"),
+        ]
+    )
+
+    assert exit_code == 0
+    assert public_path.read_text(encoding="utf-8") == VALID_FEED
+    assert not staged_path.exists()
+
+
+def test_main_does_not_publish_unusable_staged_feed_after_successful_crawl(
+    monkeypatch: pytest.MonkeyPatch, tmp_path: Path
+) -> None:
+    out_dir = tmp_path / "out"
+    public_path = feed_output_path(out_dir=out_dir, feed_slug="demo")
+    staged_path = staged_feed_output_path(out_dir=out_dir, feed_slug="demo")
+    public_path.parent.mkdir(parents=True)
+    public_path.write_text("<rss>old</rss>\n", encoding="utf-8")
+    staged_path.write_text('<rss version="2.0"/>\n', encoding="utf-8")
+
+    _patch_worker_dependencies(monkeypatch, exit_code=0)
+
+    exit_code = job_runner_module.main(
+        [
+            "--job-id",
+            "1",
+            "--execution-id",
+            "2",
+            "--db-path",
+            str(tmp_path / "republisher.db"),
+            "--out-dir",
+            str(out_dir),
+            "--stats-path",
+            str(tmp_path / "stats.jsonl"),
+        ]
+    )
+
+    assert exit_code == 1
+    assert public_path.read_text(encoding="utf-8") == "<rss>old</rss>\n"
+    assert staged_path.read_text(encoding="utf-8") == '<rss version="2.0"/>\n'
+
+
+def test_main_does_not_publish_staged_feed_after_failed_crawl(
+    monkeypatch: pytest.MonkeyPatch, tmp_path: Path
+) -> None:
+    out_dir = tmp_path / "out"
+    public_path = feed_output_path(out_dir=out_dir, feed_slug="demo")
+    staged_path = staged_feed_output_path(out_dir=out_dir, feed_slug="demo")
+    public_path.parent.mkdir(parents=True)
+    public_path.write_text("<rss>old</rss>\n", encoding="utf-8")
+    staged_path.write_text(VALID_FEED, encoding="utf-8")
+
+    _patch_worker_dependencies(monkeypatch, exit_code=1)
+
+    exit_code = job_runner_module.main(
+        [
+            "--job-id",
+            "1",
+            "--execution-id",
+            "2",
+            "--db-path",
+            str(tmp_path / "republisher.db"),
+            "--out-dir",
+            str(out_dir),
+            "--stats-path",
+            str(tmp_path / "stats.jsonl"),
+        ]
+    )
+
+    assert exit_code == 1
+    assert public_path.read_text(encoding="utf-8") == "<rss>old</rss>\n"
+    assert staged_path.read_text(encoding="utf-8") == VALID_FEED
+
+
+def _patch_worker_dependencies(
+    monkeypatch: pytest.MonkeyPatch, *, exit_code: int
+) -> None:
+    monkeypatch.setattr(
+        job_runner_module,
+        "_load_job_source_config",
+        lambda *, db_path, job_id: JobSourceConfig(
+            source_name="Demo",
+            source_slug="demo",
+            source_type="feed",
+            spider_arguments={},
+            feed_url="https://source.example/feed.rss",
+        ),
+    )
+    monkeypatch.setattr(
+        job_runner_module, "load_feed_url", lambda: "https://mirror.example"
+    )
+    monkeypatch.setattr(
+        job_runner_module,
+        "CrawlerProcess",
+        lambda settings: object(),
+    )
+    monkeypatch.setattr(
+        job_runner_module,
+        "_run_crawl",
+        lambda *, process, feed, spider_arguments: exit_code,
+    )

tests/test_postprocessing.py

@@ -0,0 +1,52 @@
+from pathlib import Path
+
+import pytest
+
+from repub.config import feed_output_path, staged_feed_output_path
+from repub.postprocessing import publish_staged_feed
+
+VALID_FEED = '<rss version="2.0"><channel><title>new</title></channel></rss>\n'
+
+
+def test_publish_staged_feed_replaces_public_feed(tmp_path: Path) -> None:
+    out_dir = tmp_path / "out"
+    public_path = feed_output_path(out_dir=out_dir, feed_slug="demo")
+    staged_path = staged_feed_output_path(out_dir=out_dir, feed_slug="demo")
+    public_path.parent.mkdir(parents=True)
+    public_path.write_text("<rss>old</rss>\n", encoding="utf-8")
+    staged_path.write_text(VALID_FEED, encoding="utf-8")
+
+    published_path = publish_staged_feed(out_dir=out_dir, feed_slug="demo")
+
+    assert published_path == public_path
+    assert public_path.read_text(encoding="utf-8") == VALID_FEED
+    assert not staged_path.exists()
+
+
+def test_publish_staged_feed_requires_staged_file(tmp_path: Path) -> None:
+    with pytest.raises(FileNotFoundError):
+        publish_staged_feed(out_dir=tmp_path / "out", feed_slug="missing")
+
+
+@pytest.mark.parametrize(
+    "staged_feed",
+    [
+        '<rss version="2.0"/>\n',
+        '<rss version="2.0"><channel></rss>\n',
+    ],
+)
+def test_publish_staged_feed_rejects_unusable_feed(
+    tmp_path: Path, staged_feed: str
+) -> None:
+    out_dir = tmp_path / "out"
+    public_path = feed_output_path(out_dir=out_dir, feed_slug="demo")
+    staged_path = staged_feed_output_path(out_dir=out_dir, feed_slug="demo")
+    public_path.parent.mkdir(parents=True)
+    public_path.write_text("<rss>old</rss>\n", encoding="utf-8")
+    staged_path.write_text(staged_feed, encoding="utf-8")
+
+    with pytest.raises(ValueError):
+        publish_staged_feed(out_dir=out_dir, feed_slug="demo")
+
+    assert public_path.read_text(encoding="utf-8") == "<rss>old</rss>\n"
+    assert staged_path.read_text(encoding="utf-8") == staged_feed