fix: restore category news collection

2026-08-04 00:58:22 +08:00
parent bdab4f5019
commit edce4baa09
10 changed files with 100 additions and 30 deletions
@@ -0,0 +1,44 @@
+from pathlib import Path
+import unittest
+
+
+ROOT = Path(__file__).resolve().parents[2]
+CRAWLER = ROOT / "docker" / "crawler"
+
+
+class NewsCollectionRegressionTest(unittest.TestCase):
+    def read(self, name: str) -> str:
+        return (CRAWLER / name).read_text(encoding="utf-8")
+
+    def test_lottery_sources_use_runtime_month(self):
+        hkjc = self.read("hkjc_lottery_news.py")
+        taiwan = self.read("taiwan_lottery_news.py")
+
+        self.assertNotIn("?ny=2026&nm=6", hkjc)
+        self.assertIn("LIST_URL_TEMPLATE", hkjc)
+        self.assertIn("def build_list_url", hkjc)
+        self.assertNotIn("from=2025-06&to=2026-06", taiwan)
+        self.assertIn("LIST_API_TEMPLATE", taiwan)
+        self.assertIn("def build_list_api", taiwan)
+
+    def test_general_lottery_task_is_registered(self):
+        tasks = (ROOT / "docker" / "crawler" / "config" / "crawler_tasks.yaml").read_text(encoding="utf-8")
+        block_start = tasks.index("  - key: lottery_news")
+        block = tasks[block_start : tasks.find("\n  - key:", block_start + 1)]
+        self.assertIn("action: lottery_news", block)
+        self.assertIn("active: true", block)
+
+    def test_content_queue_filters_non_publishable_body_not_visibility(self):
+        source = self.read("article_fetcher.py")
+        self.assertNotIn("OR is_show = 0", source)
+        self.assertIn("NOT REGEXP", source)
+        self.assertIn("MAX_RETRY = 4", source)
+
+    def test_rss_categories_allow_summary_to_be_listed(self):
+        for name in ("lottery_news.py", "crypto_news.py", "nba_news.py", "cba_news.py"):
+            source = self.read(name)
+            self.assertIn("allow_summary=True", source, name)
+
+
+if __name__ == "__main__":
+    unittest.main()